No-reference Image Quality Assessment Approach by Compressed Sensing and Mixture of Generalized Gaussian Distributions

摘要

This paper proposed a new no-reference image quality assessment approach based on compressed sensing and mixture of generalized Gaussian distribution (GGD). The image is processed by compressed sensing at first, then sparse coefficients of compressed sensing are modeled by mixture of GGD. The parameter of mixture of GGD is estimated by the parameter estimation approach and the feature vector is formed by combining the parameter of mixture of GGD. The feature vector is fed to the support vector machine for training and testing. Experiments result shows that our approach has good performance for image quality assessment.