Hybrid Partitioning Technique for Fractal Image Compression

Abstract

In fractal compression the image to be encoded is partitioned into
blocks called ranges. Each range is coded by reference to some other part
of the image called domain and by some affine transformation parameters.
The number of ranges plays an important rule in encoding parameters. In
order to obtain high compression ratio, only a small number of blocks are
allowed in encoding process. In this paper we suggested a new technique
called Composed Partitioning Technique (CPT) to reduce the number of
partitioning blocks with keeping as much as possible the quality of the
reconstructed image. CPT is used the features of both quadtree and H-V
techniques, it uses the uniformity criterion to decide the range block is to
be partitioned or not, then it uses a criterion that depends on both the pixel
and the edges information of the image to decide the type and the direction
of the partitioning and this will give a more restriction to the partitioning
process.