Article(id=1228279672945967956, tenantId=1146029695717560320, journalId=1146123166801305609, issueId=1228279664221815452, articleNumber=null, orderNo=null, doi=10.12404/j.issn.1671-1815.2408670, pmid=null, cstr=null, oa=null, hot=null, price=null, onlineType=0, articleFormat=0, articleType=null, articleTypeStr=research-article, receivedDate=1732032000000, receivedDateStr=2024-11-20, revisedDate=1747238400000, revisedDateStr=2025-05-15, acceptedDate=null, acceptedDateStr=null, onlineDate=1770774294362, onlineDateStr=2026-02-11, pubDate=1754582400000, pubDateStr=2025-08-08, doiRegisterDate=null, doiRegisterDateStr=null, onlineIssueDate=1770774294362, onlineIssueDateStr=2026-02-11, onlineJustAcceptDate=null, onlineJustAcceptDateStr=null, onlineFirstDate=null, onlineFirstDateStr=null, sourceXml=null, magXml=null, createTime=1770774294362, creator=13701087609, updateTime=1770774294362, updator=13701087609, issue=Issue{id=1228279664221815452, tenantId=1146029695717560320, journalId=1146123166801305609, year='2025', volume='25', issue='22', pageStart='9211', pageEnd='9648', issueExtLink='null', onlineDate='null', pubDate='null', beforeIssueId=null, nextIssueId=null, price=null, status=1, issueComplete=1, articleOrder=1, issueType=1, specialIssue=null, createTime=1770774292283, creator=13701087609, updateTime=1770777611996, updator=13701087609, preIssue=null, nextIssue=null, ext={EN=IssueExt(id=1228293588207992892, tenantId=1146029695717560320, journalId=1146123166801305609, issueId=1228279664221815452, language=EN, specialIssueTitle=, coverIllustrator=null, specialIssueEditor=, specialIssueAbout=), CN=IssueExt(id=1228293588207992893, tenantId=1146029695717560320, journalId=1146123166801305609, issueId=1228279664221815452, language=CN, specialIssueTitle=, coverIllustrator=null, specialIssueEditor=, specialIssueAbout=)}, issueFiles=null}, startPage=9327, endPage=9334, ext={EN=ArticleExt(id=1228279673470255980, articleId=1228279672945967956, tenantId=1146029695717560320, journalId=1146123166801305609, language=EN, title=3D Reconstruction of Underground Roadway in Coal Mine Based on Improved 2DGS, columnId=1228279673277313076, journalTitle=Science Technology and Engineering, columnName=Papers·Mining and Metallurgical Engineering, runingTitle=null, highlight=null, articleAbstract=

Underground coal mining faces hazards like gas explosions, coal dust explosions, and fires, underscoring the need for safety robotics based on 3D reconstruction. While essential for precise navigation and detection in complex environments, traditional methods fall short in data quality, accuracy, and cost. To address this, a 3D reconstruction method called 2DGS-DbTrans for underground coal mine tunnels was proposed, which is based on pure vision-based 3D reconstruction technology. To improve image resolution, a Transformer module was designed to enhance the input images, consisting of two core components: the multi-head depth convolutional axis attention mechanism and the deep convolutional gated network. In the processing workflow, sparse point clouds were first generated using Colmap, and the underground mine tunnel environment was represented by 2D Gaussian surfaces, where each Gaussian surface contains the coordinates, color, size, and orientation information of the objects. In addition, two loss functions were defined: the color loss function and the road smoothness loss function. Experimental results show that the 2DGS-DbTrans method outperforms other methods in 3D reconstruction of underground coal mine tunnels.

, correspAuthors=Chao-yu YANG, authorNote=null, correspAuthorsNote=null, copyrightStatement=null, copyrightOwner=null, extLink=null, articleAbsUrl=null, sourceXml=null, magXml=null, pdfUrl=null, pdf=null, pdfFileSize=null, pdfExtLink=null, richHtmlUrl=null, mobilePdfUrl=null, reviewReport=null, pdfFirstPage=null, abstractGraph=null, abstractGraphContent=null, abstractVideo=null, citation=null, cebUrl=null, magXmlContent=null, mapNumber=null, authorCompany=null, fund=null, authors=null, authorsList=Meng SUN, Chao-yu YANG), CN=ArticleExt(id=1228279684182508003, articleId=1228279672945967956, tenantId=1146029695717560320, journalId=1146123166801305609, language=CN, title=基于改进2DGS的煤矿井下巷道三维重建, columnId=1228279674833399960, journalTitle=科学技术与工程, columnName=论文·矿冶工程, runingTitle=null, highlight=null, articleAbstract=

煤矿井下作业面临瓦斯爆炸、煤尘爆炸及火灾等安全隐患,这使得依赖三维重建技术的安全机器人研究成为重点方向。三维重建在复杂环境中的精准导航和检测至关重要,但传统探测方法在数据质量、精度及成本方面存在不足。为此,提出了一种名为2DGS-DbTrans的煤矿井下巷道三维重建方法,这是一种基于纯视觉的三维重建技术。为提升图像分辨率,设计了一个Transformer模块对输入图像进行增强处理,其中包含两个核心组件:多头深度卷积轴注意力机制和深度卷积门控网络。在处理流程中,首先利用Colmap生成稀疏点云,并通过二维高斯曲面来表示煤矿井下巷道环境,每个高斯曲面包含物体的坐标、颜色、大小和方向信息。此外,定义了两个损失函数,即颜色损失函数和道路平滑度损失函数。实验结果表明,2DGS-DbTrans方法在煤矿井下巷道环境的三维重建中,获得了优于其他方法的性能。

, correspAuthors=杨超宇, authorNote=null, correspAuthorsNote=
* 杨超宇(1981—),男,汉族,安徽淮南人,博士,教授。研究方向:计算机视觉。E-mail:
, copyrightStatement=null, copyrightOwner=null, extLink=null, articleAbsUrl=null, sourceXml=5wj3m6ulgQr+Lia9Rwakvg==, magXml=RTrlwYjgC15Igxomi5fpmA==, pdfUrl=null, pdf=hDUJlqcVw6QKrpb8Lpi9ng==, pdfFileSize=16989691, pdfExtLink=null, richHtmlUrl=null, mobilePdfUrl=null, reviewReport=null, pdfFirstPage=null, abstractGraph=zO3f9qtxLTcBb9ZzT6kZog==, abstractGraphContent=null, abstractVideo=null, citation=null, cebUrl=null, magXmlContent=39HOjoNF6iztTf2RO3S+hQ==, mapNumber=null, authorCompany=null, fund=null, authors=

孙蒙(2000—),男,汉族,安徽马鞍山人,硕士研究生。研究方向:计算机视觉。E-mail:

, authorsList=孙蒙, 杨超宇)}, authors=[Author(id=1228369767749317214, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, orderNo=0, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=2501643875@qq.com, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1228369767854174824, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, authorId=1228369767749317214, language=EN, stringName=Meng SUN, firstName=Meng, middleName=null, lastName=SUN, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=College of Artificial Intelligence, Anhui University of Science and Technology, Huainan 232000, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1228369767984198260, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, authorId=1228369767749317214, language=CN, stringName=孙蒙, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=安徽理工大学人工智能学院, 淮南 232000, bio={"content":"

孙蒙(2000—),男,汉族,安徽马鞍山人,硕士研究生。研究方向:计算机视觉。E-mail:

"}, bioImg=null, bioContent=

孙蒙(2000—),男,汉族,安徽马鞍山人,硕士研究生。研究方向:计算机视觉。E-mail:

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1228369767615099471, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, xref=null, ext=[AuthorCompanyExt(id=1228369767640265300, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, companyId=1228369767615099471, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=College of Artificial Intelligence, Anhui University of Science and Technology, Huainan 232000, China), AuthorCompanyExt(id=1228369767652848215, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, companyId=1228369767615099471, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=安徽理工大学人工智能学院, 淮南 232000)])]), Author(id=1228369768940499584, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, orderNo=1, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=yangchy@aust.edu.cn, emailSecond=null, emailThird=null, correspondingAuthor=1, authorType=1, ext={EN=AuthorExt(id=1228369769464787591, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, authorId=1228369768940499584, language=EN, stringName=Chao-yu YANG, firstName=Chao-yu, middleName=null, lastName=YANG, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=*, address=College of Artificial Intelligence, Anhui University of Science and Technology, Huainan 232000, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1228369769582228111, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, authorId=1228369768940499584, language=CN, stringName=杨超宇, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=*, address=安徽理工大学人工智能学院, 淮南 232000, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1228369767615099471, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, xref=null, ext=[AuthorCompanyExt(id=1228369767640265300, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, companyId=1228369767615099471, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=College of Artificial Intelligence, Anhui University of Science and Technology, Huainan 232000, China), AuthorCompanyExt(id=1228369767652848215, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, companyId=1228369767615099471, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=安徽理工大学人工智能学院, 淮南 232000)])])], keywords=[Keyword(id=1228369769804526244, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, language=EN, orderNo=1, keyword=underground coal mine tunnel), Keyword(id=1228369769900995246, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, language=EN, orderNo=2, keyword=3D reconstruction), Keyword(id=1228369770010047162, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, language=EN, orderNo=3, keyword=Transformer), Keyword(id=1228369770152653510, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, language=EN, orderNo=4, keyword=point cloud), Keyword(id=1228369770286871248, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, language=EN, orderNo=5, keyword=2D Gaussian surface), Keyword(id=1228369770416894682, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, language=CN, orderNo=1, keyword=煤矿井下巷道), Keyword(id=1228369770500780769, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, language=CN, orderNo=2, keyword=三维重建), Keyword(id=1228369770727273195, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, language=CN, orderNo=3, keyword=Transformer), Keyword(id=1228369770848908020, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, language=CN, orderNo=4, keyword=点云), Keyword(id=1228369770970542844, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, language=CN, orderNo=5, keyword=二维高斯曲面)], refs=[Reference(id=1228369775747855301, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, doi=null, pmid=null, pmcid=null, year=2022, volume=22, issue=21, pageStart=9231, pageEnd=9237, url=null, language=null, rfNumber=[1], rfOrder=0, authorNames=李颖楠, 张引弟, 管奥成, journalName=科学技术与工程, refType=null, unstructuredReference=李颖楠, 张引弟, 管奥成, 等. 煤炭地下气化采出气的脱碳工艺优化及能效分析[J]. 科学技术与工程, 2022, 22 (21): 9231-9237., articleTitle=煤炭地下气化采出气的脱碳工艺优化及能效分析, refAbstract=null), Reference(id=1228369775865295821, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, doi=null, pmid=null, pmcid=null, year=2022, volume=22, issue=21, pageStart=9231, pageEnd=9237, url=null, language=null, rfNumber=[1], rfOrder=1, authorNames=Li Yingnan, Zhang Yindi, Guan Aocheng, journalName=Science Technology and Engineering, refType=null, unstructuredReference=Li Yingnan, Zhang Yindi, Guan Aocheng, et al. Decarbonization process optimization and energy efficiency analysis of coal underground gasification produced gas[J]. Science Technology and Engineering, 2022, 22(21): 9231-9237., articleTitle=Decarbonization process optimization and energy efficiency analysis of coal underground gasification produced gas, refAbstract=null), Reference(id=1228369775965959122, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, doi=null, pmid=null, pmcid=null, year=2023, volume=51, issue=1, pageStart=157, pageEnd=167, url=null, language=null, rfNumber=[2], rfOrder=2, authorNames=王卫军, 范磊, 马谕杰, journalName=煤炭科学技术, refType=null, unstructuredReference=王卫军, 范磊, 马谕杰, 等. 基于蝶形破坏理论深部巷道围岩控制技术研究[J]. 煤炭科学技术, 2023, 51(1): 157-167., articleTitle=基于蝶形破坏理论深部巷道围岩控制技术研究, refAbstract=null), Reference(id=1228369776062428120, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, doi=null, pmid=null, pmcid=null, year=2023, volume=51, issue=1, pageStart=157, pageEnd=167, url=null, language=null, rfNumber=[2], rfOrder=3, authorNames=Wang Weijun, Fan Lei, Ma Yujie, journalName=Coal Science and Technology, refType=null, unstructuredReference=Wang Weijun, Fan Lei, Ma Yujie, et al. Research on control technology of deep roadway perimeter rock based on butterfly damage theory[J]. Coal Science and Technology, 2023, 51(1): 157-167., articleTitle=Research on control technology of deep roadway perimeter rock based on butterfly damage theory, refAbstract=null), Reference(id=1228369776142119898, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, doi=null, pmid=null, pmcid=null, year=2024, volume=13, issue=14, pageStart=2850, pageEnd=2865, url=null, language=null, rfNumber=[3], rfOrder=4, authorNames=Gong W, Zhang L, Liao L, journalName=Electronics, refType=null, unstructuredReference=Gong W, Zhang L, Liao L, et al. Incremental SFM 3D reconstruction based on deep learning[J]. Electronics, 2024, 13(14): 2850-2865., articleTitle=Incremental SFM 3D reconstruction based on deep learning, refAbstract=null), Reference(id=1228369776230200288, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, doi=null, pmid=null, pmcid=null, year=2020, volume=null, issue=null, pageStart=1947, pageEnd=1956, url=null, language=null, rfNumber=[4], rfOrder=5, authorNames=Chen R, Han S, Xu J, journalName=Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, refType=null, unstructuredReference=Chen R, Han S, Xu J, et al. Fast-MVSNet: sparse-to-dense multi-view stereo with lear-ned propagation and Gauss-Newton refinement[C]// Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. New York: IEEE, 2020: 1947-1956., articleTitle=Fast-MVSNet: sparse-to-dense multi-view stereo with lear-ned propagation and Gauss-Newton refinement, refAbstract=null), Reference(id=1228369776318280675, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, doi=null, pmid=null, pmcid=null, year=2023, volume=48, issue=5, pageStart=1943, pageEnd=1956, url=null, language=null, rfNumber=[5], rfOrder=6, authorNames=明仕, 徐东, 王海川, journalName=煤炭学报, refType=null, unstructuredReference=明仕, 徐东, 王海川, 等. 特厚煤层巷道冲击破坏机理及全锚索支护技术[J]. 煤炭学报, 2023, 48(5): 1943-1956., articleTitle=特厚煤层巷道冲击破坏机理及全锚索支护技术, refAbstract=null), Reference(id=1228369776423138282, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, doi=null, pmid=null, pmcid=null, year=2023, volume=48, issue=5, pageStart=1943, pageEnd=1956, url=null, language=null, rfNumber=[5], rfOrder=7, authorNames=Ming Shi, Xu Dong, Wang Haichuan, journalName=Journal of Coal, refType=null, unstructuredReference=Ming Shi, Xu Dong, Wang Haichuan, et al. Mechanism of impact damage and all anchorcable support technology of roadway in extra-thick coal seam[J]. Journal of Coal, 2023, 48(5): 1943-1956., articleTitle=Mechanism of impact damage and all anchorcable support technology of roadway in extra-thick coal seam, refAbstract=null), Reference(id=1228369776536384495, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, doi=null, pmid=null, pmcid=null, year=2022, volume=null, issue=null, pageStart=8248, pageEnd=8258, url=null, language=null, rfNumber=[6], rfOrder=8, authorNames=Tancik M, Casser V, Yan X C, journalName=Proceedings of the IEEE/CVF Conference onComputer Vision and Pattern Recognition, refType=null, unstructuredReference=Tancik M, Casser V, Yan X C, et al. Block-nerf: scalable large scene neural view synthesis[C]// Proceedings of the IEEE/CVF Conference onComputer Vision and Pattern Recognition. New York: IEEE, 2022: 8248-8258., articleTitle=Block-nerf: scalable large scene neural view synthesis, refAbstract=null), Reference(id=1228369776645436403, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, doi=null, pmid=null, pmcid=null, year=2022, volume=null, issue=null, pageStart=12922, pageEnd=12931, url=null, language=null, rfNumber=[7], rfOrder=9, authorNames=Turki H, Ramanan D, Satya-narayanan M, journalName=Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, refType=null, unstructuredReference=Turki H, Ramanan D, Satya-narayanan M. Mega-nerf: scalableconstruction of large-scale nerfs for virtual flythroughs[C]// Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. New York: IEEE: 2022: 12922-12931., articleTitle=Mega-nerf: scalableconstruction of large-scale nerfs for virtual flythroughs, refAbstract=null), Reference(id=1228369776754488310, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, doi=null, pmid=null, pmcid=null, year=2022, volume=null, issue=null, pageStart=5470, pageEnd=5479, url=null, language=null, rfNumber=[8], rfOrder=10, authorNames=Barron J T, Mildenhall B, Verbin D, journalName=IEEE/CVF Conference on Computer Vision and Pattern Recognition, refType=null, unstructuredReference=Barron J T, Mildenhall B, Verbin D, et al. Mipnerf 360: unbounded antialiased neural radiance fields[C]// IEEE/CVF Conference on Computer Vision and Pattern Recognition. New York: IEEE, 2022: 5470-5479., articleTitle=Mipnerf 360: unbounded antialiased neural radiance fields, refAbstract=null), Reference(id=1228369776838374396, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, doi=null, pmid=null, pmcid=null, year=2023, volume=48, issue=1, pageStart=238, pageEnd=250, url=null, language=null, rfNumber=[9], rfOrder=11, authorNames=许江, 程亮, 彭守建, journalName=煤炭学报, refType=null, unstructuredReference=许江, 程亮, 彭守建, 等. 巷道结构对突出流体冲击特性的影响性试验研究[J]. 煤炭学报, 2023, 48(1): 238-250., articleTitle=巷道结构对突出流体冲击特性的影响性试验研究, refAbstract=null), Reference(id=1228369776918066176, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, doi=null, pmid=null, pmcid=null, year=2023, volume=48, issue=1, pageStart=238, pageEnd=250, url=null, language=null, rfNumber=[9], rfOrder=12, authorNames=Xu Jiang, Cheng Liang, Peng Shoujian, journalName=Journal of Coal, refType=null, unstructuredReference=Xu Jiang, Cheng Liang, Peng Shoujian, et al. Experimental study on the influence of roadway structure on the impact characteristics of protruding fluids[J]. Journal of Coal, 2023, 48(1): 238-250., articleTitle=Experimental study on the influence of roadway structure on the impact characteristics of protruding fluids, refAbstract=null), Reference(id=1228369778297991171, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, doi=null, pmid=null, pmcid=null, year=null, volume=11, issue=2, pageStart=8, pageEnd=14, url=null, language=null, rfNumber=[10], rfOrder=13, authorNames=Song L, Wang G M, Liu J M, journalName=arXiv preprint arXiv: 2309.05028, 2023, refType=null, unstructuredReference=Song L, Wang G M, Liu J M, et al. Scnerf: self-correcting neural radiance field with sparse views[J]. arXiv preprint arXiv: 2309.05028, 2023, 11(2): 8-14., articleTitle=Scnerf: self-correcting neural radiance field with sparse views, refAbstract=null), Reference(id=1228369778448986122, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, doi=null, pmid=null, pmcid=null, year=2023, volume=12, issue=2, pageStart=1, pageEnd=8, url=null, language=null, rfNumber=[11], rfOrder=14, authorNames=Zhu S, Wang G M, Blum H, journalName=arXiv preprint ar-Xiv, refType=null, unstructuredReference=Zhu S, Wang G M, Blum H, et al. Snislam: semantic neural implicit slam[J]. arXiv preprint ar-Xiv, 2023, 12(2): 1-8., articleTitle=Snislam: semantic neural implicit slam, refAbstract=null), Reference(id=1228369778532872207, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, doi=null, pmid=null, pmcid=null, year=2023, volume=13, issue=4, pageStart=1, pageEnd=14, url=null, language=null, rfNumber=[12], rfOrder=15, authorNames=Kerbl B, Kopanas G, Lei-mkÜhler T, journalName=ACM Transactions on Graphics, refType=null, unstructuredReference=Kerbl B, Kopanas G, Lei-mkÜhler T, et al. 3D Gaussian splatting for real-time radiance field rendering[J]. ACM Transactions on Graphics, 2023, 13(4): 1-14., articleTitle=3D Gaussian splatting for real-time radiance field rendering, refAbstract=null), Reference(id=1228369778662895637, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, doi=null, pmid=null, pmcid=null, year=2024, volume=8, issue=3, pageStart=1, pageEnd=12, url=null, language=null, rfNumber=[13], rfOrder=16, authorNames=Chen G K, Wang W G, journalName=arXiv preprint arXiv, refType=null, unstructuredReference=Chen G K, Wang W G. A survey on 3D Gaussian splatting[J]. arXiv preprint arXiv, 2024, 8(3): 1-12., articleTitle=A survey on 3D Gaussian splatting, refAbstract=null), Reference(id=1228369778792919066, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, doi=null, pmid=null, pmcid=null, year=2024, volume=null, issue=null, pageStart=1, pageEnd=19, url=null, language=null, rfNumber=[14], rfOrder=17, authorNames=Huang B B, Yu Z H, Chen A P, journalName=IEEE Conference on Computer Vision and Pattern Recognition, refType=null, unstructuredReference=Huang B B, Yu Z H, Chen A P, et al. 2D gaussian splatting for geometrically accurate radiance fields[C]// IEEE Conference on Computer Vision and Pattern Recognition. New York: ACM, 2024: 1-19., articleTitle=2D gaussian splatting for geometrically accurate radiance fields, refAbstract=null), Reference(id=1228369778918748190, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, doi=null, pmid=null, pmcid=null, year=2021, volume=null, issue=null, pageStart=1, pageEnd=21, url=null, language=null, rfNumber=[15], rfOrder=18, authorNames=Dosovitskiy A, Beyer L, Kolesnikov A, journalName=The International Conference on Learning Representations, refType=null, unstructuredReference=Dosovitskiy A, Beyer L, Kolesnikov A, et al. An image is worth 16×16 words: transformers for image recognition at scale[C]// The International Conference on Learning Representations. New York: IEEE, 2021: 1-21., articleTitle=An image is worth 16×16 words: transformers for image recognition at scale, refAbstract=null), Reference(id=1228369779027800099, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, doi=null, pmid=null, pmcid=null, year=2021, volume=null, issue=null, pageStart=1021, pageEnd=1025, url=null, language=null, rfNumber=[16], rfOrder=19, authorNames=Touvron H, Cord M, Douze M, journalName=International Conference on Machine Learning, refType=null, unstructuredReference=Touvron H, Cord M, Douze M, et al. Training data efficient image transformers & distillation through attention[C]// International Conference on Machine Learning. New York: IEEE, 2021: 1021-1025., articleTitle=Training data efficient image transformers & distillation through attention, refAbstract=null), Reference(id=1228369779132657701, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, doi=null, pmid=null, pmcid=null, year=2024, volume=24, issue=17, pageStart=6999, pageEnd=7013, url=null, language=null, rfNumber=[17], rfOrder=20, authorNames=许时昂, 吴海波, 欧元超, journalName=科学技术与工程, refType=null, unstructuredReference=许时昂, 吴海波, 欧元超, 等. 采煤沉陷松散层变形研究现状与分析[J]. 科学技术与工程, 2024, 24(17): 6999-7013., articleTitle=采煤沉陷松散层变形研究现状与分析, refAbstract=null), Reference(id=1228369779371733041, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, doi=null, pmid=null, pmcid=null, year=2024, volume=24, issue=17, pageStart=6999, pageEnd=7013, url=null, language=null, rfNumber=[17], rfOrder=21, authorNames=Xu Shiang, Wu Haibo, Ou Yuanchao, journalName=Science Technology and Engineering, refType=null, unstructuredReference=Xu Shiang, Wu Haibo, Ou Yuanchao, et al. Research status and analysis of loose layerdeformation under coal mining subsidence condition[J]. Science Technology and Engineering, 2024, 24(17): 6999-7013., articleTitle=Research status and analysis of loose layerdeformation under coal mining subsidence condition, refAbstract=null), Reference(id=1228369779493367864, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, doi=null, pmid=null, pmcid=null, year=2021, volume=8, issue=3, pageStart=1, pageEnd=10, url=null, language=null, rfNumber=[18], rfOrder=22, authorNames=Xie E Z, Wang W H, Yu Z D, journalName=arXiv preprint arXiv, refType=null, unstructuredReference=Xie E Z, Wang W H, Yu Z D, et al. Segformer: simple and efficient design for semantic segmentation with transformers[J]. arXiv preprint arXiv, 2021, 8(3): 1-10., articleTitle=Segformer: simple and efficient design for semantic segmentation with transformers, refAbstract=null), Reference(id=1228369779585642558, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, doi=null, pmid=null, pmcid=null, year=2024, volume=null, issue=null, pageStart=1, pageEnd=13, url=null, language=null, rfNumber=[19], rfOrder=23, authorNames=Hamdi A, Melas-Kyriazi L, Mai J J, journalName=IEEE Conference on Computer Vision and Pattern Recognition, refType=null, unstructuredReference=Hamdi A, Melas-Kyriazi L, Mai J J, et al. GES: generalized expone-ntial splatting for efficient radiance field rendering[C]// IEEE Conference on Computer Vision and Pattern Recognition. Seattle: IEEE, 2024: 1-13., articleTitle=GES: generalized expone-ntial splatting for efficient radiance field rendering, refAbstract=null), Reference(id=1228369779669528643, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, doi=null, pmid=null, pmcid=null, year=2024, volume=null, issue=null, pageStart=1, pageEnd=10, url=null, language=null, rfNumber=[20], rfOrder=24, authorNames=Yu Z H, Chen A P, Huang B B, journalName=IEEE Conference on Computer Vision and Pattern Recognition, refType=null, unstructuredReference=Yu Z H, Chen A P, Huang B B, et al. Mip-spl-atting: alias-free 3D Gaussian splatting[C]// IEEE Conference on Computer Vision and Pattern Recognition. New York: IEEE, 2024: 1-10., articleTitle=Mip-spl-atting: alias-free 3D Gaussian splatting, refAbstract=null)], funds=[Fund(id=1228369775550723005, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, awardId=52227901, language=CN, fundingSource=国家自然科学基金(52227901), fundOrder=null, country=null)], companyList=[AuthorCompany(id=1228369767615099471, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, xref=null, ext=[AuthorCompanyExt(id=1228369767640265300, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, companyId=1228369767615099471, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=College of Artificial Intelligence, Anhui University of Science and Technology, Huainan 232000, China), AuthorCompanyExt(id=1228369767652848215, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, companyId=1228369767615099471, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=安徽理工大学人工智能学院, 淮南 232000)])], figs=[ArticleFig(id=1228369771163480840, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, language=EN, label=Fig.1, caption=2DGS-DbTrans overall process, figureFileSmall=A4oKQlKT29EmQsKih/wK9Q==, figureFileBig=eNCKdZcmeOgKA/8lr+hLuw==, tableContent=null), ArticleFig(id=1228369771276727054, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, language=CN, label=图1, caption=2DGS-DbTrans整体流程, figureFileSmall=A4oKQlKT29EmQsKih/wK9Q==, figureFileBig=eNCKdZcmeOgKA/8lr+hLuw==, tableContent=null), ArticleFig(id=1228369771444499222, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, language=EN, label=Fig.2, caption=Transformer image enhancement module, figureFileSmall=H5S+SIhTRE6ARm9PIPxOfQ==, figureFileBig=LsF0RRF+9wCX2uKsZ7w89w==, tableContent=null), ArticleFig(id=1228369771549356827, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, language=CN, label=图2, caption=Transformer图像增强模块, figureFileSmall=H5S+SIhTRE6ARm9PIPxOfQ==, figureFileBig=LsF0RRF+9wCX2uKsZ7w89w==, tableContent=null), ArticleFig(id=1228369771645825824, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, language=EN, label=Fig.3, caption=Attention mechanism module for multi-headed deep convolutional axes in Transformer Block, figureFileSmall=BDF+T8i4RfVnJ5AL94sSAA==, figureFileBig=NxgApSjna3/PUmYIzS5B6w==, tableContent=null), ArticleFig(id=1228369771771654954, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, language=CN, label=图3, caption=Transformer Block中多头深度卷积轴注意力机制模块, figureFileSmall=BDF+T8i4RfVnJ5AL94sSAA==, figureFileBig=NxgApSjna3/PUmYIzS5B6w==, tableContent=null), ArticleFig(id=1228369771910066993, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, language=EN, label=Fig.4, caption=Deep convolutional gating network module in transformer block, figureFileSmall=2hYCrZkgOgTo1bokJxScZQ==, figureFileBig=1S6zVEvFYxUy4qYnhYpUmw==, tableContent=null), ArticleFig(id=1228369772019118902, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, language=CN, label=图4, caption=Transformer Block中深度卷积门控网络模块, figureFileSmall=2hYCrZkgOgTo1bokJxScZQ==, figureFileBig=1S6zVEvFYxUy4qYnhYpUmw==, tableContent=null), ArticleFig(id=1228369772115587898, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, language=EN, label=Fig.5, caption=Plot of data from the experimental section, figureFileSmall=048vriCw1Oc5V+d09o5WGQ==, figureFileBig=Osu/bS2aAgJOhjPpP+bqng==, tableContent=null), ArticleFig(id=1228369772270777154, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, language=CN, label=图5, caption=实验部分数据图, figureFileSmall=048vriCw1Oc5V+d09o5WGQ==, figureFileBig=Osu/bS2aAgJOhjPpP+bqng==, tableContent=null), ArticleFig(id=1228369772354663239, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, language=EN, label=Fig.6, caption=Reconstruction results of underground coal mine tunnel environment under different models, figureFileSmall=hXx1j6cFZARoopG053aUsQ==, figureFileBig=nNIMdnA9rlsL6LJhHO5ebA==, tableContent=null), ArticleFig(id=1228369772476298062, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, language=CN, label=图6, caption=不同模型下煤矿井下巷道环境重建结果, figureFileSmall=hXx1j6cFZARoopG053aUsQ==, figureFileBig=nNIMdnA9rlsL6LJhHO5ebA==, tableContent=null), ArticleFig(id=1228369773889778516, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, language=EN, label=Fig.7, caption=Variation of PSNR values with number of iterations for different models, figureFileSmall=5zL/tFa4JOSQ/NOnBGECiA==, figureFileBig=sOUnIQOR2DazCZFal0Tx6w==, tableContent=null), ArticleFig(id=1228369773986247515, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, language=CN, label=图7, caption=不同模型下PSNR、SSIM、LPIPS随迭代次数变化, figureFileSmall=5zL/tFa4JOSQ/NOnBGECiA==, figureFileBig=sOUnIQOR2DazCZFal0Tx6w==, tableContent=null), ArticleFig(id=1228369774078522211, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, language=EN, label=Fig.8, caption=External view of 3D reconstruction of underground tunnel in coal mine, figureFileSmall=kcV7Ntjv5gA9lTdkICaQkg==, figureFileBig=Io2vCxJtqvatYMcDkFiyLQ==, tableContent=null), ArticleFig(id=1228369774200157037, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, language=CN, label=图8, caption=煤矿井下巷道三维重建外部视角图, figureFileSmall=kcV7Ntjv5gA9lTdkICaQkg==, figureFileBig=Io2vCxJtqvatYMcDkFiyLQ==, tableContent=null), ArticleFig(id=1228369774325986166, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, language=EN, label=Fig.9, caption=2D Gaussian surface visualization of tunnel, figureFileSmall=QZkfOm1m2Y6hn4ZaU1bsaA==, figureFileBig=39LbK5NUvspL99R3TfaRcw==, tableContent=null), ArticleFig(id=1228369774414066556, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, language=CN, label=图9, caption=巷道二维高斯曲面可视化, figureFileSmall=QZkfOm1m2Y6hn4ZaU1bsaA==, figureFileBig=39LbK5NUvspL99R3TfaRcw==, tableContent=null), ArticleFig(id=1228369774489564032, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, language=EN, label=Fig.10, caption=3D point cloud of tunnel, figureFileSmall=SrwN/oPkcAhiu49uI+w4pg==, figureFileBig=odmp5HivNwd1Jt1W0tjwAQ==, tableContent=null), ArticleFig(id=1228369774577644424, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, language=CN, label=图10, caption=巷道三维点云, figureFileSmall=SrwN/oPkcAhiu49uI+w4pg==, figureFileBig=odmp5HivNwd1Jt1W0tjwAQ==, tableContent=null), ArticleFig(id=1228369774665724811, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, language=EN, label=Fig.11, caption=Plot of results of ablation experiments B and C, figureFileSmall=bPQhYndMxBQpaiflWq4kgw==, figureFileBig=l8iv80EDL2mE/QaoqK1CKw==, tableContent=null), ArticleFig(id=1228369774753805200, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, language=CN, label=图11, caption=消融实验B和实验C的结果图, figureFileSmall=bPQhYndMxBQpaiflWq4kgw==, figureFileBig=l8iv80EDL2mE/QaoqK1CKw==, tableContent=null), ArticleFig(id=1228369774837691287, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, language=EN, label=Table 1, caption=

Experiment-related environments

, figureFileSmall=null, figureFileBig=null, tableContent=
CPU intel i9-13900H
GPU NVIDIA RTX 4090
Python 3.8
Pytorch 2.0.0
CUDA 11.8
), ArticleFig(id=1228369774913188764, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, language=CN, label=表1, caption=

实验相关环境

, figureFileSmall=null, figureFileBig=null, tableContent=
CPU intel i9-13900H
GPU NVIDIA RTX 4090
Python 3.8
Pytorch 2.0.0
CUDA 11.8
), ArticleFig(id=1228369775018046370, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, language=EN, label=Table 2, caption=

Comparison of 3D reconstruction of underground coal mine roadways under different methods

, figureFileSmall=null, figureFileBig=null, tableContent=
方法 PNSR SSIM LPIPS
3DGS 28.79 0.899 0.176
GES 28.71 0.898 0.181
2DGS 29.03 0.905 0.177
Mip-Splatting 27.52 0.889 0.186
本文方法 31.38 0.935 0.135
), ArticleFig(id=1228369775127098280, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, language=CN, label=表2, caption=

煤矿井下巷道在不同方法下的三维重建比较

, figureFileSmall=null, figureFileBig=null, tableContent=
方法 PNSR SSIM LPIPS
3DGS 28.79 0.899 0.176
GES 28.71 0.898 0.181
2DGS 29.03 0.905 0.177
Mip-Splatting 27.52 0.889 0.186
本文方法 31.38 0.935 0.135
), ArticleFig(id=1228369775252927406, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, language=EN, label=Table 3, caption=

Ablation experiment result data

, figureFileSmall=null, figureFileBig=null, tableContent=
实验 PSNR↑ SSIM↑ LPIPS↓
A(2DGS) 29.03 0.905 0.177
B 29.22 0.916 0.160
C 29.98 0.924 0.153
D(Ours) 31.38 0.935 0.135
), ArticleFig(id=1228369775366173618, tenantId=1146029695717560320, journalId=1146123166801305609, articleId=1228279672945967956, language=CN, label=表3, caption=

消融实验结果数据

, figureFileSmall=null, figureFileBig=null, tableContent=
实验 PSNR↑ SSIM↑ LPIPS↓
A(2DGS) 29.03 0.905 0.177
B 29.22 0.916 0.160
C 29.98 0.924 0.153
D(Ours) 31.38 0.935 0.135
)], attaches=null, journal=Journal(id=1146119176004939786, delFlag=0, nameCn=科学技术与工程, nameEn=Science Technology and Engineering, nameHistory1=null, nameHistory2=null, issn=1671-1815, eissn=, cn=11-4688/T, coden=null, periodic=4, language=CN, oaType=是, ccby=null, superviseOffice=null, ownerOffice=null, pubOffice=null, editorOffice=null, officeType=null, aims=null, clcCode=null, officeProv=null, officeCity=null, officeAddr=null, officeZip=null, officeEmail=null, officePhone=null, editDirector=null, officeDirector=null, officeDirectorPhone=null, officeStaffNum=null, officeEmpNum=null, coverPicUrl=UKU/O7GSka5polgCTkbIIw==, journalPrice=null, startedYear=null, abbrevIsoEn=Sci Technol Eng, journalRemark=null, publicationField=null, createdTime=null, updatedTime=1754445529766, createdBy=null, updatedBy=13701087609, firstLetterCn=S, firstLetterEn=S, subjectCode=Natural Sciences, subjectName=自然科学, subjectCodeEn=Natural Sciences, subjectNameEn=null, picCn=UKU/O7GSka5polgCTkbIIw==, picEn=5hwlULoNwcbj3xUmVi9MAQ==, jcr=null, cjcr=null, exts=[JournalExt(id=1159791870395564357, language=CN, name=科学技术与工程, nameHistory1=null, nameHistory2=null, managedBy=, sponsoredBy=, publishedBy=, editorOffice=, officeProv=null, officeCity=null, officeAddr=, officeZip=, editDirector=null, officeDirector=null, officePhone=null, coverPicUrl=null, journalRemark=, submitArticleUrl=null, websiteUrl=http://www.stae.com.cn/jsygc/home, createdTime=1754445529793, updatedTime=1754445529793, createdBy=13701087609, updatedBy=13701087609, submissionGuidelinesUrl=http://www.stae.com.cn/jsygc/site/menus/20090429150146001, submissionAuthorUrl=http://www.stae.com.cn/jsygc/author/login, submissionEditorUrl=http://www.stae.com.cn/jsygc/editor/login, submissionReviewUrl=http://www.stae.com.cn/jsygc/reviewer/login, submissionCeEditorUrl=, submissionAeEditorUrl=, option={"copyright":""}), JournalExt(id=1159791870441701702, language=EN, name=Science Technology and Engineering, nameHistory1=null, nameHistory2=null, managedBy=, sponsoredBy=, publishedBy=, editorOffice=, officeProv=null, officeCity=null, officeAddr=, officeZip=, editDirector=null, officeDirector=null, officePhone=null, coverPicUrl=null, journalRemark=, submitArticleUrl=null, websiteUrl=http://www.stae.com.cn/jsygc/home, createdTime=1754445529804, updatedTime=1754445529804, createdBy=13701087609, updatedBy=13701087609, submissionGuidelinesUrl=, submissionAuthorUrl=http://www.stae.com.cn/jsygc/author/login, submissionEditorUrl=http://www.stae.com.cn/jsygc/editor/login, submissionReviewUrl=http://www.stae.com.cn/jsygc/reviewer/login, submissionCeEditorUrl=, submissionAeEditorUrl=, option={"copyright":""})], databaseList=null, tenantJournalId=1146123166801305609, websiteList=[Website(id=1148243202391400884, webName=null, webTitle=null, webDomain=null, webCopyrigh=null, webIpcNo=null, seoTitle=null, seoKeywords=null, seoDescription=null, tenantJournalId=null, journalId=1146123166801305609, journalNameCn=null, journalNameEn=null, grayFlag=null, tenantId=1146029695717560320, platformId=null, journalGroupId=null, journalGroupNameCn=null, journalGroupNameEn=null, type=1, domain=https://castjournals.cast.org.cn/joweb/kxjsygc/CN, language=CN, createTime=1751692112777, createBy=18614031015, updateTime=1753520965431, updateBy=18614031015, name=科学技术与工程-中文站点, tplId=1146099689490845704, title=科学技术与工程, delFlag=0, indexPage=/home, props=[WebsiteProps(id=1148622798802673703, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1148243202391400884, code=articleTextType, value=kx, createTime=1751782615614, updateTime=1751782615614, creator=18614031015, updator=18614031015), WebsiteProps(id=1148622798781702180, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1148243202391400884, code=banner, value=null, createTime=1751782615609, updateTime=1751782615609, creator=18614031015, updator=18614031015), WebsiteProps(id=1148622798769119267, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1148243202391400884, code=logo, value=https://castjournals.cast.org.cn/joweb/kjdb/CN/file/pic?fileId=j86gbwi+p0Idkyl5SzIlmQ==, createTime=1751782615606, updateTime=1751782615606, creator=18614031015, updator=18614031015), WebsiteProps(id=1148622798794285094, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1148243202391400884, code=picServerUrl, value=https://castjournals.cast.org.cn/joweb/kjdb/CN/file/pic, createTime=1751782615612, updateTime=1751782615612, creator=18614031015, updator=18614031015), WebsiteProps(id=1148622798790090789, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1148243202391400884, code=staticResourcePath, value=https://castjournals.cast.org.cn/joweb/cast_kjdb_cn_619/, createTime=1751782615611, updateTime=1751782615611, creator=18614031015, updator=18614031015)]), Website(id=1155914124811976731, webName=null, webTitle=null, webDomain=null, webCopyrigh=null, webIpcNo=null, seoTitle=null, seoKeywords=null, seoDescription=null, tenantJournalId=null, journalId=1146123166801305609, journalNameCn=null, journalNameEn=null, grayFlag=null, tenantId=1146029695717560320, platformId=null, journalGroupId=null, journalGroupNameCn=null, journalGroupNameEn=null, type=1, domain=https://castjournals.cast.org.cn/joweb/kxjsygc/EN, language=EN, createTime=1753521003206, createBy=18614031015, updateTime=1753521003206, updateBy=18614031015, name=科学技术与工程-英文站点, tplId=1146101810881728533, title=Science Technology and Engineering, delFlag=0, indexPage=/home, props=[WebsiteProps(id=1155914371227308235, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1155914124811976731, code=articleTextType, value=kx, createTime=1753521061952, updateTime=1753521061952, creator=18614031015, updator=18614031015), WebsiteProps(id=1155914371210531016, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1155914124811976731, code=banner, value=null, createTime=1753521061947, updateTime=1753521061947, creator=18614031015, updator=18614031015), WebsiteProps(id=1155914371202142407, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1155914124811976731, code=logo, value=https://castjournals.cast.org.cn/joweb/kjdb/CN/file/pic?fileId=j86gbwi+p0Idkyl5SzIlmQ==, createTime=1753521061945, updateTime=1753521061945, creator=18614031015, updator=18614031015), WebsiteProps(id=1155914371223113930, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1155914124811976731, code=picServerUrl, value=https://castjournals.cast.org.cn/joweb/kjdb/CN/file/pic, createTime=1753521061950, updateTime=1753521061950, creator=18614031015, updator=18614031015), WebsiteProps(id=1155914371218919625, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1155914124811976731, code=staticResourcePath, value=https://castjournals.cast.org.cn/joweb/cast_kjdb_cn_619/, createTime=1753521061949, updateTime=1753521061949, creator=18614031015, updator=18614031015)])], journalTitle=科学技术与工程, weixinUrl=null, journalUrl=null, iacademicId=null, status=0, seqNo=null, journalTitleEn=Science Technology and Engineering, journalPhotoCn=UKU/O7GSka5polgCTkbIIw==, journalPhotoEn=5hwlULoNwcbj3xUmVi9MAQ==, journalFirstLetter=S, journalRecommend=null, journalNew=null, journalCollection=null, jcrJf=null, cjcrJf=null, jcrJfStr=null, cjcrJfStr=null, submissionFirstDecision=null, sciSubjectClassification=null, casSubjectClassification=null, citeScore=null, totalCitationFrequency=null, icpCode=null, psCode=null, advertisingLicenseCode=null, copyrightInformation=null, country=null, option=null, provinceCode=null, provinceName=null, collectFlag=false), detailUrlCn=https://castjournals.cast.org.cn/joweb/kxjsygc/CN/10.12404/j.issn.1671-1815.2408670, detailUrlEn=https://castjournals.cast.org.cn/joweb/kxjsygc/EN/10.12404/j.issn.1671-1815.2408670, pdfUrlCn=https://castjournals.cast.org.cn/joweb/kxjsygc/CN/PDF/10.12404/j.issn.1671-1815.2408670, pdfUrlEn=https://castjournals.cast.org.cn/joweb/kxjsygc/EN/PDF/10.12404/j.issn.1671-1815.2408670, aliStartDate=null, aliEndDate=null, collectionFlag=false, citedCount=null, citedUrl=null, reference=null)
收藏切换
基于改进2DGS的煤矿井下巷道三维重建
收藏切换
PDF下载
孙蒙 , 杨超宇 *
科学技术与工程 | 论文·矿冶工程 2025,25(22): 9327-9334
收起
收藏切换
科学技术与工程 | 论文·矿冶工程 2025, 25(22): 9327-9334
基于改进2DGS的煤矿井下巷道三维重建
全屏
孙蒙 , 杨超宇*
作者信息
  • 安徽理工大学人工智能学院, 淮南 232000
  • 孙蒙(2000—),男,汉族,安徽马鞍山人,硕士研究生。研究方向:计算机视觉。E-mail:

通讯作者:

* 杨超宇(1981—),男,汉族,安徽淮南人,博士,教授。研究方向:计算机视觉。E-mail:
3D Reconstruction of Underground Roadway in Coal Mine Based on Improved 2DGS
Meng SUN , Chao-yu YANG*
Affiliations
  • College of Artificial Intelligence, Anhui University of Science and Technology, Huainan 232000, China
出版时间: 2025-08-08 doi: 10.12404/j.issn.1671-1815.2408670
文章导航
收藏切换

煤矿井下作业面临瓦斯爆炸、煤尘爆炸及火灾等安全隐患,这使得依赖三维重建技术的安全机器人研究成为重点方向。三维重建在复杂环境中的精准导航和检测至关重要,但传统探测方法在数据质量、精度及成本方面存在不足。为此,提出了一种名为2DGS-DbTrans的煤矿井下巷道三维重建方法,这是一种基于纯视觉的三维重建技术。为提升图像分辨率,设计了一个Transformer模块对输入图像进行增强处理,其中包含两个核心组件:多头深度卷积轴注意力机制和深度卷积门控网络。在处理流程中,首先利用Colmap生成稀疏点云,并通过二维高斯曲面来表示煤矿井下巷道环境,每个高斯曲面包含物体的坐标、颜色、大小和方向信息。此外,定义了两个损失函数,即颜色损失函数和道路平滑度损失函数。实验结果表明,2DGS-DbTrans方法在煤矿井下巷道环境的三维重建中,获得了优于其他方法的性能。

煤矿井下巷道  /  三维重建  /  Transformer  /  点云  /  二维高斯曲面

Underground coal mining faces hazards like gas explosions, coal dust explosions, and fires, underscoring the need for safety robotics based on 3D reconstruction. While essential for precise navigation and detection in complex environments, traditional methods fall short in data quality, accuracy, and cost. To address this, a 3D reconstruction method called 2DGS-DbTrans for underground coal mine tunnels was proposed, which is based on pure vision-based 3D reconstruction technology. To improve image resolution, a Transformer module was designed to enhance the input images, consisting of two core components: the multi-head depth convolutional axis attention mechanism and the deep convolutional gated network. In the processing workflow, sparse point clouds were first generated using Colmap, and the underground mine tunnel environment was represented by 2D Gaussian surfaces, where each Gaussian surface contains the coordinates, color, size, and orientation information of the objects. In addition, two loss functions were defined: the color loss function and the road smoothness loss function. Experimental results show that the 2DGS-DbTrans method outperforms other methods in 3D reconstruction of underground coal mine tunnels.

underground coal mine tunnel  /  3D reconstruction  /  Transformer  /  point cloud  /  2D Gaussian surface
孙蒙, 杨超宇. 基于改进2DGS的煤矿井下巷道三维重建. 科学技术与工程, 2025 , 25 (22) : 9327 -9334 . DOI: 10.12404/j.issn.1671-1815.2408670
Meng SUN, Chao-yu YANG. 3D Reconstruction of Underground Roadway in Coal Mine Based on Improved 2DGS[J]. Science Technology and Engineering, 2025 , 25 (22) : 9327 -9334 . DOI: 10.12404/j.issn.1671-1815.2408670
煤矿井下作业环境复杂多变,安全问题始终是煤矿生产中的重中之重。井下巷道的塌方、瓦斯泄漏等安全隐患给矿工的生命安全和矿山的正常运营带来了巨大风险。为了更有效地监测和预防这些潜在危险,现代化的矿井安全检测手段逐步引入了机器人技术。这些专门设计的机器人能够在极端环境下执行检测任务,提供实时的安全监控。然而,为了使机器人能够准确地在井下复杂的环境中进行导航、检测并作出正确反应,巷道环境的三维重建技术变得至关重要。通过三维重建,机器人可以获得精确的空间布局和环境信息,从而更高效地执行安全检测任务,提升井下作业的安全性[1]
随着煤矿开采深度的增加和开采区域的复杂性提升,传统的二维平面测量和监控手段已无法全面、准确地反映巷道内部结构和环境状态。三维重建技术通过对井下巷道进行全面的空间扫描与数据采集,能够生成精确的三维模型,为矿山管理者提供直观的巷道形态和环境信息。这不仅有助于提高矿山的安全监控能力,还可以优化巷道设计与维护,提升煤矿开采的效率与安全性。在此背景下,研究和应用井下巷道环境的三维重建技术显得尤为重要[2]
对于传统的三维重建方法,Gong等[3]侧重于几何整体结构的恢复和运动结构恢复(structure from motion,SfM),Chen等[4]侧重于多视图立体重建(multi-view stereo,MVS)方法重建场景中稀疏或半密集的点云。在煤矿井下巷道里,煤尘和粉尘容易悬浮在空气中,造成图像的对比度和清晰度降低,经常会出现孔洞和噪声[5]。文献[6-8]设计了基于Nerf(neural radiance fields)的三维重建方法可对于这些孔洞和噪声提供逼真的渲染效果。然而,这些方法很难从隐式表示中恢复巷道场景的形状[9-10],特别是在大型场景中。文献[11-13]提出基于3DGS(3D Gaussian splatting),大大提高了渲染速度,并在巷道内场景形状重建方面提供了天然优势。但3D高斯模型的体积辐射表示与表面纤细特性之间的冲突,在煤矿井下巷道三维重建过程中,3DGS也很难捕捉井下巷道中复杂的几何机器结构。Huang等[14]设计了2DGS(2D Gaussian splatting),将3D高斯重新表示为其中一个轴被折叠的2D替代方案,虽然效果上优于3DGS,但也没有解决模糊环境问题。
在此背景下,现提出一种用于煤矿井下巷道三维重建的方法,称为2DGS-DbTrans。具体来说,首先采用基于Transformer[15-16]的方法对输入的多视角图像进行预处理,高质量的RGB图像输入能够显著提升重建效果。接着利用Colmap方法获取稀疏点云,初始化后使用二维高斯曲面表示场景,每个高斯曲面包含坐标、颜色、大小和方向信息,并对其进行投影和光栅化。随后通过原始损失函数和新引入的损失函数计算误差,并通过反向传播调整参数,最终得到具有颜色信息的三维巷道环境的重建结果。
在煤矿井下巷道中,复杂的环境(如高粉尘和低光照)给场景重建带来了极大挑战。本文研究提出的2DGS-DbTrans方法能有效应对这些困难。该方法的流程如图1所示,主要由基于Transformer的图像增强模块和用于真实场景表示的2D高斯曲面模块组成。在图像增强模块中,Transformer block包含两个关键组件:多头深度卷积轴注意力机制和深度卷积门控网络。接着,使用2D高斯曲面来表达真实的井下巷道环境,最终通过反向传播不断优化参数。通过上述机制,2DGS-DbTrans能够克服煤矿井下巷道中的复杂环境,实现高效且高质量的巷道重建。
煤矿井下巷道中,由于煤矿作业会产生大量煤尘和岩粉,这些颗粒悬浮于空气中,对光线造成散射和吸收,进而导致成像质量下降,图像的对比度和清晰度降低,尤其是在强光区域更易形成明显的雾霾效果[17]。为解决此问题,本文研究设计了一个模块,旨在恢复高分辨率图像。在考虑到Transformer计算复杂度高、可能产生瓶颈的情况下,引入了多头自注意力机制和多尺度分层模块,使该模块的计算复杂度显著低于单尺度网络。
首先,使用三维卷积核对输入图像进行卷积操作,得到一系列低级特征图。接着,通过三级对称的编码器和解码器对这些低级特征图进行编码与解码,以获取深层特征。编码器从高分辨率图像开始输入,逐层降低空间分辨率并增加通道数;解码器则从低分辨率图像开始输入,逐层恢复至高分辨率。编码器和解码器的每一级均包含多个Transformer模块,并在每一级之间的上采样或下采样过程中,分别采用像素重排和像素逆重排方法。
在连接后,增加一个1×1卷积层来降维输入特征图。然后,将第一级输出的低级特征与解码器输出的高级特征聚合,以丰富特征表达,从而更好地保留原图的结构和纹理信息。最终,通过一次卷积操作将结果与原图像相加,得到恢复后的图像。整体的Transformer模块结构如图2所示。Transformer block组件包括两个核心模块:多头深度卷积轴注意力机制和深度卷积门控网络。
由于煤矿井下巷道几乎没有自然光,通常依赖矿灯和人工照明。然而,这些光源覆盖范围有限,导致巷道内存在大量阴影区域,尤其是在远离光源的地方,视觉效果显得暗淡,细节缺失。这些条件会显著影响模型对图像局部信息的捕捉和建模能力。为了解决这一问题,本文研究中引入了注意力机制,以提升对局部细节的关注度。然而,传统的自注意力机制在处理高分辨率图像时,计算复杂度和内存需求会随着图像分辨率的增加而显著提高,因此不适用于煤矿巷道这种复杂环境。为此,本文研究中提出了多头深度卷积轴注意力机制模块,该模块将卷积操作与自注意力机制相结合,以提高模型对图像局部信息的建模能力,从而更有效地应对阴影区域和细节缺失的问题。
多头深度卷积轴注意力机制模块如图3所示,先是经过一个层归一化,它首先生成Q(查询)、K(键)、V(值)投影矩阵。在某些任务中,仅仅使用全局的信息可能不够,因此这些矩阵考虑上下文信息,通过考虑局部上下文信息,能够更好地捕捉到输入数据的局部特征。然后通过1×1卷积来聚合像素级别的跨通道上下文,即通过每个像素点的所有通道的信息融合,来提取跨通道的特征。接着通过3×3深度卷积来编码通道内的空间上下文,即在每个通道内部通过3×3的空间卷积来捕捉像素之间的空间关系。这种组合方式可以高效地整合通道与空间的信息,使模型在处理图像等数据时,既能感知局部通道间的关系,又能捕捉到通道内的空间结构。最后经过一个轴注意力机制,将卷积后的特征图重新排列,使得不同的空间位置相互交互。这一步骤通过将深度卷积输出的空间维度和通道维度进行转置,实现对全局特征的捕捉。总体流程公式为
$\boldsymbol{X}^{\prime}=\boldsymbol{W}_{\mathrm{p}} \text { Attention }\left(\boldsymbol{Q}^{\prime}, \boldsymbol{K}^{\prime}, \boldsymbol{V}^{\prime}\right)+\boldsymbol{X}$
$\text { Attention }\left(\boldsymbol{Q}^{\prime}, \boldsymbol{K}^{\prime}, \boldsymbol{V}^{\prime}\right)=\operatorname{Softmax}\left(\boldsymbol{K}^{\prime} \boldsymbol{Q}^{\prime} / \alpha\right)$
式中:X'X分别为输出和输入的特征映射;Wp为线性映射的权重矩阵;Q'K'、V'分别为查询、键、值矩阵,由输入特征映射X经线性变化生成;Attention为多头注意力计算函数;α为一个可学习参数,用于缩放K'Q'点积的大小,防止注意力分数的数值过大,导致梯度消失或模型过于专注于少数位置。
将输入的通道数被划分为多个子集,称为“头”(heads),每个头独立处理一部分通道的信息,并并行学习独立的注意力矩阵。这种并行化的学习方式能够让模型在不同的注意力头中捕捉到输入数据中的不同特征或上下文信息,最后将这些头的输出组合起来,提升模型的表达能力。
在井下作业中,设备常受到振动影响,导致成像系统产生物理抖动,从而生成模糊图像。此外,巷道中常存在积水或湿滑区域,反射光容易形成光斑,进一步影响图像局部特征的捕捉。在标准的Transformer架构中,前馈网络[18](feed-forward network, FFN)是一种两层的全连接网络(multi-layer perceptron,MLP)。尽管FFN能够捕捉一定的非线性特征,但在捕捉局部空间信息方面存在局限。为解决这一问题,本文研究中引入了深度卷积门控网络,在FFN的基础上加入门控机制和深度卷积模块。这样不仅强化了局部特征的提取能力,同时增强了对复杂井下环境中微小图像细节的捕捉能力,使系统能够更好地适应振动和反射等干扰因素。
深度卷积门控网络如图4所示,先经过一个层归一化,然后进行与多头深度卷积轴注意力机制一样的1×1的卷积和3×3的深度卷积。门控机制通过两个并行的线性变换路径,其中一个路径使用非线性激活(Swish),另一个路径保持线性。最终通过逐元素相乘,动态调整特征值,控制哪些信息被传递、哪些被抑制。这种机制使模型可以动态调整特征信息的流动,从而增强特征选择的灵活性,适用于更复杂的模式捕捉。深度卷积门控网络的总体公式为
T=${W}_{\mathrm{d}}^{1}{W}_{\mathrm{p}}^{1}$[LN(X)]
h1=Swish(T)=$\frac{T}{1+{\mathrm{e}}^{-T}}$
h2=${W}_{\mathrm{d}}^{2}{W}_{\mathrm{p}}^{2}$[LN(X)]
X'=${W}_{\mathrm{p}}^{0}$h1h2+X
式中:LN为层归一化;WpWd矩阵分别为1×1卷积和3×3深度卷积的权重矩阵;☉为逐元素相乘的操作。
深度卷积门控网络结合了深度卷积和门控机制,使得模型在提取局部空间特征的同时,具备了动态控制信息流的能力。在视觉Transformer等结构中,深度卷积门控网络提高了模型在处理图像任务时的有效性,同时减少了计算成本。
使用二维高斯曲面对煤矿井下巷道进行三维重建。相比三维高斯模型,二维高斯曲面更贴近煤矿井下巷道路面的真实物理特性。首先,通过前述基于Transformer的图像增强过程获得了高质量的图像,随后利用Colmap生成稀疏点云,并基于此进行二维高斯初始化,为后续重建提供了精准的起点。
2DGS的主要思想是通过将3D高斯椭球投影到二维平面上,从而简化其表示和计算。具体来说,2DGS从3DG的协方差矩阵中去掉第3行和第3列,得到二维的协方差矩阵,从而实现3D到2D的转换,公式为
G2D(p)=exp$\left[-\frac{1}{2}(p-{p}_{k}{)}^{\mathrm{T}}{\Sigma }_{2\mathrm{D}}^{-1}(p-{p}_{k})\right]$
Σ2D=RSSTRT
式中:RS分别为旋转矩阵和缩放矩阵;Σ2D为协方差矩阵;p为当前点的信息;pk为第k个高斯椭圆盘的中心位置。
2D高斯在空间上的局部切平面如式(9),H为2D高斯的几何齐次变换矩阵,用来将2D高斯椭圆盘上的点投影到世界坐标系中,即
P(u,v)=Pk+sutuu+svtvv=H(u,v,1,1)T
H=$\left[\begin{array}{llll}{s}_{\mathrm{u}}{t}_{\mathrm{u}}& {s}_{\mathrm{v}}{t}_{\mathrm{v}}& 0& {P}_{k}\\ 0& 0& 0& 1\end{array}\right]$
式中:P(u,v)为2D切平面空间中点的位置;susv为缩放因子,决定了椭圆盘在两个方向上的大小;tutv为旋转因子,决定椭圆盘的在2D平面上的方向;uv为在2D椭圆盘中的坐标;Pk、(su,sv)、(tu,tv)为可学习参数,后面通过反向传播逐渐优化。
接着进行光栅化,在光栅化过程中,2D高斯椭圆盘被用来代表复杂的几何场景,以实现从多个视角的高质量新视图合成和几何重建。具体方法类似于3DGS技术,但通过使用2D高斯椭圆盘来更好地对齐表面细节。具体计算公式为
c(x)=$\stackrel{i-1}{\sum _{i=1}}$ciαiG'i[u(x)]$\stackrel{i-1}{\prod _{j=1}}${-αjG'j[u(x)]}
式(11)中:c(x)为位于x出的最终颜色值;ci为第i个高斯椭圆盘的颜色值;αi为第i个高斯椭圆盘不透明度值;G'[u(x)]为进过低通滤波后的高斯函数值;u(x)为x处的信号输入值。
由于三维重建的煤矿井下巷道是彩色的,以及对于巷道内的道路,要保证一定的平滑度,本文研究中新引入颜色损失函数和平滑度损失函数,具体公式为
$ L_{\mathrm{color}}=\frac{1}{\mid M\mid}\sum_{i=1}^{M}m_{i}\mid c_{i}-\bar{c}_{i}\mid$
$ L_{\mathrm{smooth}}=\frac{1}{N}\sum_{i=1}^{N}\sum_{j\in N(i)}\left\|z_{i}-z_{j}\right\|_{2}^{2}$
式中:${\stackrel{-}{c}}_{i}$为颜色基础真值; $ \mid M\mid$为有效像素的数量;Nk个最邻近的点;${\stackrel{-}{z}}_{i}$通过找到与该点在xy平面上最接近的邻居点来确定。
总损失是这些损失的加权和为
Ltotal=λ2DGSL2DGS+λcolorLcolor+λsmoothLsmooth
式(14)中:λ2DGSλcolorλsmooth为相对应的权重;L2DGS 为2DGS的损失函数。
本次实验数据来源于淮南矿务局潘二煤矿,实验设计充分结合了中国煤矿井下的典型作业条件。实验环境及数据处理方法充分考虑了煤矿井下复杂的实际情况,如巷道内的高湿度、低光照以及煤尘等因素,确保实验具有较强的实际应用价值。部分实验数据如图5所示。
本文的实验环境如表1所示。在训练过程中,总计迭代4×104次。初始的3 000次迭代仅对2D高斯模型进行训练,以获得相对稳定的位置和形状;随后,结合2D高斯和变形场进行联合训练。本文方法将井下场景表示为辐射场的三维结构,能够生成高质量的视图合成。
实验结果可视化如图6所示,展示了不同模型下的渲染效果。本文方法与当前多视图合成三维重建领域中的主流方法进行了对比,结果表明本文方法在渲染方面具有显著优势,具体数据如表2所示。对比方法包括3DGS、GES[19]、2DGS和Mip-Splatting[20],这些方法在多视图合成领域表现较为出色。评价指标采用了峰值信噪比(peak signal-to-noise ratio,PSNR)、结构相似性指数(structural similarity index measure,SSIM)和学习感知图像块相似度(learned perceptual image patch similarity,LPIPS),这是多视图合成和三维重建领域常用的质量评价标准。PSNR是一种用于评估图像或视频质量的指标,通常衡量重建图像与原始图像的差异。本文方法在PSNR上有明显提高,说明重建图像质量更加接近原始图像。SSIM用于评估图像质量,特别适合衡量两张图像之间的感知相似度。实验中,本文方法的SSIM高于其他方法,更好地反映了人眼的感知效果。LPIPS是一种关注视觉相似性的度量方法,用于衡量两张图像间的感知相似性。本文方法的LPIPS较低,说明在图像质量评估中更符合人类的视觉感知。本次实验分别取迭代第3 000、8 000、16 000、24 000、32 000、40 000次的PSNR、SSIM、LPIPS值,并绘制折线图,结果如图7所示。
图8展示了煤矿井下巷道的整体三维重建效果。从外部视角,可以清晰地看到巷道的长度和形状等宏观结构。从内部视角,重建结果同样能够准确呈现巷道环境的细节,例如内部的灯光效果、巷道旁的机器设备以及墙上的宣传海报等。此外,图9图10分别展示了在splat模式下的二维高斯曲面图和巷道的三维点云图。这些图像直观地展示了巷道结构和表面的重建效果,有助于全面理解巷道环境的空间特征。
为了进一步验证本文方法的可行性,进行了消融实验,结果如表3所示。有4组对照实验,第1组实验A,采用未经改进的2DGS模型作为基准。第2组实验B在2DGS模型上加入了颜色损失函数和平滑度损失函数。结果表明,PSNR值较基础模型提高了0.19,SSIM提高了0.011,LPIPS降低了0.017,这验证了本文所引入的损失函数在煤矿井下巷道三维重建中的有效性。第3组实验C采用2DGS并添加了基于Transformer的图像增强模块,模型性能显著提升。PSNR较基础模型提高了0.95,SSIM提高了0.019,LPIPS降低了0.024,进一步证明了本文提出的图像增强模块可以有效提升井下巷道三维重建的质量。第4组使用本文提出的完整方法2DGS-DbTrans,结果显示PSNR提高了2.35,SSIM提高了0.03,LPIPS降低了0.042,表明本文方法在煤矿井下巷道中实现了高效且高质量的三维重建。图11展示了实验B和实验C的渲染效果图和点云图。
提出了一种面向煤矿井下巷道的三维重建方法——2DGS-DbTrans。在煤矿井下作业过程中,空气中悬浮着大量煤尘和岩粉,加上光线条件和采集设备的偏差,输入图像的质量往往较低。然而,该方法能够有效应对这些挑战,实现高效且高质量的三维重建效果。
(1)设计了基于Transformer的图像增强模块,包含两个关键创新设计。首先,在多头深度卷积轴注意力机制模块中引入了轴注意力机制,通过将全局注意力分解为沿轴的局部操作来隐式构建全局上下文。其次,通过深度卷积门控网络实现受控的特征变换,增强了局部信息的表达能力。
(2)采用二维高斯曲面来表示煤矿井下巷道。每个二维高斯包含坐标、颜色、大小和方向信息,经过投影和光栅化处理,最终得到一个带有颜色的三维巷道环境重建结果。
(3)设计了颜色损失函数和平滑度损失函数,以提升巷道内颜色的表达能力和道路的平滑程度。通过反向传播不断调整模型参数,从而实现更优的重建效果。
  • 国家自然科学基金(52227901)
参考文献 引证文献
排序方式:
[1]
李颖楠, 张引弟, 管奥成, 等. 煤炭地下气化采出气的脱碳工艺优化及能效分析[J]. 科学技术与工程, 2022, 22 (21): 9231-9237.
Li Yingnan, Zhang Yindi, Guan Aocheng, et al. Decarbonization process optimization and energy efficiency analysis of coal underground gasification produced gas[J]. Science Technology and Engineering, 2022, 22(21): 9231-9237.
[2]
王卫军, 范磊, 马谕杰, 等. 基于蝶形破坏理论深部巷道围岩控制技术研究[J]. 煤炭科学技术, 2023, 51(1): 157-167.
Wang Weijun, Fan Lei, Ma Yujie, et al. Research on control technology of deep roadway perimeter rock based on butterfly damage theory[J]. Coal Science and Technology, 2023, 51(1): 157-167.
[3]
Gong W, Zhang L, Liao L, et al. Incremental SFM 3D reconstruction based on deep learning[J]. Electronics, 2024, 13(14): 2850-2865.
[4]
Chen R, Han S, Xu J, et al. Fast-MVSNet: sparse-to-dense multi-view stereo with lear-ned propagation and Gauss-Newton refinement[C]// Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. New York: IEEE, 2020: 1947-1956.
[5]
明仕, 徐东, 王海川, 等. 特厚煤层巷道冲击破坏机理及全锚索支护技术[J]. 煤炭学报, 2023, 48(5): 1943-1956.
Ming Shi, Xu Dong, Wang Haichuan, et al. Mechanism of impact damage and all anchorcable support technology of roadway in extra-thick coal seam[J]. Journal of Coal, 2023, 48(5): 1943-1956.
[6]
Tancik M, Casser V, Yan X C, et al. Block-nerf: scalable large scene neural view synthesis[C]// Proceedings of the IEEE/CVF Conference onComputer Vision and Pattern Recognition. New York: IEEE, 2022: 8248-8258.
[7]
Turki H, Ramanan D, Satya-narayanan M. Mega-nerf: scalableconstruction of large-scale nerfs for virtual flythroughs[C]// Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. New York: IEEE: 2022: 12922-12931.
[8]
Barron J T, Mildenhall B, Verbin D, et al. Mipnerf 360: unbounded antialiased neural radiance fields[C]// IEEE/CVF Conference on Computer Vision and Pattern Recognition. New York: IEEE, 2022: 5470-5479.
[9]
许江, 程亮, 彭守建, 等. 巷道结构对突出流体冲击特性的影响性试验研究[J]. 煤炭学报, 2023, 48(1): 238-250.
Xu Jiang, Cheng Liang, Peng Shoujian, et al. Experimental study on the influence of roadway structure on the impact characteristics of protruding fluids[J]. Journal of Coal, 2023, 48(1): 238-250.
[10]
Song L, Wang G M, Liu J M, et al. Scnerf: self-correcting neural radiance field with sparse views[J]. arXiv preprint arXiv: 2309.05028, 2023, 11(2): 8-14.
[11]
Zhu S, Wang G M, Blum H, et al. Snislam: semantic neural implicit slam[J]. arXiv preprint ar-Xiv, 2023, 12(2): 1-8.
[12]
Kerbl B, Kopanas G, Lei-mkÜhler T, et al. 3D Gaussian splatting for real-time radiance field rendering[J]. ACM Transactions on Graphics, 2023, 13(4): 1-14.
[13]
Chen G K, Wang W G. A survey on 3D Gaussian splatting[J]. arXiv preprint arXiv, 2024, 8(3): 1-12.
[14]
Huang B B, Yu Z H, Chen A P, et al. 2D gaussian splatting for geometrically accurate radiance fields[C]// IEEE Conference on Computer Vision and Pattern Recognition. New York: ACM, 2024: 1-19.
[15]
Dosovitskiy A, Beyer L, Kolesnikov A, et al. An image is worth 16×16 words: transformers for image recognition at scale[C]// The International Conference on Learning Representations. New York: IEEE, 2021: 1-21.
[16]
Touvron H, Cord M, Douze M, et al. Training data efficient image transformers & distillation through attention[C]// International Conference on Machine Learning. New York: IEEE, 2021: 1021-1025.
[17]
许时昂, 吴海波, 欧元超, 等. 采煤沉陷松散层变形研究现状与分析[J]. 科学技术与工程, 2024, 24(17): 6999-7013.
Xu Shiang, Wu Haibo, Ou Yuanchao, et al. Research status and analysis of loose layerdeformation under coal mining subsidence condition[J]. Science Technology and Engineering, 2024, 24(17): 6999-7013.
[18]
Xie E Z, Wang W H, Yu Z D, et al. Segformer: simple and efficient design for semantic segmentation with transformers[J]. arXiv preprint arXiv, 2021, 8(3): 1-10.
[19]
Hamdi A, Melas-Kyriazi L, Mai J J, et al. GES: generalized expone-ntial splatting for efficient radiance field rendering[C]// IEEE Conference on Computer Vision and Pattern Recognition. Seattle: IEEE, 2024: 1-13.
[20]
Yu Z H, Chen A P, Huang B B, et al. Mip-spl-atting: alias-free 3D Gaussian splatting[C]// IEEE Conference on Computer Vision and Pattern Recognition. New York: IEEE, 2024: 1-10.
2025年第25卷第22期
PDF下载
152
70
引用本文
BibTeX
文章信息
doi: 10.12404/j.issn.1671-1815.2408670
  • 接收时间:2024-11-20
  • 首发时间:2026-02-11
  • 出版时间:2025-08-08
补充材料
相关文章
文章信息
作者
出版历史
  • 收稿日期:2024-11-20
  • 修回日期:2025-05-15
基金
国家自然科学基金(52227901)
作者信息
    安徽理工大学人工智能学院, 淮南 232000

通讯作者:

* 杨超宇(1981—),男,汉族,安徽淮南人,博士,教授。研究方向:计算机视觉。E-mail:
参考文献
分享链接
https://castjournals.cast.org.cn/joweb/kxjsygc/CN/10.12404/j.issn.1671-1815.2408670
分享至
全文二维码

扫描看全文

引用本文
BibTeX
本文的引用情况
2种不同金属材料的力学参数

Family
属数
Number of
genus
种数
Number of
species
占总种数比例
Percentage of
total species (%)

Genus
种数
Number of
species
占总种数比例
Percentage of total
species (%)
鹅膏菌科Amanitaceae 2 11 5.26 鹅膏菌属 Amanita 10 4.78
小菇科 Mycenaceae 2 12 5.74 丝盖伞属 Inocybe 5 2.39
多孔菌科 Polyporaceae 8 14 6.70 蜡蘑属 Laccaria 5 2.39
红菇科 Russulaceae 3 23 11.00 小皮伞属 Marasmius 6 2.87
小菇属 Mycena 11 5.26
光柄菇属 Pluteus 5 2.39
红菇属 Russula 17 8.13
栓菌属 Trametes 5 2.39
关闭全屏