An SAR image segmentation method based on level set evolution without employing any prior information was proposed. The method was a statistical geometric active contour model in which region information was used. The step function was utilized to estimate the probability distribution function (PDF), so it was avoid to suppose a probability distribution model of images in advance, which required additional prior information. Further, a penalty term was introduced into the energy functional minimized by the level set evolution, then the costly re-initialization of level set function, which was also difficult to be implemented, was removed effectively. In addition, an iterated numerical scheme and the parameters setting were suggested, as well as the condition of terminating iteration was improved. Experiments demonstrate correct segmentation with proposed method and suggested parameters. For a few images whose segmentation is not well, correct segmentation can be achieved only by tuning one parameter simply.
Martin P,Réfrégier P,Goudail F,et al.Influence of the noise model on level set active contour segmentation[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2004,��26(6):��799-803
Ayed I,Vdiquez C,Mitiche A,et al.SAR image segmentation with active contours and level sets //Proceedings of 2004 IEEE International Conference on Image Processing (ICIP-04).Singapore:IEEE Computer Society,2004:2717-2720
Martin P,Réfrégier P,Galland F,et al.Nonparametric statistical snake based on the minimum stochastic complexity[J].IEEE Transactions on Image Processing.2006,15(9):2762-2770
Li Chunming,Xu Chenyang,Gui Changfeng,et al.Level set evolution without re-initialization:A new variational formation // Proceedings of the 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR-05).San Diego,California:IEEE Computer Society,2005:430-436
Chan T,Vese L.Active contours without edges[J].IEEE Transactions on Image Processing.2001,10(2):266-277