计算机工程

• 多媒体技术及应用 • 上一篇    下一篇

基于几何结构的自适应空域错误隐藏算法

杨 丹1,2,卢桂馥1,2,3,周鸣争1   

  1. (1. 安徽工程大学计算机与信息学院,安徽 芜湖 241000;2. 南京理工大学高维信息智能感知与系统教育部重点实验室,南京 210094;3. 东南大学信息科学与工程学院,南京 210096)
  • 收稿日期:2013-07-03 出版日期:2014-03-15 发布日期:2014-03-13
  • 作者简介:杨 丹(1973-),男,讲师、硕士,主研方向:模式识别,视频编码;卢桂馥,副教授、博士;周鸣争,教授。
  • 基金项目:
    国家自然科学基金资助项目(61231002);安徽省自然科学基金资助项目(1308085MF95);高维信息智能感知与系统教育部重点实验室(南京理工大学)开放基金资助项目(30920130122005);中国博士后科学基金资助项目(2013M531251)。

Adaptive Spatial Error Concealment Algorithm Based on Geometric Structure

YANG Dan 1,2, LU Gui-fu 1,2,3, ZHOU Ming-zheng 1   

  1. (1. School of Computer and Information, Anhui Polytechnic University, Wuhu 241000, China; 2. Key Laboratory of Intelligent Perception and Systems for High-dimensional Information, Ministry of Education, Nanjing University of Science and Technology, Nanjing 210094, China; 3. School of Information Science and Engineering, Southeast University, Nanjing 210096, China)
  • Received:2013-07-03 Online:2014-03-15 Published:2014-03-13

摘要: 针对视频传输时容易发生损坏或丢失数据的问题,提出一种基于几何结构的自适应空域错误隐藏算法,以提高恢复图像质量。利用受损宏块的相邻2层像素提取几何结构,依据受损块相邻像素区分平滑块和边缘块。对平滑块采用双线插值,对边缘块根据内外两层的转折点寻找边缘方向,从而划分区域插值运算。实验结果表明,对于不同宏块丢失率和不同的视频序列,该算法所恢复的视频序列的峰值信噪比比双线插值、方向插值算法提高了0.5 dB~3 dB,不仅避免了虚假边缘,而且也提高了方向插值的准确性,改善了恢复图像的主观效果。

关键词: 几何结构, 错误隐藏, 转折点, 双线插值, 方向插值, 空域

Abstract: Aiming at the characteristic of damage or lose data during video transmission, an algorithm of adaptive spatial error concealment based on geometric structure is proposed, which can recover the quality of image better. The two nearest surrounding layers of pixels are used to extract the local geometric structure. The proposed algorithm makes use of surrounding pixels of a damaged block to distinguish between smooth block and edge block. Bilinear Interpolation(BI) is used to process smooth block, while edge block makes use of the transition point to detect the edge direction and the lost Macro Block(MB) is partitioned into segments. Each pixel in segments directionally interpolated from the boundary pixels are adjacent to the segments. Experimental results show that the value of Peak Signal to Noise Ratio(PSNR) of reconstructed video is higher than that of BI and Directional Interpolation(DI) by 0.5 dB~3 dB for different rates of lost MB and different video sequences. The proposed algorithm can not only achieve good directional interpolation but also avoid fake edges. The image’s subjective effect has great improvement.

Key words: geometric structure, error concealment, transition point, Bilinear Interpolation(BI), Directional Interpolation(DI), spatial domain

中图分类号: