A method of background extraction and update for vehicle detection is proposed

which is special for the saturated intersection. According to the characteristics of traffic flow in intersections

a background evaluation mechanism is established by using the feedback of vehicle detection result. Then the true background image is updated from the noisy initial background iteratively combining evaluation mechanism

Median Filter and Inter-frame Difference algorithm. The experiments show that the algorithm has a faster convergence rate in saturated intersection

and satisfy complex engineering needs.