Practice and exploration of AI image recognition algorithm in digital protection and feature extraction of intangible cultural heritage

This paper explores the application of AI image recognition algorithm in digital protection and feature extraction of intangible cultural heritage. In this paper, the image of intangible cultural heritage is collected by high-resolution camera equipment, and the image segmentation, denoising and enhancement techniques in AI image recognition algorithm are used for preprocessing, which effectively improves the image quality. In the aspect of feature extraction and recognition, combined with CNN, YOLO and other algorithms, the shape, texture, color and other features of cultural relics are automatically extracted, and high-precision classification and recognition are realized. Through the algorithm, the virtual reconstruction and digital restoration of cultural relics are realized, which reduces the protection cost and shortens the protection period. Taking Suzhou embroidery as an example, U-Net, ResNet-50, CycleGAN and other models are adopted to realize the automatic segmentation, stitch feature extraction, color restoration and pattern recognition of embroidery, and the restoration effect is significantly better than that of traditional manual restoration. At the same time, the NeRF algorithm is used to generate a three-dimensional model, and the AR application is developed, which enhances the effect of cultural communication. The research shows that AI image recognition algorithm has broad application prospects in the digital protection of intangible cultural heritage, which can effectively improve the protection efficiency and promote the inheritance and development of intangible cultural heritage.