In this Sparse Representation Based Image Quality Index with Adaptive Sub-Dictionaries work, 

we have addressed a novel full-reference image quality assessment model using sparse representation.Particularly, we propose an adaptive sub-dictionary selection approach to achieve this goal. 

Instead of comparing the sparse coefficients directly, we have also proposed to construct two feature maps and evaluate image quality based on their similarity measurement. The feature maps are also adopted to construct a weighting map.The final quality score is generated by combining the sparse feature similarity with auxiliary features, including gradient, color and luminance. We have conducted extensive experiments and comparisons on five publicly available image quality databases, 

and the experimental results have demonstrated that the proposed method produces the state-of-the-art performance.