A leak detection and location system for SF6 based on infrared image is designed and developed. By using the high speed microprocessor TMS320C6416 and its configurable video ports and combined with the powerful control capability of the S3C6410, the real-time infrared image where SF6 leakage may occur is collected and processed. Combine with the frame of difference algorithm and the local entropy algorithm, detect and locate leaks. Experimental results show that the algorithm not only effectively improves the precise detecting location of frame difference algorithm and reduces calculated amount of local entropy method, but also effectively reduces the effect of diagnosis process with object shape, surface defects and environmental background noise etc, which can efficiently and accurately implement SF6 leak detection and location.