摘要
对当前智能监控领域安全帽监检测存在硬件资源要求高、准确性低、实时性差的问题,提出了一种基于颜色分割的安全帽检测方法,该方法以传统的数字图像处理为基础,首先通过监控的实际场景获取监控画面,人工标注全帽所在区域,然后将RGB颜色空间图像转化为HSV颜色空间,统计不同颜色安全帽的HS颜色分量的阈值,利用K-mesns算法获取矩形框类别,并分类统计矩形框面积和矩形度范围,最后进行边缘提取、矩形框筛选方式现实安全帽检测。通过仿真分析,对于特定的安全帽场景及通用的CPU硬件设备下,与YOLOv3、SSD等传统的深度学习算法相比较,在保持准确率相当的前提下大幅提高了检测的速度。
In the current smart surveillance field,there are problems of high hardware resource requirements,low accuracy,and poor real-time performance in the detection of safety helmets.A safety helmet detection method based on color segmentation is proposed.This method is based on traditional digital image processing.First,obtain the monitoring picture through the actual scene of the monitoring,manually mark the area where the full hat is located,then convert the RGB color space image into the HSV color space,count the thresholds of the HS color components of different color helmets,and use the k-mesns algorithm to obtain the rectangular frame category,And classify and count the area of the rectangular frame and the range of the rectangular degree,and finally perform the edge extraction and the rectangular frame screening method to realize the helmet detection.Through simulation analysis,for specific helmet scenarios and general CPU hardware devices,compared with traditional deep learning algorithms such as YOLOv3 and SSD,the detection speed is greatly improved while maintaining the same accuracy.
作者
王运辉
王景龙
权雪祺
李恭斌
Wang Yun-hui;Wang Jing-long;Quan Xue-qi;Li Gong-bin
出处
《电力系统装备》
2021年第8期183-185,共3页
Electric Power System Equipment
关键词
安全帽检测
颜色分割
HSV颜色空间
safety helmet detection
color segmentation
HSV color space
