<?xml version="1.0" encoding="ISO-8859-1"?>
<metadatalist>
	<metadata ReferenceType="Conference Proceedings">
		<site>marte2.sid.inpe.br 802</site>
		<holdercode>{isadg {BR SPINPE} ibi 8JMKD3MGPCW/3DT298S}</holdercode>
		<identifier>8JMKD3MGP6W34M/3PSMDFK</identifier>
		<repository>sid.inpe.br/marte2/2017/10.27.15.54.44</repository>
		<lastupdate>2017:10.27.15.54.44 dpi.inpe.br/marte2/2013/05.17.15.03.06 banon</lastupdate>
		<metadatarepository>sid.inpe.br/marte2/2017/10.27.15.54.45</metadatarepository>
		<metadatalastupdate>2018:06.06.03.11.45 dpi.inpe.br/marte2/2013/05.17.15.03.06 administrator {D 2017}</metadatalastupdate>
		<isbn>978-85-17-00088-1</isbn>
		<label>59355</label>
		<citationkey>ZorteaSanZadSchSte:2017:DeInEu</citationkey>
		<title>Detecting individual eucalyptus crowns in aerial photographs using template matching and classification</title>
		<format>Internet</format>
		<year>2017</year>
		<secondarytype>PRE CN</secondarytype>
		<numberoffiles>1</numberoffiles>
		<size>3980 KiB</size>
		<author>Zortea, Maciel,</author>
		<author>Santos, Marcelo Nery dos,</author>
		<author>Zadrozny, Bianca,</author>
		<author>Schoeninger, Emerson Roberto,</author>
		<author>Stetz, Cristiano Cardoso,</author>
		<electronicmailaddress>mazortea@br.ibm.com</electronicmailaddress>
		<editor>Gherardi, Douglas Francisco Marcolino,</editor>
		<editor>Aragão, Luiz Eduardo Oliveira e Cruz de,</editor>
		<e-mailaddress>daniela.seki@inpe.br</e-mailaddress>
		<conferencename>Simpósio Brasileiro de Sensoriamento Remoto, 18 (SBSR)</conferencename>
		<conferencelocation>Santos</conferencelocation>
		<date>28-31 maio 2017</date>
		<publisher>Instituto Nacional de Pesquisas Espaciais (INPE)</publisher>
		<publisheraddress>São José dos Campos</publisheraddress>
		<pages>6749-6756</pages>
		<booktitle>Anais</booktitle>
		<organization>Instituto Nacional de Pesquisas Espaciais (INPE)</organization>
		<transferableflag>1</transferableflag>
		<abstract>Collecting and analyzing forest information is critical for sustainable forest management. Here we propose a method to automatically detect the location and diameter of tree crowns at early growth stages in regularly planted forests using very high spatial resolution RGB imagery acquired by unmanned aerial vehicles (UAVs). A list of candidate detections is generated matching the multiscale convolutions of  synthetic crown templates to image objects. Strong matches are filtered using color-based rules. Local attributes describing the color and spatial information in small image patches centered in each retained detection are passed to an off-line trained Random Forest classifier that assigns a level of confidence to each tree crown detection. The method is tested on orthorectified RGB mosaics  with a pixel spacing of about 11 cm using circular templates with diameters in the range 50-200 cm. Experiments at two study sites containing about 120-day-old plantations of eucalyptus, located in Southern Brazil, suggest detections accuracies above 90% when non-overlapping adjacent crowns have a diameter larger than 6 pixels and are surrounded by mixed backgrounds such as exposed soil and  debris from the previous harvest. The automated counts of trees in 12 footprints of 1,257m2 were within 11% of the visual estimate, and within 4% when averaged for the study. Examples of challenging scenarios requiring further methodological developments are presented.  We anticipate that automated tree crown detection using the proposed prototype algorithm may complement traditional field-based tree inventory.</abstract>
		<area>SRE</area>
		<type>Processamento de imagens</type>
		<language>en</language>
		<targetfile>59355.pdf</targetfile>
		<usergroup>banon</usergroup>
		<visibility>shown</visibility>
		<mirrorrepository>urlib.net/www/2011/03.29.20.55</mirrorrepository>
		<nexthigherunit>8JMKD3MGP6W34M/3PMFNUS</nexthigherunit>
		<citingitemlist>sid.inpe.br/marte2/2017/09.25.14.55 2</citingitemlist>
		<hostcollection>dpi.inpe.br/marte2/2013/05.17.15.03.06</hostcollection>
		<username>banon</username>
		<lasthostcollection>dpi.inpe.br/marte2/2013/05.17.15.03.06</lasthostcollection>
		<url>http://marte2.sid.inpe.br/rep-/sid.inpe.br/marte2/2017/10.27.15.54.44</url>
	</metadata>
</metadatalist>