行为分析下敏感区域人群异常检测方法仿真

Based on the Analysis of the Behavior Anomaly Detection Method of Sensitive Area Population Research

对敏感区域中人群的异常行为进行准确检测,对于维护公共安全和社会稳定方面具有重要的意义。传统的算法对于图像采集设备过于依赖,只是简单的对人群行为特征进行提取,无法进行准确分类,从而影响了检测的准确性。为此,提出一种基于加权欧式距离行为分析的敏感区域人群的异常检测方法。对不同的人群异常行为特征赋予相应的数值,并进行规范化处理,能够建立人群异常行为的特征库,从而为敏感区域中人群的异常检测提供了准确的依据;根据欧几里得距离法的原理,对人群异常行为与异常行为特征数据库进行比较,从而实现对敏感区域人群异常行为的检测。实验结果表明;利用改进算法能够对敏感区域人群异常行为进行准确检测,效果令人满意。

This article proposes a method for crowd abnormal detection in sensitive area based on behavior analysis of weighted Euclidean distance. Our research gave relevant value to different abnormal behaviors of crowd and carried out normalized process, then built feature library of the abnormal behavior to provide accurate gist for the abnormal detection. According to theory of Euclidean distance, the research compared the abnormal behavior with the feature library. Thus, we achieved the crowd abnormal detection. Simulation results show that the modified algorithm can detect crowd abnormal behavior accurately. The effect is satisfactory.