一种有效的块DCT编码图像去块效应算法  

Effective de-blocking algorithm for block-DCT images

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,广东广州510632

出　　处：《计算机应用》2007年第6期1460-1462,共3页journal of Computer Applications

基　　金：广东省科技攻关计划项目(2006A10201003);暨南大学博士启动项目(51205067);广东省软科学研究项目(2006B70103011)

摘　　要：块离散余弦变换在低比特率时其恢复图像的块边界上会出现明显可见的方块效应,从而降低了图像的视觉质量。提出一种基于小波域特征分析的去块效应算法,目的是尽量消除块效应的同时充分保护图像的边缘信息。图像空域中的块效应,在小波域高频子带表现为平行线效应、竖直线效应或者网格效应。通过为各个子带设置自适应操作算子去抑制块边界系数能量聚集现象,使得空域中的块效应得到消除。仿真结果表明在不同比特率下,对不同类型的图像,新算法都能得到较好的去块效应效果,并且具有较高的运算速度。The block-Based Discrete Cosine Transform （BDCT） suffers from visible blocking artifacts between adjacent image blocks at low bit rates, which reduces the image visual quality. This paper proposed a blocking artifacts reduction algorithm based on feature analysis in wavelet domain in order to reduce blocking artifacts while preserving edge and texture information. The blocking artifacts in space domain display horizontal line-shape effect, vertical line-shape effect and grid-shape effect in high frequency sub-bands of wavelet domain. The proposed algorithm used adaptive operators for every sub-band to suppress these energy-concentrated coefficients on the block boundary to reduce blocking artifacts in space domain. The simulation result shows that it has good performance in reducing blocking effects at different bit rates for different kinds of images and the calculation speed is very fast.

关 键 词：块编码图像 块效应 小波变换 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]