The work aims to propose digital printing quality assessment method based on quaternion singular value decomposition for the defect of the human visual perception not conforming to the result of printing quality assessment. First of all, according to the color characteristics of printing images, quaternion was adopted as a printing image representation model. The real part of the quaternion matrix was the image brightness variance, and the imaginary part was red, green and blue. Thus, a full quaternion representation model of printing images was established. Singular value decomposition was carried out to the full quaternion matrix to obtain the singular value vector of the quaternion matrix corresponding to blocks. Finally, the printing image quality assessment index was measured by the linear correlation degree of image singular value feature vector. Quaternion model of printing images could completely represent digital printing images, effectively highlight sensitive structural information of human vision, and realize parallel processing of image information in printing quality assessment. Quaternion model of printing image improves and increases the accuracy of color printing image quality and has good consistency with human visual characteristics.