Based On Contourlet Transform Image Reconstruction And Image Compression Algorithm

Over the last decade, wavelets have had a growing impact on image processing. The wavelet transform has been proven to be powerful in many signal and image processing applications such as compression, noise removal, image edge enhancement, and feature extraction. Unfortunately, wavelets are good at catching point or zero-dimensional discontinuity. Wavelets lack directionality, and do not capture the geometrical smoothness of the contours. Wavelet is not optimal bases for natural image. Wavelet is not anisotropy, can't capture smooth contours in images.The Contourlet transform is one of the new geometrical image transforms, which can efficiently represent images containing contours and textures. The Contourlet transform is a new "true" two-dimensional representation for images that can capture the intrinsic geometrical structure of pictorial information. The Contourlet transform provides a flexible multi-resolution, local and directional expansion for images.The primary goal of the dissertation is to explore the theory and new applications of Contourlet transforms, particularly with respect to their applications in the image processing. The object of the paper is the structure of the Contourlet transform. The primary content is how to improve the Laplacian pyramid and the applications of Contourlet.The main work of this dissertation is summarized as follows:The paper investigating the theory and design of Contourlet transform. Based on the study of joint statistical characteristics of coefficients of Contourlet transform, an image compression algorithm is proposed. The algorithm is based on Contourlet transform and wavelets, it provides an embedded coder. The proposing embedded image coder has better performance than some popular wavelet image coder due to some above-cited improvement and innovations. Experimental results show the result is superior to that of wavelet. The reconstruction image quality outperforms SPIHT 1.16dB～2.22dB in PSNR.The paper discusses the basic idea and character of the Laplacian pyramid, and analyses the character of the lifting schemes and adaptive lifting schemes. The lifting pyramid and the adaptive lifting pyramid are applied in image processing. It can improve performance of de-noise and enhancement when compared to the standard Laplacian pyramid. Experiments results show the algorithm is superior to the traditional algorithm. The de-noise image quality outperforms the traditionalalgorithm 0.1dB～0.5dB in PSNR.In the implementation of Contourlet transforms, a crucial problem in the Contourlet construction is the design filter. The property is effect the performance of Contourlet in many image applications. In this paper first give a brief review of the early work on eigenfilter, discuss characteristic of filter in the Contourlet transform. Then according to characteristic of filter, maximally flat filter is designed by using eigenfilter. This allowed the directional filter bank to have much more efficient characteristic than with FIR filter prototypes. Simulations show that filters designed using the new criterion is better than those designed using the old criterion in least squares sense.The dissertation presents the theory and implementation of several directional filter banks. Also investigating several issues related to the realization of the two dimensional filter banks. Experimental results show the feasibility and validity of several algorithms.