Title: MSSBM and Its Application to Nature Image Coding
Abstract:Yao Zhao, Baozong Yuan, Member IEEE Institute of Information Science, Northern Jiaotong University, Beijing 100044, P. R. China Abstract In traditional fractal image coding (FIC) schemes, domain block...Yao Zhao, Baozong Yuan, Member IEEE Institute of Information Science, Northern Jiaotong University, Beijing 100044, P. R. China Abstract In traditional fractal image coding (FIC) schemes, domain blocks are constrained to be twice as large as range blocks in order to ensure the convergence of the iterative decoding stage. However, this constraint has limited the fractal encoder to fully exploit the self-similarity of the original image. In order to overcome the shortcoming, a novel scheme using multiple same-sized block mapping (MSSBM) is proposed in the paper. Meanwhile, a recursive scheme feeding the coding results back to the input during the coding procedure is used to further improve the decoded image quality. The concise coding steps of our scheme combining MSSBM with the recursive coding are as follows: 1) For an original image orig X , copy orig X as initial domain image dom X ; 2) Quadtree partition orig X into range blocks similar to Fisher’s; 3) Constructing domain blocks pool D from dom X , unlike Fisher’s scheme, D is composed of two parts, 1 D and 2 D , 1 D includes same-sized domains, 2 D includes twice larger ones; 4) For a range block, check if it can be encoded with a same-sized domain or a twice larger domain according to its coding status; 5) If the range block is encoded with 1 D , just calculate the rms (root mean square) error between the range and domains in 1 D . If rms lower than a pre-selected threshold, check its contractivity to decide its usage. The coding method using 2 D is the same as Fisher’s; 6) If the range block is encoded in the above step, update the domain image dom X by transforming dom X using all the transforms achieved for several times, go to step 3); 7) When every range block is encoded, the whole image is therefore encoded. Experiments have been done to compare the performance with Fisher’s quadtree scheme. The peak-to-peak signal to noise ratio (PSNR) of the two schemes vs. different compression ratios (CR) is shown in Fig. 1. Experimental results indicate that the scheme can achieve remarkable improvement in compression ratio and image quality.Read More
Publication Year: 2001
Publication Date: 2001-03-27
Language: en
Type: article
Access and Citation
AI Researcher Chatbot
Get quick answers to your questions about the article from our AI researcher chatbot