Insulator Recognition Method Based On Features Of Infrared Image

The reliability and operation of transmission and substation equipments directly affects the stabilization and security of electrical power system. Therefore, monitoring their status is a crucial process. Insulator is indispensable element in transmission and substation system, and its status affects the security and reliability of the power grid. Detecting status of insulator is especially important.Aiming at the characteristic of insulator in infrared image, this paper designs two recognition methods of insulator based on feature of infrared image, and realizes the localization of insulator. The major work contains two sections including the recognition method of insulator based on contour feature and the recognition method of insulator based on points matching.In the recognition method of insulator based on contour feature, this paper proposes the localization method of insulators two times. Firstly, design sliding window strategy combination with Hu moment character to recognize insulators preliminarily pointing at the characteristic of multiple insulators with different directions in infrared image. Then, aiming at the problems of original GHT, such as large storage space cost, slow operational speed, and redundant vote and so on, this paper votes by FAST feature points instead of original edge points, and uses GHT to realize precise recognition which can decrease vote accumulation of invalid points, reduce storage space and improve operating efficiency. Experimental results show that this method is beneficial to the recognition of insulator in infrared image with multiple insulators with different directions.The recognition method of insulator based on contour feature can not achieve good effect for the recognition of insulator in image with poor definition and less obvious target contour. This paper proposes and realizes a recognition method of insulators based on points feature. Extract FAST feature points of image, describe FAST points by DAISY, match template and test image by matching algorithm, and finally recognize and localize targets by affine transform.