Research On Image Denoising Method Based On Edge-Preserving Sparse Representation

Digital image process plays an important role in public security. Images may be infected by noise in the monitoring. In order to get a better monitoring image, denoising becomes a problem which the subsequent processing must face. Its main purpose is to improve the image quality by reducing image noise.Sparse representation is an important research subject in the solution of the denoising. The main theory based on is that noise of the image and original image information belong to two different structures. Original image information can be decomposed by the Sparse Coding, while the other can not, therefore, noise can be distinguished.The defect of denoising on sparse representation is to preserve the edge structure. This paper puts forward the optimization model of sparse coding which is based on sparse coding to solve the edge preserving problems in current image denoising field and improves it on the current methods. The main contents and innovative points of this paper are as follows:1) This paper proposed a LOBS edge detection operator which can solve the problems of the singularity of the edge detection operator to some extent and the absence of the robustness to noise. Related edge detection operator is usually based on the first-order derivative or second-order derivative of the gray value, of which only LOG operator and Canny operator that belong to second-order derivative are robust to noise. This paper researched the selection problem of Laplacian filter and proposed an improvement plan. Operator which is presented in this paper simulated under the environment of Matlab and compared to the LOG operator and Canny operator. Experiments had demonstrated that the proposed edge detection operator had better robust and lost less edge compared with the LOG operator and the Canny operator.2) This paper analysised the sparse representation and focused on the OMP algorithm which is used in denoising model based on sparse representation. In accordance with lost and clearness of edge, this paper improved OMP by presenting an novel operator named edge preserving operator and designed a new denoising method. Then we solved the model which joined the edge preserving operator. The comparison and analysis of the performance difference which was between the traditional method and the joined edge preserving operator has been done through the experiments.