Objective quality evaluation of stereoscopic images is useful in three-dimensional television application, and it is difficult to establish a reasonable relationship between objective evaluation and quality score. In this paper, a new stereoscopic image quality assessment method is proposed. In the method, stereoscopic features are first extracted to describe perceptual attributes of image quality and depth perception by singular value decomposition. Then, the relationship between stereoscopic features and quality score is established by using support vector regression. Finally, the objective evaluation values are tested on the symmetric database. Experimental results show that the proposed is more effective in quantifying image quality, compared with other two relevant quality evaluation metrics.