Studies On Image Denoising

Image denoising as an important pre-processing step is always one of the research focuses in image processing and keeps continuous development, which is mainly powered by deeper comprehension of the image processing and more successful introductions of new mathematical methods. So this dissertation aims at systematical study on the theory and the algorithm of image denoising, which emphasizes thorough research of some key techniques and profound exploration of some problems. The main contributions of this dissertation can be outlined as follows.At first, combining multiresolution model with total least square theory, a new method is proposed for removing noise from images with mixed noise, which allows us to consider the uncertainties in the measured data. Firstly structure detectors are used to divide an image into different regions, and then different denoising strategies are adapted to these regions respectively. This new method can shorten the runtime as well as improve efficiency of denoised arithmetic, while ensure the denoising quality, especially in the point, line, and edge regions.And then addressing SAR image speckle denoising, this dissertation proposed a new method based on bivariate shrinkage function combined with enhancement of wavelet significant coefficients. In our paper we make the speckle noise model suit the bivariate shrinkage function, and the joint probability density functions (PDF) and noisePDF could be united by MAP to de-noise image, then the wavelet coefficients are enhanced according to a rule whether the coefficient is a significant one or not.The simulation demonstrates that the new algorithms have a better denoised effect comparing with other traditional denoising methods.