基于小波变换的图像去噪方法的研究与实现-应用数学专业论文.docxVIP

华南理工大学工学硕士学位论文Abstract 华南理工大学工学硕士学位论文 Abstract Noise is the important factor of image quality．It will result in being unable to recognize some character details of image and decreasing of PSNR．How to reduce noise effectively and extract image information becomes very important．More and more research and application focus on methods for image denosing based wavelet transform．Wavelel shrinkage is method that creating certain regulation according to the different representation of signal and noise in wavelet domain and processing the wavelet coefficients．The essential lies in shrinking deleting the coefficients raised from noise and reserve those raised from signal． For any given signal，the optimal linear filter is Wiener filter，which use self—related function of signal to give optimal estimation in minimax mean—square error(MSE)sense．In this paper，we develop new method for wavelet denoising that blends wavelet shrinkage and Wiener filtering．Ima- ge with noise is proces sed by Wiener filtering in wavelet domain． How to get the estimation of signal and construct the coefficient of Wiener the key of achieving Wiener filtering in wavelet domain．We design the Wiener filter with two wavelet bases．At first，estimating signal in the fi rst wavelet domain．Then，transforming the estimate value in the second wavelet domain and COnstruct the toefficient of Wienef fi Jter．At last，reducing noise by Wiener filte ring in the second wavelet domain． In order to get reasonable estimation of signal，we use new thTesholding function as the filter in the first Wavelet domain，Which significantly reduces noise and well preserves edges at the same time， By designing VagueDenoise filter in AviSynth，we have the experiment of sialic image denoising and animat d lmage denoising with executing the AviS ynth script．The result indicate that the developed method gives better Visun J qaality and PSNR Pe rformancc than others． Key Words： image denosing；wavelet transform；wavelet shrinkage Wiener filterin