Research on vehicle object detection in traffic video stream
Xingqiang Zhang · 2010
A vision based method of vehicle detection was proposed. Firstly, a improved common region algorithm to realize the background subtraction which having a better robustness and easy to determine the background image real timely. Then a threshold segmentation method is used in object extraction stage. Compared with the existing algorithm, such an iterative algorithm based on weighted average of thresholds achieved improved accuracy on background image and increased the stability of the object extraction.