利用ecognition构建决策树

《利用ecognition构建决策树》由会员上传分享，免费在线阅读，更多相关内容在学术论文-天天文库。

1、NameVAylomalic1(shape;0.1compel:0.9]creating'NewLevef1.多尺度分割。注意：不要选择“利用矢S文件分割”，分割的尺度尽量小，参考参数:1,0.1,0.9。EditProcessAlgorithmDescnpbonApp^anoptimizationprocedurev4ichlocalsminimizestheaverageheterogeneityofmageobjectsforagivenrewbbon.Algonthmpa;arne

2、ThresholdconditionImageObiectDomaort•1rcy19o.ParameterValueOverwriteexistinglevelTes日LerelSettinesUvtlMeatMtvLevelBS會亀助Setlines53Layer1?1,1

3、&ThMiatieLayerus<(«MoSCM*osilionofhShtptCoapactMiiLoop“Cycte$7LoopwhilesomethingchangesonlyNumberofcycles12.利用矢S文件创建炎别：assignclassbythematiclayer,EditProcessN

4、ameVAvMoma(icatNewLevelat^ignclanbythematiclayerusr>gcla^s^n.Aigofthm▼AlaorthmDwcnptionmigndmttv(hemtficl«ti(obitoltiMlbtctettiedIo.dmimAthemaficaitrbutewiththehgbestcovecagt)AlaorihmpafamclennameofiheImageOb)ectDomamimageobjectlevelParameterLtvtlCitesfilttrTKrtiKoldcondition■•pMaxaunb^rofParamete

5、rTlicAalicl

6、训练哪些类也要训上!EdHprocwI鏟週”ggg[JarreJA^omalcdoAlger4hmclauihermoeob,ec«tevd1ParameterValueLtv«lClftsifillerIVusldiaccorr^fVrThresholdeoaditioa—HipFro>f«rbe3OperationC«7)ficir«tionV“only

7、日FeatureTypeobjjcljbaie^Feature【Layer1,RetJiLa^tr2,...13ClassifierTyp*DaptKobHansimplecount0Crassfolds3AlgcmlhmDetaptionTtan.,