Binocular vision based objective quality assessment method for stereoscopic images

Abstract

Human visual system (HVS) can perceive the difference between two retinal images to create a mental image with depth perception, which is the result of two binocular interactions, i.e., binocular fusion and suppression. According to perceptual attributes of binocular interactions, in this paper, a full-reference stereoscopic image quality assessment (SIQA) method is proposed based on the mechanisms of binocular fusion and suppression. There are two kinds of information in stereoscopic images: monocular information which is visible in only one view, and binocular information which is visible in two views. HVS adopts two ways to deal with the binocular information, one is binocular fusion which deals with the information with similar content and small disparity, the other is binocular suppression which deals with the information with dissimilar content or large disparity. Therefore, the proposed method firstly divides a distorted stereoscopic image into occluded, pseudo-binocular fusion and pseudo-binocular suppression regions. Then three methods are respectively adopted to assess the quality of the three regions and the three quality indices combine into one to represent the overall quality of the distorted stereoscopic image. Finally, the predictive performance of the proposed method is evaluated and compared with existing methods in terms of consistency, cross-image and cross-distortion, and robustness. Experimental results show that the proposed SIQA method outperforms other methods and can predict human visual perception of stereoscopic image more accurately.