Using on-line transmitted domain blocks ordered in variance for non-iterative fractal image coding

Abstract

Iteration in the decoding process of conventional fractal image compression not only leads to a high computation complexity but also requires a large size of memory. Moreover, the convergent criterion for the iterated image should be carefully designed. Otherwise, the iterated image will not reach a convergent state. To overcome these problems, we propose a noniterative method based on a novel domain pool designfor fractal image coding. In encoder, the domain pool is built by the use of the mean information in fractal code and is on-line transmitted to the decoder. The domain blocks are selected from the mean image according to the block variance. In computer simulation, we successfully speed up the decoding process when an image is partitioned into 8*8 range blocks. The coding performances of the test images are also well preserved. A better performance than converntional schemes is obtained when an image is partitioned into 4*4 range blocks. However, a competitive computation load is required.