Fractal Coding for Still Images

• Main Authors: Cheng, Chih-Jen, 鄭智仁
• Other Authors: Ruey-Feng Chang
• Format: Others
• Language: zh-TW
• Published: 1996
• Online Access: http://ndltd.ncl.edu.tw/handle/94597194123120247767

碩士 === 國立中正大學 === 資訊工程學系 === 84 === In recent years, there are more and more data transmitted in the Internet.It is clear to know that the bandwidth of network is always not able to affordsuch a large amount of data. Since compressing data for saving storage andtransmitted time is easier than expanding the bandwidth of network, thecompression techniques are more and more important. Among the image codingtechniques, fractal coding has been considered to be an effective method forcoding image at low bit rate. In general, image compression scheme eliminatesthe redundancy between pixels to reduce the bit rate for storing the image.However, in the fractal coding scheme all we want to do is to find a similarblock for the block that we want to encode and employ transformations todescribe the relationship between the blocks. The parameters of transformationsand location of found block are stored as the code for the encoded block. The main disadvantage of fractal coding scheme is that the encoding procedurespends very long time. In order to find a similar block, the encoding proceduremust examines all the blocks in the image and computes the distortions. A largeramount of comparisons and computations are needed for this step. In this thesis,we propose a method that classifies the blocks into several classes and encodesthose blocks with different ways according to their classes. The blocks with small variances are encoded by their mean values, and notransformations are needed to be employed. The second kind of blocks with largervariances is restricted to search similar blocks in a small region instead ofsearching the whole image and some transformations are eliminated. For theblocks that have largest variances all transformations for the fractal codingare involved and a fast searching scheme in the whole image is employed to speedup the procedure. By the way, the quality of reconstructed image is acceptableand the bit rate is smaller with shorter encoding time. Comparing to severalfractal coding schemes, the proposed method has a better reconstructed imagewith nearly the same bit rate.