COMBINING LOCAL FEATURES AND PROGRESSIVE SUPPORT VECTOR MACHINE FOR URBAN CHANGE DETECTION OF VHR IMAGES

Abstract

The difficulties about change detection of VHR images are analyzed from different perspectives. Motivated by perception and cognition mechanism of human vision, visual change detection principles are discussed, and a unified change detection framework is proposed. To address the difficulties in change detection of VHR images, a novel approach is presented within the framework, which exploits the combination of local features and change vector displacement field to represent the complex changes of VHR images and utilizes transductive SVM (Support Vector Machine) to classify change features progressively. Experiments demonstrate the effectiveness of the proposed approach.