Blind image quality assessment based on saliency extraction and contourlet transformation

摘要

In this paper, we present a blind image quality assessment based on saliency extraction and Contourlet transformation (BIQSC). The BIQSC first uses a method to extract the salient regions from the image and gives more emphasis on the salient regions. By combining Contourlet transform with a version of the hidden Markov mode--non-Gaussianty, the marginal distributions of neighbor coefficients in the Contourlet domain are modeled. With the Contourlet transform, the marginal histogram of coefficients in each subband can be well fitted by Asymmetric Generalized Gaussian Distribution (AGGD) after divisive normalization transforming. The parameters extracted from the Asymmetric Generalized Gaussian Distribution (AGGD) will be selected as the feature parameters. Experiments shows that the proposed metric has good consistency with the human subjective perception.