基于机器视觉的煤炭运输列车车厢状态智能检测方法
Intelligent Detection Method of Coal Transport Train Carriage Status Based on Machine Vision
陈小霞 1李锁弟 1朱良恺 1张东伟 1王祁峰1
作者信息
- 1. 枣庄矿业(集团)有限责任公司煤炭洗选加工中心,山东 枣庄 277000
- 折叠
摘要
提出基于机器视觉的煤炭运输列车车厢状态智能检测方法,该方法能及时发现安全隐患并进行预警,提升检车效率,有效预防事故的发生,保障煤炭运输的安全.利用线阵相机等设备采集煤炭运输列车车厢的原始图像;利用Retinex算法增强列车车厢原始图像,提升图像质量;采用索贝尔算子对图像中的车厢实施切分处理,得到每一节完整的车厢图片,用于后续的车厢状态检测;构建YOLOv5算法,并提出一种抑制异类冗余框的方法,对其实施改进,利用改进后的方法完成对煤炭运输列车车厢状态的智能检测,并将检测结果应用于车厢异常报警中.实验证明,该方法能够精准检测煤炭运输列车车厢状态,并及时发出报警信息,在mAP和FPS方面均有较好的表现.
Abstract
This paper proposes an intelligent detection method based on machine vision for the state of coal transport trains,which can detect and warn potential safety hazards in time,improve inspection efficiency,effectively prevent accidents,and ensure the safety of coal transport.Using linear array camera and other equipment to collect the original images of coal transport train cars.The Retinex algorithm is used to enhance the original image of the train car and improve the image quality.Sobel operator is applied to the segmentation of the carriages in the image,and the complete picture of each carriage is obtained,which is used for the subsequent state detection of the carriage.The YOLOv5 algorithm model is constructed,and a method to suppress heterogeneous redundant frames is proposed,which is improved.The improved model is used to complete intelligent state detection of coal transport train cars,and the detection results are applied to abnormal alarm of train cars.Experiments show that this method can accurately detect the status of coal transport train cars and send alarm information in time,and has a good performance in mAP and FPS.
关键词
机器视觉/车厢状态/智能检测/图像增强/车厢切分/YOLOv5算法Key words
machine vision/carriage status/intelligent detection/image enhancement/compartment segmentation/YOLOv5 algorithm引用本文复制引用
出版年
2024
NETL
NSTL
万方数据