Quality Assessment of 3D Video Based on Visual Salient Regions

Compare with traditional 2D video, 3D video can provide viewers with more realistic stereo experiences because of the depth information. In the paper, a new objective quality assessment algorithm for 3D videos is proposed. And experiments show that it is more accurate than peak signal-to-noise (PSNR) and structural similarity index measurement (SSIM). According to human visual system (HVS), we abstract contrast information, edge information, motion information and depth information from the test 3D video sequences and then build salient maps. The objective quality is calculated on the basis of SSIM/PSNR and salient maps. In order to test the accuracy of the algorithm, a subjective assessment experiment is designed according to related standards. Results have shown that the performance of proposed algorithm is better than original PSNR and SSIM measurements.