Fast Image Compression Algorithms Based On Fractal Theory

Image compression is an important technology to improve the efficiency of image storage and transmission.The image compression algorithm based on fractal theory has received extensive attention due to its high compression ratio,good reconstructed image quality and independence of image resolution.However,due to the huge coding time consumption of the fractal image compression during the process of searching the matching image blocks for range blocks,it is difficult to apply fractal encoding technology on realistic applications.This paper proposes two improved fractal image compression algorithms based on the basic fractal encoding framework.The first fractal encoding algorithm is based on iterative control search strategy.The proposed method uses the update times of iterative function system to control the search process in order to avoid the low-yield search in the matching process,which overcomes the problem that inefficient search or invalid search caused by the uneven distribution of the feature space of domain blocks in the fractal encoding algorithms that using domain blocks classification scheme or codebook adaptive expansion scheme,and avoids the upper bound limit of the image decoding quality caused by the application of fixed quality allowable error control search strategy.At the same time,the applications of range blocks partitioning technique that the range blocks divided into smooth block class or rough block class according to the feature standard deviation,the codebook reduction technique that the adaptive expansion of the domain blocks and the isometric sampling dimensionality reduction operation used in the step of calculating the similarity of image blocks could effectively speed up the encoding process.From the view of practice,the fractal encoding algorithm based on iterative control search strategy proposed in this paper is a better coding acceleration scheme that can achieved in the case of loss of certain image decoding quality.The second fractal encoding algorithm is based on quadtree partitioning using adaptive threshold.Comparing to the setting of fixed block size in basic fractal coding and the setting of fixed threshold in traditional quadtree segmentation,the adaptive threshold segmentation is proposed which considering the characteristics of different input image and the objective differences of different regions of the same image.The quadtree splitting method calculates a unique threshold of the image block according to the mean and standard deviation of itself.From the view of practical point,the fractal encoding algorithm based on quadtree segmentation using adaptive threshold effectively improves the decoding quality under the premise of slighting increasing the coding time,which achieves a better balance between encoding time and decoding quality.To verify the effects of proposed methods,eight common 512*512 standard gray-scale images are used for testing,and their performance are quantified by encoding time(ET)and quality of reconstructed image(PSNR).The results show that the algorithms described in this paper can effectively speed up the encoding process under the premise of a certain loss of decoded image quality,when compared with alternatives.It achieves faster encoding scheme and satisfactory image decoding quality.