作者: Tony F. Chan , Selim Esedoglu , Mila Nikolova
DOI: 10.1137/040615286
关键词: Segmentation-based object categorization 、 Image processing 、 Segmentation 、 Computer vision 、 Scale-space segmentation 、 Mathematical optimization 、 Convex optimization 、 Image segmentation 、 Computer science 、 Artificial intelligence 、 Bregman method 、 Optimization problem
摘要: … makes the initial guess for gradient descent–based algorithms sometimes critically important for obtaining satisfactory results. In this paper we propose algorithms which are guaranteed to find global minimizers of certain denoising and segmentation models that are known to have local minima. As a common feature, the models we consider involve minimizing functionals over characteristic functions of sets, which is a nonconvex collection; this feature is … In section 4, we will obtain the analogue of Proposition 1 for the piecewise constant Mumford–Shah …