摘要
将人体运动区域检测技术应用到矿井视频中可以检测矿井下矿工的运动情况,进一步可以智能检测矿工的异常行为,根据反馈的检测结果实现实时报警和联动控制,减少矿井事故的发生。针对矿井场景下的人体运动区域检测,提出了一种实现人体运动区域提取的融合方法 TD-HF(Time Difference and Haar Feature),该方法融合了时间差分法和基于Haar特征的人体检测算法。实验表明,所提方法在检测率和误识率方面均比单纯的基于AdaBoost算法的分类器更胜一筹,并且在检测时间上满足实时性要求,适用于矿井视频这种特殊场景下的人体运动区域检测。
The human motion area detection technology applied to the mine video can detect motion of miners and intelligently detect abnormal behavior of miners through further analysis.According to the results of feedback detection to achieve real-time alarm and linkage control,it obviously reduces the occurrence of mine accidents.This paper proposed a hybrid method TD-HF(Time Difference and Haar Feature)for extracting human motion area,which integrates the time difference method and the human detection algorithm based on Haar feature especially under the condition of mine.The experiment shows that this method is better than the simple classifier based on AdaBoost algorithm in the detection rate and false recognition rate,at the same time,it can satisfy the real-time requirements in detection time.It’s applicable to the detection of human motion area under the special condition of mine video.
作者
李珊
饶文碧
LI Shan ;RAO Wen-bi(School of Computer Science and Technology,Wuhan University of Technology,Wuhan 430070,China)
出处
《计算机科学》
CSCD
北大核心
2018年第4期291-295,共5页
Computer Science
基金
国家自然科学基金项目(61601337)
湖北省重大科技创新计划项目(2015BCE068)资助
作者简介
李珊(1993-),女,硕士生,主要研究方向为数据挖掘、图形图像处理,E-mail:chonger@whut.edu.cn;(通信作者)饶文碧(1967-),女,教授,博士生导师,主要研究方向为普适计算、机器学习与数据挖掘、计算机图形图像处理方法与技术,E-mail:wbrao@whut.edu.cn。
