An edge property-based neighborhood region search strategy for fractal image compression

In this paper, an edge property-based neighborhood region search method is proposed to speedup the fractal encoder. The method searches for the best matched solution in the frequency domain. A coordinate system is constructed using the two lowest discrete cosine transformation (DCT) coefficients of image blocks. Image blocks with similar edge shapes will be concentrated in some specific regions. Therefore the purpose of speedup can be reached by limiting the search space. Moreover, embedding the edge property of block into the proposed search method, the speedup rate can be lifted further. Experimental results show that, under the condition of the same PSNR, the encoding time of the proposed method is only about two-fifth of Duh’s classification method. Compared with Tseng’s method, the proposed method is near or superior to the performance of their method. Moreover, the encoding speed of the proposed method is about 120 times faster than that of the full search method, while the penalty of retrieved image quality is only decaying 0.9 dB.