基于双目测距和YOLOv5s的红外人体测温精度补偿系统  

Infrared Human Body Temperature Measurement Accuracy Compensation System Based on Binocular Ranging and YOLOv5s

作　　者：彭泽龙 李铨 PENG Zelong;LI Quan(Jiangsu Normal University Kewen College,Xuzhou Jiangsu 221132,China)

机构地区：[1]江苏师范大学科文学院,江苏徐州221132

出　　处：《信息与电脑》2022年第7期80-83,共4页Information & Computer

摘　　要：针对不同距离下人体测温系统准确率低的问题,提出一种基于双目测距和YOLOv5s的人体测温方法。该方法首先通过YOLOv5s深度学习算法检测人脸,识别成功后,双目摄像头实时测算目标深度信息,经转换后输出三维位置数据补偿红外测量精度。改进后的YOLOv5s算法识别快速、准确,双目摄像头定位精确,红外热成像温度信息通过距离补偿后有效降低了测量误差,完全具备快速精准测量人体温度的能力。Aiming at the problem of low accuracy of human body temperature measurement system at different distances, a human body temperature measurement method based on binocular ranging and YOLOv5s is proposed. Firstly, the face is detected by YOLOv5s deep learning algorithm. After successful recognition, the binocular camera measures the target depth information in real time, and outputs the three-dimensional position data after conversion to compensate the infrared measurement accuracy. The improved YOLOv5s algorithm has the advantages of fast and accurate recognition, accurate positioning of binocular camera, effective reduction of measurement error of infrared thermal imaging temperature information through distance compensation, and fully has the ability of fast and accurate measurement of human body temperature.

关 键 词：YOLOv5s 双目测距 红外热成像测温 误差补偿 

分 类 号：TN219[电子电信—物理电子学]