红外技术, 2020, 42 (3): 231, 网络出版: 2020-04-13
基于红外阵列传感器的人体行为识别系统研究
Human Behavior Recognition System Based on Infrared Array Sensors
摘要
随着人口老龄化的到来, 为了避免发生意外事故, 对老人日常活动行为进行识别和监测的安全监护系统的需求不断增长。传统的基于摄像头拍摄或者穿戴式传感器的活动状态监测系统存在着隐私保护和使用不方便等不足。为此, 本文设计一种基于红外阵列传感器的人体行为识别系统。该系统通过检测环境中的温度分布和变化情况识别人体行为, 不需要在老人身上佩戴任何设备, 尺寸小易于安装, 在黑暗环境中可正常工作, 且由于采集到的是低分辨率信息, 不会造成隐私泄露, 对比传统方案具有明显优势。从采集到的温度分布信息中提取特征并采用 K最近邻( K-Nearest Neighbor, KNN)算法实现了“走”、“坐”和“跌倒” 3种状态的识别。实验结果表明平均准确率可达到 95%, 其中跌倒准确率为 97.5%, 行走准确率高达 100%, 坐下准确率为 92.5%。
Abstract
With the increase in the aging population, the demand to identify and monitor the daily activities of the elderly is growing. A monitoring system can effectively prevent accidents of elderly people. The traditional activity monitoring system based on the use of camera or wearable sensors has issues, such as privacy violations and inconvenience of use. Therefore, this study designs a human behavior recognition system based on infrared array sensors. The system recognizes activities on different temperature distributions in the environment. There is no need for the sensor to be worn by the elderly. The sensor is small in size, easy to install indoors, and can work in the dark. In addition, the data acquired by the sensor have a low resolution; therefore, there is no privacy violation. The designed system has significant advantages over the traditional systems. The features are extracted from the obtained temperature data, and the K-nearest neighbors algorithm is used to identify the three poses of “walking,” “sitting,” and “falling.” The experimental results show that the average accuracy can reach 95%, of which the accuracies for falling, walking, and sitting are 97.5%, 100%, and 92.5%, respectively.
王召军, 许志猛, 陈良琴. 基于红外阵列传感器的人体行为识别系统研究[J]. 红外技术, 2020, 42(3): 231. WANG Zhaojun, XU Zhimeng, CHEN Liangqin. Human Behavior Recognition System Based on Infrared Array Sensors[J]. Infrared Technology, 2020, 42(3): 231.