Real-Time Tunnel Crack Analysis System via Deep Learning
通过深度学习的实时隧道裂缝分析系统
论文:http://static.tongtianta.site/paper_pdf/eea088ee-eb45-11e9-a1f6-00163e08bb86.pdf
Corresponding author: Qing Song (priv@bupt.edu.cn) I. INTRODUCTION Rail transport is a main way of transportation in society. It is important to guarantee the safety of railway to keep the personal security and property security. Repairing the rail defects is an urgent demand in railway safety including rail defects, train defects, and tunnel cracks. Our system focuses on the tunnel cracks mainly. Cracks in tunnels become an unavoidable problem which must be identified in time [26]. But the current detection methods are basically some methods with low efficiency and few technical skills. The traditional tunnel crack detection is mainly realized by manual which means the technician builds scaffolding on site [4], and then the technician observes the crack through the human vision or takes a photo of the whole tunnel for looking for cracks in the photo. It is very inefficient and labor intensive, and the accuracy of the inspection depends largely on the experience and quality of the technician. On the other hand, due to a large The associate editor coordinating the review of this manuscript and approving it for publication was Michele Nappi.
通讯作者:宋青(priv@bupt.edu.cn)引言铁路运输是社会上的主要运输方式。保证铁路安全,保持人身安全和财产安全至关重要。修复铁路缺陷是铁路安全的迫切需求,包括铁路缺陷,火车缺陷和隧道裂缝。我们的系统主要关注隧道裂缝。隧道裂缝成为不可避免的问题,必须及时确定[26]。但是,目前的检测方法基本上是一些效率低,技术技能差的方法。传统的隧道裂缝检测主要通过人工来实现,这意味着技术人员在现场搭建脚手架[4],然后技术人员通过人类的视觉观察裂缝或拍摄整个隧道的照片以寻找照片中的裂缝。它效率低下且劳动强度大,并且检查的准确性很大程度上取决于技术人员的经验和素质。另一方面,由于副手大,米歇尔·纳皮(Michele Nappi)负责协调手稿的审阅并批准出版。
64186 amount of data, the method of manual identification loses timeliness and costs a lot of money.
64186的数据量,手动识别方法失去了及时性,并花费了大量金钱。
With the development of computer science and digital image processing technology, the use of image processing to detect cracks has attracted more and more attention. It has the advantages of non-contact, high efficiency, convenience, and directness, and has gradually become the main direction of research and obtained a large number of research results [26]. However, tunnel images have some complex situations, such as water stains, pollution, structural seams, uneven illumination, numerous noises, and irregular distribution, which have brought bottlenecks to the development of traditional image processing methods. Due to the disturbance of these situations, traditional image processing methods cannot get a good performance to solve the problems.
随着计算机科学和数字图像处理技术的发展,使用图像处理来检测裂缝已引起越来越多的关注。它具有非接触,高效,便捷和直接的优点,已逐渐成为研究的主要方向并获得了大量的研究成果[26]。然而,隧道图像存在一些复杂的情况,例如水渍,污染,结构接缝,照明不均匀,大量噪声和不规则分布,这给传统图像处理方法的发展带来了瓶颈。由于这些情况的干扰,传统的图像处理方法不能很好地解决这些问题。
In recent years, the rapid development of artificial intelligence and deep learning has brought revolutionary development in computer vision. Object image classification [9], [14], [20], [22], [32], [33] and segmentation [5], [10], [11], [13] are very active tasks in computer vision. Given a picture,
近年来,人工智能和深度学习的飞速发展带来了计算机视觉的革命性发展。对象图像分类[9],[14],[20],[22],[32],[33]和分割[5],[10],[11],[13]是计算机视觉中非常活跃的任务。给一张照片
VOLUME 7, 2019 Received April 16, 2019, accepted May 3, 2019, date of publication May 13, 2019, date of current version May 30, 2019.
2019年第7卷,2019年4月16日收到,2019年5月3日接受,发布日期为2019年5月13日,当前版本为2019年5月30日。
Real-Time Tunnel Crack Analysis System via Deep Learning This work was supported in part by the Beijing University of Posts and Telecommunications, and in part by the China Academy of Railway Sciences.
通过深度学习的实时隧道裂缝分析系统这项工作部分得到北京邮电大学的支持,部分得到中国铁道科学研究院的支持。
所有论文

添加客服微信,加入用户群
蜀ICP备18016327号