New method for edge detection of infrared images based on Mumford-Shah model

Chan-Vese(C-V) model is an excellent model of simplified Mumford-Shah models. As Chan-Vese model is based on two homogenous regions, it causes failure in edge detection of infrared images. A level set method for edge detection of infrared image based on improved Mumford-Shah model is proposed. The schemes of preserving object edges, establishing the stopping criterion and speeding up the algorithm are discussed. The experiments show that this model can overcome the failure of C-V model and detect edge of infrared images efficiently.