Fractal coding technique based on different block size

Abstract

The proposed technique presents adaptive new method used to decrease the encoding time in “fractal coding algorithms”. This technique reduces the “number of transforms” used to each domain block. So, the time required for the comparison between one “range block and domain blocks” is decreased. In this technique we partitioned the image depending on the Variance Criteria. Different block sizes ranging (4-16) have been used. The results show that the number of partitions (blocks), produced by quadtree partition, depends strongly on the adopted partitioning parameters (i.e. the Standard deviation, Mean, Minimum, and Maximum block size). The selection of small values of Mean(0.1-0.4) and Standard deviation (0.5-3) had reconstructed high image quality and low compression ratio,(e.g. we get PSNR =34.8dB and CR=4.89 E.T=18s) and vice versa.