Stereo Video Quality assessment using SSIM and depth maps

摘要

Research on 3D stereoscopic images and videos has been extended quite a lot over the last 10 years. However, there is still empty space for objective quality metrics yet to come for 3D images and videos. Objective quality measurement is based on designing mathematical metrics which result in high correlation with subjective test results that take lots of time to perform in fact. This study mainly relies on SSIM (Structural SIMilarity Index) metric which has been proposed to measure structural similarities between two monoscopic images. But, the work on adaptation of SSIM to stereoscopic videos is found very limited in the literature. In this paper, we propose two approaches to adapt SSIM metric in stereoscopic videos by means of combining depth maps with local and global weighting methods. The proposed metric is evaluated using the SESVIQ (Subjective Evaluation of Stereo Video Quality) methodology for subjective assessment. The results show that our metric highly correlated with subjective scores.