In power systems, to use computer vision and image processing technology to detect zero-value insulators is very important for the safe operation of power systems. An automatic zero-value insulator detection algorithm based on infrared images is proposed. Firstly, an input image is preprocessed and the insulator is positioned precisely by combining the GLOH descriptors with machine learning. Then, the boundary points in the image are extracted by using a graphic method. The initial contour searched is used as the initial value. A GVF Snake model is used to check the precise contour of the insulator string so as to separate the insulators. Finally, automatic zero-value insulator detection is realized by analyzing the statistical histogram. The experimental results show that the proposed algorithm can detect the zero value insulators effectively.