Utpal Nandi and Jyotsna Kumar Mandal, University of Kalyani, India

The image partitioning scheme in fractal image compression is one of the important aspect for enhancement of performance. In this paper, an adaptive quardtree partitioning scheme is proposed wherethe entire image is sub-divided recursively into four sub-images. The partitioning points are selectedadaptively in a image context-dependent way instead of middle points of the image sides as in quardtree partitioning scheme. Biased successive differences of sum of pixel values of rows of the image are calculated to divide the image row-wise into two sub-images. Then, each sub-image is farther divided column-wise into two parts using biased successive differences of sum of pixel values of columns of the sub-image. Then, a fractal image compression technique is proposed based on the proposed partitioning scheme. The comparison of the compression ratio and PSNR are done between fractal image compression with quardtree and proposed adaptive quardtree partitioning schemes. The comparison of the compressiontime between the same is also done. The fractal image compression with proposed partitioning scheme offers better compression rates most of the times with comparatively improved PSNR. But, the compression time of the fractal image compression with proposed partitioning scheme is much more than its counterpart.

Fractal compression, compression ratio, Quardtree partition, Affine map, Iterated Function Systems (IFS), Partitioned Iterated Function Systems (PIFS).