Against the backdrop of rapid development of artificial intelligence technology, traditional art creation is facing a critical period of digital transformation. This study aims to explore the application value of augmented reality (AR) technology in traditional art creation, and enhance artistic expression through technological innovation. The research method adopts an image enhancement algorithm based on deep learning, focusing on solving the problem of uneven brightness in traditional art images under low-light conditions. The experimental results show that the proposed algorithm performs excellently in multiple indicators: the absolute average brightness error is reduced to 0.15 (a 42% improvement compared to traditional methods), the peak signal-to-noise ratio reaches 28.6 dB (a 23% improvement), the contrast quality index is 0.89 (a 31% improvement), and the structural similarity index reaches 0.92 (an 18% improvement). The research results indicate that AR technology can effectively enhance the expression forms of traditional art, provide new technological paths for artistic creation, and have important practical application value.