基于视频的车流量检测建模与分析

交通流量信息是智能交通控制中的关键成分。论文在对视频图像进行灰度处理、频域滤波和均衡化的基础上，结合直方图分析、背景差分、动态阈值二值化处理等图像分析过程，提出了双虚拟线平均法检测多车道车流量，并完成参数灵敏度检验。针对初识背景提取时间过长的问题，加入了跳帧法和像素块法，以提高系统效率。考虑到背景的时变性，采用选择更新法进行周期更新。实际结果表明，该方法兼顾了智能交通系统的环境适应性、检测实时性和统计精确性等要求，四个车道车流量统计的准确率均大于90％。

Modeling and Analysis of Vehicle Flow Detection Based on Video

Traffic flow information is the key component of intelligent traffic control .Based on gray processing ,fre-quency domain filtering and equalization for video ,a method with application of double virtual test lines averaging was pro-posed to detect multi-lane vehicle flow .Combining with histogram analysis ,background difference ,dynamic threshold bina-rization and so on ,parameter sensitivity testing was added .Contrary to the problem of taking long time to extract initial background ,the ways of skipping frames and using pixel blocks were adopted to improve the system efficiency .In addition , considering the time variation of background ,the selecting method was chosen to update it periodically .The actual results showed that the method had a detecting accuracy of 90% ,with environmental adaptability ,real-time detection and statistical accuracy taken into account .