Blind Quality Assessment for Screen Content Images by Texture Information

Ning Lu, Guohui Li

Abstract: Most image quality assessment (IQA) methods designed for screen content images (SCIs) require the reference information, and existing blind IQA metrics can not obtain consistent results with subjective ratings. In this study, we propose a novel blind image quality assessment method for SCIs based on orientation selectivity mechanism by which the primary visual cortex performs visual texture information extraction for scene understanding. First, we extract the orientation features to perceive the visual distortion of degraded SCIs. Second, the structure features are extracted from the derivatives as the complementary information of orientation features. Finally, we employ support vector regression (SVR) as the mapping function from the features to quality scores. Experimental results show that the proposed method can obtain better performance than other existing related methods.