一种新的基于实时视频流的车型特征提取算法

背景的实时建模、车辆存在判定及车型特征提取是基于实时视频流的车型识别系统中的关键性问题。本文首先从建模质量和建模速度两个方面对比分析了几种常用的背景建模算法。然后提出了一种新的车辆存在判定、特征提取算法“弹性松弛算法”，最后给出了实验结果。结果表明：该算法可以有效的进行车辆存在判定，并能够实时的提取车型长度特征，从而完成车型识别系统中的车辆计数工作，为车型识别奠定了良好的基础。

New Vehicle Recognition Algorithm Based on Real-Time Videos

In the vehicle recognition systems based on the real-time videos,the construction of the background,the judgment of vehicles existence and features extraction are important.Firstly,this paper compares several common background update algorithms,and points out the advantages and disadvantages.And then a new algorithm method to extract the vehicle features is proposed.The experiments show that the algorithm is very adaptive to the real-time environment,very sensitive to the existence of the judgment of vehicles has it has very good vehicle feature extraction and classification function.