基于SfM的城市电缆隧道三维重建方法优化研究

doi:10.11996/JG.j.2095-302X.2023030540

图学学报 ›› 2023, Vol. 44 ›› Issue (3): 540-550.DOI: 10.11996/JG.j.2095-302X.2023030540

• 图像处理与计算机视觉 • 上一篇下一篇

基于SfM的城市电缆隧道三维重建方法优化研究

葛海明¹(), 张维²^,³(), 王小龙¹, 朱晶晶¹, 贾非²^,³, 薛亚东²^,³

1.中国能源建设集团江苏省电力设计院有限公司，江苏南京 211102
2.同济大学土木工程学院地下建筑与工程系，上海 200092
3.同济大学岩土及地下工程教育部重点实验室，上海 200092

收稿日期:2022-09-28 接受日期:2023-02-22 出版日期:2023-06-30 发布日期:2023-06-30
通讯作者: 张维(1998-)，男，硕士。主要研究方向为三维重建、计算机视觉、隧道性能评估与预测等。E-mail：1653507333@qq.com
作者简介:
葛海明(1976-)，男，正高级工程师，本科。主要研究方向为隧道工程数字图像处理与识别。E-mail：gehaiming@jspdi.com.cn
基金资助:
中国电力工程顾问集团有限公司科研项目(GSKJ2-G03-2021)

Optimization of 3D reconstruction method for urban cable tunnel based on SfM method

GE Hai-ming¹(), ZHANG Wei²^,³(), WANG Xiao-long¹, ZHU Jing-jing¹, JIA Fei²^,³, XUE Ya-dong²^,³

1. China Energy Engineering Group Jiangsu Power Design Institute Co., Ltd, Nanjing Jiangsu 211102, China
2. Department of Geotechnical Engineering College of Civil Engineering, Tongji University, Shanghai 200092, China
3. Key Laboratory of Geotechnical and Underground Engineering (Tongji University), Ministry of Education, Shanghai 200092, China

Received:2022-09-28 Accepted:2023-02-22 Online:2023-06-30 Published:2023-06-30
Contact: ZHANG Wei (1998-), master. His main research interests cover 3D reconstruction, computer vision, tunnel performance evaluation and prediction, etc. E-mail：1653507333@qq.com
About author:
GE Hai-ming (1976-), undergraduate. His main research interests cover digital image processing and recognition for tunnel engineering. E-mail：gehaiming@jspdi.com.cn
Supported by:
Research Projects of China Power Engineering Consulting Group Co., Ltd(GSKJ2-G03-2021)

摘要/Abstract

摘要：

三维重建技术常被应用于城市地铁隧道、铁路隧道等场景，用来进行隧道形态的可视化展观以及隧道性态的分析。对于城市电缆隧道，由于其内部障碍物较多，断面较小等特点，采用三维激光扫描仪等传统的三维重建方式存在一定的困难。摄影测量由于其设备的便携性，应用于电缆隧道的三维重建具有优势。基于摄影测量理论，对影响电缆隧道三维重建效果的各种因素进行总结和分析并建立了针对电缆隧道场景三维重建的模型评价方法。于南通滨江路GIL电缆隧道内开展现场实验，对于原始图像存在的明暗不均问题，选用直方图均衡化+伽马变换进行增强化预处理，恢复隧道图像暗部的衬砌细节。基于运动恢复结构(SfM)三维重建方法，构建了具有完整内部纹理信息的电缆隧道三维模型。同时以模型评价方法为判断标准，对图像采集方法和建模参数等关键因素进行优化研究。结果表明，使用200张图像、中等质量建模能够保证较高的点云密度(1.0×10⁵ m)、合理的中心特征点数量(321个)、较低的均方根重投影误差(1.36 pix)，同时又兼顾了较高的建模效率(177 s)，满足隧道三维重建的需要。

关键词: 电缆隧道, 摄影测量, 三维重建, 运动恢复结构, 伽马变换, 直方图均衡化

Abstract:

The three-dimensional reconstruction technology is extensively applied to various scenarios such as urban subway tunnels and railway tunnels to visualize tunnel structure and analyze tunnel properties. For urban cable tunnels, due to many internal obstacles with smaller sections than subway tunnels, there are some difficulties in using the traditional three-dimensional reconstruction methods such as 3D laser scanners. Photogrammetry technology, with its portable equipment, holds significant potential in the three-dimensional reconstruction of cable tunnels. Based on photogrammetry theory, this paper summarized and analyzed various factors affecting the effect of 3D reconstruction of cable tunnels and established a model evaluation method for 3D reconstruction of cable tunnel scenes. Experiments were carried out in a gas insulated transmission line (GIL) cable tunnel in Nantong, Jiangsu Province. To mitigate the problem of uneven brightness and darkness in the original image, histogram equalization and gamma transform were employed for enhanced preprocessing, thereby restoring the lining details of the dark part of the tunnel image. Based on the structure from motion (SfM) 3D reconstruction method, a 3D model of the cable tunnel with complete internal texture information was successfully constructed. Taking the model evaluation method as the judgment standard, the key factors such as image acquisition method and modeling parameters were optimized. The results revealed that employing 200 images and moderate quality modeling can ensure high point cloud density (1.0×10⁵ m), a reasonable number of central feature points (321), a low RMS reprojection error (1.36 pix), and a high modeling efficiency (177 s), meeting the needs of tunnel inspection.

Key words: cable tunnels, photogrammetry, three-dimensional reconstruction, structure from motion, gamma transform, histogram equalization

中图分类号:

TP391

葛海明, 张维, 王小龙, 朱晶晶, 贾非, 薛亚东. 基于SfM的城市电缆隧道三维重建方法优化研究[J]. 图学学报, 2023, 44(3): 540-550.

GE Hai-ming, ZHANG Wei, WANG Xiao-long, ZHU Jing-jing, JIA Fei, XUE Ya-dong. Optimization of 3D reconstruction method for urban cable tunnel based on SfM method[J]. Journal of Graphics, 2023, 44(3): 540-550.

图/表 17

图1 SfM算法流程

Fig. 1 SfM algorithm flow

图2 电缆隧道图像直方图均衡化处理((a)原始图像和RGB像素通道直方图；(b)均衡化处理后的图像和RGB像素通道直方图)

Fig. 2 Image histogram equalization processing ((a) The original image and the RGB pixel channel histogram; (b) Balanced image and RGB pixel channel histogram)

图3 不同γ值的伽马变换曲线(所有情况c=1)

Fig. 3 Transform curves with different values (All cases: c=1)

图4 不同γ值下隧道图像的伽马变换效果

Fig. 4 The effect of the gamma transform of the tunnel image with different γ values ((a) γ=1.0; (b) γ=0.9; (c) γ=0.8; (d) γ=0.7; (e) γ=0.6; (f) γ=0.5; (g) γ=0.4; (h) γ=0.3)

图5 特征点提取结果

Fig. 5 Extraction results of feature points

图6 特征匹配结果((a)间隔0.1 m；(b)间隔0.4 m)

Fig. 6 Matching results of feature points ((a) 0.1 m interval; (b) 0.4 m interval)

图7 相机位置姿态与电缆隧道稀疏点云

Fig. 7 Camera position and cable tunnel sparse point cloud

图8 电缆隧道密集点云

Fig. 8 Cable tunnel dense point cloud

图9 电缆隧道模型重建((a)划分网格；(b)模型重建)

Fig. 9 Cable tunnel model reconstruction ((a) Mesh division; (b) Model reconstruction)

图10 纹理空间

Fig. 10 Texture space

图11 电缆隧道三维纹理模型

Fig. 11 Cable tunnel 3D texture model

图12 电缆隧道平铺纹理模型

Fig. 12 Cable tunnel tiled texture model

图13 各工况均方根重投影误差

Fig. 13 Root mean square reprojection error for each condition

图14 各工况点云密度

Fig. 14 Point cloud density for each condition

图15 各工况建模所用时间

Fig. 15 Modeling time for each condition

图16 各工况特征点个数((a) 50张图像；(b) 100张图像；(c) 200张图像；(d) 500张图像)

Fig. 16 Number of feature points for each condition ((a) 50 images; (b) 100 images; (c) 200 images; (d) 500 images)

图17 各工况特征点个数((a)低质量；(b)中质量；(c)高质量)

Fig. 17 The number of characteristic points of each condition ((a) Low quality; (b) Medium quality; (c) High quality)

参考文献 18

[1]	STENYTOUMIS C, LOUPOS K, DOULAMIS A, et al. A computer vision system for tunnel inspection[EB/OL]. [2022-07-13]. https://www.researchgate.net/publication/269317782_A_Computer_Vision_System_For_Tunnel_Inspection.
[2]	SUNG C, KIM P Y. 3D terrain reconstruction of construction sites using a stereo camera[J]. Automation in Construction, 2016, 64: 65-77. DOI URL
[3]	王资健, 马宏飞, 熊一丹, 等. 基于变交摄影的隧道3D重建方法[J]. 测绘科学, 2018, 43(6): 72-77.
	WANG Z J, MA H F, XIONG Y D, et al. The method of tunnel 3D reconstruction based on changeable photography[J]. Science of Surveying and Mapping, 2018, 43(6): 72-77. (in Chinese)
[4]	GARCÍA-LUNA R, SENENT S, JURADO-PIÑA R, et al. Structure from motion photogrammetry to characterize underground rock masses: experiences from two real tunnels[J]. Tunnelling and Underground Space Technology, 2019, 83: 262-273. DOI URL
[5]	阳军生, 张宇, 祝志恒, 等. 基于图像三维重建的隧道超欠挖检测方法研究[J]. 中南大学学报: 自然科学版, 2020, 51(3): 714-723.
	YANG J S, ZHANG Y, ZHU Z H, et al. Study on tunnel under-over break detection method based on three-dimensional image reconstruction technology[J]. Journal of Central South University: Science and Technology, 2020, 51(3): 714-723. (in Chinese)
[6]	XUE Y D, ZHANG S, ZHOU M L, et al. Novel SfM-DLT method for metro tunnel 3D reconstruction and Visualization[J]. Underground Space, 2021, 6(2): 134-141. DOI URL
[7]	曹先革, 李东海, 冯金宇, 等. 基于地面激光扫描技术获取地铁隧道3维点云数据的设计与试验[J]. 测绘与空间地理信息, 2013, 36(10): 12-14, 18.
	CAO X G, LI D H, FENG J Y, et al. Design and test of subway 3D point cloud data obtained based on ground 3D laser scanning technology[J]. Geomatics & Spatial Information Technology, 2013, 36(10): 12-14, 18. (in Chinese)
[8]	XIE X Y, LU X Z. Development of a 3D modeling algorithm for tunnel deformation monitoring based on terrestrial laser scanning[J]. Underground Space, 2017, 2(1): 16-29. DOI URL
[9]	LOWE D G. Distinctive image features from scale-invariant keypoints[J]. International Journal of Computer Vision, 2004, 60(2): 91-110. DOI URL
[10]	FISCHLER M A, BOLLES R C. Random sample consensus: a paradigm for model fitting with applications to image analysis and automated cartography[J]. Communications of the ACM, 1981, 24(6): 381-395. DOI URL
[11]	FISCHLER M A, BOLLES R C. Random sample consensus[J]. Communications of the ACM, 1981, 24(6): 381-395. DOI URL
[12]	KAZHDAN M, BOLITHO M, HOPPE H. Poisson surface reconstruction[C]// The 4th Eurographics Symposium on Geometry Processing. New York: ACM, 2006: 61-70.
[13]	KAZHDAN M, HOPPE H. Screened Poisson surface reconstruction[J]. ACM Transactions on Graphics, 2013, 32(3): 1-13.
[14]	CALAKLI F, TAUBIN G. SSD: smooth signed distance surface reconstruction[J]. Computer Graphics Forum, 2011, 30(7): 1993-2002. DOI URL
[15]	DIGNE J, MOREL J M, SOUZANI C M, et al. Scale space meshing of raw data point sets[J]. Computer Graphics Forum, 2011, 30(6): 1630-1642. DOI URL
[16]	WAECHTER M, MOEHRLE N, GOESELE M. Let there be color! Large-scale texturing of 3D reconstructions[C]// European Conference on Computer Vision. Cham: Springer International Publishin, 2014: 836-850.
[17]	张杰, 周浦城, 张谦. 基于迭代多尺度引导滤波Retinex的低照度图像增强[J]. 图学学报, 2018, 39(1): 1-11.
	ZHANG J, ZHOU P C, ZHANG Q. Low-light image enhancement based on iterative multi-scale guided filter retinex[J]. Journal of Graphics, 2018, 39(1): 1-11. (in Chinese)
[18]	毕秀丽, 邱雨檬, 肖斌, 等. 基于统计特征的图像直方图均衡化检测方法[J]. 计算机学报, 2021, 44(2): 292-303.
	BI X L, QIU Y M, XIAO B, et al. Histogram equalization detection based on statistical features in digital image[J]. Chinese Journal of Computers, 2021, 44(2): 292-303. (in Chinese)

基于SfM的城市电缆隧道三维重建方法优化研究

Optimization of 3D reconstruction method for urban cable tunnel based on SfM method

RichHTML

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

图/表 17

参考文献 18

相关文章 15

编辑推荐

Metrics

本文评价

[1]	薛皓玮, 王美丽. 融合生物力学约束与多模态数据的手部重建 [J]. 图学学报, 2023, 44(4): 794-800.
[2]	王江安, 庞大为, 黄乐, 秦林珍. 基于多尺度特征递归卷积的稠密点云重建网络 [J]. 图学学报, 2022, 43(5): 875-883.
[3]	白静, 孟庆亮, 徐昊, 范有福, 杨瞻源. ST-Rec3D：基于结构和目标感知的三维重建[J]. 图学学报, 2022, 43(3): 469-477.
[4]	王浩 1，邵堃 1，霍星 2，杨鹏 1，檀结庆 2 . 一种基于极大特征点的三维椎骨分割方法[J]. 图学学报, 2019, 40(1): 40-45.
[5]	田建勇，石林江 . 局部线性嵌入与模糊 C-均值聚类的红外图像彩色化算法[J]. 图学学报, 2018, 39(5): 917-925.
[6]	邹晓1，陈正鸣1,2，朱红强1，童晶1,2. 基于移动平台的三维虚拟试发型系统实现及应用[J]. 图学学报, 2018, 39(2): 309-316.
[7]	罗昌平1,2，夏海波1，赵琼3. 基于运动视差的三维重建技术在自导引小车中的应用[J]. 图学学报, 2017, 38(4): 615-622.
[8]	缪永伟1,2，冯小红2，林海斌2，张旭东2，高飞2. 2D-3D 转换:由二维线画图生成三维复杂零部件[J]. 图学学报, 2017, 38(2): 162-169.
[9]	陈培兴，王明泉，李世虎，侯慧玲，王玉. 基于形态学和Otsu 的固体火箭发动机CT 缺陷三维分割[J]. 图学学报, 2015, 36(4): 581-586.
[10]	张佳佳，孟宪罗，蒋涛，刘宇. 气象计量自动化温度检定系统中的图像预处理研究[J]. 图学学报, 2015, 36(3): 352-355.
[11]	孙正. 血管内超声图像后处理综述[J]. 图学学报, 2015, 36(3): 468-476.
[12]	孙修恩，张典华，孙昕萍. 三维数字重建在青铜器修复中的应用研究[J]. 图学学报, 2014, 35(6): 912-917.
[13]	李雪，高满屯，赵军. 由物体线框模型构建表面模型研究综述[J]. 图学学报, 2014, 35(5): 663-668.
[14]	江平，张锦. 一种结合CDD 模型和Criminisi 算法的图像修复算法[J]. 图学学报, 2014, 35(5): 741-746.
[15]	王承博，朱登明. 数据驱动的大规模水面动画合成方法[J]. 图学学报, 2014, 35(4): 491-497.