随着近年来计算机视觉、激光雷达、深度相机等技术的迅猛发展，为煤矿开采带来了变革性的空间信息获取手段。虽然这些技术尚且存在数据量巨大、数据传输困难、数据处理算法复杂等难题，但随着矿山井下高速环网等基础信息设施的建设，使得深度相机之类的新型空间数据采集技术在煤矿采煤工作面的应用成为可能，这也为智能矿山地测空间信息的获取提供了一种新的技术途径，对智慧矿山的建设具有重要意义。

（1）针对相机畸变对模型精度的影响，使用张正友标定法对RealSense L515进行标定，得到的平均重投影误差为 0.0636，符合井下建模精度要求，进而求取了相机内参。由于实际采煤工作面水雾、煤尘情况会出现采集的图像质量差的问题，本研究通过实验对比改进前后的去雾效果与运行速度，发现改进的去雾算法更适合井下SLAM系统。为了进一步提高数据质量，采用了点云体素滤波方法对点云数据进行了降采样滤波处理，这使得点云数据更加清晰和易于处理。

（2）针对井下图像模糊、缺少纹理的情况，本文采用了适合井下特殊环境的特征点法作为前端视觉里程计。通过实验分析对比SIFT、SURF和ORB这三种算法，发现ORB算法比较适合井下环境特征提取；在进行特征匹配时，选取暴力匹配算法和RANSAC 算法进行误匹配剔除，提高了系统在井下的鲁棒性。

（3）针对矿井环境的复杂性，本研究搭建了适合井下特殊环境的软件开发环境和硬件实验平台，在此基础上使用L515相机采集房间内环境数据、模拟巷道环境数据和模拟工作面环境数据，对采集的数据三维点云进行了重建。重建结果表明，生成的巷道与工作面的三维模型精度能达到厘米级别，然后采用三维八叉树地图方法对生成的点云模型转换成八叉树地图，最后把点云模型发布到Web端进行三维可视化展示。

综上所述，本研究提出了一种基于深度相机的矿井三维重建系统，使用RealSense L515雷达相机作为传感器获取数据，在模拟煤矿实验室环境进行实验验证，实现对井下场景的三维重建，为矿山的智能化开采提供了理论创新性和一定的应用价值。

With the rapid development of computer vision, LiDAR, depth cameras, and other technologies in the field of artificial intelligence in recent years, revolutionary means of acquiring spatial information have been brought to the coal mining industry. Although these technologies still face challenges such as massive data volume, difficulties in data transmission, and complex data processing algorithms, the construction of foundational infrastructure such as high-speed networks in underground mines has made it possible to apply new spatial data acquisition technologies like depth cameras in coal mining working faces. This provides a new high technology approach for obtaining spatial information in intelligent mine surveying, which holds significant importance for the development of smart mines.

(1) To address the impact of camera distortion on model accuracy, this study used the Zhang Zhengyou calibration method to calibrate the RealSense L515, the average reprojection error obtained was 0.0636, which met the accuracy requirements for underground modeling, and then obtained the camera internal reference to prepare for the next step of data acquisition.

To address the issue of poor image quality caused by water mist and coal dust in actual mining faces, this study conducted experiments to compare the before and after effects of improved haze removal algorithms in terms of both haze reduction and computational speed. The results revealed that the improved haze removal algorithm was more suitable for underground SLAM (Simultaneous Localization and Mapping) systems. To further enhance data quality, a point cloud voxel filtering method was employed to downsample and filter the point cloud data, which made the point cloud data clearer and easier to process.

(2)To address the blurriness and lack of texture in underground images, this paper adopted a feature-based method suitable for the unique conditions of underground environments as the front-end visual odometry. Through experimental analysis and comparison of three algorithms—SIFT, SURF, and ORB, it was found that the ORB algorithm was more suitable for feature extraction in underground environments. During feature matching, a combination of the brute-force matching algorithm and RANSAC algorithm was selected to eliminate incorrect matches, thereby improving the system's robustness in underground settings.

(3) For the complexity of the mine environment, this study built a software development environment and hardware experiment platform suitable for the special underground environment, based on which the L515 camera was used to collect environmental data in the laboratory room, simulated roadway environmental data and simulated working face environmental data, and the 3D point cloud of the collected data was reconstructed, the reconstruction results showed that the generated 3D models of tunnels and working faces achieved accuracy at the centimeter level. Subsequently, the point cloud model was transformed into an octree map by establishing a 3D octree map. Finally, the point cloud models were published to the web for 3D visualization.

In summary, this study proposed a mining 3D reconstruction system based on depth cameras. The RealSense L515 depth camera was used as the sensor for data acquisition. Through experimental verification in a simulated mining laboratory environment, the system achieved 3D reconstruction of underground scenes, providing theoretical innovation and certain application value for intelligent mining in mines.

[1] 王国法, 刘峰, 孟祥军, 等. 煤矿智能化(初级阶段)研究与实践[J]. 煤炭科学技术, 2019, 47(8): 1-36.

[2] 张操. 我国能源消费结构调整背景下煤炭企业的应对策略[J]. 煤炭与化工, 2017, 40(5): 133-136.

[3] 王国法. 加快煤矿智能化发展建设智能+绿色煤炭工业新体系[J]. 中国煤炭工业, 2020, 398(4): 8-15.

[4] 国家安全监管总局关于开展“机械化换人、自动化减人”科技强安专项行动的通知[J]. 国家安全生产监督管理总局国家煤矿安全监察局公告, 2015, 162(7): 13-18.

[5] 国家发展改革委 国家能源局 应急管理部 国家煤矿安监局 工业和信息化部 财政部 科技部 教育部. 关于加快煤矿智能化发展的指导意见[J]. 中国安全生产, 2020, 15(3): 5.

[6] 河南省人民政府办公厅关于印发河南省煤矿智能化建设三年行动方案(2021—2023年)的通知[J]. 河南省人民政府公报, 2021, 424(4): 13-15.

[7] 胡青松, 钱建生, 李世银, 等. 智能煤矿技术研究与政策制定现状[J]. 工矿自动化, 2021, 47(3): 1-8.

[8] 李鹏, 程建远. 回采工作面煤层三维建模技术及其在智能开采中的应用[J]. 煤矿安全, 2021, 52(8): 156-161.

[9] 王国法, 杜毅博. 智慧煤矿与智能化开采技术的发展方向[J]. 煤炭科学技术, 2019, 47(1): 1-10.

[10] 丁恩杰, 俞啸, 夏冰, 等. 矿山信息化发展及以数字孪生为核心的智慧矿山关键技术[J]. 煤炭学报, 2022, 47(1): 564-578.

[11] 姜浩. 基于激光雷达的地下巷道三维建图方法研究[D]. 北京: 北方工业大学, 2022.

[12] 李猛钢, 胡而已, 朱华. 煤矿移动机器人LiDAR/IMU紧耦合SLAM方法[J]. 工矿自动化, 2022, 48(12): 68-78.

[13] Civera J, Davison A J, Montiel J. Inverse depth parametrization for monocular SLAM[J]. IEEE Transactions on Robotics, 2008, 24(5): 932-945.

[14] 陈鹏. 基于激光雷达与视觉融合的SLAM系统设计与实现[D]. 南京: 南京邮电大学, 2021.

[15] 王国法, 王虹, 任怀伟, 等. 智慧煤矿2025情景目标和发展路径[J]. 煤炭学报, 2018, 43(02): 295-305.

[16] 李杰. 智慧煤矿与智能化开采技术的发展方向[J]. 中国石油和化工标准与质量, 2020, 40(18): 221-222.

[17] 王国法, 赵国瑞, 任怀伟. 智慧煤矿与智能化开采关键核心技术分析[J]. 煤炭学报, 2019, 44(1): 34-41.

[18] 徐静, 谭章禄. 智慧矿山系统工程与关键技术探讨[J]. 煤炭科学技术, 2014, 42(4): 79-82.

[19] 韩茜. 智慧矿山信息化标准化系统关键问题研究[D]. 北京: 中国矿业大学(北京), 2016.

[20] 方新秋, 梁敏富, 李爽, 等. 智能工作面多参量精准感知与安全决策关键技术[J]. 煤炭学报, 2020, 45(1): 493-508.

[21] 王国法, 张德生. 煤炭智能化综采技术创新实践与发展展望[J]. 中国矿业大学学报, 2018, 47(3): 459-467.

[22] 张申, 胡青松, 王刚. 矿山物联网标准建设的思考[J]. 工矿自动化, 2018, 44(1): 1-5.

[23] Durrant-Whyte H, Bailey T. Simultaneous localization and mapping: part I[J]. IEEE robotics& automation magazine, 2006, 13(2): 99-108.

[24] 王霞, 左一凡. 视觉SLAM研究进展[J]. 智能系统学报, 2020, 15(5): 825-834.

[25] Bailey T, Durrant-Whyte H. Simultaneous localization and mapping (SLAM): part II[J]. IEEE Robotics & Automation Magazine, 2006, 13(3): 108-117.

[26] Li A , Wang J , Xu M , et al. DP-SLAM: A Visual SLAM with Moving Probability Towards Dynamic Environments[J]. Information Sciences, 2021, 556: 128-142.

[27] 邸凯昌, 万文辉, 赵红颖, 等. 视觉SLAM技术的进展与应用[J]. 测绘学报, 2018, 47(6): 770-779

[28] Davison A J , Reid I D , Molton N D , et al. MonoSLAM: real-time single camera SLAM[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2007, 29(6): 1052-1067.

[29] Belter D, Skrzypczynski P. Precise self-localization of a walking robot on rough terrain using parallel tracking and mapping[J]. Industrial Robot, 2013, 40(3): 229-237.

[30] Endres F, Hess J, Sturm J, et al. 3-D mapping with an RGB-D camera[J]. IEEE Transactions on Robotics, 2014, 30(1): 177-187.

[31] Mur-Artal R, Montiel J M M, Tardos J D. ORB-SLAM: a versatile and accurate monocular SLAM system[J]. IEEE Transactions on Robotics, 2015, 31(5): 1147-1163.

[32] Mur-ArtalRaúl, D Tardósjuan. ORB-SLAM2: An Open-Source SLAM System for Monocular, Stereo, and RGB-D Cameras[J]. IEEE Transactions on Robotics, 2017, 33(5): 1255-1262.

[33] Dai A, Niener M, Zollhfer M, et al. BundleFusion: Real-Time Globally Consistent 3D Reconstruction Using On-the-Fly Surface Reintegration[J]. ACM Transactions on Graphics, 2017, 36(3): 17-34.

[34] Campos C, Elvira R, Juan J. Gómez Rodríguez, et al. ORB-SLAM3: An Accurate Open-Source Library for Visual, Visual-Inertial and Multi-Map SLAM[J]. IEEE Transactions on Robotics, 2020, 37(6): 1874 - 1890.

[35] Zhou H, Zou D , Pei L , et al. StructSLAM: Visual SLAM With Building Structure Lines[J]. IEEE Transactions on Vehicular Technology, 2015, 64(4): 1364-1375.

[36] Qin Tong, Li Peiliang, Shen Shaojie. VINS-Mono: A Robust and Versatile Monocular Visual-Inertial State Estimator[J]. IEEE Transactions on Robotics, 2018, 34(4): 1004-1020.

[37] Cui J, Zhang F, Feng D, et al. An improved SLAM based on RK-VIF: Vision and inertial information fusion via Runge-Kutta method[J]. Defence Technology, 2023, 21(03): 133-146.

[38] 蒋林, 刘林锐, 周安娜, 等. 基于运动预测的改进ORB-SLAM算法[J]. 浙江大学学报(工学版), 2023, 57(1): 170-177.

[39] 石信肖, 王健, 王磊, 等. 点云数据下的矿山巷道三维建模[J]. 遥感信息, 2019, 34(06): 99-104.

[40] 史瑶. 基于草图输入的煤矿巷道建模方法的研究[D]. 北京: 中国矿业大学, 2020.

[41] Anonymous. CAE Sells Datamine to Constellation Software[J]. Engineering and Mining Journal, 2015, 216(10): 77.

[42] 车德福, 吴立新. 数字矿山基础平台-GeoMo3D的功能结构与应用[C]// 第七届全国矿山测量学术会议.第七届全国矿山测量学术会议论文集. 唐山: 矿山测量. 2007: 142-148.

[43] 胡冬冬. 基于图像纹理信息的三维重建研究[D]. 武汉: 华中科技大学, 2017.

[44] Li YAN, Jicheng DAI, Junxiang TAN, et al. Global Fine Registration of Point Cloud in LiDAR SLAM Based on Pose Graph[J]. 测绘学报（英文）, 2020, 003(2): 26-35.

[45] Guclu Oguzhan, Can Ahmet Burak. Integrating global and local image features for enhanced loop closure detection in RGB-D SLAM systems[J]. The Visual Computer, 2020, 36(6): 1271-1290.

[46] Dibyendu Ghosh, Biswajit Samanta, Debashish Chakravarty. Multi sensor data fusion for 6D pose estimation and 3D underground mine mapping using autonomous mobile robot[J]. International Journal of Image and Data Fusion, 2016, 8(2): 173-187.

[47] 王保岩. 矿井巷道三维重建及安全监测技术研究[D]. 山东: 山东大学, 2018.

[48] 杜雨馨. 矿井悬臂式掘进机位姿感知及定位方法研究[D]. 北京: 中国矿业大学, 2019.

[49] 杨文娟, 张旭辉, 张超, 等. 基于三激光束标靶的煤矿井下长距离视觉定位方法[J]. 煤炭学报, 2022, 47(2): 986-1001.

[50] 王成龙, 周国丽. 煤矿钻锚机器人Kinect视觉系统三维重建[J]. 煤矿机械, 2022, 43(7): 181-183.

[51] 高翔. 视觉SLAM十四讲: 从理论到实践[M]. 北京: 电子工业出版社, 2019: 19.

[52] 权美香, 朴松昊, 李国. 视觉SLAM综述[J]. 智能系统学报, 2016, 11(6): 768-776.

[53] Younes G , Asmar D , Shammas E , et al. Keyframe-based monocular SLAM: design, survey, and future directions[J]. Robotics and Autonomous Systems, 2017, 98: 67-88.

[54] 刘浩敏, 章国锋, 鲍虎军. 基于单目视觉的同时定位与地图构建方法综述[J]. 计算机辅助设计与图形学学报, 2016, 28(6): 855-868.

[55] 许智宾, 李宏伟, 张斌, 等. 双目视觉与惯导融合的移动机器人定位方法[J]. 测绘学报, 2021, 50(11): 1512-1521.

[56] 张恒, 侯家豪, 刘艳丽. 基于动态区域剔除的RGB-D视觉SLAM算法[J]. 计算机应用研究, 2022, 39(3): 675-680.

[57] 刘三毛, 朱文球, 孙文静, 等. 基于RGB-D单目视觉的室内场景三维重建[J]. 微型机与应用, 2017, 36(1): 44-47.

[58] 杨凤. 基于视觉SLAM的室内定位算法研究[D]. 南京: 东南大学, 2020.

[59] 杨宇, 赵成星, 张晓玲. 鱼眼相机的视觉标定及畸变校正[J]. 激光杂志, 2020, 41(9): 20-23.

[60] Xiaoyan Li, Xiaofeng Su, Zhuoyue Hu, et al. Improved distortion correction method and applications for large aperture infrared tracking cameras[J]. Infrared Physics and Technology, 2019, 98: 82-88.

[61] 英特尔最新激光雷达相机技术重新定义了计算机视觉[J]. 传感器世界, 2019, 25(12): 41.

[62] Zhang Z. A Flexible New Technique for Camera Calibration[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2000, 22(11): 1330-1334.

[63] 王启明, 李季. 煤矿井下高清图像快速去雾算法研究[J]. 小型微型计算机系统, 2018, 39(11): 2557-2560.

[64] He Kaiming, Sun Jian, Tang Xiaoou. Single Image Haze Removal Using Dark Channel Prior.[J]. IEEE transactions on pattern analysis and machine intelligence, 2011, 33(12): 1956-1963.

[65] 张彬, 熊传兵. 基于体素下采样和关键点提取的点云自动配准[J]. 激光与光电子学进展, 2020, 57(4): 109-117.

[66] 谭涌, 潘树国, 高旺, 等. 融合光度参数估计的单目直接法视觉里程计[J]. 测绘工程, 2021, 30(3): 59-65.

[67] 于雅楠, 史敦煌, 华春杰. 特征点法SLAM视觉里程计自适应优化算法（英文）[J]. 系统仿真学报, 2022,34(1): 104-112.

[68] Lowe D G. Distinctive Image Features from Scale-Invariant Keypoints[J]. International Journal of Computer Vision, 2004, 60(2): 91-110.

[69] Bay H, Ess A , Tuytelaars T, et al. SURF: Speeded up robust features[J]. Computer Vision and Image Understanding, 2008, 110(3): 346–359.

[70] 冉峰, 李天, 季渊, 等. 基于改进SURF算法的单目视觉里程计[J]. 电子测量技术, 2017, 40(5): 185-188.

[71] Rublee E, Rabaud V, Konolige K, et al. ORB: an efficient alternative to SIFT or SURF[C]// IEEE International Conference on Computer Vision.Proceedings of the 2011 International Conference on Computer Vision, United States: IEEE Computer Society, 2011: 2564-2571.

[72] 程珉, 陈临强, 杨全鑫. 基于改进ORB特征的单目视觉SLAM算法[J]. 计算机应用与软件, 2021, 38(10): 242-248.

[73] Wenyi Xiong, Weiming Tian, Zhijun Yang, et al. Improved FAST corner-detection method[J]. The Journal of Engineering, 2019, 2019(19): 5493-5497.

[74] Xiao Ling Ding, Qiang Zhao, Yi Bin Li, et al. A Rapid and Efficient Feature Point Detection and Matching Algorithm[J]. Applied Mechanics and Materials, 2014, 3343(607): 641-646.

[75] Huang S, Sun G, Li M. FAST and FLANN for feature matching based on SURF[C]//中国自动化学会信息物理系统控制与决策专业委员会.第33届中国控制与决策会议论文集(9), 沈阳: 控制与决策, 2021: 212-217.

[76] 任彬, 宋海丽, 赵增旭, 等. 基于RANSAC的视觉里程计优化方法研究[J]. 仪器仪表学报, 2022, 43(6): 205-212.

[77] Galvez-Lpez D, Tardos J D. Bags of Binary Words for Fast Place Recognition in Image Sequences[J]. IEEE Transactions on Robotics, 2012, 28(5): 1188-1197.

[78] Hornung A, Wurm K M, Bennewitz M, et al. OctoMap: An efficient probabilistic 3D mapping framework based on octrees[J]. Autonomous Robots, 2013, 34(3): 189-206.

[79] 叶梦轩, 危双丰, 张冬梅. 基于HTML5和WebGL的三维点云可视化方法[J]. 工程勘察, 2017, 45(1): 40-44.

[80] 程泽华. 基于WebGL的地质三维模型构建及可视化方法研究[D]. 北京: 中国地质大学(北京), 2020.