全景钻孔图像自动识别技术在工程实践中的应用研究  

Application of Automatic Recognition Technique for the Panoramic Borehole Images Obtained in Engineering Practice

作　　者：邹先坚[1] 王川婴[1] 宋欢 ZOU Xianjian;WANG Chuanying;SONG Huan(State Key Laboratory of Geomechanics and Geotechnical Engineering,Institute of Rock and Soil Mechanics,Chinese Academy of Sciences,Wuhan 430071,China;Center for Computational&Exploration Geophysics,Institute of Geodesy and Geophysics,Chinese Academy of Sciences,Wuhan 430071,China;School of Electronic Information,Wuhan University,Wuhan 430072,China)

机构地区：[1]中国科学院武汉岩土力学研究所,岩土力学与工程国家重点实验室,湖北武汉430071 [2]中国科学院测量与地球物理研究所计算与勘探地球物理研究中心,湖北武汉430077 [3]武汉大学电子信息学院,湖北武汉430072

出　　处：《应用基础与工程科学学报》2022年第1期246-256,共11页Journal of Basic Science and Engineering

基　　金：国家自然科学基金项目(41731284,42007266);湖北省自然科学基金项目(2019CFB345)。

摘　　要：数字全景摄像技术在钻孔工程实践中已经得到广泛应用.利用数字全景钻孔摄像系统获得的大量高清全景钻孔图像,采用基于聚类和特征函数的自动识别方法进行了智能分析和自动识别研究,并成功应用在乌东德水电站获得的实际全景钻孔图像中.结果表明,该自动识别方法能够全面快速地全自动识别出整个钻孔图像中大部分重要结构面及其位置、倾向、倾角和隙宽等特征参数信息,并对结构面特征进行了初步归类和统计分析.在12个约50m深的钻孔当中,每个孔所获全景钻孔图像主要结构面的自动识别率约为90%、准确率约为85%、全孔耗时约为3h,精度偏差小于4%.全景钻孔图像自动识别技术的应用极大地提高了实践工作效率,缩短了工程时间,避免了较多的人为因素干扰,实现了全孔图像结构面形态参数的自动识别与实践应用,有利于推动钻孔摄像技术智能化的进一步发展.The digital panoramic camera technology has been widely applied in the borehole engineering.For the obtained massive high-definition images by this panoramic camera system,we use a clustering and characteristic function-based method to realize the intelligent analysis of borehole images and research its application of automatic recognition technique,which had been successfully used in the analysis of borehole images from the Wudongde Hydroelectric Power Station project.Results show that the method can rapidly and automatically identify most major structural planes and their parameters including locations,dip directions,dip angles and gap widths,and provide an initial classification and statistical analysis of such features.For 12 boreholes with about 50 meters in depth every one,the method has a 90%detection rate for major structural planes for each borehole,with an accuracy rate of 85%and a deviation less than 4%,in a total period of about 3 hours.The method has greatly improved actual work efficiency,reduced engineering time,avoided the risk of human error,and realized the automatic recognition of feature parameters for full-borehole structural planes and its actual engineering application.It can promote the further development of intelligent borehole camera technology.

关 键 词：钻孔图像 自动识别 结构面 工程应用 聚类 正弦曲线 乌东德水电站 

分 类 号：P631.8[天文地球—地质矿产勘探] TB811.2[天文地球—地质学]