摘要
摔倒检测是视频监控系统应用研究中的一个重要问题,快速有效地获得摔倒信息可使摔倒人得到及时救助,降低进一步的伤害.提出一种改进的自动摔倒检测算法,该算法在利用人体宽高比特征判断的基础上,采用有效面积比和中心变化率两个特征对判断结果进行修正,有效地防止误判,提高了检测结果的正确率.该算法计算复杂度低,易于实现.实验结果表明该算法具有较好的鲁棒性.
Fall detection is an important problem in the application research of video surveillance. Quick and effective ways of getting the information of the people falling to the ground can help the people get timely assistance to reduce further injury. This paper proposed an improved algorithm of automatic fall detection. Based on using human aspect ratio features to judge human movements, two features--effective area ratio and center variation rate were adopted in this algorithm to modify detection results. Misjudgments could be effectively prevented and the accuracy of detection results could be greatly increased. This algorithm had a less computing complexity and was easy to implement. Experimental results showed that the algorithm had good robustness.
出处
《安徽大学学报（自然科学版）》
CAS
北大核心
2012年第6期57-61,共5页
Journal of Anhui University(Natural Science Edition)
基金
安徽省教育厅自然科学基金重点资助项目(KJ2009A60)
安徽大学博士科研启动基金资助项目(33190049)
关键词
视频监控
摔倒检测
人体宽高比
有效面积比
中心变化率
video surveillance
fall detection
human aspect ratio
effective area ratio
center variation rate
