计算机应用 ›› 2012, Vol. 32 ›› Issue (03): 715-718.DOI: 10.3724/SP.J.1087.2012.00715

• 图形图像技术 • 上一篇    下一篇

基于HVS时频特性的视频质量评价方法

王海峰1,2   

  1. 1.临沂大学 信息学院,山东 临沂 276002;
    2.上海理工大学 管理学院,上海 200093
  • 收稿日期:2011-09-08 修回日期:2011-12-12 发布日期:2012-03-01 出版日期:2012-03-01
  • 通讯作者: 王海峰
  • 作者简介:王海峰(1976),男,山东临沂人,副教授,博士,主要研究方向:视频分析、并行计算。

Video quality evaluation based on temporal feature of HVS

WANG Hai-feng1,2   

  1. 1.School of Information, Linyi University, Linyi Shandong 276002, China;
    2.School of Management, University of Shanghai for Science and Technology, Shanghai 200093, China
  • Received:2011-09-08 Revised:2011-12-12 Online:2012-03-01 Published:2012-03-01
  • Contact: Hai-feng WANG

摘要: 对于内容变化快的视频,现有仿真人类视觉系统的质量评价方法效果较差。因为主观评价者具有带通和掩蔽两种重要时域特征,现有方法中并未考虑这类与运动变化相关的视觉特性,导致客观与主观评价存在较大偏差。为提高运动变化快视频的评价准确性,利用统计学习的方法确定人类视觉阈值并建立模拟视觉系统的带通滤波模型;通过衰减权函数仿真人眼掩蔽特性。与基于信号特征的峰值信噪比(PSNR)方法、常权特征组合法、基于规则的变权特征评价法进行实验比较,在丢包率小于5%情况下获得最优评价效果。在带通模型的滤波作用下评价算法的执行效率得到提高。总之,该方法用简单方式有效地模拟复杂人类视觉特性,不仅提高评价性能而且降低计算复杂度。

关键词: 视频质量评价, 人类视觉系统, 带通滤波, 衰减权函数

Abstract: For the videos that have fast changing motion scenes, the existing simulation of the human visual system is less effective in quality assessment. Due to the bandpass and masking features of the subjective testers, the objective evaluation model ignores the two temporal features of the Human Visual System (HVS), which leads to the deviation between the subjective and objective evaluation results. To improve the evluation performance of the rapidly changing videos, the visual threshold was determined by using the statistic learning method and the filtering of the HVS was built up. The masking of human eyes was emulated through a new attenuation-weight function. The experimental results demonstrate that the proposed method obtained the best performance when the lost packets rates was lower than 5 percent compared with the Peak Signal-to-Noise Ratio (PSNR) method, the constant-weight evaluation model and the rule evaluation model.With the filter function of the bandpass model, it improved the execution efficiency. In brief, the proposed method not only improves the evaluation performance but also reduces the computational complexity.

Key words: video quality metric, Human Visual System (HVS), bandpass filtering, attenuation-weight function

中图分类号: