In order to solve the problems of fuzzy detail and low contrast of infrared image of electrical equipment, an infrared image processing method based on BEMD(Bidimensional empirical mode decomposion algorithm,BEMD) and bat adaptive nonlocal mean denoising is proposed. Firstly, the original infrared image is decomposed by BEMD. The image can be divided into high-frequency component (bimf) and low-frequency component (residual component). Bat adaptive nonlocal mean denoising is adopted for high frequency component, and only the details and edges of the image are preserved. The low frequency component contains the information and details of the original image, which is linearly enhanced. Finally, the components are reconstructed by inverse BEMD. After comparison and application, this algorithm can effectively highlight the hot area in the infrared image of power equipment, which is convenient for earlier fault detection and fault location.