An innovative development model for combining visual communication and art design based on the integration of multilevel features

Art design for visual communication is an innovative model that combines art, design and communication, aiming to present artworks through visual elements and design principles. This paper utilizes level 5 features extracted from pillar networks as a rough unimodal feature to extract saliency targets in art design images. Improving their performance can be achieved by cascading different unimodal features. Design the multi-branch group fusion module, which is divided into multi- and single-branch fusion branches according to the number of groups to extract different numbers of image features. Determine the classification of art images by their original and stroke information features. Design a quantitative analysis scheme for image data to analyze the role of art design from the perspective of RGB, HSV, and image visual information communication effect. In the image HSV analysis, the hue distribution region of the sample original image is mainly concentrated in [0,0.1] and [0.5,0.6], the distribution is more concentrated, and 1/3 of the maximum value of the saturation is under 0.8, which shows the visual characteristics mainly through the hue component. A better communication effect can be attributed to the mean value of the color conciseness score reaching 4.212 in the image visual communication effect.