模糊ISODATA聚类结合直方图熵值算法的异常行为检测

摘要

针对传统智能视频监控系统不能及时准确地发现危险或者发现危险不能及时报警的问题,提出一种基于模糊迭代自组织数据分析聚类结合直方图熵值算法的异常行为智能检测方法.该方法通过模糊迭代自组织数据分析聚类方法获取视频关键帧,根据分类结果采用直方图熵值法对异常行为进行判断.实验结果表明,所提算法可以对具有复杂背景的监控区域有效地实现人体的检测,并且能准确地识别出人体异常的动态行为,有效地减少了住宅小区和养老院的安全隐患.%Since the traditional intelligent video surveillance system can′ t discover the danger accurately and send the alarm timely,an abnormal behavior intelligent detection method combining fuzzy iterative self-organizing data analysis(ISODATA) clustering algorithm with histogram entropy method is proposed. The key frame of video is acquired with fuzzy ISODATA cluster-ing method. According to the classification result,the histogram entropy method is adopted to judge the abnormal behavior. The experimental results show that the proposed algorithm can detect the human body in the monitoring area with complex back-ground effectively,identify the abnormal dynamic behavior of the human body accurately,and reduce the safe hidden trouble in residential district and nursing home effectively.