Quality assessment of digital television picture based on local feature matching

Abstract

This paper presents a novel, full-reference image quality measurement method for video quality assessment in television systems. Particularly, the proposed metric targets the problem of automatic assessment of digital picture degradations. The performance of proposed algorithm has been evaluated on different image sequences with various distortions of interest. The results show that the proposed scheme for quality assessment outperforms other image quality measures: PSNR, SSIM and VIF in terms of more efficient quality inspection based on local feature matching as well as robustness against various distortions such as luminance or contrast change, negligible picture misalignment, etc.