Font Size: a A A

Research On Image Denoising Algorithms Based On Contourlet Transform

Posted on:2009-08-12Degree:MasterType:Thesis
Country:ChinaCandidate:R WangFull Text:PDF
GTID:2298360245489018Subject:Communication and Information System
Abstract/Summary:PDF Full Text Request
A number of facts have approved that wavelet mainly applies to isotropic singularity objects. But for the anisotropic singularity ones, such as the edge of image and the wirelike character, wavelet is not the best tool. It is the reason why the blur of image edges and details occur during image denoising.But edge and details are the most information of an image, A multi-resolution, multi-directional transform -- Contourlet transform is focused on, and its application to image denoising is studied in this paper.At first Contourlet transform theory is studied , analysis pointed out during the transform the frequencies suffers from aliasing, thus weakening their directional selectivity, and two improved algorithms are put forward. An algorithm based on steerable pyramid and the other based on promoted Laplacian pyramid are proposed. Steerable pyramid Contourlet of better directional selectivity is steerable pyramid combined with direction filter bank, so that aliasing is avoided.Promoted Laplacian pyramid is laplacian pyramid which has been reset combined with direction filter bank,so that it can suppress the artifacts around the edge effectively.On this basis, image denoising algorithm based on steerable pyramid Contourlet and denoising algorithm based on improved laplacian pyramid are proposed. We can decompose an image by these two methods and then denoises with adaptive thresholding. Experiments show that the two proposed method outperforms the wavelet transform and the contourlet transform in image denoising.
Keywords/Search Tags:Image denoising, Multiscale Geometric Analysis, Contourlet transform, Steerable pyramid, Laplacian pyramid
PDF Full Text Request
Related items