客流数据的实时采集是交管部门对车辆管理研究的关键问题之一，利用图像处理方法来进行人群监测在国内外已有研究，但在复杂场景的监测并不令人满意。本文利用普通CCD采集图像，综合帧间图像差异和消除背景图像的处理技术来有效检测图像中的运动人群，克服传统方法中存在的测量误差，并依据所检测的前景图像区域面积与整个图像有效面积的比例来估测当前图像中人群数目-Real-time traffic data collection is the traffic control department of vehicle management research one of the key issues, using image processing methods for crowd monitoring study at home and abroad have been, but in the monitoring of complex scenes is not satisfactory. In this paper, ordinary CCD acquisition image, an integrated inter-fr a me image differences and eliminate the background image of the technology to effectively deal with detection of image movement groups, to overcome the traditional methods that exist in the measurement error, and based on the prospects for the detection of images with the entire image area of effective area in proportion to estimate the current number of images in the crowd