Method of calculation of a multimode descriptor for the analysis of a fixed or moving pictures

摘要

The method involves segmenting a fixed image or a video image, and determining an invariant region of a segmented region. The isotropic normalization is performed on the region. Isotropic area is extracted from the normalized region, where the isotropic area includes central area obtained by segmentation, an environmental area defined according to the scale of the central area, and the circumference. One or a set of spatial and temporal descriptors for each of the central area, the environmental area and the circumference, is/are calculated to separately describe each area.