Efficiency of adaptive fractal image compression with archetype classification and its modifications

Abstract

The DCT-based JPEG image compression technique is a standard for lossy image compression. But the technique is not suitable at high compression ratios. Again, it is resolution dependent. Alternative compression techniques are introduced to reduce these problems. One such popular technique is fractal-based image compression. In this paper, efficiency of a fractal-based image compression technique is discussed that applies adaptive quadtree partitioning and archetype classification schemes. The adaptive quadtree partitioning scheme is dependent on image pixel context to share some self-similar structure of the image and improves the decoded image quality significantly. The archetype classification scheme tries to base the technique on finding better covering of range by domain that suites better in fractal image compression techniques. Then, modification of the same is also done to enhance the performance by applying dictionary-based loss-less data compression techniques, i.e. OLZW on the fractal compressed image that is basically a collection of affine transforms. The affine transform consists of a number of parameters that are compressed using OLZW technique. After that, the same experiment is carried out by a variant of OLZW, i.e. MOLZW. The experimental results are compared among existing and modified techniques. The modified versions yield not only better image quality but also high encoding speed than their counterparts.