Dianxin kexue (Mar 2018)
A quality assessment metric for stereoscopic image in free viewpoint video system
Abstract
A quality assessment metric of stereoscopic image in free viewpoint video system was proposed.After the reference and distorted images being decomposed by wavelet then weighted by contrast sensitivity function respectively,the content score was obtained by calculating the average structural similarity.After the center region being found in the binocular matching region,the average phase similarity was calculated on the region to obtain depth perception.The distortion region perceived by the human eye in the matching region by using the threshold,the average structural similarity was calculated between this region and the corresponding region of the other view image by disparity mapping,then the binocular characteristic score was obtained.The final image quality can be obtained by fusing score of each part.The experimental results show that the proposed algorithm has better consistency with subjective perception and it is more in line with human visual characteristics.
