No-reference image quality assessment based on BNB measurement

In this paper, we present a no-reference image quality assessment method, which we call BNB (an acronym for Blurriness, Noisiness, Blockiness). Our BNB method quantifies blurriness, noisiness and blockiness of a given image, which are considered three critical factors that affect users´ quality of experience (QoE). The well designed BNB metrics are based on the observation that the difference between any two adjacent pixel values follows a Laplace distribution with mean zero, and the Laplace distribution will change differently under different artifacts, i.e., blurriness, noisiness and blockiness. Then we use supervised learning to map the three BNB metrics of an image to a human perception score. Experimental results show that the image quality score obtained by our BNB method has higher correlation with human perceptual score and our method needs much less computation, compared to existing no-reference image quality assessment methods.