基于视频图像处理的交通流检测方法

针对视频交通流的特点，提出了一种无需进行视场标定的交通流检测方法。该方法选用检测线上的点及其近似点作为检测区域，通过对背景模板上检测区域中的突变点的分析和统计，可以实时地检测车辆的驶入、存在和驶出，提高了检测速度。在此基础上，设计和实现了视频交通流信息采集系统，并用软件实现检测算法和图像处理算法。试验结果表明，其检测速度和检测精度都能够满足实际要求，达到了用最小的系统规模实现交通流信息采集的目的，能够为交通管理与控制提供信息数据。

Traffic flow detection method based on video images processing

Based on the characteristics of video traffic flow, a new method without the calibration of vision field about traffic flow detection is put forward. This method selects the points on the detection lines and the points near it as detection area to improve the detection speed. It can detect the vehicles that are going into, being in and going out by analyzing and counting change-pixel in virtual detection area of the background image. A video traffic flow information collection system is developed based on this method, its detection arithmetic and image processing arithmetic are designed in software. The experiment results indicate the detection speed and detection precision of the system can satisfy the real requirements. This system can provide useful information for traffic management, control and correlation research. 2 tabs, 5 figs, 6 refs.