SAR imagery segmentation based on integrated active contour

Image segmentation is an important problem on image processing technique. Based on active contour model, a novel integrated active contour (IAC) model for SAR imagery segmentation has been developed. In this model, the edge information is extracted by edge detection operators based on ratio of average framework, and the region statistical information is distilled by the maximization of likelihood function of different regions. And an unconditionally stable numerical scheme is used by means of additional operator splitting arithmetic. In the end, some segmentation tests have been done using MSTAR, Radarsat-2 and domestic spaceborne SAR images. Results show: proposed model has a good adaptability to complicated SAR images segmentation, and gives an accurate and fast partition for different regions in the images; implementing method is robust which is insensitive to parameters setting and initializations.