基于绝对矩块截断编码融合Clifford代数的图像压缩  

Research of Image Compression Based on Fusion of AMBTC and Clifford Algebra

作　　者：赵丽[1,2] 齐兴斌[1] 李雪梅[1] 田涛[3] 

机构地区：[1]山西大学计算机工程系,山西太原030013 [2]北京航空航天大学计算机学院,北京100191 [3]北京师范大学教育信息技术协同创新中心,北京100875

出　　处：《电视技术》2015年第6期13-17,31,共6页Video Engineering

基　　金：国家自然科学基金项目(61202163);山西省自然科学基金项目(2013011017-2)

摘　　要：针对现有的图像压缩方法很难兼顾压缩效率和压缩后图像质量的问题,提出了一种基于绝对矩块截断编码和Clifford代数的融合方法。首先,将原始图像分为若干大小相等且互不重叠的局部小块;然后,利用绝对矩块截断编码保留每个子块的第一和第二矩;最后,利用Clifford代数将图像矩阵表示为最大完全平方和,并利用解码器重构图像。压缩实验结果表明,该方法的峰值信噪比可接近100 d B,结构相似度接近1,相比其他几种较新的方法,该方法取得了更好的压缩图像质量,并且降低了压缩耗时。For the issue that it is hard to juggle compression efficiency and the quality of compressed image when using traditional compressed methods,a fusion method based on absolute moment block truncation coding and Clifford algebra is proposed. Firstly, original images are divided into a number of non-overlapping local small pieces with equal size. Then, absolute moment block truncation coding is used to retain the first and second moment of each sub-block. Finally, Clifford algebra is used to represent image matrix to be the largest completely squares sum, and decoder is used to reconstruct image. Experimental results show that peak signal noise ratio and structural similarity of proposed method can achieve lOOdB and 1 respectively, it has better compression quality of image and less time-consuming than several other advanced methods.

关 键 词：图像压缩 绝对矩块截断编码 CLIFFORD代数 峰值信噪比 加权峰值信噪比 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]