一种基于改进的卡尔曼滤波的人群密度检测算法

A Crowd Density Estimation Based on Improved Kalman Filtering

摘要: 提出了一种改进的基于卡尔曼滤波技术的人群密度检测算法，该算法通过减背景法识别人群，使用改进的基于卡尔曼滤波技术进行背景更新，通过阈值统计的方法增强识别出的前景人群的效果，最后在此基础上基于像素统计信息进行人群密度计算。实验结果表明，该算法实现简单，可以有效的解决卡尔曼滤波技术更新背景时的空洞现象，比较准确地计算出人群密度。

Abstract: The crowd density estimation is the basis for the abnormal behavior detection of crowed scene or group events in intelligent video surveillance system. A crowd density estimation algorithm is given which is based on improved Kalman filtering. The given algorithm will detect crowd by using background subtraction, and update the background by the improved Kalman filtering, and then it will enhance the foreground crowd by statistic method. Last, it will calculate the crowd density. The experimental result shows that it is simple and can solve the holes in the background caused by Kalman filtering effectively, and the crowd density calculated is accurate.