基于Canny边缘帧差法的公交客流量统计

Counting for passenger flow based on Canny edge frame subtraction algorithm

为了提高公交客流高峰期时的客流统计准确率,提出了一种基于机器视觉的客流统计方法。该方法是以对乘客的头部轮廓提取边缘信息后进行帧间差法做为主要手段,对于出现的多目标相连的情况,利用适应的投影法对其进行多目标分割。计算各目标的中心位置、面积等特征值,建立目标跟踪链,分析目标运动特点,并最终获得乘客上下车的个数。通过实验表明,应用该方法进行公交客流统计,大大提高了准确率。

To improve the accuracy rate of passenger flow counting during rush hour,a machine vision-based method for counting passenger flow was presented.Pick up the edge information of passenger＇s head contour,then do the frame subtraction is the main instrument of the method.As to connected target,projection method was adopted to separate them.Figure out each target＇s feature values such as central coordinates,area,establish tracking chains,analyze moving-target＇s features and finally get passenger number.The result of test shows that the accuracy rate of passenger flow counting is enhanced using this methods.