基于红外识别的简易升降机违规载人监测  

Monitoring of simple lift manned operation against rules based on infrared identification

作　　者：李贤 王国贤[1] 朱春东[1,2] 程茁 Li Xian;Wang Guoxian;Zhu Chundong;Cheng Zhuo

机构地区：[1]武汉理工大学,武汉430070 [2]随州武汉理工大学工业研究院,随州441300

出　　处：《起重运输机械》2024年第14期67-73,共7页Hoisting and Conveying Machinery

摘　　要：文中采用红外摄像头拍摄不同情况下货厢内部的图片作为样本,提出一种HOG特征进行人形分类特征提取并进行降维处理的方法,选择常用于可见光识别领域的支持向量机作为优化分类模型,结合粒子群优化算法寻找支持向量机的核函数的核参数。最终成功实现了对简易升降机违规载人行为的监测,识别正确率达到了97%。In this paper,taking the pictures of the interior of the cargo compartment taken by infrared cameras under different conditions as samples,a method of HOG feature extraction and dimensionality reduction is proposed.Support vector machine,which is commonly used in the field of visible light recognition,was selected as the optimal classification model,and the kernel parameters of the kernel function of support vector machine were found by combining the particle swarm optimization algorithm.Finally,the monitoring of manned operations against rules of the simple lift was realized,with a correct recognition rate of 97%.

关 键 词：简易升降机 红外识别 HOG特征 SVM 粒子群优化算法 

分 类 号：TH211.6[机械工程—机械制造及自动化]