Video quality assessment using content-partitioned approach

YAO Jie,XIE Yongqiang,TAN Jianming,LI Dong,TANG Chao,WANG Fuhua
2013-01-01
Abstract:Current structural similarity based image quality assessment algorithm is generally the overall image quality analysis.However, different regions in image have different structural characteristics and visual perceptions, and the overall quality analysis can not reflect these differences effectively. In this view, a content-partitioned structural similarity image quality assessment algorithm is presented, which partitions an image into four regions according to their different gradient magnitudes and assesses the qualities of these regions respectively. A frame motion estimation weighted approach is used to extend this approach to video quality assessment. The experiments show that the proposed is more accurate than several modern popular algorithms.
What problem does this paper attempt to address?