In order to evaluate the quality of a stereoscopic virtual view image,an objective assessment based on three-dimensio nal perception is proposed.Two major distortion types of stereoscopic virtual vi ew images,single view rendering distortion and stereo view mismatch distortion,are considered.Fo r the single view rendering distortion, firstly,the different region between the original image and the distorted image will be extracted;secondly,the mean structural similarity index will be calculated for this region;finally,the left and right viewpoints will be averagely pooled as the monocular texture feature.For the stereo view mismatch d istortion,firstly,the left and right view distorted images will be mapped to each other respectively;secondly,the di fferent region between the mapped and the original images will be extracted as binocular mismatch region;thirdly, the mean structural similarity index will be calculated for this region;finally,the left and right viewpoints will be averagely pooled as the binocular competition feature;lastly,the two features will be converged as the objective q uality of stereoscopic virtual view image.Experimental results show that our method matches the DMOS effectively,the Pearson linear correlation coefficient is 0.911and the Spearman rank order correlation coefficient is 0.900,so the stereoscopic virtual view image quality is accurately reflected.