An FFT-based technique for fast fractal image compression

Abstract

We present a new FFT-based technique of fractal image compression which can significantly speed up the process of matching domain and range blocks. In this method the isometric transformations of a domain block are chosen to permit block matching through cross correlation of range and domain blocks, which is implemented through the FFT. A suboptimal choice of domain blocks by performing a magnitude Fourier domain comparison of the range and domain blocks prior to correlation further speeds up the block matching process.