Joint horizontal-vertical enhancement and tracking scheme for robust contact-point detection from pantograph-catenary infrared images

It is a key point to accurately and stably detect the pantograph-catenary contact point for its geometric parameter measurement from infrared images under complex backgrounds (e.g., cloud, cross-bridge, curved contact wire, in particular, the contact wire switching time). In order to complete the task of accurate and robust contact point detection, this paper proposes an progressive detection strategy, called as joint horizontal-vertical enhancement and tracking (JHVET), including three crucial parts: Firstly, an input infrared image is decomposed into a horizontal image layer and a vertical image layer by a horizontal-vertical enhancement operator, and the potential contact point is located by an extensive random sample consensus (RANSAC) algorithm. Secondly, an updated tracking approach is proposed to deal with the contact point detection problem in the contact wire switching time. Finally, an initial scheme is proposed to determine the first contact wire after switching the contact wire. Experimental results verify the effectiveness of the proposed JHVET method in both quantization and qualification. Especially, the performance with high detection accuracy (98.23%), low average pixel error (0.523 pixel), and satisfactory detection rate (over 108 fps) yielded by the proposed JHVET method is very suitable for its extensive application.