Image De-noising Based On Wavelet Shrinkage

The image is usually corrupted by noise in its collection, acquisition or transmission. All kinds of noises influence the necessary information extracted form the image greatly. At present, wavelet analysis is very popular in the world. Meanwhile it is the front question and study hotspot of signal processing. So the image de-noising based on wavelet theory is getting more attention of experts and scholars, and has a very good effect.This dissertation mainly studies the wavelet theory and its appliance in image de-noising, and analyzes the choice of wavelet shrinkage threshold. At the beginning, the dissertation introduces the meaning and the purpose of image de-noising, the history and prospect of the wavelet de-noising. Then it presents the basic knowledge of the image de-noising and the evaluation criterion of images. After the introduction of methods of the traditional image de-noising, the dissertation states the theory of wavelet image de-noising and introduces three wavelet de-noising methods--- wavelet shrinkage method,modulus maxima method,correlation method. The dissertation compares their amount of calculation, the stability, the effects of de-noising, the scopes of application and so on by the means of quality. Besides, the selecting process of wavelet base is analyzed by examples. Finally, on the study of the threshold function and its selecting methods, the dissertation suggests a wavelet shrinkage threshold selecting method on image characteristic. This method uses the root-mean-square of the wavelet modulus to determine the variable coefficient of the adjustment formula threshold with which a sizing model of variable coefficient will be determined to get a new formula of the wavelet shrinkage. In the experiment, the dissertation uses the Donoho threshold and the shrinkage threshold derived from the threshold formula calculated in the dissertation to de-noise the white Gaussian noise and pepper-and-salt noise of some images. The experiment proves that this method has a better effect on de-noising.