Study on the Path of Visual Symbol Extraction and Design Transformation of Traditional Culture in Western Guangdong by Integrating Computer Vision Technology under the Background of Digital Intelligence

Abstract

In this paper, an adaptive convolutional neural network (ACNN) model is proposed to accurately recognize traditional cultural images of western Guangdong in a complex environment. The DDPM pattern symbol reconstruction technique is used to realize the translation and analysis of the Lijin pattern, and the color clustering method is introduced to effectively mine the shapes and colors in the cultural materials and transform them into design elements. Through the factor analysis method, it is clear which design materials are prioritized and used for product design practice. The study shows that the image recognition accuracy of the ACNN model is 83.2%. A total of six aspects of traditional cultural visual symbols of western Guangdong represented by Nuo dance, paper-cutting, lion dance, floating color, jumping flower hut and architecture are translated, and five representative main colors and 17 types of symmetrical patterns are obtained. The comprehensive evaluation value of charm, artistry and locality of the architectural aspect is 4.27, which is most suitable to be applied in the design of seal products.