Evaluation of Compressing Haar Wavelet transformed images With Fast Fractal Image Compression

Abstract

This paper proposed some methods for applying fast fractal image compression (FFIC) on haar wavelet transformed images. The received red, green and blue (RGB) color image is first converted to YCbCr color space, Then Haar wavelet transform is applied to each of the subbands Y, Cb and Cr separately. This produces four smaller filtered images or subbands: LL, HL, LH, and HH. Three m are conducted to test the effect of applying fractal compression on these four subbands. In each method the subbands are treated in different way by applying the FFIC on some parts and leave the others without any changes to find the best compression method. The FFIC is speeded up by using the centralized moment descriptors which are applied on each range and domain block, then sort the domain blocks to determine the suitable symmetry case without trying the eight symmetry cases when searching for the best match in the domain blocks. The subbands (HL, LH, and HH) in Cb and Cr components are not saved at all to increase the compression because these parts do not contain important information that affects the quality of the image while the LL part and all Y component parts are managed in different way in each of the three suggested methods. Quantization is applied to reduce the saved data. Finally the approach is tested on Lena’s images using the PSNR to test the quality, compression ratio and the compression time parameters.