Velocity Estimation of Metro Train Using Hough Transform Based Feature Extraction

In this paper, a Hough transform based feature extraction in image is proposed to evaluate the velocity of a metro train, which is a fundamental quantity for the metro safety monitoring system. A rough trajectory of the metro train is acquired from its waterfall graph constructed by embedded fiber optical arrays. To block non-linearity of trajectory induced by time-varying velocity, its entire trajectory is divided into several linear sections, where straight-line features are obtained by Hough transform. By fitting the endpoints collected from straight-line into a polynomial curve, the metro train's trajectory is approximated and its velocity is estimated as well. The experimental results detected in Wuhan Metro Line 7 demonstrate feasibility of our method.