A perceptual blurring metric, based on the variance difference between the original and compressed video sequence, is proposed. The human contrast sensitivity function is used to define different spatial frequency bands in which the variance is measured. The region of interest area is used for masking video content to further improve the accuracy of the metric. The proposed metric is evaluated on subjective blurring scores, and yields a high correlation coefficient of 0.934. Robustness of the metric is tested and revealed correlations higher than 0.78, when all video fragments were used. ©2009 IEEE.
|Title of host publication||1st International Conference on Information Science and Engineering (ICISE), 26-28 December 2009, Nanjing|
|Place of Publication||Nanjing|
|Publication status||Published - 2009|