Fuzzy segmentation of urban areas in panchromatic remote sensing images
全色遥感图像中城区的快速模糊分割算法

A fast fuzzy algorithm for the segmentation of urban areas especially in panchromatic remote sensing images was proposed. The algorithm was supposed to be an image previous process for the small targets recognition in immense remote sensing images. Based on the fuzzy set theory, a membership function was introduced according to Bayesian criteria. The process of segmentation was achieved gradually by using appropriate features of urban areas. The experiment was implemented on SPOT-5 PAN images and a reliability analysis of the fuzzy membership function was made after the segmentation. Compared with the region growing segmentation method and the algorithm based on multi-scale wavelet geometry information, the experimental results show that this algorithm has good performance with low complexity and high accuracy, and it is an efficient preprocessing technique for some important remote sensing applications.