交通视频采集后的图像预处理。由于硬件及环境等因素的影响，摄像头提取的图像不可避免地包含噪声，图像的质量会受到不同程度的失真。因此，我们要对采集到的视频序列进行图像的预处理。

车辆检测算法。概括性的分析了基于图像处理的几种常用的车辆检测算法：帧差法、光流法，总结了各种方法的优缺点，并采用背景差法对车流量进行检测。该算法自适应能力强，计算量小，可正确判断有无车辆、完成车辆计数，实现车流量计算，为交通监管系统提供实时有效的交通参数。

车辆计数。分析常用的车辆计数方法：虚拟线圈法和目标跟踪法。并结合具体的实际应用采用基于检测带的车辆计数方式。结果表明，该方法能很好的进行车流量统计。

-Image pre-processing traffic after video capture. Due to hardware and environmental factors, the camera inevitably contain the extracted image noise, image quality will be subject to different degrees of distortion. Therefore, we want to capture the video sequence image preprocessing.

Vehicle detection algorithm. General analysis of several common vehicle detection algorithm based on image processing: frame difference, optical flow, summed up the advantages and disadvantages of each method, and using background subtraction to detect traffic flow. The algorithm is adaptive ability, a small amount of calculation can correctly determine whether the vehicle, the vehicle should count to achieve traffic flow calculation provides effective real-time traffic parameters for a traffic monitoring system.

Vehicle counting. Analysis of common vehicle counting methods: virtual coil method and target tracking method. Combined with concrete and practical applications based on vehicle detection with