Forest Cover Classification by Optimal Segmentation of High Resolution Satellite Imagery
Identifieur interne : 000245 ( Pmc/Corpus ); précédent : 000244; suivant : 000246Forest Cover Classification by Optimal Segmentation of High Resolution Satellite Imagery
Auteurs : So-Ra Kim ; Woo-Kyun Lee ; Doo-Ahn Kwak ; Greg S. Biging ; Peng Gong ; Jun-Hak Lee ; Hyun-Kook ChoSource :
- Sensors (Basel, Switzerland) [ 1424-8220 ] ; 2011.
Abstract
This study investigated whether high-resolution satellite imagery is suitable for preparing a detailed digital forest cover map that discriminates forest cover at the tree species level. First, we tried to find an optimal process for segmenting the high-resolution images using a region-growing method with the scale, color and shape factors in Definiens® Professional 5.0. The image was classified by a traditional, pixel-based, maximum likelihood classification approach using the spectral information of the pixels. The pixels in each segment were reclassified using a segment-based classification (SBC) with a majority rule. Segmentation with strongly weighted color was less sensitive to the scale parameter and led to optimal forest cover segmentation and classification. The pixel-based classification (PBC) suffered from the “salt-and-pepper effect” and performed poorly in the classification of forest cover types, whereas the SBC helped to attenuate the effect and notably improved the classification accuracy. As a whole, SBC proved to be more suitable for classifying and delineating forest cover using high-resolution satellite images.
Url:
DOI: 10.3390/s110201943
PubMed: 22319391
PubMed Central: 3274007
Links to Exploration step
PMC:3274007Le document en format XML
<record><TEI><teiHeader><fileDesc><titleStmt><title xml:lang="en">Forest Cover Classification by Optimal Segmentation of High Resolution Satellite Imagery</title><author><name sortKey="Kim, So Ra" sort="Kim, So Ra" uniqKey="Kim S" first="So-Ra" last="Kim">So-Ra Kim</name><affiliation><nlm:aff id="af1-sensors-11-01943"> Division of Environmental Science and Ecological Engineering, Korea University, Seoul 136-701, Korea; E-Mails:<email>allwhile@korea.ac.kr</email> (S.-R.K.);<email>tulip96@korea.ac.kr</email> (D.-A.K.)</nlm:aff></affiliation></author><author><name sortKey="Lee, Woo Kyun" sort="Lee, Woo Kyun" uniqKey="Lee W" first="Woo-Kyun" last="Lee">Woo-Kyun Lee</name><affiliation><nlm:aff id="af1-sensors-11-01943"> Division of Environmental Science and Ecological Engineering, Korea University, Seoul 136-701, Korea; E-Mails:<email>allwhile@korea.ac.kr</email> (S.-R.K.);<email>tulip96@korea.ac.kr</email> (D.-A.K.)</nlm:aff></affiliation></author><author><name sortKey="Kwak, Doo Ahn" sort="Kwak, Doo Ahn" uniqKey="Kwak D" first="Doo-Ahn" last="Kwak">Doo-Ahn Kwak</name><affiliation><nlm:aff id="af1-sensors-11-01943"> Division of Environmental Science and Ecological Engineering, Korea University, Seoul 136-701, Korea; E-Mails:<email>allwhile@korea.ac.kr</email> (S.-R.K.);<email>tulip96@korea.ac.kr</email> (D.-A.K.)</nlm:aff></affiliation></author><author><name sortKey="Biging, Greg S" sort="Biging, Greg S" uniqKey="Biging G" first="Greg S." last="Biging">Greg S. Biging</name><affiliation><nlm:aff id="af2-sensors-11-01943"> Department of Environmental Science, Policy and Management, University of California at Berkeley, Mulford Hall, Berkeley, CA 94720, USA; E-Mails:<email>biging@berkeley.edu</email> (G.S.B.);<email>penggong@berkeley.edu</email> (P.G.);<email>jhlee@berkeley.edu</email> (J.-H.L.)</nlm:aff></affiliation></author><author><name sortKey="Gong, Peng" sort="Gong, Peng" uniqKey="Gong P" first="Peng" last="Gong">Peng Gong</name><affiliation><nlm:aff id="af2-sensors-11-01943"> Department of Environmental Science, Policy and Management, University of California at Berkeley, Mulford Hall, Berkeley, CA 94720, USA; E-Mails:<email>biging@berkeley.edu</email> (G.S.B.);<email>penggong@berkeley.edu</email> (P.G.);<email>jhlee@berkeley.edu</email> (J.-H.L.)</nlm:aff></affiliation></author><author><name sortKey="Lee, Jun Hak" sort="Lee, Jun Hak" uniqKey="Lee J" first="Jun-Hak" last="Lee">Jun-Hak Lee</name><affiliation><nlm:aff id="af2-sensors-11-01943"> Department of Environmental Science, Policy and Management, University of California at Berkeley, Mulford Hall, Berkeley, CA 94720, USA; E-Mails:<email>biging@berkeley.edu</email> (G.S.B.);<email>penggong@berkeley.edu</email> (P.G.);<email>jhlee@berkeley.edu</email> (J.-H.L.)</nlm:aff></affiliation></author><author><name sortKey="Cho, Hyun Kook" sort="Cho, Hyun Kook" uniqKey="Cho H" first="Hyun-Kook" last="Cho">Hyun-Kook Cho</name><affiliation><nlm:aff id="af3-sensors-11-01943"> Division of Forest Resources Information, Korean Forest Research Institute, Seoul 136-012, Korea; E-Mail:<email>hcho@forest.go.kr</email> (H.-K.C.)</nlm:aff></affiliation></author></titleStmt><publicationStmt><idno type="wicri:source">PMC</idno><idno type="pmid">22319391</idno><idno type="pmc">3274007</idno><idno type="url">http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3274007</idno><idno type="RBID">PMC:3274007</idno><idno type="doi">10.3390/s110201943</idno><date when="2011">2011</date><idno type="wicri:Area/Pmc/Corpus">000245</idno><idno type="wicri:explorRef" wicri:stream="Pmc" wicri:step="Corpus" wicri:corpus="PMC">000245</idno></publicationStmt><sourceDesc><biblStruct><analytic><title xml:lang="en" level="a" type="main">Forest Cover Classification by Optimal Segmentation of High Resolution Satellite Imagery</title><author><name sortKey="Kim, So Ra" sort="Kim, So Ra" uniqKey="Kim S" first="So-Ra" last="Kim">So-Ra Kim</name><affiliation><nlm:aff id="af1-sensors-11-01943"> Division of Environmental Science and Ecological Engineering, Korea University, Seoul 136-701, Korea; E-Mails:<email>allwhile@korea.ac.kr</email> (S.-R.K.);<email>tulip96@korea.ac.kr</email> (D.-A.K.)</nlm:aff></affiliation></author><author><name sortKey="Lee, Woo Kyun" sort="Lee, Woo Kyun" uniqKey="Lee W" first="Woo-Kyun" last="Lee">Woo-Kyun Lee</name><affiliation><nlm:aff id="af1-sensors-11-01943"> Division of Environmental Science and Ecological Engineering, Korea University, Seoul 136-701, Korea; E-Mails:<email>allwhile@korea.ac.kr</email> (S.-R.K.);<email>tulip96@korea.ac.kr</email> (D.-A.K.)</nlm:aff></affiliation></author><author><name sortKey="Kwak, Doo Ahn" sort="Kwak, Doo Ahn" uniqKey="Kwak D" first="Doo-Ahn" last="Kwak">Doo-Ahn Kwak</name><affiliation><nlm:aff id="af1-sensors-11-01943"> Division of Environmental Science and Ecological Engineering, Korea University, Seoul 136-701, Korea; E-Mails:<email>allwhile@korea.ac.kr</email> (S.-R.K.);<email>tulip96@korea.ac.kr</email> (D.-A.K.)</nlm:aff></affiliation></author><author><name sortKey="Biging, Greg S" sort="Biging, Greg S" uniqKey="Biging G" first="Greg S." last="Biging">Greg S. Biging</name><affiliation><nlm:aff id="af2-sensors-11-01943"> Department of Environmental Science, Policy and Management, University of California at Berkeley, Mulford Hall, Berkeley, CA 94720, USA; E-Mails:<email>biging@berkeley.edu</email> (G.S.B.);<email>penggong@berkeley.edu</email> (P.G.);<email>jhlee@berkeley.edu</email> (J.-H.L.)</nlm:aff></affiliation></author><author><name sortKey="Gong, Peng" sort="Gong, Peng" uniqKey="Gong P" first="Peng" last="Gong">Peng Gong</name><affiliation><nlm:aff id="af2-sensors-11-01943"> Department of Environmental Science, Policy and Management, University of California at Berkeley, Mulford Hall, Berkeley, CA 94720, USA; E-Mails:<email>biging@berkeley.edu</email> (G.S.B.);<email>penggong@berkeley.edu</email> (P.G.);<email>jhlee@berkeley.edu</email> (J.-H.L.)</nlm:aff></affiliation></author><author><name sortKey="Lee, Jun Hak" sort="Lee, Jun Hak" uniqKey="Lee J" first="Jun-Hak" last="Lee">Jun-Hak Lee</name><affiliation><nlm:aff id="af2-sensors-11-01943"> Department of Environmental Science, Policy and Management, University of California at Berkeley, Mulford Hall, Berkeley, CA 94720, USA; E-Mails:<email>biging@berkeley.edu</email> (G.S.B.);<email>penggong@berkeley.edu</email> (P.G.);<email>jhlee@berkeley.edu</email> (J.-H.L.)</nlm:aff></affiliation></author><author><name sortKey="Cho, Hyun Kook" sort="Cho, Hyun Kook" uniqKey="Cho H" first="Hyun-Kook" last="Cho">Hyun-Kook Cho</name><affiliation><nlm:aff id="af3-sensors-11-01943"> Division of Forest Resources Information, Korean Forest Research Institute, Seoul 136-012, Korea; E-Mail:<email>hcho@forest.go.kr</email> (H.-K.C.)</nlm:aff></affiliation></author></analytic><series><title level="j">Sensors (Basel, Switzerland)</title><idno type="eISSN">1424-8220</idno><imprint><date when="2011">2011</date></imprint></series></biblStruct></sourceDesc></fileDesc><profileDesc><textClass></textClass></profileDesc></teiHeader><front><div type="abstract" xml:lang="en"><p>This study investigated whether high-resolution satellite imagery is suitable for preparing a detailed digital forest cover map that discriminates forest cover at the tree species level. First, we tried to find an optimal process for segmenting the high-resolution images using a region-growing method with the scale, color and shape factors in Definiens<sup>®</sup> Professional 5.0. The image was classified by a traditional, pixel-based, maximum likelihood classification approach using the spectral information of the pixels. The pixels in each segment were reclassified using a segment-based classification (SBC) with a majority rule. Segmentation with strongly weighted color was less sensitive to the scale parameter and led to optimal forest cover segmentation and classification. The pixel-based classification (PBC) suffered from the “salt-and-pepper effect” and performed poorly in the classification of forest cover types, whereas the SBC helped to attenuate the effect and notably improved the classification accuracy. As a whole, SBC proved to be more suitable for classifying and delineating forest cover using high-resolution satellite images.</p></div></front><back><div1 type="bibliography"><listBibl><biblStruct></biblStruct><biblStruct><analytic><author><name sortKey="Achard, F" uniqKey="Achard F">F. Achard</name></author><author><name sortKey="Eva, H" uniqKey="Eva H">H. Eva</name></author><author><name sortKey="Mayaux, P" uniqKey="Mayaux P">P. Mayaux</name></author></analytic></biblStruct><biblStruct><analytic><author><name sortKey="Dralle, K" uniqKey="Dralle K">K. Dralle</name></author></analytic></biblStruct><biblStruct><analytic><author><name sortKey="Key, T" uniqKey="Key T">T. Key</name></author><author><name sortKey="Warner, T A" uniqKey="Warner T">T.A. Warner</name></author><author><name sortKey="Mcgraw, J B" uniqKey="Mcgraw J">J.B. McGraw</name></author><author><name sortKey="Fajvan, M" uniqKey="Fajvan M">M. Fajvan</name></author></analytic></biblStruct><biblStruct><analytic><author><name sortKey="Brandtberg, T" uniqKey="Brandtberg T">T. Brandtberg</name></author></analytic></biblStruct><biblStruct><analytic><author><name sortKey="Walker, J S" uniqKey="Walker J">J.S. Walker</name></author><author><name sortKey="Briggs, J M" uniqKey="Briggs J">J.M. Briggs</name></author></analytic></biblStruct><biblStruct><analytic><author><name sortKey="Walker, J S" uniqKey="Walker J">J.S. Walker</name></author><author><name sortKey="Briggs, J M" uniqKey="Briggs J">J.M. Briggs</name></author></analytic></biblStruct><biblStruct><analytic><author><name sortKey="Cho, H K" uniqKey="Cho H">H.K. Cho</name></author><author><name sortKey="Lee, W K" uniqKey="Lee W">W.K. Lee</name></author><author><name sortKey="Lee, S H" uniqKey="Lee S">S.H. Lee</name></author></analytic></biblStruct><biblStruct><analytic><author><name sortKey="Morales, R M" uniqKey="Morales R">R.M. Morales</name></author><author><name sortKey="Miura, T" uniqKey="Miura T">T. Miura</name></author><author><name sortKey="Idol, T" uniqKey="Idol T">T. Idol</name></author></analytic></biblStruct><biblStruct><analytic><author><name sortKey="Couturier, S" uniqKey="Couturier S">S. Couturier</name></author><author><name sortKey="Gastellu Etchegorry, J P" uniqKey="Gastellu Etchegorry J">J.P. Gastellu-Etchegorry</name></author><author><name sortKey="Patino, P" uniqKey="Patino P">P. Patino</name></author><author><name sortKey="Martin, E" uniqKey="Martin E">E. Martin</name></author></analytic></biblStruct><biblStruct><analytic><author><name sortKey="Ridley, H" uniqKey="Ridley H">H. Ridley</name></author><author><name sortKey="Atkinson, P" uniqKey="Atkinson P">P. Atkinson</name></author><author><name sortKey="Alpin, P" uniqKey="Alpin P">P. Alpin</name></author><author><name sortKey="Muller, J P" uniqKey="Muller J">J.P. Muller</name></author><author><name sortKey="Dowman, I" uniqKey="Dowman I">I. Dowman</name></author></analytic></biblStruct><biblStruct><analytic><author><name sortKey="Bauer, T" uniqKey="Bauer T">T. Bauer</name></author><author><name sortKey="Steinnocher, K" uniqKey="Steinnocher K">K. Steinnocher</name></author></analytic></biblStruct><biblStruct><analytic><author><name sortKey="Van Der Sande, C J" uniqKey="Van Der Sande C">C.J. Van der Sande</name></author><author><name sortKey="De Jong, S M" uniqKey="De Jong S">S.M. De Jong</name></author><author><name sortKey="De Roo, A P J" uniqKey="De Roo A">A.P.J. De Roo</name></author></analytic></biblStruct><biblStruct><analytic><author><name sortKey="Arroyo, L A" uniqKey="Arroyo L">L.A. Arroyo</name></author><author><name sortKey="Pascual, C" uniqKey="Pascual C">C. Pascual</name></author><author><name sortKey="Manzanera, J A" uniqKey="Manzanera J">J.A. Manzanera</name></author></analytic></biblStruct><biblStruct><analytic><author><name sortKey="Chung, K H" uniqKey="Chung K">K.H. Chung</name></author><author><name sortKey="Lee, W K" uniqKey="Lee W">W.K. Lee</name></author><author><name sortKey="Kim, K H" uniqKey="Kim K">K.H. Kim</name></author><author><name sortKey="Lee, S H" uniqKey="Lee S">S.H. Lee</name></author></analytic></biblStruct><biblStruct><analytic><author><name sortKey="Lee, J H" uniqKey="Lee J">J.H. Lee</name></author><author><name sortKey="Lee, W K" uniqKey="Lee W">W.K. Lee</name></author></analytic></biblStruct><biblStruct><analytic><author><name sortKey="Kim, S W" uniqKey="Kim S">S.W. Kim</name></author><author><name sortKey="Lee, K S" uniqKey="Lee K">K.S. Lee</name></author><author><name sortKey="Lee, J M" uniqKey="Lee J">J.M. Lee</name></author></analytic></biblStruct><biblStruct><analytic><author><name sortKey="Ham, B Y" uniqKey="Ham B">B.Y. Ham</name></author><author><name sortKey="Lee, W K" uniqKey="Lee W">W.K. Lee</name></author><author><name sortKey="Chong, J S" uniqKey="Chong J">J.S. Chong</name></author><author><name sortKey="Lee, J H" uniqKey="Lee J">J.H. Lee</name></author></analytic></biblStruct><biblStruct><analytic><author><name sortKey="Gong, P" uniqKey="Gong P">P. Gong</name></author><author><name sortKey="Marceau, D J" uniqKey="Marceau D">D.J. Marceau</name></author><author><name sortKey="Howarth, P J" uniqKey="Howarth P">P.J. Howarth</name></author></analytic></biblStruct><biblStruct><analytic><author><name sortKey="Kabir, S" uniqKey="Kabir S">S. Kabir</name></author><author><name sortKey="He, D C" uniqKey="He D">D.C. He</name></author><author><name sortKey="Sanusi, M A" uniqKey="Sanusi M">M.A. Sanusi</name></author><author><name sortKey="Wan Hussina, W M A" uniqKey="Wan Hussina W">W.M.A. Wan Hussina</name></author></analytic></biblStruct><biblStruct><analytic><author><name sortKey="Shekhar, S" uniqKey="Shekhar S">S. Shekhar</name></author><author><name sortKey="Schrater, P R" uniqKey="Schrater P">P.R. Schrater</name></author><author><name sortKey="Vatsavai, R R" uniqKey="Vatsavai R">R.R. Vatsavai</name></author><author><name sortKey="Weili, W" uniqKey="Weili W">W. Weili</name></author><author><name sortKey="Chawla, S" uniqKey="Chawla S">S. Chawla</name></author></analytic></biblStruct><biblStruct><analytic><author><name sortKey="Liu, D" uniqKey="Liu D">D. Liu</name></author><author><name sortKey="Kelly, M" uniqKey="Kelly M">M. Kelly</name></author><author><name sortKey="Gong, P" uniqKey="Gong P">P. Gong</name></author></analytic></biblStruct><biblStruct><analytic><author><name sortKey="Sun, W X" uniqKey="Sun W">W.X. Sun</name></author><author><name sortKey="Heidt, V" uniqKey="Heidt V">V. Heidt.</name></author><author><name sortKey="Gong, P" uniqKey="Gong P">P. Gong</name></author><author><name sortKey="Xu, G" uniqKey="Xu G">G. Xu</name></author></analytic></biblStruct><biblStruct><analytic><author><name sortKey="Dong, R" uniqKey="Dong R">R. Dong</name></author><author><name sortKey="Dong, J" uniqKey="Dong J">J. Dong</name></author><author><name sortKey="Wu, G" uniqKey="Wu G">G. Wu</name></author><author><name sortKey="Deng, H" uniqKey="Deng H">H. Deng</name></author></analytic></biblStruct><biblStruct><analytic><author><name sortKey="Yu, Q" uniqKey="Yu Q">Q. Yu</name></author><author><name sortKey="Gong, P" uniqKey="Gong P">P. Gong</name></author><author><name sortKey="Clinton, N" uniqKey="Clinton N">N. Clinton</name></author><author><name sortKey="Biging, G" uniqKey="Biging G">G. Biging</name></author><author><name sortKey="Kelly, M" uniqKey="Kelly M">M. Kelly</name></author><author><name sortKey="Schirokauer, D" uniqKey="Schirokauer D">D. Schirokauer</name></author></analytic></biblStruct><biblStruct><analytic><author><name sortKey="Mathieu, R" uniqKey="Mathieu R">R. Mathieu</name></author><author><name sortKey="Aryal, J" uniqKey="Aryal J">J. Aryal</name></author></analytic></biblStruct><biblStruct><analytic><author><name sortKey="Hay, G J" uniqKey="Hay G">G.J. Hay</name></author><author><name sortKey="Castilla, G" uniqKey="Castilla G">G. Castilla</name></author><author><name sortKey="Wulder, M A" uniqKey="Wulder M">M.A. Wulder</name></author><author><name sortKey="Ruiz, J R" uniqKey="Ruiz J">J.R. Ruiz</name></author></analytic></biblStruct><biblStruct><analytic><author><name sortKey="Van Coillie, F M B" uniqKey="Van Coillie F">F.M.B. Van Coillie</name></author><author><name sortKey="Verbeke, L P C" uniqKey="Verbeke L">L.P.C. Verbeke</name></author><author><name sortKey="De Wulf, R R" uniqKey="De Wulf R">R.R. De Wulf</name></author></analytic></biblStruct><biblStruct><analytic><author><name sortKey="Pal, S K" uniqKey="Pal S">S.K. Pal</name></author><author><name sortKey="Ghosh, A" uniqKey="Ghosh A">A. Ghosh</name></author><author><name sortKey="Shankar, B U" uniqKey="Shankar B">B.U. Shankar</name></author></analytic></biblStruct><biblStruct></biblStruct><biblStruct><analytic><author><name sortKey="Lillesand, T M" uniqKey="Lillesand T">T.M. Lillesand</name></author><author><name sortKey="Kiefer, R W" uniqKey="Kiefer R">R.W. Kiefer</name></author></analytic></biblStruct><biblStruct></biblStruct><biblStruct></biblStruct><biblStruct><analytic><author><name sortKey="Meinel, G" uniqKey="Meinel G">G. Meinel</name></author><author><name sortKey="Neubert, M" uniqKey="Neubert M">M. Neubert</name></author><author><name sortKey="Reder, J" uniqKey="Reder J">J. Reder</name></author></analytic></biblStruct><biblStruct><analytic><author><name sortKey="Congalton, R G" uniqKey="Congalton R">R.G. Congalton</name></author></analytic></biblStruct><biblStruct><analytic><author><name sortKey="Fenstermaker, L" uniqKey="Fenstermaker L">L. Fenstermaker</name></author></analytic></biblStruct><biblStruct><analytic><author><name sortKey="Plourde, L" uniqKey="Plourde L">L. Plourde</name></author><author><name sortKey="Congalton, R G" uniqKey="Congalton R">R.G. Congalton</name></author></analytic></biblStruct><biblStruct><analytic><author><name sortKey="Meyer, P" uniqKey="Meyer P">P. Meyer</name></author><author><name sortKey="Staenz, K" uniqKey="Staenz K">K. Staenz</name></author><author><name sortKey="Itten, K I" uniqKey="Itten K">K.I. Itten</name></author></analytic></biblStruct><biblStruct><analytic><author><name sortKey="Gong, P" uniqKey="Gong P">P. Gong</name></author><author><name sortKey="Howarth, P J" uniqKey="Howarth P">P.J. Howarth</name></author></analytic></biblStruct><biblStruct><analytic><author><name sortKey="Blaschke, T" uniqKey="Blaschke T">T. Blaschke</name></author><author><name sortKey="Strobl, J" uniqKey="Strobl J">J. Strobl</name></author></analytic></biblStruct><biblStruct><analytic><author><name sortKey="Blaschke, T" uniqKey="Blaschke T">T. Blaschke</name></author></analytic></biblStruct><biblStruct><analytic><author><name sortKey="Heyman, O" uniqKey="Heyman O">O. Heyman</name></author><author><name sortKey="Gaston, G G" uniqKey="Gaston G">G.G. Gaston</name></author><author><name sortKey="Kimerling, A J" uniqKey="Kimerling A">A.J. Kimerling</name></author><author><name sortKey="Campbell, J T" uniqKey="Campbell J">J.T. Campbell</name></author></analytic></biblStruct><biblStruct><analytic><author><name sortKey="Xiao, X" uniqKey="Xiao X">X. Xiao</name></author><author><name sortKey="Boles, S" uniqKey="Boles S">S. Boles</name></author><author><name sortKey="Liu, J" uniqKey="Liu J">J. Liu.</name></author><author><name sortKey="Zuang, D" uniqKey="Zuang D">D. Zuang</name></author><author><name sortKey="Liu, M" uniqKey="Liu M">M. Liu</name></author></analytic></biblStruct><biblStruct><analytic><author><name sortKey="Jobin, B" uniqKey="Jobin B">B. Jobin</name></author><author><name sortKey="Labrecque, S" uniqKey="Labrecque S">S. Labrecque</name></author><author><name sortKey="Grenier, M" uniqKey="Grenier M">M. Grenier</name></author><author><name sortKey="Falardeau, G" uniqKey="Falardeau G">G. Falardeau</name></author></analytic></biblStruct></listBibl></div1></back></TEI><pmc article-type="research-article"><pmc-dir>properties open_access</pmc-dir>
<front><journal-meta><journal-id journal-id-type="nlm-ta">Sensors (Basel)</journal-id><journal-title-group><journal-title>Sensors (Basel, Switzerland)</journal-title></journal-title-group><issn pub-type="epub">1424-8220</issn><publisher><publisher-name>Molecular Diversity Preservation International (MDPI)</publisher-name></publisher></journal-meta><article-meta><article-id pub-id-type="pmid">22319391</article-id><article-id pub-id-type="pmc">3274007</article-id><article-id pub-id-type="doi">10.3390/s110201943</article-id><article-id pub-id-type="publisher-id">sensors-11-01943</article-id><article-categories><subj-group subj-group-type="heading"><subject>Article</subject></subj-group></article-categories><title-group><article-title>Forest Cover Classification by Optimal Segmentation of High Resolution Satellite Imagery</article-title></title-group><contrib-group><contrib contrib-type="author"><name><surname>Kim</surname><given-names>So-Ra</given-names></name><xref ref-type="aff" rid="af1-sensors-11-01943"><sup>1</sup></xref></contrib><contrib contrib-type="author"><name><surname>Lee</surname><given-names>Woo-Kyun</given-names></name><xref ref-type="aff" rid="af1-sensors-11-01943"><sup>1</sup></xref><xref ref-type="corresp" rid="c1-sensors-11-01943">*</xref></contrib><contrib contrib-type="author"><name><surname>Kwak</surname><given-names>Doo-Ahn</given-names></name><xref ref-type="aff" rid="af1-sensors-11-01943"><sup>1</sup></xref></contrib><contrib contrib-type="author"><name><surname>Biging</surname><given-names>Greg S.</given-names></name><xref ref-type="aff" rid="af2-sensors-11-01943"><sup>2</sup></xref></contrib><contrib contrib-type="author"><name><surname>Gong</surname><given-names>Peng</given-names></name><xref ref-type="aff" rid="af2-sensors-11-01943"><sup>2</sup></xref></contrib><contrib contrib-type="author"><name><surname>Lee</surname><given-names>Jun-Hak</given-names></name><xref ref-type="aff" rid="af2-sensors-11-01943"><sup>2</sup></xref></contrib><contrib contrib-type="author"><name><surname>Cho</surname><given-names>Hyun-Kook</given-names></name><xref ref-type="aff" rid="af3-sensors-11-01943"><sup>3</sup></xref></contrib></contrib-group><aff id="af1-sensors-11-01943"><label>1</label> Division of Environmental Science and Ecological Engineering, Korea University, Seoul 136-701, Korea; E-Mails:<email>allwhile@korea.ac.kr</email> (S.-R.K.);<email>tulip96@korea.ac.kr</email> (D.-A.K.)</aff><aff id="af2-sensors-11-01943"><label>2</label> Department of Environmental Science, Policy and Management, University of California at Berkeley, Mulford Hall, Berkeley, CA 94720, USA; E-Mails:<email>biging@berkeley.edu</email> (G.S.B.);<email>penggong@berkeley.edu</email> (P.G.);<email>jhlee@berkeley.edu</email> (J.-H.L.)</aff><aff id="af3-sensors-11-01943"><label>3</label> Division of Forest Resources Information, Korean Forest Research Institute, Seoul 136-012, Korea; E-Mail:<email>hcho@forest.go.kr</email> (H.-K.C.)</aff><author-notes><corresp id="c1-sensors-11-01943"><label>*</label> Author to whom correspondence should be addressed; E-Mail: <email>leewk@korea.ac.kr</email>; Tel.: +82-2-3290-3016; Fax: +82-2-3290-3470.</corresp></author-notes><pub-date pub-type="collection"><year>2011</year></pub-date><pub-date pub-type="epub"><day>1</day><month>2</month><year>2011</year></pub-date><volume>11</volume><issue>2</issue><fpage>1943</fpage><lpage>1958</lpage><history><date date-type="received"><day>20</day><month>12</month><year>2010</year></date><date date-type="rev-recd"><day>28</day><month>1</month><year>2011</year></date><date date-type="accepted"><day>30</day><month>1</month><year>2011</year></date></history><permissions><copyright-statement>© 2011 by the authors; licensee MDPI, Basel, Switzerland.</copyright-statement><copyright-year>2011</copyright-year><license><license-p><pmc-comment>CREATIVE COMMONS</pmc-comment>This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution license (<ext-link ext-link-type="uri" xlink:href="http://creativecommons.org/licenses/by/3.0/">http://creativecommons.org/licenses/by/3.0/</ext-link>).</license-p></license></permissions><abstract><p>This study investigated whether high-resolution satellite imagery is suitable for preparing a detailed digital forest cover map that discriminates forest cover at the tree species level. First, we tried to find an optimal process for segmenting the high-resolution images using a region-growing method with the scale, color and shape factors in Definiens<sup>®</sup> Professional 5.0. The image was classified by a traditional, pixel-based, maximum likelihood classification approach using the spectral information of the pixels. The pixels in each segment were reclassified using a segment-based classification (SBC) with a majority rule. Segmentation with strongly weighted color was less sensitive to the scale parameter and led to optimal forest cover segmentation and classification. The pixel-based classification (PBC) suffered from the “salt-and-pepper effect” and performed poorly in the classification of forest cover types, whereas the SBC helped to attenuate the effect and notably improved the classification accuracy. As a whole, SBC proved to be more suitable for classifying and delineating forest cover using high-resolution satellite images.</p></abstract><kwd-group><kwd>digital forest cover map</kwd><kwd>high resolution</kwd><kwd>satellite image</kwd><kwd>pixel-based classification</kwd><kwd>segment-based classification</kwd></kwd-group></article-meta></front><body><sec><label>1.</label><title>Introduction</title><p>Forest cover maps containing spatial information about tree species, age and diameter class, and density of each forest type are widely used for forest resource management. In Korea, analog forest cover maps were conventionally produced on topographic maps at a scale of 1:25,000, through visual interpretation of 1:15,000 scale aerial photographs [<xref ref-type="bibr" rid="b1-sensors-11-01943">1</xref>]. Until the middle of the 1990s, the forest cover maps were prepared with a series of complex processes, namely digitization by visual interpretation using a mechanical plotting instrument. This approach is, however, time-consuming and labor-intensive for making forest cover maps of the entire forested area in South Korea.</p><p>A promising alternative is the direct mapping of forest cover types from high-resolution satellite images, which has the advantage of covering relatively large land areas on potentially regular repeating cycles. A number of earlier studies have used coarse spatial resolution satellite imagery for forest cover mapping [<xref ref-type="bibr" rid="b2-sensors-11-01943">2</xref>], which allowed the classification of forest cover into broad categories, such as coniferous, deciduous, and mixed forests. While broad category forest cover classifications are useful in some instances, they are not suitable for providing detailed forest cover information according to tree species. Tree species can be identified or classified using high-resolution aerial images [<xref ref-type="bibr" rid="b3-sensors-11-01943">3</xref>–<xref ref-type="bibr" rid="b7-sensors-11-01943">7</xref>] and high-resolution multispectral satellite images [<xref ref-type="bibr" rid="b8-sensors-11-01943">8</xref>–<xref ref-type="bibr" rid="b10-sensors-11-01943">10</xref>]. Since high-resolution multispectral satellite image such as IKONOS and Quickbird image has been commercialized, researchers have classified land cover or land use patterns at a detailed level [<xref ref-type="bibr" rid="b11-sensors-11-01943">11</xref>–<xref ref-type="bibr" rid="b14-sensors-11-01943">14</xref>].</p><p>In Korea, researchers have also utilized high-resolution multispectral satellite image to identify several tree species and classify forest stands [<xref ref-type="bibr" rid="b15-sensors-11-01943">15</xref>–<xref ref-type="bibr" rid="b18-sensors-11-01943">18</xref>]. Higher spatial resolution images have been shown to reduce the proportion of mixed pixels and provide the possibility for better interpretation [<xref ref-type="bibr" rid="b11-sensors-11-01943">11</xref>,<xref ref-type="bibr" rid="b12-sensors-11-01943">12</xref>]. Pixel-based classification (PBC) approaches are traditional methods for image classification but suffer from several problems when applied to high spatial resolution images. Due to the shade and gaps, high spatial resolution data can also increase the spectral variation, structural composition and heterogeneity within a class, which might cause confusion among distinct land use categories [<xref ref-type="bibr" rid="b12-sensors-11-01943">12</xref>], or may also lead to unwanted details [<xref ref-type="bibr" rid="b12-sensors-11-01943">12</xref>]. Consequently, the accuracy assessment shows low performance, accompanied by the salt-and-pepper effect which is not suitable for the stand level forest management. To overcome such problems, some pixel-based methods have already been implemented, mainly consisting of three categories: (a) image pre-processing, such as low-pass filter and texture analysis [<xref ref-type="bibr" rid="b19-sensors-11-01943">19</xref>,<xref ref-type="bibr" rid="b20-sensors-11-01943">20</xref>], (b) contextual classification, such as Markov random field [<xref ref-type="bibr" rid="b21-sensors-11-01943">21</xref>,<xref ref-type="bibr" rid="b22-sensors-11-01943">22</xref>], and (c) post-classification processing, such as mode filtering, morphological filtering and rule-based processing [<xref ref-type="bibr" rid="b23-sensors-11-01943">23</xref>,<xref ref-type="bibr" rid="b24-sensors-11-01943">24</xref>]. These techniques incorporate spatial information to characterize each class using neighborhood relationships. However, despite their considerable improvement in classification, these techniques suffer significant disadvantages. First, the pre-defined neighborhood window size may favor all the land-cover types evenly since different classes reach their maximum accuracies at different pixel window sizes [<xref ref-type="bibr" rid="b25-sensors-11-01943">25</xref>]. These processes have blur effects and cannot produce accurate results at the boundaries of different land-cover units [<xref ref-type="bibr" rid="b25-sensors-11-01943">25</xref>].</p><p>Object-based classification (OBC) may be a good alternative to the pixel-based methods. It is useful for analyzing groups of contiguous pixels as objects instead of using the conventional PBC unit [<xref ref-type="bibr" rid="b25-sensors-11-01943">25</xref>]. The OBC approach uses contextual relations between neighboring objects [<xref ref-type="bibr" rid="b26-sensors-11-01943">26</xref>], shape features, such as surface area, length and width of the objects [<xref ref-type="bibr" rid="b13-sensors-11-01943">13</xref>], form and size [<xref ref-type="bibr" rid="b27-sensors-11-01943">27</xref>], structural composition of land use [<xref ref-type="bibr" rid="b12-sensors-11-01943">12</xref>], and a combination of spectral and spatial information [<xref ref-type="bibr" rid="b7-sensors-11-01943">7</xref>,<xref ref-type="bibr" rid="b14-sensors-11-01943">14</xref>], in addition to spectral information.</p><p>OBC has been found to be suitable for segmenting and classifying urban or agricultural areas whose objects usually have regular shapes [<xref ref-type="bibr" rid="b7-sensors-11-01943">7</xref>,<xref ref-type="bibr" rid="b12-sensors-11-01943">12</xref>,<xref ref-type="bibr" rid="b24-sensors-11-01943">24</xref>,<xref ref-type="bibr" rid="b28-sensors-11-01943">28</xref>]. The assumption underlying this approach is that the land use or cover can be distinguished on the basis of the differences in the shapes and sizes [<xref ref-type="bibr" rid="b12-sensors-11-01943">12</xref>]. Accordingly, assuming the image needs to be divided into meaningful homogeneous forest types, as defined by a forest classification scheme, the object-based image classification approach was used in this study. However, difficulties might be encountered when trying to segment complex forest stand patterns using an object-based approach because the objects (forest stands) may be quite dissimilar in size and shape, and their spatial location on the landscape given can vary from random to clustered. Therefore, a new approach needs to be applied for segmenting forest cover.</p><p>This study proposes suitable criteria for classifying Korea’s forest cover and uses objects as minimum and maximum classification units to overcome the problem of the salt-and-pepper effect resulting from pixel-based-classification methods. Segment-based classification (SBC) is applied using objects (segments) with pixel-based classification result. SBC has been considered as an alternative to classifying high-resolution satellite image [<xref ref-type="bibr" rid="b13-sensors-11-01943">13</xref>,<xref ref-type="bibr" rid="b29-sensors-11-01943">29</xref>,<xref ref-type="bibr" rid="b30-sensors-11-01943">30</xref>] and has been evaluated as being suitable for preparing a polygon-based forest cover map.</p><p>This study explored the potential of high-resolution IKONOS images, in an SBC, for mapping detailed forest cover types. The IKONOS images were evaluated for identifying tree species, with the ultimate aim of providing a suitable method for the direct preparation of polygon-based forest cover maps.</p></sec><sec sec-type="materials|methods"><label>2.</label><title>Materials and Methods</title><sec sec-type="materials|methods"><label>2.1.</label><title>Study Area and Materials</title><p>The study area was a representative rural landscape, comprised mostly of forest land, with a few agricultural and residential areas. This area represents well the typical forest cover types and topography of central Korea. <italic>Quercus</italic> spp. (Oak) and other deciduous tree species evenly dominated the entire area, as typically found in central Korea, with some planted coniferous stands of <italic>Pinus koraiensis</italic> (Korean pine), <italic>Larix leptolepis</italic> (Japanese larch) and <italic>Pinus rigida</italic> (Pitch pine) distributed at relatively lower elevations. Elevation in the study area ranged from 74 to 560 m, while about 60% of the slopes were over 20 degrees, with aspects evenly distributed in all directions. The study area was selected between 127°39′54″E, 37°29′47″N and 127°41′26″E, 37°28′26″N, covering an area of 592 ha. The corresponding portion of the IKONOS image is shown in <xref ref-type="fig" rid="f1-sensors-11-01943">Figure 1</xref>.</p><p>The image was acquired on 8th May, 2000, of an 11 km × 11 km area in central Korea, and used to test various classification algorithms for mapping forest types. In general, the IKONOS image was composed of four spectral bands with 4 m spatial resolution and one panchromatic band with 1 m spatial resolution. A panchromatic band with 1m spatial resolution and four multispectral bands were fused through Intensity, Hue, and Saturation (IHS) transformation with a 4-3-2 band combination for RGB color. A fused pan-sharpened image was then used for classification.</p><p>Segments, which are partial areas characterized into the same tree species, were used as the unit of the training dataset. Based on field visits and visual observations, 240 segments were selected as training areas. The segment size ranged from 25 to 7,239 m<sup>2</sup> according to the tree species and stand condition, with a mean of 702 m<sup>2</sup>. This large range of segment sizes was attributed to the fact that the stands were not spectrally homogeneous in the high spatial resolution image, even though they can be considered naturally homogeneous. In each classification class 409 test areas were randomly and independent selected. We performed visual interpretation of the image to identify potential accuracy assessment areas, and investigated each of these areas in the field.</p></sec><sec sec-type="methods"><label>2.2.</label><title>Methods</title><sec><label>2.2.1.</label><title>Classification Scheme</title><p>The number and type of classes in a map are generally dependent on the information requirements and availability of certain types of remotely sensed data used for classification. To create a forest cover map, a suitable classification scheme, which has a sufficient level of detail for the intended use, is required. The goal here was to classify the forest into these individual tree species, and to prepare a detailed forest cover map. During the early growing season in the study area, tree species, especially deciduous trees, begin to grow and display different leaf and crown colors. The IKONOS image also enabled us to optically classify the forest cover for some deciduous as well as coniferous tree species. Eight classes of forest area and one class of non-forest area were used for legends in this study (<xref ref-type="table" rid="t1-sensors-11-01943">Table 1</xref>).</p></sec><sec sec-type="methods"><label>2.2.2.</label><title>Classification Methods</title><p>Firstly, the IKONOS image was classified using a conventional PBC with a maximum likelihood classifier [<xref ref-type="bibr" rid="b31-sensors-11-01943">31</xref>], using ERDAS Imagine 9.2 program. For the PBC, spectral values of red (R), green (G), blue (B) and near infrared (NIR) bands were used. Results of the PBC were used for SBC using a majority rule under different segmentation options.</p><sec><title>Segmentation</title><p>For the SBC [<xref ref-type="bibr" rid="b12-sensors-11-01943">12</xref>,<xref ref-type="bibr" rid="b13-sensors-11-01943">13</xref>], the IKONOS image was firstly segmented using Definiens<sup>®</sup> Professional 5 Program [<xref ref-type="bibr" rid="b32-sensors-11-01943">32</xref>], which uses a region-growing segmentation approach, where the segment size is determined firstly with a scale parameter measuring the maximum possible homogeneity. The higher the value, the larger the resulting image objects [<xref ref-type="bibr" rid="b3-sensors-11-01943">3</xref>]. Homogeneity criteria were set up using color and shape parameters, which define the total relative homogeneity of the resulting image objects. Color refers to the digital value of the resulting image objects, and shape to the textural homogeneity of the resulting image objects. The shape criterion was composed of two parameters: smoothness was used to optimize objects with regard to the smoothness of the borders, and compactness to optimize image objects with regard to their compactness [<xref ref-type="bibr" rid="b33-sensors-11-01943">33</xref>].</p><p>The selection of the optimal combination of parameters for the segmentation is dependent on the segmentation goal and data type. In the absence of any generally accepted criteria for segmenting forest area [<xref ref-type="bibr" rid="b8-sensors-11-01943">8</xref>], tests must be run until the appropriate segmentation parameters have been found [<xref ref-type="bibr" rid="b34-sensors-11-01943">34</xref>].</p><p>Different pairs of weights were employed to determine the optimal criteria for segmenting forest cover (<xref ref-type="table" rid="t2-sensors-11-01943">Table 2</xref>). In step I, extreme weights (0.1 or 0.9) were assigned to color and shape. In step II, the average weights (0.5) were assigned to color and shape, after which the average and extreme weights (0.1, 0.9) were assigned to smoothness. In step III, weights of 0.75 and 0.25 were given to color, with average weights given to smoothness and compactness. Different scales, from 30 to 280, were applied to each pair of weights for segmenting the study area.</p></sec><sec><title>Segment-Based Classification (SBC) Using a Majority Rule</title><p>An SBC method, utilizing the majority rule, was employed for reclassifying the segments using the results of the PBC. In this approach, the pixels of each class classified by the pixel-based method were counted using the GRID module of ArcInfo, with a particular class occupying the majority of a segment assigned to the entire segment.</p></sec></sec><sec><label>2.2.3.</label><title>Verification of Segmentation and Classification</title><p>The classification result was verified using the independent reference unit or test area. As reference units, we employed a cluster of pixels [<xref ref-type="bibr" rid="b35-sensors-11-01943">35</xref>]. Regarding the cluster size, the area for accuracy assessment must be larger than the spatial resolution of image [<xref ref-type="bibr" rid="b36-sensors-11-01943">36</xref>], and that of the error of the GPS receiver, Trimble Pathfinder XR, should be within approximately ±3 m in a forested area. Therefore, we determined that an appropriate size of a cluster would be 9 × 9 pixels, which is large enough to compensate for such errors.</p><p>We used both homogeneous and majority rule for assigning tree species to the cluster and distributing the test area [<xref ref-type="bibr" rid="b37-sensors-11-01943">37</xref>]. Based on the points randomly distributed in the study area, clusters with 9 × 9 pixels were placed such that they completely belong to a homogeneous segment and a tree species, which occupied a majority in the cluster, was assigned to the cluster.</p><p>The number of test areas in each class was established as 50 according to the rule of thumb [<xref ref-type="bibr" rid="b35-sensors-11-01943">35</xref>] and adjusted by considering the occupation area of each class. A total of 409 test areas were established for accuracy assessment of nine classification classes. Kappa values or K-HAT statistics [<xref ref-type="bibr" rid="b27-sensors-11-01943">27</xref>,<xref ref-type="bibr" rid="b31-sensors-11-01943">31</xref>] were used for comparing segmentation alternatives and deciding optimal segment size. The results of PBC and SBC with the optimal segment size found in this study were compared with the error matrix and map.</p></sec></sec></sec><sec><label>3.</label><title>Results and Discussions</title><sec sec-type="methods"><label>3.1.</label><title>Signature Analysis</title><p><xref ref-type="fig" rid="f2-sensors-11-01943">Figure 2</xref> shows the average signature diagram for the various forest type classes of training areas obtained from the IKONOS image. While the first three channels show only modest spectral differences between the classes, the fourth near the infrared channel shows relatively good spectral separation between classes [<xref ref-type="bibr" rid="b10-sensors-11-01943">10</xref>]. Infrared has been considered a useful band for separating coniferous and deciduous trees [<xref ref-type="bibr" rid="b31-sensors-11-01943">31</xref>,<xref ref-type="bibr" rid="b38-sensors-11-01943">38</xref>]. As in previous works [<xref ref-type="bibr" rid="b31-sensors-11-01943">31</xref>,<xref ref-type="bibr" rid="b38-sensors-11-01943">38</xref>], the coniferous and deciduous trees of our study also showed distinctive differences in the infrared band. Deciduous trees had relatively higher infrared spectral values than coniferous trees. Japanese Chestnut (<italic>Castanea crenata</italic>) had a lower spectral value than other deciduous trees because its leaves are not fully developed by early May when the scene was acquired. Pitch pine (<italic>Pinus rigida</italic>) and Korean pine (<italic>Pinus koraiensis</italic>) had similar mean spectral characteristics.</p></sec><sec><label>3.2.</label><title>Optimal Segment size</title><p><xref ref-type="fig" rid="f3-sensors-11-01943">Figure 3(a)</xref> shows the kappa values of the classification, with strong weights assigned to color or shape (Step I). C<sub>9</sub>S<sub>1</sub>(M<sub>9</sub>O<sub>1</sub>) and C<sub>9</sub>S<sub>1</sub>(M<sub>1</sub>O<sub>9</sub>) showed a similar accuracy when strong weights were applied for color and compactness. The accuracies of the SBCs increased as the scale parameter increased up to around 140, but then slightly decreased. The accuracy of the classifications using strongly weighted colors was relatively high with large scale parameters, as they segmented objects largely according to the spectral value, despite the large scale parameter.</p><p>C<sub>1</sub>S<sub>9</sub>(M<sub>9</sub>O<sub>1</sub>) and C<sub>1</sub>S<sub>9</sub>(M<sub>1</sub>O<sub>9</sub>), with strong weights for shape, had similar accuracies to the C<sub>9</sub>S<sub>1</sub>(M<sub>9</sub>O<sub>1</sub>) and C<sub>9</sub>S<sub>1</sub>(M<sub>1</sub>O<sub>9</sub>) with low scale parameters up to 100. However, the accuracy of C<sub>1</sub>S<sub>9</sub>(M<sub>9</sub>O<sub>1</sub>) and C<sub>1</sub>S<sub>9</sub>(M<sub>1</sub>O<sub>9</sub>) abruptly decreased as the scale parameter increased over 100. Therefore, the accuracy of C<sub>1</sub>S<sub>9</sub>(M<sub>9</sub>O<sub>1</sub>) and C<sub>1</sub>S<sub>9</sub>(M<sub>1</sub>O<sub>9</sub>), which are strongly weighted for shape, was low with large scale parameters. In C<sub>1</sub>S<sub>9</sub>(M<sub>9</sub>O<sub>1</sub>) and C<sub>1</sub>S<sub>9</sub>(M<sub>1</sub>O<sub>9</sub>), the shape strongly influenced the segmentation, and segments having similar shapes could be merged, despite the spectral variations in the segments.</p><p><xref ref-type="fig" rid="f3-sensors-11-01943">Figure 3(b)</xref> shows kappa values for the classification with average color and shape weights of 0.5 (Step II), and 0.25 or 0.75 (Step III). Overall, the accuracies tended to decrease with increasing scale parameter, regardless of the color weight values.</p><p>When the color was weakly weighted and the shape was strongly weighted [C<sub>1</sub>S<sub>9</sub>(M<sub>9</sub>O<sub>1</sub>), C<sub>1</sub>S<sub>9</sub>(M<sub>1</sub>O<sub>9</sub>)], the accuracies tended to decrease dramatically as the scale parameter increased. The higher the weighting of the shape, the greater the decrease in the accuracies as the scale parameter increased. Segmentation with high weight for the shape parameter can cause the merging of segments of different tree species with different spectral values, leading to low segmentation accuracy by the kappa value.</p><p>Considering weight for color, the segmentation with strong weight for color was less sensitive to the scale parameter and had better accuracy. Our results demonstrated that segmentation with color parameters from 0.5 to 0.9 and scale parameter from 120 to 200 can lead to optimal forest cover segmentation and classification in central Korea with IKONOS satellite image (<xref ref-type="fig" rid="f4-sensors-11-01943">Figure 4</xref>).</p></sec><sec><label>3.3.</label><title>Forest Cover Map</title><sec><label>3.3.1.</label><title>PBC-Based Forest Cover Map</title><p><xref ref-type="fig" rid="f5-sensors-11-01943">Figure 5(a)</xref> shows the forest cover map produced by the PBC method. The high spatial resolution makes it possible to classify forest cover by tree species. However, the high spatial resolution revealed high variability of the pixel values within a particular class.</p><p>The conventional PBC with high spatial resolution image is inevitably associated with the “salt-and-pepper” effect, which results in small inclusion of other classes within a polygon [<xref ref-type="bibr" rid="b12-sensors-11-01943">12</xref>]. A closer inspection of <xref ref-type="fig" rid="f5-sensors-11-01943">Figure 5</xref> shows that the salt-and-pepper effect can be observed on the classification map. This salt-and-pepper effect makes it difficult to integrate the classification result into a polygon-based forest cover map. It has proven difficult to classify high-resolution images on a pixel-by-pixel basis due to the high level of information and increased intra-class complexity [<xref ref-type="bibr" rid="b31-sensors-11-01943">31</xref>,<xref ref-type="bibr" rid="b34-sensors-11-01943">34</xref>,<xref ref-type="bibr" rid="b39-sensors-11-01943">39</xref>,<xref ref-type="bibr" rid="b40-sensors-11-01943">40</xref>].</p></sec><sec><label>3.3.2.</label><title>SBC-Based Forest Cover Map</title><p>Several attempts to apply the object-based approach for classifying forest cover revealed the difficulty in formalizing the contextual, formal, spatial and structural relationships for the forest cover classification, as required in Definiens<sup>®</sup> Professional 5.0.</p><p>In the SBC, a segmented image is classified by membership rule or maximum likelihood. Even though the SBC is an appropriate method for dividing the forest area, it results in misclassification due to the mean value of pixels in the segmented image. On the other hand, even though a precise classification with pixel units can be conducted by PBC, the gap or shadow in the forested area causes misclassification results. Therefore, in this study, when our method fuses the results of PBC and segment, it supplemented each weak point of PBC and SBC by applying the majority rule.</p><p>The SBC in this study produced a forest cover map without any salt-and-pepper effect (<xref ref-type="fig" rid="f5-sensors-11-01943">Figure 5(b)</xref>), as expected from the previous SBC [<xref ref-type="bibr" rid="b9-sensors-11-01943">9</xref>,<xref ref-type="bibr" rid="b12-sensors-11-01943">12</xref>,<xref ref-type="bibr" rid="b41-sensors-11-01943">41</xref>,<xref ref-type="bibr" rid="b42-sensors-11-01943">42</xref>]. Each error matrix for the SBC using a majority rule and PBC using maximum likelihood is shown in <xref ref-type="table" rid="t3-sensors-11-01943">Tables 3</xref> and <xref ref-type="table" rid="t4-sensors-11-01943">4</xref>. The use of the SBC avoided the salt-and-pepper effects of the PBC. In addition, the overall accuracy was improved from 64% in the PBC to 77% and the kappa value from 0.59 to 0.73. This detailed and reliable information about forest cover will be useful for several applications related to forestry planning and management, natural resources management, carbon cycle studies and for biogeochemistry, hydrology and climate monitoring [<xref ref-type="bibr" rid="b43-sensors-11-01943">43</xref>].</p></sec><sec sec-type="methods"><label>3.3.3.</label><title>Comparison Analysis</title><p>For verifying the feasibleness of our methodology, the results were compared with previous methods that used the traditional majority rule-based filter with a 3 × 3 window and object-based approach. Objects were prepared using Definiens<sup>®</sup> Professional 5 Program. The procedure of object-based classification (OBC) can achieve the same objective based on the segmentation of spectral bands of the image creating homogeneous polygons with regard to the single pixel value, shape, texture, and pixel spatial continuity [<xref ref-type="bibr" rid="b44-sensors-11-01943">44</xref>].</p><p>However, in this study, the accuracy of OBC was the lowest due to the over-segmentation when compared with other classification methods (<xref ref-type="fig" rid="f6-sensors-11-01943">Figure 6</xref>). This was attributed to the spectral similarity between the tree species by which the forest type might be misclassified, especially for <italic>Pinus koraiensis</italic> and <italic>Pinus rigida</italic>. Moreover, the sensitivity for stand shape and texture can lead to over-segmentation because of the various and irregular shapes of the forest stands (objects) in study area.</p><p>Majority rule-based 3 × 3 filter is one of the post-processing methods on PBC. Although these techniques can improve the classification accuracy considerably, their disadvantages are apparent when applied to high spatial resolution images (1 m to 10 m) [<xref ref-type="bibr" rid="b25-sensors-11-01943">25</xref>]. This process suffers blur effects and cannot produce accurate results at the boundaries of different land-cover units. SBC reduces the local spectral variation caused by crown textures, gaps, and shadows. In addition, with spectrally homogeneous segments of images, both spectral values and spatial properties, such as size and shape, can be explicitly utilized as features for further classification. Therefore, the result of majority rule-based classification was analyzed to be lower than that of the segment-based method, despite being higher than other approaches.</p></sec></sec></sec><sec><label>4.</label><title>Conclusions</title><p>This study investigated the suitability of high spatial resolution IKONOS images for preparing precise forest cover maps for central Korea. The experimental results demonstrated the potential capability of IKONOS images in discriminating forest cover types at the tree species level.</p><p>Segmentation weighted for shape was sensitive to the scale parameter, but tended to merge segments of different tree species with different spectral values. Based on a step-by-step analysis, this study suggested that segmentation, with color strongly weighted, was less sensitive to the scale parameter, leading to optimal forest cover segmentation and classification.</p><p>Pixel-based maximum likelihood classification produced an undesirable “salt-and-pepper” effect. SBC delivered higher accuracy and more homogeneous classification results that were free from the “salt-and-pepper” effect.</p></sec></body><back><ref-list><title>References</title><ref id="b1-sensors-11-01943"><label>1.</label><element-citation publication-type="book"><person-group person-group-type="author"><collab>KFRI</collab></person-group><source>Criteria for National Forest Inventory</source><comment>Available online: <ext-link ext-link-type="uri" xlink:href="http://book.kfri.go.kr/forest/resources/html/stockmap_01.jsp">http://book.kfri.go.kr/forest/resources/html/stockmap_01.jsp</ext-link>,</comment><publisher-name>Korea Forest Research Institute</publisher-name><publisher-loc>Seoul</publisher-loc><comment>(accessed on 29 July 2009).</comment></element-citation></ref><ref id="b2-sensors-11-01943"><label>2.</label><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Achard</surname><given-names>F.</given-names></name><name><surname>Eva</surname><given-names>H.</given-names></name><name><surname>Mayaux</surname><given-names>P.</given-names></name></person-group><article-title>Tropical forest mapping from coarse spatial resolution satellite data: production and accuracy assessment issues</article-title><source>Int. J. Remote Sens</source><year>2001</year><volume>22</volume><fpage>2741</fpage><lpage>2762</lpage></element-citation></ref><ref id="b3-sensors-11-01943"><label>3.</label><element-citation publication-type="book"><person-group person-group-type="author"><name><surname>Dralle</surname><given-names>K.</given-names></name></person-group><article-title>Locating Trees by Digital Image Processing of Aerial Photos</article-title><comment>Ph.D. Dissertation,</comment><publisher-name>Royal Veterinary and Agricultural University</publisher-name><publisher-loc>Frederiksberg, Denmark</publisher-loc><year>1997</year></element-citation></ref><ref id="b4-sensors-11-01943"><label>4.</label><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Key</surname><given-names>T.</given-names></name><name><surname>Warner</surname><given-names>T.A.</given-names></name><name><surname>McGraw</surname><given-names>J.B.</given-names></name><name><surname>Fajvan</surname><given-names>M.</given-names></name></person-group><article-title>A comparison of multispectral and multitemporal information in high spatial resolution imagery for classification of individual tree species in a temperate hardwood forest</article-title><source>Remote Sens. Environ</source><year>2001</year><volume>75</volume><fpage>100</fpage><lpage>112</lpage></element-citation></ref><ref id="b5-sensors-11-01943"><label>5.</label><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Brandtberg</surname><given-names>T.</given-names></name></person-group><article-title>Individual tree-based species classification in high spatial resolution aerial images of forests using fuzzy sets</article-title><source>Fuzzy Sets System</source><year>2002</year><volume>132</volume><fpage>371</fpage><lpage>387</lpage></element-citation></ref><ref id="b6-sensors-11-01943"><label>6.</label><element-citation publication-type="confproc"><person-group person-group-type="author"><name><surname>Walker</surname><given-names>J.S.</given-names></name><name><surname>Briggs</surname><given-names>J.M.</given-names></name></person-group><article-title>An object-oriented classification of an arid urban forest with true-color aerial photography</article-title><conf-name>Proceedings of the ISPRS Joint Conference</conf-name><conf-loc>Tempe, AZ, USA</conf-loc><conf-date>March 2005</conf-date></element-citation></ref><ref id="b7-sensors-11-01943"><label>7.</label><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Walker</surname><given-names>J.S.</given-names></name><name><surname>Briggs</surname><given-names>J.M.</given-names></name></person-group><article-title>An object-oriented approach to urban forest mapping in phoenix</article-title><source>Photogramm. Eng. Remote Sensing</source><year>2007</year><volume>73</volume><fpage>577</fpage><lpage>583</lpage></element-citation></ref><ref id="b8-sensors-11-01943"><label>8.</label><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Cho</surname><given-names>H.K.</given-names></name><name><surname>Lee</surname><given-names>W.K.</given-names></name><name><surname>Lee</surname><given-names>S.H.</given-names></name></person-group><article-title>Mapping of vegetation cover using segment based classification of IKONOS imagery</article-title><source>Korean J. Ecol. Soc</source><year>2002</year><volume>1</volume><fpage>165</fpage><lpage>169</lpage></element-citation></ref><ref id="b9-sensors-11-01943"><label>9.</label><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Morales</surname><given-names>R.M.</given-names></name><name><surname>Miura</surname><given-names>T.</given-names></name><name><surname>Idol</surname><given-names>T.</given-names></name></person-group><article-title>An assessment of Hawaiian dry forest condition with fine resolution remote sensing</article-title><source>Forest Ecol. Manag</source><year>2008</year><volume>255</volume><fpage>2524</fpage><lpage>2532</lpage></element-citation></ref><ref id="b10-sensors-11-01943"><label>10.</label><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Couturier</surname><given-names>S.</given-names></name><name><surname>Gastellu-Etchegorry</surname><given-names>J.P.</given-names></name><name><surname>Patino</surname><given-names>P.</given-names></name><name><surname>Martin</surname><given-names>E.</given-names></name></person-group><article-title>A model-based performance test for forest classifiers on remote-sensing imagery</article-title><source>Forest Ecol. Manag</source><year>2009</year><volume>257</volume><fpage>23</fpage><lpage>37</lpage></element-citation></ref><ref id="b11-sensors-11-01943"><label>11.</label><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Ridley</surname><given-names>H.</given-names></name><name><surname>Atkinson</surname><given-names>P.</given-names></name><name><surname>Alpin</surname><given-names>P.</given-names></name><name><surname>Muller</surname><given-names>J.P.</given-names></name><name><surname>Dowman</surname><given-names>I.</given-names></name></person-group><article-title>Evaluating the potential of the forthcoming commercial US high-resolution satellite sensor imagery at the Ordnance survey</article-title><source>Photogramm. Eng. Remote Sensing</source><year>1997</year><volume>63</volume><fpage>997</fpage><lpage>1005</lpage></element-citation></ref><ref id="b12-sensors-11-01943"><label>12.</label><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Bauer</surname><given-names>T.</given-names></name><name><surname>Steinnocher</surname><given-names>K.</given-names></name></person-group><article-title>Per-parcel land use classification in urban areas applying a rule-based technique</article-title><source>GeoBIT/GIS</source><year>2001</year><volume>6</volume><fpage>24</fpage><lpage>27</lpage></element-citation></ref><ref id="b13-sensors-11-01943"><label>13.</label><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Van der Sande</surname><given-names>C.J.</given-names></name><name><surname>De Jong</surname><given-names>S.M.</given-names></name><name><surname>De Roo</surname><given-names>A.P.J.</given-names></name></person-group><article-title>A segmentation and classification approach of IKONOS imagery for land cover mapping to assist flood risk and flood damage assessment</article-title><source>Int. J. Appl. Earth Obs. Geoinf</source><year>2003</year><volume>4</volume><fpage>217</fpage><lpage>229</lpage></element-citation></ref><ref id="b14-sensors-11-01943"><label>14.</label><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Arroyo</surname><given-names>L.A.</given-names></name><name><surname>Pascual</surname><given-names>C.</given-names></name><name><surname>Manzanera</surname><given-names>J.A.</given-names></name></person-group><article-title>Fire models and methods to map fuel types: The role of remote sensing</article-title><source>Forest Eco. Manag</source><year>2008</year><volume>256</volume><fpage>1239</fpage><lpage>1252</lpage></element-citation></ref><ref id="b15-sensors-11-01943"><label>15.</label><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Chung</surname><given-names>K.H.</given-names></name><name><surname>Lee</surname><given-names>W.K.</given-names></name><name><surname>Kim</surname><given-names>K.H.</given-names></name><name><surname>Lee</surname><given-names>S.H.</given-names></name></person-group><article-title>Classification of forest type using high resolution imagery of satellite IKONOS</article-title><source>Korean J. Ecol. Soc</source><year>2001</year><volume>17</volume><fpage>275</fpage><lpage>284</lpage></element-citation></ref><ref id="b16-sensors-11-01943"><label>16.</label><element-citation publication-type="confproc"><person-group person-group-type="author"><name><surname>Lee</surname><given-names>J.H.</given-names></name><name><surname>Lee</surname><given-names>W.K.</given-names></name></person-group><article-title>Spatial characteristic of oak species distribution on the basis of geomorphological factors and IKONOS satellite imagery</article-title><conf-name>Proceedings of the VIII International Congress Ecology</conf-name><conf-loc>Seoul, Korea</conf-loc><conf-date>August 2002</conf-date><fpage>160</fpage></element-citation></ref><ref id="b17-sensors-11-01943"><label>17.</label><element-citation publication-type="confproc"><person-group person-group-type="author"><name><surname>Kim</surname><given-names>S.W.</given-names></name><name><surname>Lee</surname><given-names>K.S.</given-names></name><name><surname>Lee</surname><given-names>J.M.</given-names></name></person-group><article-title>Applicability of multispectral IKONOS imagery for the interpretation of forest stand characteristics</article-title><conf-name>Proceedings of Geographic Information System Association of Korea, Advanced Institute of Science and Technology Hall</conf-name><conf-loc>Seoul, Korea</conf-loc><conf-date>April 2003</conf-date><fpage>139</fpage><lpage>144</lpage></element-citation></ref><ref id="b18-sensors-11-01943"><label>18.</label><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Ham</surname><given-names>B.Y.</given-names></name><name><surname>Lee</surname><given-names>W.K.</given-names></name><name><surname>Chong</surname><given-names>J.S.</given-names></name><name><surname>Lee</surname><given-names>J.H.</given-names></name></person-group><article-title>Estimation of spatial distribution and occurrence probability of oak species using IKONOS satellite imagery and GIS</article-title><source>Korean J. Forest Meas</source><year>2004</year><volume>7</volume><fpage>74</fpage><lpage>84</lpage></element-citation></ref><ref id="b19-sensors-11-01943"><label>19.</label><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Gong</surname><given-names>P.</given-names></name><name><surname>Marceau</surname><given-names>D.J.</given-names></name><name><surname>Howarth</surname><given-names>P.J.</given-names></name></person-group><article-title>A comparison of spatial feature extraction. Algorithms for land-use classification with SPOT HRV data</article-title><source>Remote Sens. Environ</source><year>1992</year><volume>40</volume><fpage>137</fpage><lpage>151</lpage></element-citation></ref><ref id="b20-sensors-11-01943"><label>20.</label><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Kabir</surname><given-names>S.</given-names></name><name><surname>He</surname><given-names>D.C.</given-names></name><name><surname>Sanusi</surname><given-names>M.A.</given-names></name><name><surname>Wan Hussina</surname><given-names>W.M.A.</given-names></name></person-group><article-title>Texture analysis of IKONOS satellite imagery for urban land use and land cover classification</article-title><source>Imag. Sci. J</source><year>2010</year><volume>58</volume><fpage>163</fpage><lpage>170</lpage></element-citation></ref><ref id="b21-sensors-11-01943"><label>21.</label><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Shekhar</surname><given-names>S.</given-names></name><name><surname>Schrater</surname><given-names>P.R.</given-names></name><name><surname>Vatsavai</surname><given-names>R.R.</given-names></name><name><surname>Weili</surname><given-names>W.</given-names></name><name><surname>Chawla</surname><given-names>S.</given-names></name></person-group><article-title>Spatial contextual classification and prediction mining geospatial data</article-title><source>IEEE Trans. Multimed</source><year>2002</year><volume>4</volume><fpage>174</fpage><lpage>188</lpage></element-citation></ref><ref id="b22-sensors-11-01943"><label>22.</label><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Liu</surname><given-names>D.</given-names></name><name><surname>Kelly</surname><given-names>M.</given-names></name><name><surname>Gong</surname><given-names>P.</given-names></name></person-group><article-title>A spatial-temporal approach to monitoring forest disease spread using multi-temporal high spatial resolution imagery</article-title><source>Remote Sens. Environ</source><year>2006</year><volume>101</volume><fpage>167</fpage><lpage>180</lpage></element-citation></ref><ref id="b23-sensors-11-01943"><label>23.</label><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Sun</surname><given-names>W.X.</given-names></name><name><surname>Heidt.</surname><given-names>V.</given-names></name><name><surname>Gong</surname><given-names>P.</given-names></name><name><surname>Xu</surname><given-names>G.</given-names></name></person-group><article-title>Information fusion for rural land-use classification with high-resolution satellite imagery</article-title><source>IEEE Trans. Geosci. Remote Sens</source><year>2003</year><volume>42</volume><fpage>883</fpage><lpage>890</lpage></element-citation></ref><ref id="b24-sensors-11-01943"><label>24.</label><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Dong</surname><given-names>R.</given-names></name><name><surname>Dong</surname><given-names>J.</given-names></name><name><surname>Wu</surname><given-names>G.</given-names></name><name><surname>Deng</surname><given-names>H.</given-names></name></person-group><article-title>Optimization of post-classification processing of high-resolution satellite image: A case study</article-title><source>Sci. China: Series E Tech. Sci</source><year>2006</year><volume>49</volume><fpage>98</fpage><lpage>107</lpage></element-citation></ref><ref id="b25-sensors-11-01943"><label>25.</label><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Yu</surname><given-names>Q.</given-names></name><name><surname>Gong</surname><given-names>P.</given-names></name><name><surname>Clinton</surname><given-names>N.</given-names></name><name><surname>Biging</surname><given-names>G.</given-names></name><name><surname>Kelly</surname><given-names>M.</given-names></name><name><surname>Schirokauer</surname><given-names>D.</given-names></name></person-group><article-title>Object-based detailed vegetation classification with airborne high spatial resolution remote sensing imagery</article-title><source>Photogramm. Eng. Remote Sensing</source><year>2006</year><volume>72</volume><fpage>799</fpage><lpage>811</lpage></element-citation></ref><ref id="b26-sensors-11-01943"><label>26.</label><element-citation publication-type="confproc"><person-group person-group-type="author"><name><surname>Mathieu</surname><given-names>R.</given-names></name><name><surname>Aryal</surname><given-names>J.</given-names></name></person-group><article-title>Object-oriented classification and Ikonos multispectral imagery for mapping vegetation communities in urban areas</article-title><conf-name>Proceedings of 17th Annual Colloquium of the Spatial Information Research Centre (SIRC 2005: A Spatio-temporal Workshop)</conf-name><conf-loc>Dunedin, New Zealand</conf-loc><conf-date>November 2005</conf-date><fpage>181</fpage><lpage>188</lpage></element-citation></ref><ref id="b27-sensors-11-01943"><label>27.</label><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Hay</surname><given-names>G.J.</given-names></name><name><surname>Castilla</surname><given-names>G.</given-names></name><name><surname>Wulder</surname><given-names>M.A.</given-names></name><name><surname>Ruiz</surname><given-names>J.R.</given-names></name></person-group><article-title>An automated object-based approach for the multiscale image segmentation of forest scenes</article-title><source>Int. J. Appl. Earth Obs. Geoinf</source><year>2005</year><volume>7</volume><fpage>339</fpage><lpage>359</lpage></element-citation></ref><ref id="b28-sensors-11-01943"><label>28.</label><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Van Coillie</surname><given-names>F.M.B.</given-names></name><name><surname>Verbeke</surname><given-names>L.P.C.</given-names></name><name><surname>De Wulf</surname><given-names>R.R.</given-names></name></person-group><article-title>Feature selection by genetic algorithms in object-based classification of IKONOS imagery for forest mapping in Flanders, Belgium</article-title><source>Remote Sens. Environ</source><year>2007</year><volume>110</volume><fpage>476</fpage><lpage>487</lpage></element-citation></ref><ref id="b29-sensors-11-01943"><label>29.</label><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Pal</surname><given-names>S.K.</given-names></name><name><surname>Ghosh</surname><given-names>A.</given-names></name><name><surname>Shankar</surname><given-names>B.U.</given-names></name></person-group><article-title>Segmentation of remotely sensed images with Fuzzy thresholding, and quantitative evaluation</article-title><source>Int. J. Remote Sens</source><year>2000</year><volume>21</volume><fpage>2269</fpage><lpage>2300</lpage></element-citation></ref><ref id="b30-sensors-11-01943"><label>30.</label><element-citation publication-type="book"><person-group person-group-type="author"><collab>eCognition</collab></person-group><source>User’s Guide</source><publisher-name>Definiens Imaging GmBH</publisher-name><publisher-loc>Munich, Germany</publisher-loc><year>2002</year><fpage>65</fpage></element-citation></ref><ref id="b31-sensors-11-01943"><label>31.</label><element-citation publication-type="book"><person-group person-group-type="author"><name><surname>Lillesand</surname><given-names>T.M.</given-names></name><name><surname>Kiefer</surname><given-names>R.W.</given-names></name></person-group><source>Remote Sensing and Image Interpretation</source><edition>4th ed</edition><publisher-name>John Wiley and Sons</publisher-name><publisher-loc>New York, NY, USA</publisher-loc><year>2000</year><fpage>724</fpage></element-citation></ref><ref id="b32-sensors-11-01943"><label>32.</label><element-citation publication-type="book"><person-group person-group-type="author"><collab>Definiens</collab></person-group><source>Cognition User’s guide</source><publisher-name>Definiens AG</publisher-name><publisher-loc>Munich, Germany</publisher-loc><year>2002</year><fpage>65</fpage></element-citation></ref><ref id="b33-sensors-11-01943"><label>33.</label><element-citation publication-type="book"><person-group person-group-type="author"><collab>Definiens</collab></person-group><source>Definiens Professional 5 User Guide</source><publisher-name>Definiens AG</publisher-name><publisher-loc>Munich, Germany</publisher-loc><year>2006</year><fpage>15</fpage></element-citation></ref><ref id="b34-sensors-11-01943"><label>34.</label><element-citation publication-type="confproc"><person-group person-group-type="author"><name><surname>Meinel</surname><given-names>G.</given-names></name><name><surname>Neubert</surname><given-names>M.</given-names></name><name><surname>Reder</surname><given-names>J.</given-names></name></person-group><article-title>The potential use of very high resolution satellite data for urban areas—First experiences with IKONOS data, their classification and application in urban planning and environmental monitoring</article-title><conf-name>Proceedings of the Symposium on Remote Sensing of Urban Areas</conf-name><conf-loc>Regensburg, Germany</conf-loc><conf-date>June 2001</conf-date><fpage>196</fpage><lpage>205</lpage></element-citation></ref><ref id="b35-sensors-11-01943"><label>35.</label><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Congalton</surname><given-names>R.G.</given-names></name></person-group><article-title>A review of assessing the accuracy of classifications of remotely sensed data</article-title><source>Remote Sens. Environ</source><year>1991</year><volume>37</volume><fpage>35</fpage><lpage>46</lpage></element-citation></ref><ref id="b36-sensors-11-01943"><label>36.</label><element-citation publication-type="confproc"><person-group person-group-type="author"><name><surname>Fenstermaker</surname><given-names>L.</given-names></name></person-group><article-title>A proposed approach for national to global scale error assessments</article-title><conf-name>Proceedings of Geographic Information Systems and Land Information Systems,</conf-name><conf-loc>Atlanta, GA, USA</conf-loc><conf-date>28 October–1 November 1991</conf-date><fpage>293</fpage><lpage>300</lpage></element-citation></ref><ref id="b37-sensors-11-01943"><label>37.</label><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Plourde</surname><given-names>L.</given-names></name><name><surname>Congalton</surname><given-names>R.G.</given-names></name></person-group><article-title>Sampling method sample placement: How do they affect the accuracy of remotely sensed maps?</article-title><source>Photogramm. Eng. Remote Sensing</source><year>2003</year><volume>69</volume><fpage>289</fpage><lpage>297</lpage></element-citation></ref><ref id="b38-sensors-11-01943"><label>38.</label><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Meyer</surname><given-names>P.</given-names></name><name><surname>Staenz</surname><given-names>K.</given-names></name><name><surname>Itten</surname><given-names>K.I.</given-names></name></person-group><article-title>Semi-automated procedure for tree species identification in high spatial resolution data from digitized colour infrared–aerial photography</article-title><source>ISPRS J. Photogramm. Remote Sens</source><year>1996</year><volume>51</volume><fpage>5</fpage><lpage>16</lpage></element-citation></ref><ref id="b39-sensors-11-01943"><label>39.</label><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Gong</surname><given-names>P.</given-names></name><name><surname>Howarth</surname><given-names>P.J.</given-names></name></person-group><article-title>Frequency-based contextual classification and grey-level vector reduction for land-use identification</article-title><source>Photogram. Eng. Remote Sensing</source><year>1992</year><volume>58</volume><fpage>423</fpage><lpage>437</lpage></element-citation></ref><ref id="b40-sensors-11-01943"><label>40.</label><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Blaschke</surname><given-names>T.</given-names></name><name><surname>Strobl</surname><given-names>J.</given-names></name></person-group><year>2002</year><article-title>What’s wrong with pixels? Some recent developments interfacing remote sensing and GIS</article-title><source>GeoBIT/GIS</source><year>1992a</year><volume>6</volume><fpage>12</fpage><lpage>17</lpage></element-citation></ref><ref id="b41-sensors-11-01943"><label>41.</label><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Blaschke</surname><given-names>T.</given-names></name></person-group><article-title>Ohne Salz und Pfeffer—Objektorientierte Bildanalyse—Eine Revolution in der Fernerkundung</article-title><source>GeoBIT</source><year>2000</year><volume>2</volume><fpage>30</fpage><lpage>32</lpage></element-citation></ref><ref id="b42-sensors-11-01943"><label>42.</label><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Heyman</surname><given-names>O.</given-names></name><name><surname>Gaston</surname><given-names>G.G.</given-names></name><name><surname>Kimerling</surname><given-names>A.J.</given-names></name><name><surname>Campbell</surname><given-names>J.T.</given-names></name></person-group><article-title>A Per-Segment Approach to Improve Aspen Mapping from High-Resolution Remote Sensing Imagery</article-title><source>J. For</source><year>2003</year><volume>101</volume><fpage>29</fpage><lpage>33</lpage></element-citation></ref><ref id="b43-sensors-11-01943"><label>43.</label><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Xiao</surname><given-names>X.</given-names></name><name><surname>Boles</surname><given-names>S.</given-names></name><name><surname>Liu.</surname><given-names>J.</given-names></name><name><surname>Zuang</surname><given-names>D.</given-names></name><name><surname>Liu</surname><given-names>M.</given-names></name></person-group><article-title>Characterization of forest types in Northeastern China, using multi-temporal SPOT-4 VEGETSTION sensor data</article-title><source>Remote Sens. Environ</source><year>2002</year><volume>82</volume><fpage>335</fpage><lpage>348</lpage></element-citation></ref><ref id="b44-sensors-11-01943"><label>44.</label><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Jobin</surname><given-names>B.</given-names></name><name><surname>Labrecque</surname><given-names>S.</given-names></name><name><surname>Grenier</surname><given-names>M.</given-names></name><name><surname>Falardeau</surname><given-names>G.</given-names></name></person-group><article-title>Object-based classification as an alternative approach to the traditional pixel-based classification to identify potential habitat of the Grasshopper Sparrow</article-title><source>Environ. Manage</source><year>2008</year><volume>41</volume><fpage>20</fpage><lpage>31</lpage><pub-id pub-id-type="pmid">17985180</pub-id></element-citation></ref></ref-list></back><floats-group><fig id="f1-sensors-11-01943" position="float"><label>Figure 1.</label><caption><p>The study area with the pan-sharpened IKONOS image.</p></caption><graphic xlink:href="sensors-11-01943f1"></graphic></fig><fig id="f2-sensors-11-01943" position="float"><label>Figure 2.</label><caption><p>Average spectral signature of the IKONOS image for a selected training area of each class (P.rig: <italic>Pinus rigida</italic>, P.kor: <italic>Pinus koraiensis</italic>, L.lep: <italic>Larix leptolepis</italic>, Q.mon: <italic>Quercus mongolica</italic>, Q.var: <italic>Quercus variabilis</italic>, Q.acu: <italic>Quercus acutissima</italic>, C.cre: <italic>Castanea crenata</italic>, grass land, non forest).</p></caption><graphic xlink:href="sensors-11-01943f2"></graphic></fig><fig id="f3-sensors-11-01943" position="float"><label>Figure 3.</label><caption><p>The kappa values of SBCs: <bold>(a)</bold> strong weights for color or shape, and <bold>(b)</bold> medium weights for color and shape.</p></caption><graphic xlink:href="sensors-11-01943f3"></graphic></fig><fig id="f4-sensors-11-01943" position="float"><label>Figure 4.</label><caption><p>Optimal weights for segmentation.</p></caption><graphic xlink:href="sensors-11-01943f4"></graphic></fig><fig id="f5-sensors-11-01943" position="float"><label>Figure 5.</label><caption><p>Forest cover map: <bold>(a)</bold> forest cover map of the pixel-based classification (PBC) using the maximum likelihood method, and <bold>(b)</bold> segment-based classification (SBC) with majority principle (color parameter of 0.5, scale parameter of 160).</p></caption><graphic xlink:href="sensors-11-01943f5"></graphic></fig><fig id="f6-sensors-11-01943" position="float"><label>Figure 6.</label><caption><p>Forest cover maps generated by pixel-based (PBC), majority rule-based, object-based (OBC), with a color parameter of 0.5 and a scale parameter of 160 (C<sub>5</sub>S<sub>5</sub>(M<sub>5</sub>O<sub>5</sub>)), and segment-based (SBC) classification methods.</p></caption><graphic xlink:href="sensors-11-01943f6"></graphic></fig><table-wrap id="t1-sensors-11-01943" position="float"><label>Table 1.</label><caption><p>Classification schema for preparing forest cover map.</p></caption><table frame="hsides" rules="groups"><thead><tr><th align="center" valign="bottom" rowspan="1" colspan="1"><bold>Class description</bold></th><th align="center" valign="bottom" rowspan="1" colspan="1"><bold>Abbreviation</bold></th><th align="center" valign="bottom" rowspan="1" colspan="1"><bold>Common name</bold></th><th align="center" valign="bottom" rowspan="1" colspan="1"><bold>Large scale class</bold></th></tr></thead><tbody><tr><td align="left" valign="middle" rowspan="1" colspan="1"><italic>Pinus rigida</italic>–stand</td><td align="center" valign="middle" rowspan="1" colspan="1"><italic>P.rig</italic></td><td align="center" valign="middle" rowspan="1" colspan="1">Pitch pine</td><td align="center" valign="middle" rowspan="8" colspan="1">Forest area</td></tr><tr><td align="left" valign="middle" rowspan="1" colspan="1"><italic>Pinus koraiensis</italic>–stand</td><td align="center" valign="middle" rowspan="1" colspan="1"><italic>P.kor</italic></td><td align="center" valign="middle" rowspan="1" colspan="1">Korean pine</td></tr><tr><td align="left" valign="middle" rowspan="1" colspan="1"><italic>Larix leptolepis</italic>–stand</td><td align="center" valign="middle" rowspan="1" colspan="1"><italic>L.lep</italic></td><td align="center" valign="middle" rowspan="1" colspan="1">Japanese larch</td></tr><tr><td align="left" valign="middle" rowspan="1" colspan="1"><italic>Quercus mongolica</italic>–stand</td><td align="center" valign="middle" rowspan="1" colspan="1"><italic>Q.mon</italic></td><td align="center" valign="middle" rowspan="1" colspan="1">Mongolian oak</td></tr><tr><td align="left" valign="middle" rowspan="1" colspan="1"><italic>Quercus variabilis</italic>–stand</td><td align="center" valign="middle" rowspan="1" colspan="1"><italic>Q.var</italic></td><td align="center" valign="middle" rowspan="1" colspan="1">Cork oak</td></tr><tr><td align="left" valign="middle" rowspan="1" colspan="1"><italic>Quercus acutissima</italic>–stand</td><td align="center" valign="middle" rowspan="1" colspan="1"><italic>Q.acu</italic></td><td align="center" valign="middle" rowspan="1" colspan="1">Sawtooth oak</td></tr><tr><td align="left" valign="middle" rowspan="1" colspan="1"><italic>Castanea crenata</italic>–stand</td><td align="center" valign="middle" rowspan="1" colspan="1"><italic>C.cre</italic></td><td align="center" valign="middle" rowspan="1" colspan="1">Chestnut</td></tr><tr><td align="left" valign="middle" rowspan="1" colspan="1">Grass land</td><td align="center" valign="middle" rowspan="1" colspan="1">grass</td><td align="center" valign="middle" rowspan="1" colspan="1">-</td></tr><tr><td colspan="4" align="center" valign="middle" rowspan="1"><hr></hr></td></tr><tr><td align="left" valign="middle" rowspan="1" colspan="1">Agricultural area (dry, wet), Water, House, Road</td><td align="center" valign="middle" rowspan="1" colspan="1">Non-forest</td><td align="center" valign="middle" rowspan="1" colspan="1">-</td><td align="center" valign="middle" rowspan="1" colspan="1">Non-forest area</td></tr></tbody></table></table-wrap><table-wrap id="t2-sensors-11-01943" position="float"><label>Table 2.</label><caption><p>Different weights for finding optimal segment size. The notation conveys information about the weights for color, shape, smoothness or compactness (C = color, S = shape, M = smoothness, O = compactness, subscript 1 = 0.1, 3 = 0.25, 5 = 0.5, 7 = 0.75 and 9 = 0.9).</p></caption><table frame="hsides" rules="groups"><thead><tr><th align="center" valign="bottom" rowspan="1" colspan="1"><bold>Step</bold></th><th align="center" valign="bottom" rowspan="1" colspan="1"><bold>Case</bold></th><th align="center" valign="bottom" rowspan="1" colspan="1"><bold>Color</bold></th><th align="center" valign="bottom" rowspan="1" colspan="1"><bold>Shape(smoothness/compactness)</bold></th></tr></thead><tbody><tr><td align="center" valign="middle" rowspan="4" colspan="1">I</td><td align="center" valign="middle" rowspan="1" colspan="1">C<sub>9</sub>S<sub>1</sub>(M<sub>9</sub>O<sub>1</sub>)</td><td align="center" valign="middle" rowspan="1" colspan="1">0.9</td><td align="center" valign="middle" rowspan="1" colspan="1">0.1(0.9/0.1)</td></tr><tr><td align="center" valign="middle" rowspan="1" colspan="1">C<sub>9</sub>S<sub>1</sub>(M<sub>1</sub>O<sub>9</sub>)</td><td align="center" valign="middle" rowspan="1" colspan="1">0.9</td><td align="center" valign="middle" rowspan="1" colspan="1">0.1(0.1/0.9)</td></tr><tr><td align="center" valign="middle" rowspan="1" colspan="1">C<sub>1</sub>S<sub>9</sub>(M<sub>9</sub>O<sub>1</sub>)</td><td align="center" valign="middle" rowspan="1" colspan="1">0.1</td><td align="center" valign="middle" rowspan="1" colspan="1">0.9(0.9/0.1)</td></tr><tr><td align="center" valign="middle" rowspan="1" colspan="1">C<sub>1</sub>S<sub>9</sub>(M<sub>1</sub>O<sub>9</sub>)</td><td align="center" valign="middle" rowspan="1" colspan="1">0.1</td><td align="center" valign="middle" rowspan="1" colspan="1">0.9(0.1/0.9)</td></tr><tr><td colspan="4" align="center" valign="middle" rowspan="1"><hr></hr></td></tr><tr><td align="center" valign="middle" rowspan="3" colspan="1">II</td><td align="center" valign="middle" rowspan="1" colspan="1">C<sub>5</sub>S<sub>5</sub>(M<sub>5</sub>O<sub>5</sub>)</td><td align="center" valign="middle" rowspan="1" colspan="1">0.5</td><td align="center" valign="middle" rowspan="1" colspan="1">0.5(0.5/0.5)</td></tr><tr><td align="center" valign="middle" rowspan="1" colspan="1">C<sub>5</sub>S<sub>5</sub>(M<sub>9</sub>O<sub>1</sub>)</td><td align="center" valign="middle" rowspan="1" colspan="1">0.5</td><td align="center" valign="middle" rowspan="1" colspan="1">0.5(0.9/0.1)</td></tr><tr><td align="center" valign="middle" rowspan="1" colspan="1">C<sub>5</sub>S<sub>5</sub>(M<sub>1</sub>O<sub>9</sub>)</td><td align="center" valign="middle" rowspan="1" colspan="1">0.5</td><td align="center" valign="middle" rowspan="1" colspan="1">0.5(0.1/0.9)</td></tr><tr><td colspan="4" align="center" valign="middle" rowspan="1"><hr></hr></td></tr><tr><td align="center" valign="middle" rowspan="2" colspan="1">III</td><td align="center" valign="middle" rowspan="1" colspan="1">C<sub>7</sub>S<sub>3</sub>(M<sub>5</sub>O<sub>5</sub>)</td><td align="center" valign="middle" rowspan="1" colspan="1">0.75</td><td align="center" valign="middle" rowspan="1" colspan="1">0.25(0.5/0.5)</td></tr><tr><td align="center" valign="middle" rowspan="1" colspan="1">C<sub>3</sub>S<sub>7</sub>(M<sub>5</sub>O<sub>5</sub>)</td><td align="center" valign="middle" rowspan="1" colspan="1">0.25</td><td align="center" valign="middle" rowspan="1" colspan="1">0.75(0.5/0.5)</td></tr></tbody></table></table-wrap><table-wrap id="t3-sensors-11-01943" position="float"><label>Table 3.</label><caption><p>Error matrix of the pixel-based classification (PBC) using a maximum likelihood method.</p></caption><table frame="hsides" rules="groups"><thead><tr><th align="center" valign="middle" rowspan="2" colspan="2"><bold>Classes</bold></th><th colspan="9" align="center" valign="bottom" rowspan="1"><bold>Reference data</bold><hr></hr></th><th align="center" valign="middle" rowspan="2" colspan="1"><bold>Sum</bold></th><th align="center" valign="middle" rowspan="2" colspan="1"><bold>User’s Accuracy (%)</bold></th></tr><tr><th align="center" valign="middle" rowspan="1" colspan="1"><bold><italic>P. rig</italic></bold></th><th align="center" valign="middle" rowspan="1" colspan="1"><bold><italic>P. kor</italic></bold></th><th align="center" valign="middle" rowspan="1" colspan="1"><bold><italic>L. lep</italic></bold></th><th align="center" valign="middle" rowspan="1" colspan="1"><bold><italic>Q. mon</italic></bold></th><th align="center" valign="middle" rowspan="1" colspan="1"><bold><italic>Q. var</italic></bold></th><th align="center" valign="middle" rowspan="1" colspan="1"><bold><italic>Q. acu</italic></bold></th><th align="center" valign="middle" rowspan="1" colspan="1"><bold><italic>C. cre</italic></bold></th><th align="center" valign="middle" rowspan="1" colspan="1"><bold>Grass</bold></th><th align="center" valign="middle" rowspan="1" colspan="1"><bold>Non-forest</bold></th></tr></thead><tbody><tr><td align="center" valign="middle" rowspan="9" colspan="1">Classified data</td><td align="center" valign="middle" rowspan="1" colspan="1"><italic>P. rig</italic></td><td align="center" valign="middle" rowspan="1" colspan="1"><bold>13</bold></td><td align="center" valign="middle" rowspan="1" colspan="1">13</td><td align="center" valign="middle" rowspan="1" colspan="1">1</td><td align="center" valign="middle" rowspan="1" colspan="1">2</td><td align="center" valign="middle" rowspan="1" colspan="1">0</td><td align="center" valign="middle" rowspan="1" colspan="1">1</td><td align="center" valign="middle" rowspan="1" colspan="1">1</td><td align="center" valign="middle" rowspan="1" colspan="1">1</td><td align="center" valign="middle" rowspan="1" colspan="1">0</td><td align="center" valign="middle" rowspan="1" colspan="1">33</td><td align="center" valign="middle" rowspan="1" colspan="1">41.0</td></tr><tr><td align="center" valign="middle" rowspan="1" colspan="1"><italic>P. kor</italic></td><td align="center" valign="middle" rowspan="1" colspan="1">3</td><td align="center" valign="middle" rowspan="1" colspan="1"><bold>25</bold></td><td align="center" valign="middle" rowspan="1" colspan="1">3</td><td align="center" valign="middle" rowspan="1" colspan="1">3</td><td align="center" valign="middle" rowspan="1" colspan="1">0</td><td align="center" valign="middle" rowspan="1" colspan="1">1</td><td align="center" valign="middle" rowspan="1" colspan="1">1</td><td align="center" valign="middle" rowspan="1" colspan="1">0</td><td align="center" valign="middle" rowspan="1" colspan="1">1</td><td align="center" valign="middle" rowspan="1" colspan="1">36</td><td align="center" valign="middle" rowspan="1" colspan="1">67.5</td></tr><tr><td align="center" valign="middle" rowspan="1" colspan="1"><italic>L. lep</italic></td><td align="center" valign="middle" rowspan="1" colspan="1">2</td><td align="center" valign="middle" rowspan="1" colspan="1">6</td><td align="center" valign="middle" rowspan="1" colspan="1"><bold>40</bold></td><td align="center" valign="middle" rowspan="1" colspan="1">5</td><td align="center" valign="middle" rowspan="1" colspan="1">0</td><td align="center" valign="middle" rowspan="1" colspan="1">3</td><td align="center" valign="middle" rowspan="1" colspan="1">0</td><td align="center" valign="middle" rowspan="1" colspan="1">0</td><td align="center" valign="middle" rowspan="1" colspan="1">1</td><td align="center" valign="middle" rowspan="1" colspan="1">57</td><td align="center" valign="middle" rowspan="1" colspan="1">70.7</td></tr><tr><td align="center" valign="middle" rowspan="1" colspan="1"><italic>Q. mon</italic></td><td align="center" valign="middle" rowspan="1" colspan="1">0</td><td align="center" valign="middle" rowspan="1" colspan="1">1</td><td align="center" valign="middle" rowspan="1" colspan="1">3</td><td align="center" valign="middle" rowspan="1" colspan="1"><bold>34</bold></td><td align="center" valign="middle" rowspan="1" colspan="1">1</td><td align="center" valign="middle" rowspan="1" colspan="1">10</td><td align="center" valign="middle" rowspan="1" colspan="1">0</td><td align="center" valign="middle" rowspan="1" colspan="1">0</td><td align="center" valign="middle" rowspan="1" colspan="1">1</td><td align="center" valign="middle" rowspan="1" colspan="1">49</td><td align="center" valign="middle" rowspan="1" colspan="1">67.9</td></tr><tr><td align="center" valign="middle" rowspan="1" colspan="1"><italic>Q. var</italic></td><td align="center" valign="middle" rowspan="1" colspan="1">0</td><td align="center" valign="middle" rowspan="1" colspan="1">1</td><td align="center" valign="middle" rowspan="1" colspan="1">1</td><td align="center" valign="middle" rowspan="1" colspan="1">2</td><td align="center" valign="middle" rowspan="1" colspan="1"><bold>33</bold></td><td align="center" valign="middle" rowspan="1" colspan="1">5</td><td align="center" valign="middle" rowspan="1" colspan="1">1</td><td align="center" valign="middle" rowspan="1" colspan="1">4</td><td align="center" valign="middle" rowspan="1" colspan="1">5</td><td align="center" valign="middle" rowspan="1" colspan="1">53</td><td align="center" valign="middle" rowspan="1" colspan="1">63.0</td></tr><tr><td align="center" valign="middle" rowspan="1" colspan="1"><italic>Q. acu</italic></td><td align="center" valign="middle" rowspan="1" colspan="1">1</td><td align="center" valign="middle" rowspan="1" colspan="1">1</td><td align="center" valign="middle" rowspan="1" colspan="1">1</td><td align="center" valign="middle" rowspan="1" colspan="1">9</td><td align="center" valign="middle" rowspan="1" colspan="1">7</td><td align="center" valign="middle" rowspan="1" colspan="1"><bold>27</bold></td><td align="center" valign="middle" rowspan="1" colspan="1">1</td><td align="center" valign="middle" rowspan="1" colspan="1">1</td><td align="center" valign="middle" rowspan="1" colspan="1">3</td><td align="center" valign="middle" rowspan="1" colspan="1">52</td><td align="center" valign="middle" rowspan="1" colspan="1">51.2</td></tr><tr><td align="center" valign="middle" rowspan="1" colspan="1"><italic>C. cre</italic></td><td align="center" valign="middle" rowspan="1" colspan="1">1</td><td align="center" valign="middle" rowspan="1" colspan="1">2</td><td align="center" valign="middle" rowspan="1" colspan="1">0</td><td align="center" valign="middle" rowspan="1" colspan="1">1</td><td align="center" valign="middle" rowspan="1" colspan="1">3</td><td align="center" valign="middle" rowspan="1" colspan="1">3</td><td align="center" valign="middle" rowspan="1" colspan="1"><bold>14</bold></td><td align="center" valign="middle" rowspan="1" colspan="1">2</td><td align="center" valign="middle" rowspan="1" colspan="1">2</td><td align="center" valign="middle" rowspan="1" colspan="1">28</td><td align="center" valign="middle" rowspan="1" colspan="1">51.6</td></tr><tr><td align="center" valign="middle" rowspan="1" colspan="1">Grass</td><td align="center" valign="middle" rowspan="1" colspan="1">1</td><td align="center" valign="middle" rowspan="1" colspan="1">1</td><td align="center" valign="middle" rowspan="1" colspan="1">2</td><td align="center" valign="middle" rowspan="1" colspan="1">1</td><td align="center" valign="middle" rowspan="1" colspan="1">4</td><td align="center" valign="middle" rowspan="1" colspan="1">4</td><td align="center" valign="middle" rowspan="1" colspan="1">2</td><td align="center" valign="middle" rowspan="1" colspan="1"><bold>16</bold></td><td align="center" valign="middle" rowspan="1" colspan="1">3</td><td align="center" valign="middle" rowspan="1" colspan="1">35</td><td align="center" valign="middle" rowspan="1" colspan="1">47.2</td></tr><tr><td align="center" valign="middle" rowspan="1" colspan="1">Non-forest</td><td align="center" valign="middle" rowspan="1" colspan="1">0</td><td align="center" valign="middle" rowspan="1" colspan="1">1</td><td align="center" valign="middle" rowspan="1" colspan="1">0</td><td align="center" valign="middle" rowspan="1" colspan="1">1</td><td align="center" valign="middle" rowspan="1" colspan="1">2</td><td align="center" valign="middle" rowspan="1" colspan="1">1</td><td align="center" valign="middle" rowspan="1" colspan="1">0</td><td align="center" valign="middle" rowspan="1" colspan="1">1</td><td align="center" valign="middle" rowspan="1" colspan="1"><bold>59</bold></td><td align="center" valign="middle" rowspan="1" colspan="1">66</td><td align="center" valign="middle" rowspan="1" colspan="1">89.4</td></tr><tr><td colspan="13" align="center" valign="middle" rowspan="1"><hr></hr></td></tr><tr><td align="center" valign="middle" rowspan="1" colspan="1"></td><td align="center" valign="middle" rowspan="1" colspan="1">Sum</td><td align="center" valign="middle" rowspan="1" colspan="1">22</td><td align="center" valign="middle" rowspan="1" colspan="1">50</td><td align="center" valign="middle" rowspan="1" colspan="1">53</td><td align="center" valign="middle" rowspan="1" colspan="1">57</td><td align="center" valign="middle" rowspan="1" colspan="1">51</td><td align="center" valign="middle" rowspan="1" colspan="1">56</td><td align="center" valign="middle" rowspan="1" colspan="1">20</td><td align="center" valign="middle" rowspan="1" colspan="1">26</td><td align="center" valign="middle" rowspan="1" colspan="1">74</td><td align="center" valign="middle" rowspan="1" colspan="1"><bold>409</bold></td><td align="center" valign="middle" rowspan="1" colspan="1"></td></tr><tr><td colspan="12" align="center" valign="middle" rowspan="1"><hr></hr></td><td align="center" valign="middle" rowspan="1" colspan="1"></td></tr><tr><td align="center" valign="middle" colspan="2" rowspan="1">Producer’s Accuracy (%)</td><td align="center" valign="middle" rowspan="1" colspan="1">60.6</td><td align="center" valign="middle" rowspan="1" colspan="1">49.2</td><td align="center" valign="middle" rowspan="1" colspan="1">76.5</td><td align="center" valign="middle" rowspan="1" colspan="1">58.9</td><td align="center" valign="middle" rowspan="1" colspan="1">64.9</td><td align="center" valign="middle" rowspan="1" colspan="1">47.9</td><td align="center" valign="middle" rowspan="1" colspan="1">71.4</td><td align="center" valign="middle" rowspan="1" colspan="1">63.1</td><td align="center" valign="middle" rowspan="1" colspan="1">79.6</td><td align="center" valign="middle" colspan="2" rowspan="1">Overall accuracy: 63.9% Kappa value: 0.59</td></tr></tbody></table></table-wrap><table-wrap id="t4-sensors-11-01943" position="float"><label>Table 4.</label><caption><p>Error matrix of segment-based classification (SBC) using a majority rule, with a color parameter of 0.5 and a scale parameter of 160 (C<sub>5</sub>S<sub>5</sub>(M<sub>5</sub>O<sub>5</sub>)).</p></caption><table frame="hsides" rules="groups"><thead><tr><th align="center" valign="middle" rowspan="2" colspan="2"><bold>Classes</bold></th><th colspan="9" align="center" valign="middle" rowspan="1"><bold>Reference data</bold><hr></hr></th><th align="center" valign="middle" rowspan="2" colspan="1"><bold>Sum</bold></th><th align="center" valign="middle" rowspan="2" colspan="1"><bold>User’s Accuracy (%)</bold></th></tr><tr><th align="center" valign="middle" rowspan="1" colspan="1"><bold><italic>P. rig</italic></bold></th><th align="center" valign="middle" rowspan="1" colspan="1"><bold><italic>P. kor</italic></bold></th><th align="center" valign="middle" rowspan="1" colspan="1"><bold><italic>L. lep</italic></bold></th><th align="center" valign="middle" rowspan="1" colspan="1"><bold><italic>Q. mon</italic></bold></th><th align="center" valign="middle" rowspan="1" colspan="1"><bold><italic>Q. var</italic></bold></th><th align="center" valign="middle" rowspan="1" colspan="1"><bold><italic>Q. acu</italic></bold></th><th align="center" valign="middle" rowspan="1" colspan="1"><bold><italic>C. cre</italic></bold></th><th align="center" valign="middle" rowspan="1" colspan="1"><bold>Grass</bold></th><th align="center" valign="middle" rowspan="1" colspan="1"><bold>Non-forest</bold></th></tr></thead><tbody><tr><td align="center" valign="middle" rowspan="9" colspan="1">Classified data</td><td align="center" valign="middle" rowspan="1" colspan="1"><italic>P. rig</italic></td><td align="center" valign="middle" rowspan="1" colspan="1"><bold>19</bold></td><td align="center" valign="middle" rowspan="1" colspan="1">5</td><td align="center" valign="middle" rowspan="1" colspan="1">0</td><td align="center" valign="middle" rowspan="1" colspan="1">0</td><td align="center" valign="middle" rowspan="1" colspan="1">0</td><td align="center" valign="middle" rowspan="1" colspan="1">1</td><td align="center" valign="middle" rowspan="1" colspan="1">0</td><td align="center" valign="middle" rowspan="1" colspan="1">2</td><td align="center" valign="middle" rowspan="1" colspan="1">0</td><td align="center" valign="middle" rowspan="1" colspan="1">27</td><td align="center" valign="middle" rowspan="1" colspan="1">69.1</td></tr><tr><td align="center" valign="middle" rowspan="1" colspan="1"><italic>P. kor</italic></td><td align="center" valign="middle" rowspan="1" colspan="1">1</td><td align="center" valign="middle" rowspan="1" colspan="1"><bold>28</bold></td><td align="center" valign="middle" rowspan="1" colspan="1">0</td><td align="center" valign="middle" rowspan="1" colspan="1">4</td><td align="center" valign="middle" rowspan="1" colspan="1">0</td><td align="center" valign="middle" rowspan="1" colspan="1">0</td><td align="center" valign="middle" rowspan="1" colspan="1">0</td><td align="center" valign="middle" rowspan="1" colspan="1">0</td><td align="center" valign="middle" rowspan="1" colspan="1">3</td><td align="center" valign="middle" rowspan="1" colspan="1">36</td><td align="center" valign="middle" rowspan="1" colspan="1">77.0</td></tr><tr><td align="center" valign="middle" rowspan="1" colspan="1"><italic>L. lep</italic></td><td align="center" valign="middle" rowspan="1" colspan="1">1</td><td align="center" valign="middle" rowspan="1" colspan="1">8</td><td align="center" valign="middle" rowspan="1" colspan="1"><bold>44</bold></td><td align="center" valign="middle" rowspan="1" colspan="1">3</td><td align="center" valign="middle" rowspan="1" colspan="1">3</td><td align="center" valign="middle" rowspan="1" colspan="1">5</td><td align="center" valign="middle" rowspan="1" colspan="1">0</td><td align="center" valign="middle" rowspan="1" colspan="1">0</td><td align="center" valign="middle" rowspan="1" colspan="1">1</td><td align="center" valign="middle" rowspan="1" colspan="1">66</td><td align="center" valign="middle" rowspan="1" colspan="1">67.2</td></tr><tr><td align="center" valign="middle" rowspan="1" colspan="1"><italic>Q. mon</italic></td><td align="center" valign="middle" rowspan="1" colspan="1">0</td><td align="center" valign="middle" rowspan="1" colspan="1">4</td><td align="center" valign="middle" rowspan="1" colspan="1">7</td><td align="center" valign="middle" rowspan="1" colspan="1"><bold>40</bold></td><td align="center" valign="middle" rowspan="1" colspan="1">0</td><td align="center" valign="middle" rowspan="1" colspan="1">4</td><td align="center" valign="middle" rowspan="1" colspan="1">1</td><td align="center" valign="middle" rowspan="1" colspan="1">0</td><td align="center" valign="middle" rowspan="1" colspan="1">0</td><td align="center" valign="middle" rowspan="1" colspan="1">56</td><td align="center" valign="middle" rowspan="1" colspan="1">71.9</td></tr><tr><td align="center" valign="middle" rowspan="1" colspan="1"><italic>Q. var</italic></td><td align="center" valign="middle" rowspan="1" colspan="1">0</td><td align="center" valign="middle" rowspan="1" colspan="1">1</td><td align="center" valign="middle" rowspan="1" colspan="1">0</td><td align="center" valign="middle" rowspan="1" colspan="1">1</td><td align="center" valign="middle" rowspan="1" colspan="1"><bold>37</bold></td><td align="center" valign="middle" rowspan="1" colspan="1">1</td><td align="center" valign="middle" rowspan="1" colspan="1">0</td><td align="center" valign="middle" rowspan="1" colspan="1">0</td><td align="center" valign="middle" rowspan="1" colspan="1">0</td><td align="center" valign="middle" rowspan="1" colspan="1">40</td><td align="center" valign="middle" rowspan="1" colspan="1">92.0</td></tr><tr><td align="center" valign="middle" rowspan="1" colspan="1"><italic>Q. acu</italic></td><td align="center" valign="middle" rowspan="1" colspan="1">1</td><td align="center" valign="middle" rowspan="1" colspan="1">0</td><td align="center" valign="middle" rowspan="1" colspan="1">1</td><td align="center" valign="middle" rowspan="1" colspan="1">8</td><td align="center" valign="middle" rowspan="1" colspan="1">9</td><td align="center" valign="middle" rowspan="1" colspan="1"><bold>44</bold></td><td align="center" valign="middle" rowspan="1" colspan="1">0</td><td align="center" valign="middle" rowspan="1" colspan="1">2</td><td align="center" valign="middle" rowspan="1" colspan="1">3</td><td align="center" valign="middle" rowspan="1" colspan="1">69</td><td align="center" valign="middle" rowspan="1" colspan="1">64.4</td></tr><tr><td align="center" valign="middle" rowspan="1" colspan="1"><italic>C. cre</italic></td><td align="center" valign="middle" rowspan="1" colspan="1">0</td><td align="center" valign="middle" rowspan="1" colspan="1">1</td><td align="center" valign="middle" rowspan="1" colspan="1">0</td><td align="center" valign="middle" rowspan="1" colspan="1">0</td><td align="center" valign="middle" rowspan="1" colspan="1">0</td><td align="center" valign="middle" rowspan="1" colspan="1">0</td><td align="center" valign="middle" rowspan="1" colspan="1"><bold>18</bold></td><td align="center" valign="middle" rowspan="1" colspan="1">0</td><td align="center" valign="middle" rowspan="1" colspan="1">0</td><td align="center" valign="middle" rowspan="1" colspan="1">19</td><td align="center" valign="middle" rowspan="1" colspan="1">94.6</td></tr><tr><td align="center" valign="middle" rowspan="1" colspan="1">Grass</td><td align="center" valign="middle" rowspan="1" colspan="1">0</td><td align="center" valign="middle" rowspan="1" colspan="1">2</td><td align="center" valign="middle" rowspan="1" colspan="1">0</td><td align="center" valign="middle" rowspan="1" colspan="1">0</td><td align="center" valign="middle" rowspan="1" colspan="1">2</td><td align="center" valign="middle" rowspan="1" colspan="1">1</td><td align="center" valign="middle" rowspan="1" colspan="1">0</td><td align="center" valign="middle" rowspan="1" colspan="1"><bold>21</bold></td><td align="center" valign="middle" rowspan="1" colspan="1">4</td><td align="center" valign="middle" rowspan="1" colspan="1">31</td><td align="center" valign="middle" rowspan="1" colspan="1">69.0</td></tr><tr><td align="center" valign="middle" rowspan="1" colspan="1">Non-forest</td><td align="center" valign="middle" rowspan="1" colspan="1">0</td><td align="center" valign="middle" rowspan="1" colspan="1">0</td><td align="center" valign="middle" rowspan="1" colspan="1">0</td><td align="center" valign="middle" rowspan="1" colspan="1">0</td><td align="center" valign="middle" rowspan="1" colspan="1">0</td><td align="center" valign="middle" rowspan="1" colspan="1">0</td><td align="center" valign="middle" rowspan="1" colspan="1">1</td><td align="center" valign="middle" rowspan="1" colspan="1">1</td><td align="center" valign="middle" rowspan="1" colspan="1"><bold>63</bold></td><td align="center" valign="middle" rowspan="1" colspan="1">65</td><td align="center" valign="middle" rowspan="1" colspan="1">96.0</td></tr><tr><td colspan="13" align="center" valign="middle" rowspan="1"><hr></hr></td></tr><tr><td align="center" valign="middle" rowspan="1" colspan="1"></td><td align="center" valign="middle" rowspan="1" colspan="1">Sum</td><td align="center" valign="middle" rowspan="1" colspan="1">22</td><td align="center" valign="middle" rowspan="1" colspan="1">50</td><td align="center" valign="middle" rowspan="1" colspan="1">53</td><td align="center" valign="middle" rowspan="1" colspan="1">57</td><td align="center" valign="middle" rowspan="1" colspan="1">51</td><td align="center" valign="middle" rowspan="1" colspan="1">56</td><td align="center" valign="middle" rowspan="1" colspan="1">20</td><td align="center" valign="middle" rowspan="1" colspan="1">26</td><td align="center" valign="middle" rowspan="1" colspan="1">74</td><td align="center" valign="middle" rowspan="1" colspan="1"><bold>409</bold></td><td align="center" valign="middle" rowspan="1" colspan="1"></td></tr><tr><td colspan="12" align="center" valign="middle" rowspan="1"><hr></hr></td><td align="center" valign="middle" rowspan="1" colspan="1"></td></tr><tr><td align="center" valign="middle" colspan="2" rowspan="1">Producer's Accuracy (%)</td><td align="center" valign="middle" rowspan="1" colspan="1">84.8</td><td align="center" valign="middle" rowspan="1" colspan="1">55.5</td><td align="center" valign="middle" rowspan="1" colspan="1">84.5</td><td align="center" valign="middle" rowspan="1" colspan="1">70.8</td><td align="center" valign="middle" rowspan="1" colspan="1">72.0</td><td align="center" valign="middle" rowspan="1" colspan="1">78.9</td><td align="center" valign="middle" rowspan="1" colspan="1">88.5</td><td align="center" valign="middle" rowspan="1" colspan="1">81.4</td><td align="center" valign="middle" rowspan="1" colspan="1">84.7</td><td align="center" valign="middle" colspan="2" rowspan="1">Overall accuracy: 76.8%Kappa value: 0.73</td></tr></tbody></table></table-wrap><table-wrap id="t5-sensors-11-01943" position="float"><label>Table 5.</label><caption><p>Accuracy assessments by classification methods.</p></caption><table frame="hsides" rules="groups"><thead><tr><th align="center" valign="bottom" rowspan="1" colspan="1"></th><th align="center" valign="bottom" rowspan="1" colspan="1"><bold>Overall accuracy (%)</bold></th><th align="center" valign="bottom" rowspan="1" colspan="1"><bold>Kappa value</bold></th></tr></thead><tbody><tr><td align="left" valign="middle" rowspan="1" colspan="1">Pixel-based classification</td><td align="center" valign="middle" rowspan="1" colspan="1">63.9</td><td align="center" valign="middle" rowspan="1" colspan="1">0.59</td></tr><tr><td align="left" valign="middle" rowspan="1" colspan="1">Majority rule-based 3 × 3 filter</td><td align="center" valign="middle" rowspan="1" colspan="1">67.6</td><td align="center" valign="middle" rowspan="1" colspan="1">0.63</td></tr><tr><td align="left" valign="middle" rowspan="1" colspan="1">Object-based classification</td><td align="center" valign="middle" rowspan="1" colspan="1">53.1</td><td align="center" valign="middle" rowspan="1" colspan="1">0.42</td></tr><tr><td align="left" valign="middle" rowspan="1" colspan="1">Segment-based classification</td><td align="center" valign="middle" rowspan="1" colspan="1">76.8</td><td align="center" valign="middle" rowspan="1" colspan="1">0.73</td></tr></tbody></table></table-wrap></floats-group></pmc></record>Pour manipuler ce document sous Unix (Dilib)
EXPLOR_STEP=$WICRI_ROOT/Wicri/Bois/explor/CheneBelgiqueV1/Data/Pmc/Corpus
HfdSelect -h $EXPLOR_STEP/biblio.hfd -nk 000245 | SxmlIndent | more
Ou
HfdSelect -h $EXPLOR_AREA/Data/Pmc/Corpus/biblio.hfd -nk 000245 | SxmlIndent | more
Pour mettre un lien sur cette page dans le réseau Wicri
{{Explor lien
|wiki= Wicri/Bois
|area= CheneBelgiqueV1
|flux= Pmc
|étape= Corpus
|type= RBID
|clé= PMC:3274007
|texte= Forest Cover Classification by Optimal Segmentation of High Resolution Satellite Imagery
}}
Pour générer des pages wiki
HfdIndexSelect -h $EXPLOR_AREA/Data/Pmc/Corpus/RBID.i -Sk "pubmed:22319391" \
| HfdSelect -Kh $EXPLOR_AREA/Data/Pmc/Corpus/biblio.hfd \
| NlmPubMed2Wicri -a CheneBelgiqueV1
| This area was generated with Dilib version V0.6.27. |  |