Using structural information for reduced reference image quality assessment

Reduced-reference (RR) image quality assessment (IQA) aims to achieve higher evaluation accuracy by using only partial information about the reference image. In this paper we propose a novel Reduced-reference image quality assessment method which uses the Singular Value Decomposition (SVD) algorithm for reducing the amount of information. Using the SVD components makes the proposed quality metric enable to evaluate the image quality in the spatial domain and get more accuracy while reducing a high amount of information. The novel approach employs the multi-scale structural similarity index (MS-SSIM). Therefore structure information is used for achieving superior prediction accuracy. In this way, the proposed method can reach similar accuracy as the widely used full-reference (FR) image quality metric MS-SSIM.