Abstract: This study addressed the current low level of intelligence in firefighting helmets, which fail to provide visual assistance to firefighters. Augmented reality (AR) was combined with infrared thermal imaging technology to design an intelligent firefighting helmet. Because thermal infrared images often suffer from blurring, this study explored the optimization of the Scharr edge detection algorithm to enhance the edge extraction of thermal infrared images, which could then be presented to firefighters via an AR display. The specific algorithm process included the use of Gaussian filtering to smooth the image and reduce the impact of noise on the subsequent edge detection, adding filtering in six directions (0°, 45°, 90°, 135°, 225°, and 315°) to the traditional Scharr operator, generating multiple edge images, and performing maximum value synthesis and normalization to ensure consistency and clarity of the edge information. The experimental results showed that the proposed algorithm meets the real-time and practical application requirements. Compared to the traditional Scharr algorithm, the proposed algorithm improved the structural similarity and feature similarity indices by 15% and 7.6%, respectively, and reduced the mean squared error by 3.3%.