分形图像编码算法的参数选择对算法性能的影响  

The impact of parameter selection on fractal image coding algorithm performance

作　　者：崔朝辉[1] 刘冀伟[1] 王志良[1] 曲波[1] 

机构地区：[1]北京科技大学信息工程学院,北京100083

出　　处：《智能系统学报》2010年第3期233-239,共7页CAAI Transactions on Intelligent Systems

基　　金：国家"863"计划资助项目(2007AA01Z160);北京市重点学科建设资助项目(XK100080537)

摘　　要：实际应用中的分形图像编码算法有众多参数需要确定,参数选择的恰当与否直接影响算法的性能,而如何确定参数的最佳值是每个研究者和使用者需要首先面对的问题.通过对基本的分形图像编解码算法的分析,发现解码图像的质量(PSNR)不仅跟值域块的分块大小相关,还跟图像的活跃度(IAM)相关.实验进一步表明,对应每一种分块大小,PSNR与IAM均存在对数关系;而且在解码过程中,仅需要6次迭代,解码图像就进入稳定状态.A number of parameters need to be determined before a fractal image coding algorithm can be used.The performance of the algorithm is directly affected by the selection of parameters.Determining the optimal values is the first problem that each researcher and user faces.Analysis of basic fractal image encoding and decoding algo-rithms showed the quality of a decoded image,as measured by the peak signal-to-noise ratio （ PSNR）,is not only related to the size of range blocks,but also the Image activity measure （ IAM）.Experiments showed that,for each range block,the relationship between PSNR and IAM is logarithmic.Moreover,the experiments indicated that only 6 iterations were required in the decoding process before an image went to a steady state.

关 键 词：分形图像编码 图像活跃度量 PSNR估计 参数选择 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]