Hybrid image coding based on partial fractal mapping

Abstract

The fractal image compression technique models a natural image using a contractive mapping called fractal mapping in the image space. In this paper, we demonstrate that the fractal image coding algorithm is compatible with other image coding methods. In other words, we can encode only part of the image using fractal technique and model the remaining part using other algorithms. According to such an idea, a new mapping in the image space called partial fractal mapping is proposed. Furthermore, a general framework of fractal-based hybrid image coding encoding/decoding systems is presented. The framework provides us with much flexibility for real implementations. Many different hybrid image coding schemes can be derived from it. Finally, a new hybrid image coding scheme is proposed where non-fractal coded regions are used to help the encoding of fractal coded regions. Experiments show that the proposed system performs better than the quadtree-based fractal image coding algorithm and the JPEG image compression standard at high compression ratios larger than 30.