Wear Detection of Metro Catenary Based on Binocular Vision

Abstract

Compared with the traditional catenary wear detection method, the new non-contact image detection method has the characteristics of high efficiency and high precision. In this paper, binocular vision 3D reconstruction is used to detect the abrasion of metro catenary, and in the process of self-calibration and stereo correction, topological structure constraints are proposed to improve the accuracy of the SURF feature point matching. The experimental results show that the improved feature point matching has high accuracy and good real-time performance.