Vehicle detection and tracking based on optical field

Abstract

Vehicle detection and tracking are of great significance on computer vision and practical applications. The main task of it is to pick out vehicles from the images in a realtime video and tag them so as to achieve goals like traffic flow calculation and the driving direction estimate. In this essay, we chose Horn-Schunck method based on optical field to detect the vehicles. Without knowing any background information, this method can precisely process the video in a real time, picking out the statistics of the moving car and count them. The algorithm used in the essay can achieve the goal of vehicle detection and tracking, calculating and show vehicle flow precisely and avoid the interference of pedestrians and other irrelevant factors.