Quality assessment for real out-of-focus blurred images ☆

摘要: Abstract Images are vulnerable to different kinds of distortions, such as blur, noise, blockiness etc, which all degrade the image quality. Among distorted images, out-of-focus blurred images frequently-encountered and occupy a large proportion. However, few efforts have been done quality evaluation for these images. In this paper, we devise dedicated scheme automatically infer is named GPSQ (Gradient magnitude Phase congruency-based Saliency-guided Quality model). GPSQ, pair low-level features, including gradient (GM) phase congruency (PC), extracted characterize local blurriness. Then saliency detection performed on generate corresponding map. Finally, weight structure map with estimate visual image. Experimental results demonstrate proposed delivers high consistency subjective results.