Fast Fractal Image Encoder Using Non-overlapped Block Classification and Simplified Isometry Testing Scheme

Abstract

This paper aims at reducing the encoding time of a fractal image encoder. For this purpose, a non-overlapped block classification method and a simplified isometry testing scheme are proposed. The non-overlapped block classification method avoids the repeated classification operations needed for finding the domain blocks having the same type with the range block, by memorizing the classification results of the domain blocks and using them for the overlapped blocks in a new searching area. For reducing the time required for calculating a similarity between blocks, a simplified isometry testing scheme is used. It tests the isometry between a domain block and a range block using only those types of isometry having the similar features with the type of the range block. For speeding up the calculation time, the SOFM neural network is used as the block classifier and the spiral searching scheme is used. The experimental results have shown that the proposed algorithm reduces the encoding time by 50% on average while maintaining the same PSNR and bit rate, compared to the other’s recent approaches.