ATM Intelligent Video Surveillance Based on Hand-track Recognition
We present a method based on hand-track recognition in video surveillance of Automatic Teller Machine (ATM).What it solves is that abnormal hand movements are recognized automatically in ATM,i.e.password theft.The ATM camera can detect whether there exists any abnormal activities by analyzing human hand movement in videos.As to video image,Gaussian-mixture background modeling method is adopted to establish a background model.Human body information is obtained through background subtraction and tracking algorithm.Abnormal behavior will be captured by the changes proportion of human body area.As to the video image without abnormal behaviors,we adopt skin detection algorithm based on color space and location constraint to detect hand moving track.By identifying segmented hand moving trajectory one by one through Hidden Markov Model (HMM),the camera can decide whether any abnormal behavior exists.At last,the experiment indicates that the rate of identification reaches more than 88％.
ATM Abnormal behavior Gaussian-mixture background modeling Hidden Markov Model hand-track recognition
Qiong Chen Bin Yu
School of Computer Science and Technology Xidian University Xian,China
国际会议
太原
英文
557-561
2011-02-26（万方平台首次上网日期，不代表论文的发表时间）