Fractal coding is a potential image compression scheme, which has the advantages of relatively high compression ratios and good reconstruction fidelity. However, the high computational complexity of fractal image encoding greatly restricts its application. This paper proposes a DCT-based fractal image coding method, which improves the self-similarities exploiting scheme. The range and domain blocks are divided into three classes based on their DCT lower frequency coefficients, and only the domain blocks with the same class to the range block are calculated during best match exploiting process. Experimental results show that compared with standard fractal coding scheme, the encoding time is significantly reduced and the PSNR of the reconstructed image is also improved.
展开▼
