Traffic surveillance cameras are located in urban traffic network system in abundance. These videos are mainly used for public safety and traffic accident monitoring. With the increase in urban traffic congestion, the road section queue detection technology becomes the insistent demands for urban traffic management. Considering the characteristics of the surveillance cameras in city intersection area, the paper used maximum similarity matching standard method which is based on the fixed background area window to correct the deviation between the lane region in real-time image collection and the lane region in setting, then adopted the AOI region partition method for mapping the distance in video image to the actual distance, at the same time, based on the road section multi-lane queuing model analyzed the features of queuing lane. The veins' differences between having vehicle lane and empty lane are obvious, therefore the paper selected the vein unit to express traffic vein image. Simple vein unit reflected the difference on gray level in neighborhood, and effectively reflected the veins' differences between having vehicle lane and empty lane. After removing the noise, the empty lane region at the end of queue could be detected, and the real-time length of the road section vehicle queue can also be calculated. The validity of the algorithm is validated by checking the actual intersection video images and actual queue length data.
展开▼
