复杂环境下的车辆目标检测方法研究

基于视频触发方式对复杂环境下的运动车辆进行监测,利用800万像素高清摄像机24 h全天候检测车辆,最多可以检测5个车道。在传统虚拟线圈的画法基础上增添识别区和触发区,增大传输速率,这样可以省略图片处理和筛选过程,节约资源,提高采集环节的整体效率。采用背景消减法中可自适应的混合高斯模型原理,建立模型及模型后期更新,得到车辆目标检测结果,分析其实验结果得到：基于混和高斯背景差分法对于动态背景的建立十分有效。

Study on Vehicle Target Detection Method in Gas Station Complex Environment

In this paper,moving vehicles were detected in complicated environment of gas station based on video trigger mode. An 8 million-pixel camera was used to monitor the moving vehicles 24 hours a day,5 traffic lanes at most. Based on the traditional virtual coil brushwork,transmission rate was increased and storage space was saved by adding identifying zone and trigger zone,which can save the image process and selection process,and improve the entire efficiency of collection. The adaptive Gaussian mixture modeling method in background subtraction method was used to construct model and model update to obtain the detection results of vehicle targets. Results showed that the difference method based on Gaussian is effective in constructing dynamic background.