Based On The Multi-scale Wavelet Domain Correlation Image Denoising And Enhancement Methods Of Research,

Image denoising and enhancement are the basis of the image compression, image segmentation and image recognition. Additive noise and distortion often appear in the process of image obtaining and transmission, so image denoising and enhancement are necessary to improve the visual effect or make it easy for people and machine to analyse and understand the image. In this thesis, an image denoising and enhancement method based on the theory of multi- scale wavelet transform correlationship is emphatically researched.This thesis focuses on establishing the wavelet coefficient Gauss probabilistic model for edge regions and homogeneous regions respectively according to the wavelet coefficient distribution characteristics, based on discrete wavelet decomposition to the picture. And then, we obtain the initialization shrinking factor by Bayes formula. The function of shrinking factor is to reduce the wavelet factor which is related to the noise, and to enhance and preserve the wavelet factor related to the edge. On the base of this, we revise the shrinking factor under the principle of keeping the exchange power unchanged. Considering scale consistency and Geometric consistency, we improve the shrinking function forward and the better denoising effect is obtained. At last, we study an enhancement algorithm that has a good edge enhancement characteristic, and achieve the edge enhancement based on the above shrinking factors.