摘要
在很多施工现场等密集场景下,头盔佩戴是否符合规范直接关系到工作人员的生命安全,因此要求头盔佩戴必须符合标准。本文研究一种头盔佩戴智能检测方法,运用人工神经网络算法对采集到的头盔佩戴图像进行预处理,包括图像灰度化、图像滤波、图像增强、图像背景分割等四部分,利用方向梯度直方图(HOG)进行头盔佩戴图像特征提取,构建随机森林分类器,对头盔佩戴状态智能检测。结果表明:利用该方法检测正确率达到预期目标。
Whether the helmet is worn in accordance with the specification is directly related to the life safety of the staff.Therefore,in many dense scenes such as construction sites,it is required that the helmet must conform to the standard.In this context,a smart detection method for helmet wearing is studied.artificial neural network algorithms in artificial intelligence technology is used to preprocess the collected helmet wearing images,including image graying,image filtering,image enhancement,image background segmentation,etc.Directional gradient histogram(HOG)is used to extract features of helmet for the construction of a random forest classifier and the intelligent detection of the helmet wearing state.The results show that the proposed method for intelligent detection of helmet wearing in dense scenes reaches the expected goal.
作者
陈闯闯
胡绍方
CHEN Chuangchuang;HU Shaofang(School of Network Engineering,Zhoukou normal university,Zhoukou Henan 466000,China)
出处
《智能计算机与应用》
2020年第9期223-224,共2页
Intelligent Computer and Applications
关键词
密集场景
佩戴状态
图像滤波
dense scene
wearing state
image filtering