Based on the application of color language in graphic design and the principle of geometric analysis of design, this paper proposes to design the layout and elemental color of graphic visual images on the basis of optimized probability density function estimation algorithm. With reference to the container-based geometric feature modeling method, a clustering method is added to the modeling of elemental geometric features. The problem of designing the color of a print advertisement image is converted into finding the optimal sequence of features from the joint probability density distribution of the color features of the main body, background and copy elements. On the basis of the main color features obtained from the probability estimation, the algorithm reconstructs the elements in color, so that the color design task of the plane visual image can be accomplished. The color enhancement model of planar visual image is proposed, and the model effect analysis is carried out. The results of color enhancement of planar visual images show that the method of this paper gets better performance in four evaluation indexes of planar images, namely, mean, variance, information entropy and clarity.