Abstract: To deal with the low efficiency and low accuracy of current contact network geometric parameter detection equipment, a dynamic detection system of OCS geometric parameters has been developed using binocular 3D vision imaging, structured light image acquisition, and image processing. This system mainly consists of a OCS image acquisition device, a carboy offset image acquisition device, an incremental encoder, an image data conversion device, a host, and detection software. Installed on a vehicle at a speed of 0～120 km/h, it can achieve dynamic non-contact detection of the OCS's height and pull-out values and under strong light or low light, and automatically identify anomalies in the height and pull-out values of the OCS, generate anomalies report and annotated document to guide the fine adjustment and daily inspection of the OCS. Its equipment is lightweight, easy to install, and has effective calibration methods. A testing platform is built in the laboratory to conduct tests on lead height, pull-out value, and vehicle vibration displacement correction, and conducted preliminary dynamic tests on a operating line of the Beijing subway. The test results show that the system can stably and dynamically collect high-definition images, with measurement accuracy errors of less than 5 mm for the height and pull-out values of the overhead contact system. It can meet the needs of precision adjustment operations for newly constructed OCSs of railway, subway, or light rail lines as well as daily inspection operations for the OCSs of operating lines.