Research On Infrared Image Segmentation And Recognition For Substation

With the increasing power grid capacity and the voltage level upgrades,power supply reliability becomes more and more important. The monitoring and alarming on the temperature of electrical equipment in substation play an important role in ensuring the safety of the power grid. This paper researches the pre-image processing technique used in the substation monitoring system. By segmenting and recognizing the infrared image of the substation, necessary features and data are easy to extract. Thus, providing information for the expert system.According to the features of substation infrared image, such as low SNR and low contrast, the paper proposes two modified methods. One is edge detection and mathematical morphology combined segmentation method, the other is the regional growth method of color image. The former method focuses on the description of the image contours. The latter is easy to detect the high temperature region and extract the graphic of the electrical equipment. The experiment results show that the latter method is fit for the application in substation.By starting with the recognition method of template matching, the paper figures out the limitation of existing matching method. In order to apply the method in the fact, the paper recognizes the image based on its invariant moment which is independence of translation, scale and rotation. The Hu and Zernike moment is analyzed, and a lot of moment values are constructed by using higher order characteristics of Zernike moment. It's proved that the Zernike moment could be preferably used in the recognition in substation infrared image by correlation coefficient.