自适应编码与混合模型联合改进图像压缩算法  

Improved Image Compression Combined with Adaptive Coding and Hybrid Model

作　　者：李耀莹[1] 孙娟 LI Yao-ying;SUN Juan(Shanxi Nonnal University Linfen Collega,Shanxi Linfen 041000,China;Linfen People's Hospital,Shanxi Linfen 041000,China)

机构地区：[1]山西师范大学临汾学院,山西临汾041000 [2]临汾市人民医院,山西临汾041000

出　　处：《机械设计与制造》2023年第1期179-182,共4页Machinery Design & Manufacture

基　　金：2020年度山西省教育科学“十三五”规划专项课题(HLW-20181)。

摘　　要：为准确描述残差图像的概率分布特征,以实现图像的自适应无损压缩,提出了一种自适应算术编码与有限混合几何模型相联合的分块无损图像压缩算法。算法首先对图像的残差图像进行非重叠分块,并通过联合分布估计各分块的统计模型;然后通过最大似然估计对分块的混合参数进行估计;最后在预测误差块中使用直方图截断算法减少算术编码中的符号数量,从而减少零次符号的出现概率。实验结果表明,在实验图像集基础上,所提图像压缩算法的压缩比优于常用的标准JPEG-2000和JPEG-LS。In order to obtain more accurate residual image probability distribution,so as to achieve better lossless image compression,an applicable algorithm and geometric finite mixture model block lossless image compression scheme based on the proposed.First,the method will residual image into several non-overlapping blocks,and then use the expectation maximization algorithm,the mixing parameters were estimated through maximum likelihood estimation,and through the parameters of the hybrid geometric distribution statistics for each module of the model.In addition,the histogram truncation method is used for each prediction error block,so as to reduce the number of symbols in the coding algorithm,thus reducing the impact of the emergence of zero symbols.Experimental results show that the proposed compression method is better than the advanced standard JPEG-2000and JPEG-LS in the compression ratio of image set.

关 键 词：自适应算术编码 几何混合模型 无损图像压缩 期望最大化算法 非重叠分块 

分 类 号：TH16[机械工程—机械制造及自动化] TN919.8[电子电信—通信与信息系统]