文档介绍：第 卷 第 期 测绘与空间地理信息 ,
1. Jiangsu Province Surveying and Mapping Engineering Institute Nanjing 210013 China
( , , ;
2. School of Geomatics Engineering Nanjing Tech University Nanjing 211816 China
, , , )
Abstract
:In o rder to give full pla y to the advanta ges of multiple fe atures i n remo te sensing images and th e differen ces of fe ature s a-
mong diffe rent image obje cts, an object- oriented c hang e detec tion method b ased on feat ure diffe rences is propo sed by u sin g the r an-
dom forest algorith m with high p redic tion accuracy a nd stable per formanc e. Firstly, th e cha ng e vector ana lysis is use d to de tect the
pixe l level c hange of t he imag e, and the mu lti-s cale segm entation r esults a re obta ined. The n i t extract s the s pect rum, te xtu re features
and feature d iff erenc e of each object o n the fr ont an d b ack temp oral i mages as t he input data of r and o