Research On Fractal Image Compression And Its Improved Algorithms

Vision is the most important way for human brain to acquire information.Nowadays,with the development and popularity of the technologies of image acquisition and computer network,a tremendous amount of digital images have become part of our daily life,and image compression has also become an important research direction.Since the first generation image compression technologies,which attempt to reduce image's statistical correlations,can no longer meet our demands,more and more attentions have been paid to the second generation image compression technologies which use wavelet,neural networks,and fractal theories,etc.Fractal Image Compression(FIC)utilizes the self-similarities between local and local or local and whole of natural image.Image compression is achieved by eliminating geometrical redundancy of natural image to the most degree.FIC is accepted as one of the most potential image compression techniques for its high compression ratio.This thesis reviewed literatures on FIC and proposed two improved algorithms to overcome the drawbacks of baseline FIC.The core works and innovations of the thesis include:1.The decoded image of FIC often appears "block effect" which goes against Human Visual System(HVS).To improve this drawback,a new FIC algorithm is proposed in this thesis.In particular,the HVS based Feature SIMilarity(FSIIM)index is modified,and the modified FSIM is employed as the criterion in searching the most similar codebook for each of the image patches.Experimental simulation shows that,since the "block effect" is depressed to a certain degree by the proposed algorithm,the perceptual quality of the decoded images is significantly improved.2.The high computation complexity,which is caused by searching the best similar codebook block for each of the image patches,is another drawback of,FIC.The thesis proposes to use Grover's quantum search algorithm to search the best similar codebook block for each of the image patches.Simulation results show that the time complexity of FIC can be reduced by the proposed algorithm while the perceptual quality of the decoded images is slightly degenerated.