LAUSR

In order to effectively evaluate the quality of stereoscopic images, we propose a no-reference stereoscopic image quality assessment (SIQA) method. Firstly, considering the characteristics of binocular fusion, binocular rivalry, and binocular suppression of human visual system, we propose a new color cyclopean image which is suitable for symmetric and asymmetric distortion images. And then, based on the importance of the disparity map, the enhanced image is generated according to the cyclopean image and the disparity map. Next, the natural statistical features are extracted from the enhanced image and the cyclopean image weighted by the gradient of disparity map (named weighted cyclopean image) in the spatial domain. The kurtosis and skewness are extracted from disparity map. Finally, the extracted features are fused and the quality of stereoscopic image is obtained by support vector regression. Experimental results show that the proposed algorithm is superior to most existing objective SIQA methods and can maintain a high degree of consistency with the subjective scores.