If oil vapor leakage occurs in gas stations,it will cause enormous damage.Aiming at the characteristics of low concentration and dynamic development of oil vapor leakage from refueling gun,a method of oil vapor leakage detection based on infrared video is proposed.The video signal of the working area of the refueling gun is collected by infrared camera,and a mixed Gaussian background model is established.The suspected leakage image is obtained by combining frame difference method and background elimination method.Morphological and filtering operations are performed on the image to extract the shape and diffusion features of the frame image.And the signal noise interference is eliminated according to the characteristic range of oil vapor leakage.The mapping equation between image gray scale and oil vapor concentration is established.And according to the relevant provisions of the national oil vapor leakage concentration standard,whether the leakage exceeds the standard is judged.In this study,water vapor was used as a simulated experimental object to test,and the results show that this method can accurately judge the leakage situation.