Research On The Fractal Image Compression Algorithm And Its Application

Digital Image as the most common form for information storage and manifestation is vivid and intuitive, but the data volume is excessive large, in order to fulfil the high-speed digital image transmission and storage, it must be compressed effectively. Today, there are around 100 image compression methods, some international standards have also been developed, like JPEG, MPEG, and H.26X, but the compressed image quality, ratio, and coding and decoding time still cannot meet the needs of practical applications. The traditional compression algorithms have relatively low compression, poor decoded image quality, and other relative shortcomings. Therefore, as being a new thought, potential, and fast developed algorithm, fractal image compression method has been growing quickly in the recent 10 years. It has high compression ratio and decompression speed superiorities, performance assessment has been extensively in attention. Although fractal image coding has many advantages, however, in the no artificial intervention condition, the coding is very time-consuming, which greatly restricts its development. By focusing on the shortcoming, this thesis provides the exhaustive study on how to reduce the encoding time issue as well as ensure the quality of decoding image. The major work can be divided into two aspects:(1) After comparison of the similarity between range block and domain block, and lots of search algorithm experiments, this thesis conducts the analysis on possible position of the optimal matched domain block D of range block R according to the law, a kind of adaptive non-search fractal image compression coding is put forward. This algorithm assigns specific domain block as matching block, thus search is required and coding is accelerated. Besides, this algorithm adopts the method of range block adaptive decomposition and combination, and can solve problems like part of range blocks incapable of matching and low compression ratio of non-search method. Experiments indicate that this algorithm is better than search fractal image compression algorithm and JPEG algorithm.(2) Based on the analysis of gray image fractal compression algorithm and true color image in RGB color model, this thesis provides one kind of true color image fractal compression algorithm. This method puts forward a new method of color image gray level: The median method. This improved methods using R, G, B color component similarity, and reduce the color-component matching block (SFC method) that required for searching and storing from 3 to 1, which speeds up the image compression and enhances the image compression ratio.(3) Seen true color image pixel as a three-dimensional vector, by analysising three dimensional vector, this thesis provides another kind of true color image fractal compression algorithm. The algorithm uses the concept of vector similarity, measure the errors by using distance measurement in two-dimensional array from vector similarity in the three-dimensional array, and base on the best error search match block. The experiment result indicates, the true color image fractal compression algorithm is better than to the SFC method and standard JPEG method.