Bayesian Estimation Based Mumford-Shah Regularization for Blur Identification and Segmentation in Video Sequences

We present an extended Mumford-Shah (MS) regularization for blind image deconvolution and segmentation in the context of Bayesian estimation. The extended MS functional is added to have costs for the identification of blur via a newly introduced prior solution space. The functional is minimized using Gamma-convergence approximation by projecting iterations onto a newly designed embedded alternating minimization within Neumann conditions. Image segmentation is closely related to accurate blur identification and restoration, that is, the problem of estimating an image based on its degraded observation. Experiments show that the proposed algorithm is efficient and robust in that it can handle images that are formed in different environments with different types and amounts of blur and noise