Reaserch Of Image Denoising Based-On The Multiwavelet

In the real world, image needs to be denoised because it is always disturbed by varies of terms. So, image denoising is an important aspect in the image processing field.This paper mainly propose the novel image denoising algorithm based on high-order statistics, the ramp threshold denoising based on multiwavelet, the adaptive weighted threshold denoising algorithm on multiwavelet.The novel image denoising algorithm based on high-order statistics, according to the charactics that the higher-order cumulants of Gaussian processes is zero, designs the window-diagonal smoothing method, and succeeds in raising the SNR of image.In addition, in the recent years, the multiwavelet image denoising algorithm developing on the basis of wavelet is still a matter of concern, which is of great significance in both theory and practice.This thesis is based on the focus of multiwavelet image denoising algorithm. In this part, the paper first introduced the traditional alogorithm in multiwavelet denoising, and then performes the comparison and analysis of the results. Furture more, pointing to the obviously Gibbs effect after processing the denoised images, this paper propose a ramp threshold function to improve the original algorithm.To better applicate multiwavelet in image denoising filed, paper analyzes the multiwavelet coefficients'dependency between intra-scale and inter-scale, to get an appropriate threshold weights. we obtains a complete adaptive image denoising algorithm.This paper simulates the above denoising algorithm with Matlab. The results show that the image denoising algorithm based on higher-order statistics can get very good subjective and objective image quality with the disadvantage of much calculation. Compared with other algorithms, the ramp threshold denoising based on multiwavelet and the adaptive weighted threshold denoising algorithm on multiwavelet, can not only denoise the image effectively, but also retain most of the details of the image information.