中文摘要：

在海量的监控视频中,快速、准确地识别车辆对公安破案和追踪具有重要的研究意义。通过提取车辆的类Haar特征,采用AdaBoost方法构建分类器可以实现监控视频中的车辆识别。针对原始算法误检率较高的问题,提出了采用背景差分去除背景干扰,以及采用目标对象差分法进行二次识别的两种改进算法。实验结果表明,两种改进算法都能够有效地降低误检率,提高检测率,并且对不同交通场景下的监控视频具有很好的检测效果。

英文摘要：

It is quite important for solving crimes and tracking the suspect to find the vehicle quickly and accurately from huge volume video records.By extracting Haar-like features and adopting AdaBoost algorithm to construct classifier,one can identify vehicle in surveillance video.Aiming at the high false alarm rate of the original algorithm,this paper proposes two improved methods：the one adopts the background difference algorithm to remove the interference of background,and the other utilizes the target object difference algorithm to achieve the second identification.Experimental results have shown that the proposed algorithms can reduce the false alarm rate and improve the detection rate.Moreover,the two algorithms have better detection results for surveillance videos in different traffic scenes.