机器视觉技术估算边坡连续形变体积

金属矿山 ›› 2021, Vol. 50 ›› Issue (09): 60-64.

机器视觉技术估算边坡连续形变体积

叶春阳^1，2许传华³ 孙国权³ 聂闻^3，4

1. 中国科学院海西研究院泉州装备制造研究所，福建泉州 362000；2. 中北大学电气与控制工程学院，山西太原 030000；3.中钢集团马鞍山矿山研究总院股份有限公司，安徽马鞍山 243000；4. 江西理工大学资源与环境工程学院，江西赣州 341000

出版日期:2021-09-15 发布日期:2021-10-07
基金资助:
国家自然科学基金青年基金项目（编号：41702327）；国家自然科学基金面上项目（编号：51874268）；国家自然科学基金项目（编号：41867033）

Estimation of Continuous Deformation Volume of Slope Using Machine Vision Technique

YE Chunyang^1，2 XU Chuanhua³ SUN Guoquan³ NIE Wen^3，4

1. Quanzhou Institute of Equipment Manufacturing， Haixi Institutes， Chinese Academy of Sciences， Quanzhou 362000，China； 2. School of Electrical and Control Engineering， North University of China， Taiyuan 030000， China；3. Sinosteel Maanshan General Institute of Mining Research Co.， Ltd.， Maanshan 243000， China； 4. School of Resources and Environmental Engineering， Jiangxi University of Science and Technology， Ganzhou 341000， China

Online:2021-09-15 Published:2021-10-07

摘要/Abstract

摘要： 自动化”是实现矿山数字化、智能化的关键一步，同时也是矿山生产安全的重要保障。由于矿区环境复杂，造成边坡发生形变产生的原因、形变持续时间存在差异，使得准确提取边坡连续形变信息变得极为困难。为提升该类信息提取的准确度，提出一种基于深度信息的边坡连续形变信息提取方法。首先通过试验模拟矿山边坡，利用深度相机记录边坡在降雨条件下的破坏过程以获取图像数据和三维点云数据；然后采用相机标定建立图像数据和三维点云数据之间的关系并基于坐标变换和最近邻插值技术实现点云缺失值的补充；最后利用背景差分算法和去噪算法识别边坡形变区域，根据边坡形变前后点云的坐标变化估算边坡形变产生的体积。研究表明：该方法能够精确识别边坡形变区域，平均识别精度达92.39%。通过机器视觉技术能够以较高精度实现边坡形变区域自动识别与边坡形变体积信息提取，减少人工参与。

关键词: 矿山边坡, 机器视觉, 形变识别, 体积估算

Abstract: "Automation" is a key step to realize the digitization and intelligence of mines， and it is also an important guarantee for mine production safety. Due to the complex environment of the mining area， there are differences in the causes and duration of the slope deformation， which makes it extremely difficult to accurately extract the information of the slope continuous deformation. Therefore， in order to improve the accuracy of these information extraction， a method for extracting slope continuous deformation information based on depth information is proposed. Firstly， the mine slope is simulated through experiments， and the depth camera is used to record the failure process of the slope under rainfall conditions to obtain image data and three-dimensional point cloud data. Then， the camera calibration is used to establish the relationship between the image data and the 3D point cloud data， and the missing value of the point cloud is supplemented based on coordinate transformation and nearest neighbor interpolation technology.Finally， the background difference algorithm and denoising algorithm are used to identify the slope deformation area， and according to the coordinate changes of the point cloud before and after the slope deformation, the volume generated by the slope deformation is estimated. Research shows that this method can accurately identify the slope deformation area， with an average recognition accuracy of 92.39%. Machine vision technology can realize automatic identification of slope deformation area and extraction of slope deformation volume information with high accuracy， reducing manual participation.

Key words: mine slope, machine vision, deformation recognition, volume estimation

叶春阳, 许传华, 孙国权, 聂闻. 机器视觉技术估算边坡连续形变体积[J]. 金属矿山, 2021, 50(09): 60-64.

YE Chunyang, XU Chuanhua, SUN Guoquan, NIE Wen. Estimation of Continuous Deformation Volume of Slope Using Machine Vision Technique[J]. Metal Mine, 2021, 50(09): 60-64.

[1]	陶志刚, 孙滢滢, 席思达, 张未然, 朱春明, 隋麒儒. NPR 锚索对白云岩矿失稳边坡控制效应研究[J]. 金属矿山, 2024, 53(7): 220-.
[2]	张宏伟, 曼茂立, 王　宇, 刘　磊. 基于机器视觉与图卷积网络的矿区无人驾驶车辆异常行为检测[J]. 金属矿山, 2024, 53(10): 182-.
[3]	卢才武, 宋义良, 江松, 章赛, 王懋纪凡, . 基于改进 U-net 的少样本煤岩界面图像分割方法[J]. 金属矿山, 2024, 53(01): 149-157.
[4]	王斌, 田宝雄, 赵明辉. 融合场景理解与机器视觉的矿山巡检机器人导航避障研究[J]. 金属矿山, 2023, 52(11): 221-227.
[5]	曲伟霞, 潘伟, 郑飞, 许芝光, 崔恒嘉, 赵威, 孙伟芳, 侯令超, 王秋玲. 黄金矿山溜井运输AI视觉技术分析研究[J]. 金属矿山, 2023, 52(04): 174-179.
[6]	尤耿明, 王光进, 刘文连, 孔祥云, 胡航, 刘俊新, 朱宝龙, 樊晓一 . 矿山边坡软弱夹层赋存状态及蠕变特性对边坡稳定性影响研究[J]. 金属矿山, 2021, 50(12): 200-206.
[7]	贾玉跃, 高毓山. 矿山开采诱发边坡形变规律分析及控制对策[J]. 金属矿山, 2017, 46(06): 156-159.
[8]	吴永博, 高谦, 王明. 矿山边坡变形动态综合监测技术及稳定性预测[J]. 金属矿山, 2008, 38(06): 119-122.

扫码分享