IFS Coding for Zero-Mean Image Blocks

Abstract

In this research work a new fractal coding scheme based on IFS-transform for zero-mean range-domain blocks is investigated. Some improvements were performed on the IFS-matching stage, these improvements implies the use of moment indexing as a criteria to filter the domain blocks that suitable to match the range blocks, in addition to the use of stopping search condition based on the monitoring the minimum matching error, these additional coding steps will reduce the required long fractal coding time. The test results indicated that the proposed improvements had reduced the required coding time to more than 10 times without significant degradation in image quality (MSE or PSNR) level. Also the results indicated that the brightness preserved IFS-method gives compression results better than the traditional fractal coding method, and the offset coefficients values could be better encoded by using DPCM.