中文摘要：

提出了一种新的道路交通自适应背景更新算法，该算法采用虚拟线圈检测车辆平均速度，并按平均速度将交通状态区分为“畅通”和“拥堵”，以此作为背景更新的切换条件；当车辆处于畅通状态时，主要以光照变化作为背景更新的条件，并采用混合高斯模型与连续帧间差背景更新模型相结合的融合模型进行更新；当车辆处于拥堵状态时，则不进行更新；算法中引入了背景更新的评价函数，使得背景更新能够自动停止；实验结果表明该算法具有传统背景建模的优点，并且提高了实时更新的效果与减少了计算量。

英文摘要：

It proposes a new algorithm of adaptive background update for traffic road. This algorithm uses method of virtual loops to detect average speed of vehicles, divides traffic situation into ＂smooth＂ and ＂congestion＂ by average speed of vehicles, and using traffic situation is regarded as condition for background update. On the situation of smooth, according to the situation of light changes, using integrate model combined Gaussian mixture model and background subtraction updating model updates the background, On the situation of congestion, back-ground is not updated. And this algorithm also introduces the evaluation function of background updates, make background update automatically stop. Experimental results show that the technology has the advantages of traditional methods, improve effect of real--time update and reduce time of computation.