Research on algorithm of automatically recognizing and positioning road manhole covers based on vehicle-mounted sensors
基于车载传感器的路面井盖自动定位识别算法研究*

The fast recognizing and positioning of municipal manhole covers is an important problem needed to be addressed for promoting digital management of modern cities. In view of the above problem, this paper proposed an algorithm, to automatically recognize and position road manhole covers under complex background in natural scene based on vehicle-mounted sensors. Taking the elliptical geometrical characteristic of manhole cover in perspective image captured by vehicle-mounted sensors as criterion, the algorithm firstly extracted edge information by employing vector edge detection method. Secondly constructed a contour list with boundary through contour tracing, then imitated and quickly generated all the possible elliptical targets in the contour list by means of least square fitting method. Subsequently eliminated the elliptical targets without relationship corresponding to road manhole covers according to their shape features, and last formed accurate results of recognition and position. Experiment shows that generally the algorithm is able to achieve real-time manhole cover targets recognition rapidly and effectively for images conforming to quality standard of data capturing.