Automatic traffic monitoring by -OTDR data and Hough transform in a real-field environment.

In this paper, we demonstrate automatic vehicle detection and counting by processing data acquired using a phase-sensitive optical time-domain reflectometer (-OTDR) distributed optical fiber sensor. The acquired data are processed using the Hough transform, which detects the lines in the images formed by representing the acquired data in the space-time domain. A rough classification of the vehicles (heavy versus light vehicles) is also proposed, based on the amplitude of the vibration data along the detected lines. The method has been experimentally tested by performing -OTDR measurements along a telecommunication fiber cable running in a buried conduit along the state road SS18 (province of Salerno, Italy), opened to normal traffic. Comparison with ground-truth data, manually generated by inspecting video recordings, allowed us to estimate a vehicle detection success rate up to 73%, while heavy vehicles were fully detected.