A New Total Variationl Model Based On I-divergence Image Restoration

For optical coherence tomography(OCT), ultrasound, synthetic-aperture radar, and other coherent ranging methods, speckle can cause images contamination that detracts from the utility of the image. In order to improve the quality of images for further application, it is necessary to suppress the speckle noise in the images. Inspired by the classical I-divergence model, this paper focuses on the research of image restoration methods for removing the speckle noise. The main work of this paper is to improve the recognition degree of edge information, while eliminating the speckle noise and blurring of the image.The relevant principles of image restoration technology and noise category are present first. And then we analyze two classical models for restoring images with multiplicative noise: RLO model and AA model. Since the total variation-based methods can remove noise while preserving edge features,many of TV regularization models had been successfully used in the noise image restoration. By the regularization method of minimizing Csiszar's I-divergence fidelity term and a new quadratic penalty term based on the statistical characteristics of the noise, we propose a new model to restore the blurred images with speckle noise.We discuss several important properties of the new model. And under the given convergence condition, we develop a numerical primal-dual algorithm to solve the new model. Compared with the efficient method, the DZ model(Y. Dong, et.al., SIAM J. Imaging Sci, 2013, 6:.. 1598-1625), the experimental results show the new method performances better in image restoration quality and CPU time.