位置:成果数据库 > 期刊 > 期刊详情页
立体压缩图像的客观质量评价方法
  • 期刊名称:天津大学学报
  • 时间:2009.12.12
  • 页码:1084-1088
  • 分类:TP391.4[自动化与计算机技术—计算机应用技术;自动化与计算机技术—计算机科学与技术]
  • 作者机构:[1]天津大学电子信息工程学院,天津300072
  • 相关基金:国家高技术研究发展计划(863计划)资助项目(2007AA012304;2008AA01A320;2009AA012306);国家自然科学基金资助项目(60672077);天津市科技支撑计划资助项日(08ZCKFJC27900).
  • 相关项目:立体信息特征感知、质量评价基础理论与显示关键技术
中文摘要:

针对立体图像的客观质量评价问题,提出了一种适合于立体压缩图像的质量评价方法.该方法基于平面图像的人类视觉系统评价模型,考虑了人类视觉系统的多通道特性和对比度敏感度特性,使用Contourlet变换实现多通道分解,将立体图像对的差值信号进行评价.实验结果表明,该方法得到的质量评价结果符合人眼视觉特性,与主观质量评价方法的一致性较好,能够客观反映压缩立体图像的质量.

英文摘要:

For the objective measurement of stereo images,a method was proposed which was suitable for compressed images. The method was based on human vision system models of planar images for objective quality measurements,taking the characteristics of multi-channel and contrast sensitivity of human visual system into consideration. Contourlet transform was applied to realize multi-channel decomposition. The absolute differences of stereo images were used for quality measurement. Experimental results show that the method basically accords with human vision characteristics,and can reflect the quality of stereo compressed images objectively. It is also consistent with subjective evaluation preferably.

同期刊论文项目
同项目期刊论文