| 摘要 |
<p><P>PROBLEM TO BE SOLVED: To automatically perform subjective quality estimation of the quality of video for a multimedia application without judging the quality by a human subject. <P>SOLUTION: A feature quantity extracting part 1 extracts as feature quantities, the block skewness of an evaluation image y with respect to an original image x, PSNR global time variability and PSNR local time variability, and a weighted sum calculating part 2 calculates a weighted sum of the feature quantities and calculate an objective image quality index. A frame rate detecting part 3 detects a frame rate of the evaluation image y. A correcting part 4 corrects the objective image quality index on the basis of the frame rate detected by the frame rate detecting part 3. An objective evaluation scale-subjective image quality mapping part 5 applies the corrected objective image quality index Q<SB>obj</SB>to a correlation between preimparted objective image quality index and subjective image quality and derives a subjective image quality estimation value DMOS. <P>COPYRIGHT: (C)2008,JPO&INPIT</p> |
