Development and Application of Plane Image Generation System Based on Perceptual Features

The continuous development of big data, human–computer interaction, and other technologies promotes the combination of design and intelligent technology. Using computer instead of manual production can more effectively help designers complete design and development tasks. It is an important direction of the new generation of intelligent design research. As a very important design form, plane image can be seen everywhere in life. A large number of low value and consumable design requirements have become a difficult problem. In this regard, by studying the basic rules of graphic design, a hierarchical graphic design feature model is proposed. Based on the relationship between pixels and elements, a feature probability estimation model is established, fitting the characteristic probability density distribution to predict the best element position and color. This method can generate plane images with specific layout style, specific color style, and specific advertising theme and help the model output image generation results with more prominent layout style and richer color content.