Localization of Multiple Power Line Insulators Based on Shape Feature Points and Equidistant Model in Aerial Images

Insulators are important equipments on the power transmission line, and the failure may cause severe influences to power system. So localizing insulators in aerial image to avert further damage is an important research topic for automatic fault diagnosis. In order to localize multiple insulators with variant directions in aerial image, this paper proposes a novel method to extract shape points, summarizes and utilizes the in-a-line constraint and equidistant constraint of shape feature points, and proposes a localization method of insulators based on shape points and equidistant model. The localization method extracts shape points, detects the possible direction of insulators based on in-a-line constraint by RANdom SAmple Consensus (RANSAC), proposes the equidistant model to obtain the positive points of insulator, and localizes insulators in aerial image effectively. Compared with other methods, the proposed method is 3 times faster than previous methods, and can obtained 33.85% more efficient points. Experimental results show that the proposed method achieves favorable effects in the localization of insulators in real aerial images.