The Research Of Wavelet Transform On Image Denoising

The noise exists inevitably in digital images. Searching a method of denoising effectively and keeping the edge information simultaneously is a goal people pursued all the time. Traditional methods are hard to give attention to two. But wavelet has good localizing quality at the time domain and the frequency domain simultaneously and the characteristic of multiresolution analysis, it can fulfill different kinds of filtering needs such as low-pass, high-pass, sink wave, random noise denoising. Compared with traditional denoising methods, wavelet has incomparable advantage. It has become effective analysis tool and is known as math microscope of signal analysis.The research content of this dissertation consists of three aspects:This paper summarized different kinds of image denoising methods, contrasted their advantages and disadvantages, educed the wavelet transform, and revising the wavelet transform theory, wavelet denoising developing line and the methods of denoising with wavelet transform.The performance of image denoising algorithms using wavelet shrinkage can be improved significantly by taking into account the statistical dependencies among wavelet coefficients. This paper established the wavelet coefficients' model, analyzed their dependencies and proposed three new algorithms: the adaptive thresholding based on local characteristic, the adaptive thresholding based on neighborhood characteristic, and the multi-scale, multi-direction denoising algorithm. The experiment results show the new algorithm is more effective than classical algorithms.Because of good localizing quality at the time domain and the frequency domain simultaneously and the characteristic of multiresolution analysis, wavelet transform applied universally in the image denoising, but due to the less of characteristic of anisotropy, it is not sparse when it comes to the representation of singularity of line. A new image representation method, Contourlet transform, is proposed. The Contourlet transform possess not only spatial and frequency locality and multi-resolution but also directionality and anisotropy. This paper we introduced contourlet simply and applied it to image denoising. The results show the denoising method based on Contourlet transform is better than wavelet transform, especially to the images which consist of abundant edge.