Study On Image Denoising Algorithm Based On Texture Details

In image processing process, during collection, transform, and transmission,there is always much noise and jam taken in by imaging equipment andcircumstance disturbing. Image noise will bring much effect to continuousprocessing to digital image. Thus it is important to do the image denoisingbefore.Although the common denoising method can wipe off noise in gently grayscaled images, they can not give ideal result among texture image denoising.Thus, to resolve the question of keeping details and smoothing at one time, thisthesis does the algorithmic research in transform domain and spatial domainseparately. Then, according to the situation that there are too many algorithmsaiming at image denoising, this thesis studys the estimate method to theperformance of actral denoising algorithm.Once studying the image denoising method in transform domain, it presentsthe texture image denoising method based on wavelet packet transform. Thismethod can wipe off most noise of texture images quickly with well-kept textureinformation.Once studying the image denoising method in spatial domain, it presentsthe spatial domain mask denoising method based on texture analysis. Thismethod preserves image details and denoises well.In the last part of this thesis, it makes the algorithm estimation between thetexture image denoising method based on wavelet packet transform, the spatialdomain mask denoising method based on texture analysis and the WT algorithmand the SC algorithm. Through this calculation, we can see that the twoalgorithms from this thesis have so much superiority, and their PSNRs are muchhigher than algorithms before. When we compare these two methods, we cansee obviously that there is superiority of the spatial domain mask denoisingmethod based on texture analysis, it will must be used well later.