A Knowledge Graph-based Algorithm that Integrates Traditional Intelligent Pattern Retrieval with Modern Design

Traditional patterns as cultural symbols embossed with the imprint of Chinese civilization, are the result of thousands of years of craftsmanship and national aesthetic. Their digital preservation and fusion with contemporary design, is an important channel of cultural inheritance. However, existing traditional pattern retrieval places great emphasis on keyword matches and results in semantic gaps and lack of cultural ties. Furthermore, integration with modern design is often done by hand and is therefore low in efficiency, stylistically disjointed, and unable to accommodate diverse needs. To solve this problem, a traditional pattern knowledge graph by visual features, cultural attributes and craft information is constructed in this paper to overcome the solution problem of knowledge fragmentation. Next, a multimodal intelligent retrieval algorithm with the knowledge graph is designed to achieve accurate retrieval and the tracing of the cultural origin. Finally, a fusion algorithm is proposed to complete the style adaptation, element reconstruction, and intelligent generation. Results show that the proposed algorithm does achieve an average retrieval accuracy of 94% which is 32%, 19% and 13% higher than text-based retrieval, image feature-based retrieval and hybrid retrieval without a knowledge graph respectively. This authenticates the feasibility and practicality of the algorithm, which gives technical support to the living inheritance of traditional patterns and modern design innovation.