应用科学学报 ›› 2009, Vol. 27 ›› Issue (4): 353-358.

• 信号与信息处理 • 上一篇    下一篇

图像块状瑕疵消除及边缘作用的分析

朱方 时龙兴   

  1. 东南大学国家专用集成电路系统工程技术研究中心,南京210096
  • 收稿日期:2009-02-13 修回日期:2009-04-16 出版日期:2009-07-30 发布日期:2009-07-30
  • 作者简介:朱方,博士生,研究方向:嵌入式多媒体和多媒体系统芯片,E-mail: villion.zhu@delphi.com; 时龙兴,博士,教授,博导,研究方向:射频器件、功率器件设计、片上系统设计,E-mail: lxshi@seu.edu.cn
  • 基金资助:
    国家自然科学基金(No.60676011)资助项目

Reduction of Blocking Artifacts and Analysis of Edge Map Contribution

  • Received:2009-02-13 Revised:2009-04-16 Online:2009-07-30 Published:2009-07-30

摘要:

块状瑕疵是基于离散余弦变换的图像压缩常见问题之一,会造成图像质量的明显退化. 文中提出一种新的块状瑕疵消除(解块滤波)方法. 通过对近期解块滤波中边缘图提取方法及其作用的分析,引入压缩数据块的类型判断标准进行分划并提取边缘图,采用自适应复合方法消除图像中块状瑕疵. 所有的计算都在压缩域中进行,因此对原始数据的压缩特性没有任何影响. 实验结果表明,与现有其他方法相比,该方法覆盖范围广,输出稳定,处理后图像视觉效果好.

关键词: 边缘检测, 离散余弦变换, 块状瑕疵

Abstract:

Blocking artifacts are a major problem in discrete cosine transform (DCT)-based image compression algorithms, which degrades the image’s visual quality. In this paper, a new method of blocking artifact reduction is proposed. Contribution of edge maps in the deblocking filtering is analyzed. A criterion used to classify block types is introduced according to the distribution of DCT coefficients both in half-block-shifted image blocks and in the original source image blocks. Using such classification, an adaptive combination method is used to reduce blocking effects. Since all computation is performed in the DCT domain, compressibility of the original image is not affected. By comparing with some existing methods, the experiment results verified effectiveness and robustness of the proposed method.

Key words: edge identification, DCT, blocking artifact

中图分类号: