期刊名称：ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences
印刷版ISSN：2194-9042
电子版ISSN：2194-9050
出版年度：2010
卷号：XXXVIII - Part 8
页码：769-774
出版社：Copernicus Publications
摘要：In this paper, we propose a novel method of change detection for high-resolution remote sensing imagery. The method has three primary processes: image segmentation, image difference and change detection. The pre-processed multi-temporal images are segmented into objects by a multi-resolution segmentation algorithm. Then a difference image is generated according to the pair of segmented maps with an interval. Each object is represented by the mean value of some spectral absolute-valued differences, which are calculated for each pixel in it. Finally, by defining the iterative degree, a genetic algorithm (GA) was employed to search the optimal binary change detection map. In the searching procedure, the crossover and mutation was applied to produce new individuals. According to our experiments using the QuickBird imagery, the Overall Errors of the proposed method decreased by more than 2500 pixels compared with the pixel-based change detection using a GA. Meanwhile, it decreased by more than 1000 pixels compared with the object-oriented change vector analysis (CVA).
