As a national intangible cultural heritage, the Maonan Flower Bamboo Hat is a cultural symbol inherited by the Maonan for thousands of years. The weaving process of flower bamboo hats is exquisite and complex, and the patterns are rich in ethnic characteristics. In the process of modern social development, traditional handcraft skills face challenges such as the reduction of inheritors and low efficiency of innovative design. The development of artificial intelligence technology provides a new path for the digital protection of non-heritage, and the combination of intelligent algorithms and traditional crafts can realize the innovative inheritance of cultural symbols. Purpose: To construct the Maonan flower bamboo hat design translation model based on artificial intelligence algorithm to realize the innovation of rattan weaving art and cultural symbols inheritance. Methods: Collecting and organizing Maonan flower bamboo hat image data, using histogram equalization technology for preprocessing, proposing PSO-PCNN (Particle Swarm Optimization-Parallel Convolutional Neural Network) image style migration model, and realizing design translation through multi-feature fusion and granularity matching. The model contains five convolutional modules, uses Gram matrix to extract style features, and constructs fusion loss function to optimize image generation. RESULTS: The PSO-PCNN model reaches 0.784 in MSSIM, 19.61 in PSNR, 75.43 in FID, and 0.67 in KID, which are better than the comparison model in all indicators. User evaluation showed that 88% of the subjects thought that the translated pattern continued the traditional style (mean score of 4.18), and 80% thought that it was in line with contemporary aesthetics (mean score of 4.06). In the participatory design satisfaction survey, the satisfaction of all four cultural and creative products exceeded 70%. Conclusion: The PSO-PCNN model effectively realizes the innovative translation of the cultural symbols of the Maonan Flower Bamboo Hat, which not only retains the traditional characteristics but also meets the modern aesthetic needs, providing a feasible solution for the digital inheritance of non-heritage.
