Abstract: Flashover can be reduced if insulator contamination grade is accurately discriminated. Infrared images and visible images show temperature and color features of insulators respectively, which describe insulator contamination grades from different perspective. This paper presented a method to discriminate insulator contamination grades, using information fusion of infrared images and visible images. Firstly, heating models of contaminated wetted insulators were established and solved by numerical analysis method. From the models, the surface temperature distribution was found out. Secondly, infrared and visible insulator images were obtained in laboratory. Thirdly, after image processing, temperature and color features were extracted and screened by Fisher criterion, after which feature vectors consist of relative humidity, illumination and the screened features were obtained. Finally, Bayes theorem was used for information fusion on feature level with the feature vectors, and insulator contamination grades were discriminated. Experimental results show that information fusion of infrared and visible images achieve much higher accuracy in insulator contamination grades discrimination. Furthermore, on-site tests verify the feasibility of this method. This paper provides a new and workable method for accurate discrimination of insulator contamination grades.