智能视频的公交客流量统计算法

A statistical algorithm for bus passenger flow based on intelligent video

为了解决复杂背景及运动状态下城市公交客流量的准确检测、识别与统计问题,提出了运动矢量跟踪方法,研究了公交乘客头部运动特征及客流量统计算法.传统的帧间差分、边缘检测等目标检测算法很难处理紧邻的多目标情况.利用类圆形处理提取类似圆形的目标连通域,对目标特征点进行运动矢量跟踪,实现了上、下公交车人数的准确统计.

In order to achieve the accurate detection, identification and count of bus passengers under the complex background and the motion state, this paper adopted the motion vector tracking method, and studied the passenger head movement characteristics and traffic statistics algorithm. The traditional target detection algorithm, such as frame subtraction and edge detection, is very difficult to deal with the situation of immediate multi-objective. In this paper, the grid-density clustering and line segment clustering, extraction and segmentation target connected domain that is handled to be a circular, tracking the motion vector of target feature point, finally, achieving accurate statistics on the number of people getting on and off.