Research Into Fast Search Method Based On Fractal Image Coding

With the development of information technology, the multimedia technology based on theimage filled in our life, which makes people pay more and more attention to how to process imageinformation more effectively. It should be considered that how to transmit and store image withfewer bits and less store space at the premise of high image quality, so all sorts of imagecompression coding technology arise at the historic moment. At present, Fractal Image Coding（FIC）is considered as one of most promising coding methods. However, the practicability of FICis restricted because of long encoding time and complex calculation. So this paper mainly studies onhow to reduce the encoding time while guaranteeing the reconstructed image quality.Firstly, it introduces the theory of encoding and decoding in basic fractal algorithm, and italso verifies that it only need to iterate about ten times to make the decoding image stable and stayclose to the original image through analyzing several different standard test image.Secondly, a fast FIC algorithm based on the relative error is proposed, which mainly focus onthe optimization of the search process. The proposed algorithm converts the range-domain blockmatching problem into the k-neighborhood search problem in the sense of the Mean Square Error(MSE) by using an inequality linking MSE and relative error. It will avoid the unrelated matchingbetween range block and domain block and short the encoding time in the search process. Theexperimental result shows that the proposed algorithm improve the encoding speed whileguaranteeing the decoding image quality, and is better than the basic fractal coding method.Thirdly, a fast FIC algorithm based on the eliminating condition of the relative error is given.This method reuses the inequality between relative error and root mean square to reduce the codepool and improve the encoding speed by setting up a eliminate condition and two judgment valuesto establish allowable codes. The experimental result shows that the proposed algorithm improvethe encoding speed while guaranteeing the decoding image quality, and is better than the basicfractal coding method and the fast FIC algorithm based on the relative error.