Introducing the parametric design method into the field of graphic design has become an important topic in future graphic design research. In order to improve the effective integration of graphic design education and science and technology, this paper applies the color perception evaluation model based on computer vision to graphic design education. Aiming at the problems that traditional algorithms can’t adaptively extract the main color of images and the lack of evaluation indexes for intelligent color collocation effect, this paper proposes a color collocation evaluation method that integrates visual perception and similarity measurement, and an intelligent color collocation algorithm that integrates visual aesthetics. Through quantitative detection and subjective evaluation, it is shown that the test results of this method have high palette similarity with the source palette and have good effects in graphic design education, which can provide relatively good technical support for subsequent graphic design education. Compared with the Pix2Pix network model that does not integrate visual aesthetics, the method proposed in this paper has a certain improvement in both comprehensive evaluation index and peak signal-to-noise ratio. Meanwhile, the method proposed in this article has improved subjective preference compared to the Pix2Pix network model that does not integrate visual aesthetics and is of certain significance to promote the intelligent development of subsequent graphic design education.