基于热红外图像的服务器设备检测  

Detection of server equipment based on thermal infrared image

作　　者：刘航[1] 谢婷 冉建 宋先霖 王维娜 LIU Hang;XIE Ting;RAN Jian;SONG Xian-lin;WANG Wei-na(School of Electronic Science and Technology,Dalian University of Technology,Dalian 116024,Liaoning Province,China;Jilin Huanqiu Hechuang Machinery Manufacture ＆ Cleaning Co.,Ltd.,Jilin 132021,Jilin Province,China)

机构地区：[1]大连理工大学电子科学与技术学院,辽宁大连116024 [2]吉林寰球和创机械制造清洗有限公司,吉林吉林132021

出　　处：《信息技术》2018年第8期48-51,54,共5页Information Technology

基　　金：国家自然科学基金资助项目(64102074)

摘　　要：数据中心能耗巨大,文中基于热成像技术,自动监测每个服务器的表面温度分布,并提取机柜中服务器的轮廓区域,为下一步的服务器状态诊断提供数据支持,从而更好地推进数据中心节能工作。文中采用方向边缘可控滤波器求得待检测服务器横竖边缘在红外图像中的最优角度,然后利用Hough直线检测和线性拟合方法,完成红外图像中的服务器设备自动检测。实验结果表明:相比于传统的Hough直线检测,在相同的拍摄角度下,文中提出的方法可以更准确地提取服务器所在区域的温度分布。Due to the great energy consumption in data center,this paper proposed a method to monitor temperature distribution of a server and extract the server area automatically based on thermal infrared image. It provides data for the next server state detection,so as to a better promotion of the energy saving in data center. First,it obtains the optimal angle of the server＇s horizontal and vertical edges by steerable filter. Second,it completes server devices detection automatically through Hough line detection and linear fitting method. The experiment result shows that compared with the traditional Hough line detection,the proposed method can extract a more accurate temperature area of servers in the different shooting angles.

关 键 词：热红外图像 数据中心 方向可控边缘滤波 

分 类 号：TM32[电气工程—电机]