Moving Vehicle Detection and Tracking using Discriminative Robust Local Ternary Pattern Edge Extraction for Traffic Surveillance

An important aspect of traffic monitoring is the traffic surveillance. Vision based traffic surveillance has become the most popular method to perceive the traffic flow. This paper provides an effective way for detecting and tracking vehicles from a video. Primarily background subtraction is done to detect the foreground objects from the video using frame differencing method. Then the edge is extracted by Discriminative Robust Local Ternary Pattern (DRLTP) for removing shadows in the foreground and to detect objects from illumination variation, and the edge of the moving vehicle is detected. Finally the vehicle is detected using Histogram of Oriented Gradient (HOG) and Relative Discriminative Histogram of Oriented Gradient (RDHOG). Then the detected vehicle is tracked using particle filter.