Summary: | 碩士 === 國立雲林科技大學 === 電機工程技術研究所 === 88 === In this paper, we implement the SPIHT algorithm that is used in still image compression by MATLAB programming language. Our results after program running comparison the coding performance with the curve of efficiency that derived from SPIHT paper is equal approximately. so, it is able to be a tool for simulation experiment about SPIHT algorithm. We propose a method that test the Visually Lossless (VLL) for reconstructed image, we call " Alternate test method ". Furthermore, we propose a much more precise subjective image fidelity estimated criterion based above method, can use to test image quality and some JND models. This paper propose also a objective estimated JND_PSNR criterion based visually weight, it is verified to Alternate test method that wants to instead of the too redundancy method with subjective estimated criterion. We propose the JND_SPIHT algorithm for improve the disadvantage of SPIHT that will spend too much time and get low coding efficiency when it is demanded to reconstruct a Lossless image, it must receive all bit stream of all Map that transmitted by encoder. it is verified by practice experiment, JND_PSNR value is infinite while bit rate is 1.0 bpp and the program running time will decrease about 93 % for encoder, will decrease about 91 % for decoder. So that, increasing the coding efficiency and decreasing the program running time are achieved by JND_SPIHT algorithm.
|