Most of electrical equipment failure are associated with overheating and have the characteristics of regional in power systems, temperature and grey value has the nonlinear mapping relationship in infrared image. In this paper, in order to extract image edge features, the infrared image would be preprocessed, then we can achieve identification mark of higher image gray value and automatic achieve the positioning of electrical equipments high temperature area by the topology matrix correction, and then using relative temperature difference method to judge electrical equipment failure. Experimental results show that this method can quickly and effectively locating and identify suspected overheat fault of electrical equipment.