摘要
随着对信息安全的要求越来越高,对变电站巡检检修及相关运维工作过程中区域入侵防范的技术研究就很有必要。提出了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;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;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。
With the increasing requirement of infcirmation security, it is essential to research the regional intru- sion prevention technology in the proces of substation inspection, maintenance and the related operations. This paper proposes a gait recognition method used in the intrusion prevention system. This method is based on foot pressure information, and extracts the gait feature by means of convolution neural network model. First use stress test board to collect pedestrians＂ plantar pressure information and make the corresponding pretreatment; obtain the characteristic expression of plantar pressure information in combination with K-means clustering and convolution neural network self-learning characteristics; tures. Comparison experiments on typical data sets show make classification recognition of the obtained fea the effectiveness of this algorithm.
出处
《电力与能源》
2016年第2期211-214,共4页
Power & Energy
关键词
入侵防范
足底压力信息
卷积神经网络
intrusion prevention
plantar pressure in^ormation
convolution neural network
作者简介
杨春生（1985），男，硕士，工程师，从事带电作业管理工作。
