关键词: 矢量图, 变化检测, 像斑, G统计量, 类别异质度, 最大熵

Abstract: In order to solve the land use change detection problem in vector map and remote sensing imagery, this paper puts forward a change detection method of vector map and remote sensing imagery based on category. The remote sensing image segmentation is carried out to obtain image patches under the constraint of vector map. Then the histogram features of remote sensing imagery patches are extracted and the feature distance between image patches is measured by using G statistic. Thus the class heterogeneity of image patches on single wave is constructed by using the feature distance between image patches and other similar image patches, adaptive weighting combines the class heterogeneity of image patches on each band to establish the class heterogeneity of image patches. Then according to the max entropy method, the threshold of the corresponding heterogeneity of each object category is obtained and the change verification is applied based on category classification to get the change information. The proposed method's effectiveness is verified by the experimental results on the QuickBird images and it realizes the automatic change detection of vector map and remote sensing imagery.

Key words: vector map, change detection, image patches, G statistic, class heterogeneity, max entropy