Research On Contourlet Transform And Its Application On Image Processing

The purpose of contourlet transform is to obtain sparse presentation of images, it inherits wavelets' local supporting and the multiresolution property. Besides, it has anisotropy property. Contourlet transform decomposes images into band-pass directional subband using PDFB. Contourlet transform approximates images using contourlets. Contourlet can present edges and texture of images more effectively than wavelet transform.Contourlet transform and its application to image de-noising and image fusion are systemically studied in this paper. Including:1. The theory, implementation of 2-D separable wavelet and its application on image processing was particularly studied. The contourlet transform's theory and implementation was particular studied. We show the contourlets' exceeding to the wavelets through some experiments.2. The theory of image de-noising by thresholding based on wavelet transform, the selection of thresholding function and the threshold was particular studied. A method for image de-noising based on the contourlet transform was proposed taking advantage of the contourlets' good directionality and anisotropy.3. The theory of multi-resolution image fusion based on wavelet transform and the design of the fusion rule is particular studied. A method of multiresolution image fusion using the wavelet-based contourlet transform was proposed taking advantage of the contourlets' good directionality and anisotropy and wavelets' good decorrelation.4. Due to the downsampling contourlet transform and wavelet based contourlet transform was shift variance. In my study, Cycle Spinning and Recursive Cycle Spinning was introduced to suppress the artifacts due to the shift variance of the transform. Experimental results show that the method can notably improve the visual effect, and retain the image's details and texture.