改进Hu矩的ATM机异常行为识别研究

提出一种基于改进Hu矩和隐马尔可夫模型相结合的ATM机异常行为识别方法。对ATM机前用户存(取)款行为的视频序列用改进Hu变换提取运动目标的行为特征,采用Baum-Welch算法对用户的正常行为进行训练,并建立隐马尔可夫模型;最后通过模型输出测试样本序列的概率来识别异常行为。采用Matlab对ATM机用户运动行为的模拟视频进行实验仿真,结果表明:该方法对ATM机前的用户行为具有较高的识别率。

Research on Abnormal Behavior Recognition of ATM Based on Improved Hu Transform

This paper presents a recognition approach about ATM abnormal behavior based on the improved Hu transform and Hidden Markov Models.It extracts behavior characteristics of moving object by the improved Hu transform,trains and builds Hidden Markov Models of normal behavior by the Baum-Welch algorithm.Finally it can recognize whether the ATM user’s behavior is abnormal behavior or not by the output probability of test sample sequence.The simulation video of ATM users’ behavior simulatd with Matlab software.The results show that the method can improve recognition rate in the scenes of ATM.