A Hybrid Image Compression Algorithm Based on Fractal and Wavelet Transforms

植基於碎型與小波轉換之混合型影像壓縮演算法

林俊宏(Chun-Hong Lin) ； 陳飛欽(Fei-Chin Chen) ； 謝宗煌(Tsung-Huang Hsieh)

《中華技術學院學報》 33期 (2005/12) Pp. 1-10

https://doi.org/10.7095/JCIT.200512.0001

摘要

由於碎型轉換存在自我相似的性質,影像的碎形壓縮變得可行。另外，影像的小波轉換提供多解析度的小波係數，而各個小波次頻帶間存在著相似度的特性使得碎形轉換與小波轉換間有著緊密的關聯。本文旨在提出的混合型影像壓縮演算法乃植基於碎型轉換與小波轉換：本方法取出小波轉換後所得到的相同方向的次頻帶間的相似性，用於碎型編碼之區塊預測；並以實驗的方式評估每個次頻帶的最佳值域區塊大小，將此值域區塊大小值代入混合型影像壓縮演算法。實驗結果顯示：本文提出的方法得到的PSNR值為29.2，CR值為19.8，不論主觀的視覺品質或客觀的均方誤差，都明顯優於傳統的碎形影像編碼。

關鍵字

碎型編碼 ； 小波轉換 ； 混合型影像壓縮 ； 相似性

並列摘要

The self-similarity feature of fractal compression makes it work based on iterated function system. Discrete Wavelet transform of an image provides a set of wavelet coefficients which represent the image at multiresolution. There exists a connection between wavelet transform and fractal compression because the self-similarity existed in wavelet transform could be exploited by fractal theory. The purpose of this paper is to propose a hybrid image coder based on fractal coding and wavelet coding schemes. The similarity among different subbands of the same orientation in a wavelet decomposition of the image is exploited for block prediction in fractal coding scheme. The optimal sizes of range blocks for each subband have been evaluated by experiments. The evaluated data is used in the hybrid image coding scheme. The experimental results show that PSNR and CR are 29.2, 19.8 respectively in our method. the performance of our hybrid image coder is superior to that obtained from baseline fractal image coder for both visual quality and mean square error.

並列關鍵字

fractal coding ； wavelet transform ； hybrid image compression ； similarity