基于人类视觉系统的DCT编码图像后处理算法  

Post-processing algorithm for DCT-coded image based on human visual system

作　　者：石敏[1] 易清明[1] 刘金梅[1] 

机构地区：[1]暨南大学电子工程系,广州510632

出　　处：《计算机工程与应用》2007年第15期63-65,共3页Computer Engineering and Applications

基　　金：广东省科技攻关计划项目(No.2006A10201003);暨南大学博士启动项目(No.51205067);广东省软科学研究项目(No.2006B70103011)。

摘　　要：分析了块离散余弦变换(BDCT)图像编码的块效应产生原因,基于马尔可夫随机场最大后验估计(MRF-MAP)框架提出了一种有效的自适应去块效应算法。利用人类视觉系统对块效应有掩盖性的特点引入一个块效应可见度函数,根据转换块的边缘信息自适应调整势能函数的阈值,从而尽可能地消除块效应并保护图像的边缘信息。仿真结果表明了该算法的有效性。The reason that Block Discrete Cosine Transform （BDCT） brings blocking artifacts is analyzed in this paper.An efficiently adaptive de-blocking algorithm is proposed based on maximum a posteriori estimation of Markov Random Field （MRF- MAP） Frame.For Human Visual System has the masking characteristic,a blocking-artifact visible function is introduced.The threshold of the potential energy function is adaptively adjusted by edge information of the shifted block so as to reduce the blocking artifacts as much as possible and protect image edge information.The simulation results prove the efficiency of the proposed method.

关 键 词：块离散余弦变换 方块效应 图像边缘 

分 类 号：TP911[自动化与计算机技术]