Parameter Research Of Image Denoising Model Based On Variational Principle

Image denoising has been a hot issue in image processing. Nowadays many denoising theories have been widely used in the image processing field.This thesis introduces the research background, the significance and the present situation of variation denoising, and gives the theoretical basis and the derivation, which the energy function minimization problem transforms the corresponding partial differential equation. And then we have a detailed analysis of variational denoising model, introduces several methods about regular parameter's choice.Rudin had choosed a regular parameter which could satisfy two constraints based on the ROF model. On this basis this thesis proposes a new initial value to solve this model, and has determined a new method of define parameter in a total variational denoising model. Ones do select not the noise image but the reconstructed image as the initial value, under the wavelet decomposition of the noise image, keeping the low frequency coefficient and only setting up a threshold for high frequency coefficient.Finally, this thesis takes the example of the total variation denoising model to solve the corresponding Euler equation by the gradient method, and realize the mean and the variance estimate formula simultaneously by adopting this method. With the newly-selected initial value and different test images, this thesis solves PDE by the finite difference iteration and stimulates the equation numerically. The experiment results show that this method has a better denoising effect.