基于机器视觉的非接触式滑坡体监测技术研究
Research on non-contact landslide monitoring technology based on computer vision
徐海涛 1李登华 2林泊锟 3邱先志 1丁勇3
作者信息
- 1. 国家能源集团新疆开都河流域水电开发有限公司,新疆 库尔勒 841000
- 2. 南京水利科学研究院,江苏 南京 210024;水利部水库大坝安全重点实验室,江苏 南京 210024
- 3. 南京理工大学理学院,江苏 南京 210094
- 折叠
摘要
为降低监测成本有效提升滑坡体险情感知能力,文章提出了一种基于机器视觉的非接触式监测技术,并使用大型室内滑坡模拟装置对其可靠性进行验证.该技术采用特定标识物作为滑坡体的监测对象,并设计了以HSV色彩空间与Hough圆检测算法为基础的标识物提取定位算法,融合了单目测距技术对标识物的位移进行定量计算.采用该技术监测分析大型室内滑坡模拟装置的滑坡体险情发展全过程,结果表明,此技术能够较为精确地反映滑坡体的位移情况,为滑坡监测提供了一种经济、全面的监测方案.
Abstract
In order to reduce monitoring costs and effectively enhance the emotional awareness of landslide risk,this paper proposes a non-contact monitoring technology based on machine vision,and verifies its reliability using a large indoor landslide simulation device.This technology uses specific markers as monitoring objects for landslide masses,and designs a marker extraction and positioning algorithm based on HSV color space and Hough circle detection algorithm.It integrates monocular ranging technology to quantitatively calculate the displacement of markers.The use of this technology to monitor and analyze the entire process of landslide hazard development in large-scale indoor landslide simulation devices shows that this technology can accurately reflect the displacement of the landslide body,providing an economical and comprehensive monitoring scheme for landslide monitoring.
关键词
滑坡体/标识物/监测/单目测距/模型试验Key words
landslide mass/identification materials/monitoring/monocular ranging/model test引用本文复制引用
基金项目
国家重点研发计划(2022YFC3005502)
国家自然科学基金长江水科学研究联合基金(U2240221)
国家自然科学基金(51979174)
出版年
2024
NETL
NSTL
万方数据