摘要
提出一种可用于入侵防范系统中的步态识别方法。该方法以足底压力信息为基础,采用卷积神经网络模型进行步态特征提取。首先,用压力测试板采集行人的足底压力信息并作相应的预处理;然后,结合K均值聚类和卷积神经网络方法的自学习特性得到足底压力信息的特征表示;最后,对所获得的特征表示进行分类识别。在典型数据集上的比较试验表明了该算法的有效性。
A gait recognition algorithm for intrusion prevention is presented. This method is based on plantar pressure information, adopting convolution neural network model to extract teature. Firstly, the preprocessmg oI me evaiuateu data collected from the test board of plantar pressure is performed. Secondly, the deep learning technique and k-means Clustering are used to automatically extract features from the plantar pressure distribution diagram. Finally, we use the features for the recognition task. The comparative experiments conducted on the typical datasets demonstrate the effectiveness of the proposed approach.
作者
杨春生
化晨冰
王鸣镝
黄振华
邵晓东
过其峰
Yang Chunsheng Hua Chenbing Wang Mingdi Huang Zhenhua Shao Xiaodong Guo Qifeng(Linyi Power Supply Company,Linyi 276003, Shandong, China Anhui Jiyuan Electric Network Technology Co. , Ltd of NARI, Hefei 230088 ,Anhui, China)
出处
《计算机应用与软件》
2017年第3期248-251,共4页
Computer Applications and Software
关键词
入侵防范
足底压力信息
卷积神经网络
特征学习
步态识别
Intrusion prevention Plantar pressure information Convolution neural network Feature extraction Gait recognition
