分形与压缩感知理论相结合的图像编码算法研究  

Study on image coding algorithm combining fractal and compressed sensing theory

作　　者：李高平 LI Gao-ping(School of Mathematics,Southwest Minzu University,Chengdu 610041,China)

机构地区：[1]西南民族大学数学学院,四川成都610041

出　　处：《西南民族大学学报（自然科学版）》2022年第3期297-302,共6页Journal of Southwest Minzu University(Natural Science Edition)

基　　金：西南民族大学中央高校基本科研业务费专项资金资助(2020NYB17)。

摘　　要：为了将分形图像编码与图像压缩感知两种编码方法结合起来,研究性能更优的混合编码算法,首先把压缩感知理论中的测量矩阵引入到分形图像编码算法中,以去除图像子块的冗余信息来采样其有效信息;其次,利用采样图像子块所得到的测量值来定义其均值标准差乘积特征(简称均标积特征);最后,依据均标积特征设置一个剔除条件,将全搜索匹配过程转换为均标积特征下的邻域搜索.四幅图像的仿真结果表明,所提算法与全搜索分形算法、压缩感知算法相比,在主观质量评价指标SSIM值几乎没有变化、以PSNR度量的重建图像质量分别降低约0.8 dB、2.58 dB的情况下,所需时间平均分别为相应算法的9.36%与19.83%.In order to combine fractal image coding and compressed sensing,and study the hybrid coding algorithm with better performance,firstly,the measurement matrix of compressed sensing theory was introduced into fractal image coding algorithm to remove the redundant information of image sub-block to sample its effective information.Secondly,the mean standard deviation product feature(MSDPF)of image sub-blocks was defined by the measured values obtained from sampling image sub-blocks.Finally,a culling condition was set according to the MSDPF,and the whole search of matching process was transformed into neighborhood search.The simulation results of four images showed that,compared with the full search fractal algorithm and the compressed sensing algorithm,the SSIM value of subjective quality evaluation index hardly changed,and the reconstructed image quality measured by PSNR decreased by about 0.8 dB and 2.58 dB,respectively.The average time required by the proposed algorithm was 9.36%and 19.83%,respectively.

关 键 词：图像压缩 分形图像编码 压缩感知 均标积特征 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]