Research On Image Denoising Based On Sparse Representation

Image denoising is an important image processing technique, which can improve the visual effects of images and prepare for the post-processing of images. Image denoising can reduce the influence of noise in the images, thus it is an indispensable part of image processing. Recently sparse representation has attracted researchers’ attention, therefore image denoising based on sparse representation becomes one of the frontier issues in signal processing. Sparse representation has been shown to be more effective than traditional denoising methods which significantly enhance the performance of denoising. In the field of image denoising applications, researchers have made some progress.This paper gives a detailed description of the traditional image denoising methods and sparse representation firstly, and then a new denoising model based on sparse renpresentation is presented. At the same time, several new denoising algorithms are proposed which achieves very good results in experiments. The main contributions of this paper are as follows:(1)This paper analyzes and compares the advantages and disadvantages of several classical traditional denoising algorithms which are introduced. Based on the denoising methods by soft thresholding and hard thresholding which put forward by D.L.Dohono, two new thresholding functions are proposed that give better images’denoising effect in simulation experiments than the other functions do.(2)Then we make a research of the sparse representation which played a fundamental role in image denoising areas in the past decades. We deeply analyze some classical denoising algorithms based on sparse representation. Specifically the complexity and the detail of dictionary updating and signal reconstruction are focused. Furthermore, we describe the dictionary updating algorithm for optimization named K-SVD.(3)At the same time, this paper presents a method that derives discrete tight frame systems from the input image to provide a better sparse approximation to the image. The algorithm is proposed based on discrete tight frames. Detailed description of the algorithm is given to introduce the denoising principle and specific denoising process.(4)At last, new denoising algorithms which proposed in this paper are used to do denoising simulation experiments on the specified images, which achieve very good results in experiments. The experimental results demonstrate that the proposed algorithm based on tight frame performs better in image denoising than the K-SVD algorithm and other traditional algorithms do.