Abstract:  As the base of further image processing, color image quality assessment is assumed to be identical to the subjective assessment of human vision systems. However, the traditional assessment methods may result in evaluations quite different from the subjective visual assessment. To solve this problem, this paper analyzed the characteristics of human vision systems and the quaternion, discussed difficulties in converting color images into gray ones and finally proposed a full-reference color image quality assessment method called HVS-QSVD, which integrated the characteristics of human vision systems and the quaternion. Also, this paper constructed evaluations of spatial locations, local variance, the complexity of the texture edge and the color information based on the human vision characteristics. By utilizing the above four evaluations as the elements of quaternions, a model is finally proposed to assessment the quality of color images. Experimental results on LIVE image database and other methods confirm that the proposed model is more effective to get an evaluation result identical to human vision. The proposed assessment model could also be used in applications such as Shanghai-Nanjing expressway video detection and industrial tile classification detection.