Parallel Computation strategies for Fractal Compression

Fractal coding is a lossy compression scheme and had an extra edge in the early 90s because of its ability to consider the self-similarities in an image. But the high computational cost involved in its encoding process made it impractical for commercial applications. This is due to the hard task of calculating various affine transformations in the comparison of range and domain pools. But the advantages in using FIC algorithms are less complex decoding process, high PSNR, better SSIM, dimensionless scheme and high compression ratio. Recent improvements in the computing techniques put back the FIC on track by reducing the computational cost involved in the encoding process. This paper suggests the importance and use of parallel computation techniques such as General Purpose Graphical Processing Unit (GPGPU) and Compute Unified Device Architecture (CUDA) in Fractal based image compression algorithms. Analysis of the empirical results shows that the parallel computational approaches could significantly reduce the encoding time as the calculation of fractal affine transformations (AT) can be largely parallelized.