Research On Denoisingalgorithm For MR Image Based On Contourlet Transform

Contourlet transformation is a new Multiresolution Geometry Analysis method which can represent images most effectively. It becomes a hot topic of researches by the scholars allover the world. This transformation decomposes images into bandpass directional subband using Pyramidal Directional Filter Bank, which can approximates original images using contourlets and present the images well. Contourlet transformation inherits wavelet's local supporting and the mutliresolution property. Besides, it has anisotropy property. So in the image processing, contourlet transformation can present edges and texture of medical images more effectively than wavelet. Contourlet transformation and its application in image de-noising are systemically studied in this paper as following:1. The theory and implementation of the wavelet transformation was particularly studied. And then, the contourlet transformation was introduced. From their application on image processing, Our experiments show that the results of contourlet transformation surpass the results of the wavelet transformation.2. The directional filter is the fundation of the contourlet transformation. In this paper, the diamond filter and the fan filter were studied in advance, which is the fundation of the directional filter.3. The theory and the structure of the contourlet transformation was analysed in this paper. The algorithm of the laplacian pyramid decomposition and the directional filter banks was constructed.4. The contourlet transformation was combined with the soft thresholding and the cycle spinning algorithm. In my paper, the MR images was de-noised by wavelet transformation, contourlet transformation, cycle spinning wavelet transformation and cycle spinning contourlet transformation separately, Our experiments show that the cycle spinning contourlet transform can better present the images than other transforms.