A Video-Based Method for Traffic Flow Detection of Multi-lane Road

A novel algorithm for video-based traffic monitoring of multi-lane road using multiple statistical parameters is proposed to tackle the illumination condition change and adjacent moving shadow problems. This algorithm uses normalized cross correlation (NCC) of the traffic video image as the main detecting feature along with the homogeneity based parameter as the secondary parameter. The traffic flow is calculated based on the decrease-increase variation of the main detecting feature in the observation window. Blocking processing and homogeneity feature are employed to distinguish the shadow and vehicle. The results of the video test suggest that the proposed algorithm has higher efficiency and accuracy as well as better robustness against illumination condition change and adjacent moving shadow comparing to the traditional approach.