A Research Of Denoising Method Based On Contourlet Transform

Image denoising in image processing is a classic problem, the best denoising result is that the edge of the image and detail information can be well retained and getting a new image that is close to the natural image. Because of the wavelet transform in denoising the limitations, there’s super-improved wavelet transform-Contourlet transform. Compared with the wavelet transform, It can descript the details of the image and the edge more detailly.The nature of Contourlet transform is that sparsely representing the image through the singular line and the plane. It is called super-wavelet transform, because it is inherited from a fundamental analysis of wavelet multi-resolution time-frequency characteristics, but also be better in anisotropy. In image processing, in order to analyze and identify the image better, usually preprocess the image to reduce noise impact for the after work of image processing, and bring greater convenience for the work. Contourlet Laplacian pyramid transformation, including decomposition and directional filter; Laplacian pyramid decomposes the image into low-pass image and the difference image. The direction of the image filter to decompose the high frequency sub-band multi-direction image group. Laplacian pyramid decomposition is directional, directional filter can be a very good high-frequency part of the decomposition, the low frequency part is not, so combine the two to use to highlight their respective advantages, make each other’s shortcomings.This aticle does a systematic study and analysis in the Contourlet decomposition and reconstruction, but also analyzes the Contourlet transform and wavelet transform in denoising’s strengths and weaknesses, describes Contourlet thresholding and cyclic shift denoising, while panning in the loop thresholding denoising and scale based on the scale proposed Contourlet denoising improved, and the basis of the algorithm are discussed.