Article(id=1263514355325387636, tenantId=1146029695717560320, journalId=1263187241531621409, issueId=1263514351571428296, articleNumber=null, orderNo=null, doi=10.11996/JG.j.2095-302X.2026010120, pmid=null, cstr=null, oa=null, hot=null, price=null, onlineType=0, articleFormat=0, articleType=null, articleTypeStr=research-article, receivedDate=1748534400000, receivedDateStr=2025-05-30, revisedDate=null, revisedDateStr=null, acceptedDate=1756310400000, acceptedDateStr=2025-08-28, onlineDate=1779174897271, onlineDateStr=2026-05-19, pubDate=1772208000000, pubDateStr=2026-02-28, doiRegisterDate=null, doiRegisterDateStr=null, onlineIssueDate=1779174897271, onlineIssueDateStr=2026-05-19, onlineJustAcceptDate=null, onlineJustAcceptDateStr=null, onlineFirstDate=null, onlineFirstDateStr=null, sourceXml=null, magXml=null, createTime=1779174897271, creator=13701087609, updateTime=1779174897271, updator=13701087609, issue=Issue{id=1263514351571428296, tenantId=1146029695717560320, journalId=1263187241531621409, year='2026', volume='47', issue='1', pageStart='1', pageEnd='233', issueExtLink='null', onlineDate='null', pubDate='null', beforeIssueId=null, nextIssueId=null, price=null, status=1, issueComplete=1, articleOrder=1, issueType=1, specialIssue=null, createTime=1779174896376, creator=13701087609, updateTime=1779174963943, updator=13701087609, preIssue=null, nextIssue=null, ext={EN=IssueExt(id=1263514635077039012, tenantId=1146029695717560320, journalId=1263187241531621409, issueId=1263514351571428296, language=EN, specialIssueTitle=, coverIllustrator=null, specialIssueEditor=, specialIssueAbout=), CN=IssueExt(id=1263514635077039013, tenantId=1146029695717560320, journalId=1263187241531621409, issueId=1263514351571428296, language=CN, specialIssueTitle=, coverIllustrator=null, specialIssueEditor=, specialIssueAbout=)}, issueFiles=null}, startPage=120, endPage=130, ext={EN=ArticleExt(id=1263514356273300347, articleId=1263514355325387636, tenantId=1146029695717560320, journalId=1263187241531621409, language=EN, title=Defect detection of aero-engine blades based on dynamic vision sensors, columnId=1263514354654262248, journalTitle=Journal of Graphics, columnName=Image Processing and Computer Vision, runingTitle=null, highlight=null, articleAbstract=

Aeroengine blades are core components of engines; tiny surface defects can lead to serious safety accidents. Traditional vision detection technology is limited by motion blur, low dynamic range, background redundancy, and so forth. To address these challenges, a method of aeroengine blade defect detection based on Dynamic Vision Sensor (DVS) was proposed. Dynamic vision sensor produced data in an asynchronous event-stream format, and were therefore referred to as event camera, which exhibited the advantages of large dynamic range, high frame rate, and strong ability to capture small targets. Firstly, a defect detection platform based on DVS was built, and its imaging characteristics and advantages were explored. On this basis, the first Event-based Defect Detection Dataset of Aeroengine Blade (EDD-AB) dataset based on DVS was constructed, covering nearly 6 000 images of scratches, point marks and edge damage, with approximately 12 000 finely annotated target labels. The dataset was released as open source (link: https://github. com/NiBieZhouMei5520/EDD-AB.git). Furthermore, a multi-scale defect-detection algorithm based on asynchronous event-stream frame aggregation (AEAF-ABDD) was proposed: event streams were visualized through frame aggregation technology using a fixed time window; a Multi-Resolution Adaptive Feature Pyramid Network (MRAFPN) was developed to enhance multi-scale defect feature extraction capability; a lightweight SimAM attention mechanism was incorporated to strengthen focus on key regions; a star-convolution module (StarNet) was fused to improve the efficiency of high-dimensional nonlinear feature mapping, enabling accurate detection of multi-scale defects on complex curved workpieces. Experiments demonstrated that AEAF-ABDD achieved a mean Average Precision (mAP) of 97.7% on the EDD-AB dataset and a detection speed of 105 frames per second, substantially outperforming mainstream algorithms. An efficient solution for automated quality inspection of highly reflective curved workpieces was thereby provided, promoting the application of DVS in the field of industrial inspection.

, correspAuthors=null, authorNote=null, correspAuthorsNote=
CHEN Haiyong,E-mail:
, 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=Xingshun ZHANG, Haiyong CHEN), CN=ArticleExt(id=1263514364649325540, articleId=1263514355325387636, tenantId=1146029695717560320, journalId=1263187241531621409, language=CN, title=基于动态视觉传感器的航发叶片缺陷检测, columnId=1263514354893337578, journalTitle=图学学报, columnName=图像处理与计算机视觉, runingTitle=null, highlight=null, articleAbstract=

航空发动机叶片作为发动机核心零部件,其表面微小缺陷可能导致严重安全事故,传统视觉检测技术受限于运动模糊、动态范围低及背景冗余等问题。针对上述挑战,提出一种基于动态视觉传感器(DVS)的航发叶片缺陷检测方法。动态视觉传感器数据格式为异步事件流,故也被称作事件相机,具有动态范围大、高帧率和微小目标捕捉能力强等优势。首先搭建基于DVS的缺陷检测平台,探索总结了其成像特点及优势。在此基础上,构建首个基于DVS的航发叶片缺陷检测数据集(EDD-AB),涵盖划痕、点痕、边缘损伤3类缺陷近6 000张图像,精细标注近1.2万个目标标签,数据集已开源(链接: https://github.com/NiBieZhouMei5520/EDD-AB.git)。进一步提出基于异步事件流帧聚合的多尺度缺陷检测算法(AEAF-ABDD):通过固定时间窗的帧聚合技术实现事件流可视化;构建多分辨率自适应特征金字塔网络(MRAFPN)增强多尺度缺陷特征提取能力;引入轻量级SimAM注意力机制强化关键区域聚焦;融合星形卷积模块(StarNet)提升高维非线性特征映射效率,实现复杂曲面工件多尺度缺陷的精准检测。实验表明,AEAF-ABDD在EDD-AB数据集上的平均精度均值(mAP)达97.7%,检测速度达105帧/秒,显著优于主流算法,为高反光曲面工件的自动化质检提供了高效解决方案,推动了DVS在工业检测领域的应用。

, correspAuthors=null, authorNote=null, correspAuthorsNote=
陈海永,E-mail:
, copyrightStatement=null, copyrightOwner=null, extLink=null, articleAbsUrl=null, sourceXml=a1Lek1Ea8L72bexZ9qSe8A==, magXml=w+4fhNck7gZsq5O8JMj/9Q==, pdfUrl=null, pdf=XKcbTS6ade2VQg/S5MSvzA==, pdfFileSize=5029272, pdfExtLink=null, richHtmlUrl=null, mobilePdfUrl=null, reviewReport=null, pdfFirstPage=null, abstractGraph=5KObrRute1n78B43iituLg==, abstractGraphContent=null, abstractVideo=null, citation=null, cebUrl=null, magXmlContent=JxtFhwR8ZTrF99F6lHuztw==, mapNumber=null, authorCompany=null, fund=null, authors=null, authorsList=张行顺, 陈海永)}, authors=[Author(id=1263550825104794462, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, orderNo=0, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1263550826006569832, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, authorId=1263550825104794462, language=EN, stringName=Xingshun ZHANG, firstName=Xingshun, middleName=null, lastName=ZHANG, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=School of Artificial Intelligence and Data Science, Hebei University of Technology, Tianjin 300401, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1263550827537490799, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, authorId=1263550825104794462, language=CN, stringName=张行顺, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=河北工业大学人工智能与数据科学学院, 天津 300401, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1263550824567923537, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, xref=null, ext=[AuthorCompanyExt(id=1263550824584700756, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, companyId=1263550824567923537, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=School of Artificial Intelligence and Data Science, Hebei University of Technology, Tianjin 300401, China), AuthorCompanyExt(id=1263550824626643797, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, companyId=1263550824567923537, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=河北工业大学人工智能与数据科学学院, 天津 300401)])]), Author(id=1263550828921611124, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, orderNo=1, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=haiyong.chen@hebut.edu.cn, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1263550829798220670, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, authorId=1263550828921611124, language=EN, stringName=Haiyong CHEN, firstName=Haiyong, middleName=null, lastName=CHEN, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=School of Artificial Intelligence and Data Science, Hebei University of Technology, Tianjin 300401, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1263550830502863751, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, authorId=1263550828921611124, language=CN, stringName=陈海永, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=河北工业大学人工智能与数据科学学院, 天津 300401, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1263550824567923537, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, xref=null, ext=[AuthorCompanyExt(id=1263550824584700756, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, companyId=1263550824567923537, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=School of Artificial Intelligence and Data Science, Hebei University of Technology, Tianjin 300401, China), AuthorCompanyExt(id=1263550824626643797, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, companyId=1263550824567923537, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=河北工业大学人工智能与数据科学学院, 天津 300401)])])], keywords=[Keyword(id=1263550833086555031, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, language=EN, orderNo=1, keyword=dynamic vision sensor), Keyword(id=1263550834655224743, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, language=EN, orderNo=2, keyword=aeroengine blades), Keyword(id=1263550835552805809, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, language=EN, orderNo=3, keyword=defect detection), Keyword(id=1263550835951264699, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, language=EN, orderNo=4, keyword=asynchronous event stream), Keyword(id=1263550836400055233, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, language=EN, orderNo=5, keyword=multi-scale feature), Keyword(id=1263550838060999624, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, language=CN, orderNo=1, keyword=动态视觉传感器), Keyword(id=1263550839256376273, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, language=CN, orderNo=2, keyword=航空发动机叶片), Keyword(id=1263550840112014298, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, language=CN, orderNo=3, keyword=缺陷检测), Keyword(id=1263550840615330786, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, language=CN, orderNo=4, keyword=异步事件流), Keyword(id=1263550842657956840, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, language=CN, orderNo=5, keyword=多尺度特征融合)], refs=[Reference(id=1263550870025789642, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, doi=null, pmid=null, pmcid=null, year=2021, volume=47, issue=5, pageStart=1017, pageEnd=1034, url=null, language=null, rfNumber=[1], rfOrder=0, authorNames=陶显, 侯伟, 徐德, journalName=自动化学报, refType=null, unstructuredReference=陶显, 侯伟, 徐德. 基于深度学习的表面缺陷检测方法综述[J]. 自动化学报, 2021, 47(5): 1017-1034., articleTitle=基于深度学习的表面缺陷检测方法综述, refAbstract=null), Reference(id=1263550870352945359, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, doi=null, pmid=null, pmcid=null, year=2021, volume=47, issue=5, pageStart=1017, pageEnd=1034, url=null, language=null, rfNumber=[1], rfOrder=1, authorNames=TAO X, HOU W, XU D, journalName=Acta Automatica Sinica, refType=null, unstructuredReference=TAO X, HOU W, XU D. A survey of surface defect detection methods based on deep learning[J]. Acta Automatica Sinica, 2021, 47(5): 1017-1034 (in Chinese)., articleTitle=A survey of surface defect detection methods based on deep learning, refAbstract=null), Reference(id=1263550870566854871, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, doi=null, pmid=null, pmcid=null, year=2023, volume=72, issue=null, pageStart=5029013, pageEnd=null, url=null, language=null, rfNumber=[2], rfOrder=2, authorNames=CHEN Y Q, PAN J W, LEI J Y, journalName=IEEE Transactions on Instrumentation and Measurement, refType=null, unstructuredReference=CHEN Y Q, PAN J W, LEI J Y, et al. EEE-Net: efficient edge enhanced network for surface defect detection of glass[J]. IEEE Transactions on Instrumentation and Measurement, 2023, 72: 5029013., articleTitle=EEE-Net: efficient edge enhanced network for surface defect detection of glass, refAbstract=null), Reference(id=1263550870722044122, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, doi=null, pmid=null, pmcid=null, year=2021, volume=17, issue=8, pageStart=5485, pageEnd=5494, url=null, language=null, rfNumber=[3], rfOrder=3, authorNames=JIANG W B, LIU M, PENG Y N, journalName=IEEE Transactions on Industrial Informatics, refType=null, unstructuredReference=JIANG W B, LIU M, PENG Y N, et al. HDCB-Net: a neural network with the hybrid dilated convolution for pixel-level crack detection on concrete bridges[J]. IEEE Transactions on Industrial Informatics, 2021, 17(8): 5485-5494., articleTitle=HDCB-Net: a neural network with the hybrid dilated convolution for pixel-level crack detection on concrete bridges, refAbstract=null), Reference(id=1263550870927565026, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, doi=null, pmid=null, pmcid=null, year=2016, volume=null, issue=null, pageStart=21, pageEnd=37, url=null, language=null, rfNumber=[4], rfOrder=4, authorNames=LIU W, ANGUELOV D, ERHAN D, journalName=The 14th European Conference on Computer Vision, refType=null, unstructuredReference=LIU W, ANGUELOV D, ERHAN D, et al. SSD: single shot MultiBox detector[C]// The 14th European Conference on Computer Vision. Cham: Springer, 2016: 21-37., articleTitle=SSD: single shot MultiBox detector, refAbstract=null), Reference(id=1263550871191806185, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, doi=null, pmid=null, pmcid=null, year=2016, volume=null, issue=null, pageStart=779, pageEnd=788, url=null, language=null, rfNumber=[5], rfOrder=5, authorNames=REDMON J, DIVVALA S, GIRSHICK R, journalName=2016 IEEE Conference on Computer Vision and Pattern Recognition, refType=null, unstructuredReference=REDMON J, DIVVALA S, GIRSHICK R, et al. You only look once: unified, real-time object detection[C]// 2016 IEEE Conference on Computer Vision and Pattern Recognition. New York: IEEE Press, 2016: 779-788., articleTitle=You only look once: unified, real-time object detection, refAbstract=null), Reference(id=1263550871346995437, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, doi=null, pmid=null, pmcid=null, year=2014, volume=null, issue=null, pageStart=580, pageEnd=587, url=null, language=null, rfNumber=[6], rfOrder=6, authorNames=GIRSHICK R, DONAHUE J, DARRELL T, journalName=2014 IEEE Conference on Computer Vision and Pattern Recognition, refType=null, unstructuredReference=GIRSHICK R, DONAHUE J, DARRELL T, et al. Rich feature hierarchies for accurate object detection and semantic segmentation[C]// 2014 IEEE Conference on Computer Vision and Pattern Recognition. New York: IEEE Press, 2014: 580-587., articleTitle=Rich feature hierarchies for accurate object detection and semantic segmentation, refAbstract=null), Reference(id=1263550871682539764, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, doi=null, pmid=null, pmcid=null, year=2021, volume=null, issue=null, pageStart=6354, pageEnd=6363, url=null, language=null, rfNumber=[7], rfOrder=7, authorNames=ZHENG Y J, ZHENG L X, YU Z F, journalName=2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition, refType=null, unstructuredReference=ZHENG Y J, ZHENG L X, YU Z F, et al. High-speed image reconstruction through short-term plasticity for spiking cameras[C]// 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition. New York: IEEE Press, 2021: 6354-6363., articleTitle=High-speed image reconstruction through short-term plasticity for spiking cameras, refAbstract=null), Reference(id=1263550871934198009, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, doi=null, pmid=null, pmcid=null, year=2017, volume=null, issue=null, pageStart=251, pageEnd=260, url=null, language=null, rfNumber=[8], rfOrder=8, authorNames=XU X Y, SUN D Q, PAN J S, journalName=2017 IEEE International Conference on Computer Vision, refType=null, unstructuredReference=XU X Y, SUN D Q, PAN J S, et al. Learning to super-resolve blurry face and text images[C]// 2017 IEEE International Conference on Computer Vision. New York: IEEE Press, 2017: 251-260., articleTitle=Learning to super-resolve blurry face and text images, refAbstract=null), Reference(id=1263550873699999996, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, doi=null, pmid=null, pmcid=null, year=2021, volume=44, issue=6, pageStart=1258, pageEnd=1286, url=null, language=null, rfNumber=[9], rfOrder=9, authorNames=李家宁, 田永鸿, journalName=计算机学报, refType=null, unstructuredReference=李家宁, 田永鸿. 神经形态视觉传感器的研究进展及应用综述[J]. 计算机学报, 2021, 44(6): 1258-1286., articleTitle=神经形态视觉传感器的研究进展及应用综述, refAbstract=null), Reference(id=1263550874106847491, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, doi=null, pmid=null, pmcid=null, year=2021, volume=44, issue=6, pageStart=1258, pageEnd=1286, url=null, language=null, rfNumber=[9], rfOrder=10, authorNames=LI J N, TIAN Y H, journalName=Chinese Journal of Computers, refType=null, unstructuredReference=LI J N, TIAN Y H. Recent advances in neuromorphic vision sensors: a survey[J]. Chinese Journal of Computers, 2021, 44(6): 1258-1286 (in Chinese)., articleTitle=Recent advances in neuromorphic vision sensors: a survey, refAbstract=null), Reference(id=1263550874400448775, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, doi=null, pmid=null, pmcid=null, year=2006, volume=null, issue=null, pageStart=2060, pageEnd=2069, url=null, language=null, rfNumber=[10], rfOrder=11, authorNames=LICHTSTEINER P, POSCH C, DELBRUCK T, journalName=2006 IEEE International Solid State Circuits Conference-Digest of Technical Papers, refType=null, unstructuredReference=LICHTSTEINER P, POSCH C, DELBRUCK T. A 128 × 128 120db 30mw asynchronous vision sensor that responds to relative intensity change[C]// 2006 IEEE International Solid State Circuits Conference-Digest of Technical Papers. New York: IEEE Press, 2006: 2060-2069., articleTitle=A 128 × 128 120db 30mw asynchronous vision sensor that responds to relative intensity change, refAbstract=null), Reference(id=1263550874509500681, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, doi=null, pmid=null, pmcid=null, year=2019, volume=null, issue=null, pageStart=1682, pageEnd=1683, url=null, language=null, rfNumber=[11], rfOrder=12, authorNames=CHEN S S, GUO M H, journalName=2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops, refType=null, unstructuredReference=CHEN S S, GUO M H. Live demonstration: CeleX-V: a 1M pixel multi-mode event-based sensor[C]// 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops. New York: IEEE Press, 2019: 1682-1683., articleTitle=Live demonstration: CeleX-V: a 1M pixel multi-mode event-based sensor, refAbstract=null), Reference(id=1263550874631135501, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, doi=null, pmid=null, pmcid=null, year=2014, volume=null, issue=null, pageStart=686, pageEnd=689, url=null, language=null, rfNumber=[12], rfOrder=13, authorNames=BRANDLI C, MULLER L, DELBRUCK T, journalName=2014 IEEE International Symposium on Circuits and Systems, refType=null, unstructuredReference=BRANDLI C, MULLER L, DELBRUCK T. Real-time, high-speed video decompression using a frame- and event-based DAVIS sensor[C]// 2014 IEEE International Symposium on Circuits and Systems. New York: IEEE Press, 2014: 686-689., articleTitle=Real-time, high-speed video decompression using a frame- and event-based DAVIS sensor, refAbstract=null), Reference(id=1263550874765353233, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, doi=null, pmid=null, pmcid=null, year=2024, volume=50, issue=12, pageStart=2407, pageEnd=2419, url=null, language=null, rfNumber=[13], rfOrder=14, authorNames=马居坡, 陈周熠, 吴金建, journalName=自动化学报, refType=null, unstructuredReference=马居坡, 陈周熠, 吴金建. 基于动态视觉传感器的铝基盘片表面缺陷检测[J]. 自动化学报, 2024, 50(12): 2407-2419., articleTitle=基于动态视觉传感器的铝基盘片表面缺陷检测, refAbstract=null), Reference(id=1263550874912153878, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, doi=null, pmid=null, pmcid=null, year=2024, volume=50, issue=12, pageStart=2407, pageEnd=2419, url=null, language=null, rfNumber=[13], rfOrder=15, authorNames=MA J P, CHEN Z Y, WU J J, journalName=Acta Automatica Sinica, refType=null, unstructuredReference=MA J P, CHEN Z Y, WU J J. Dynamic vision sensor based defect detection for the surface of aluminum disk[J]. Acta Automatica Sinica, 2024, 50(12): 2407-2419 (in Chinese)., articleTitle=Dynamic vision sensor based defect detection for the surface of aluminum disk, refAbstract=null), Reference(id=1263550875138646294, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, doi=null, pmid=null, pmcid=null, year=2024, volume=22, issue=2, pageStart=140, pageEnd=145, url=null, language=null, rfNumber=[14], rfOrder=16, authorNames=吕和平, 洪流, journalName=数字自造科学, refType=null, unstructuredReference=吕和平, 洪流. 基于事件相机的铝基片划痕检测研究[J]. 数字自造科学, 2024, 22(2): 140-145., articleTitle=基于事件相机的铝基片划痕检测研究, refAbstract=null), Reference(id=1263550875381915929, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, doi=null, pmid=null, pmcid=null, year=2024, volume=2, issue=22, pageStart=140, pageEnd=145, url=null, language=null, rfNumber=[14], rfOrder=17, authorNames=LV H P, HONG L, journalName=Digital Manufacture Science, refType=null, unstructuredReference=LV H P, HONG L. Research on scratch detection of aluminum substrate based on event camera[J]. Digital Manufacture Science, 2024, 2(22): 140-145 (in Chinese)., articleTitle=Research on scratch detection of aluminum substrate based on event camera, refAbstract=null), Reference(id=1263550875532910876, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=[15], rfOrder=18, authorNames=邓坚, 万小芳, journalName=一种基于事件相机的缺陷检测装置: 中国, 217212305U, refType=null, unstructuredReference=邓坚, 万小芳. 一种基于事件相机的缺陷检测装置: 中国, 217212305U[P]. 2022-08-16., articleTitle=null, refAbstract=null), Reference(id=1263550875683905826, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=[15], rfOrder=19, authorNames=DENG J, WAN X F, journalName=Defect detection device based on event camera: China, 217212305U, refType=null, unstructuredReference=DENG J, WAN X F. Defect detection device based on event camera: China, 217212305U[P]. 2022-08-16 (in Chinese)., articleTitle=null, refAbstract=null), Reference(id=1263550875813929253, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://proceedings.mlr.press/v139/yang21o.html, language=null, rfNumber=[16], rfOrder=20, authorNames=YANG L X, ZHANG R Y, LI L D, journalName=null, refType=null, unstructuredReference=YANG L X, ZHANG R Y, LI L D, et al. SimAM: a simple, parameter-free attention module for convolutional neural networks[EB/OL]. [2025-03-30]. https://proceedings.mlr.press/v139/yang21o.html., articleTitle=SimAM: a simple, parameter-free attention module for convolutional neural networks, refAbstract=null), Reference(id=1263550875985895722, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, doi=null, pmid=null, pmcid=null, year=2022, volume=44, issue=1, pageStart=154, pageEnd=180, url=null, language=null, rfNumber=[17], rfOrder=21, authorNames=GALLEGO G, DELBRUCK T, ORCHARD G, journalName=IEEE Transactions on Pattern Analysis and Machine Intelligence, refType=null, unstructuredReference=GALLEGO G, DELBRUCK T, ORCHARD G, et al. Event-based vision: a survey[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2022, 44(1): 154-180., articleTitle=Event-based vision: a survey, refAbstract=null), Reference(id=1263550876212388142, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, doi=null, pmid=null, pmcid=null, year=2009, volume=null, issue=null, pageStart=775, pageEnd=778, url=null, language=null, rfNumber=[18], rfOrder=22, authorNames=CHEN S S, MARTINI B, CULURCIELLO E, journalName=2009 IEEE International Symposium on Circuits and Systems, refType=null, unstructuredReference=CHEN S S, MARTINI B, CULURCIELLO E. A bio-inspired event-based size and position invariant human posture recognition algorithm[C]// 2009 IEEE International Symposium on Circuits and Systems. New York: IEEE Press, 2009: 775-778., articleTitle=A bio-inspired event-based size and position invariant human posture recognition algorithm, refAbstract=null), Reference(id=1263550876464046386, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, doi=null, pmid=null, pmcid=null, year=2024, volume=null, issue=null, pageStart=5701, pageEnd=5711, url=null, language=null, rfNumber=[19], rfOrder=23, authorNames=AYDIN A, GEHRIG M, GEHRIG D, journalName=2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops, refType=null, unstructuredReference=AYDIN A, GEHRIG M, GEHRIG D, et al. A hybrid ANN-SNN architecture for low-power and low-latency visual perception[C]// 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops. New York: IEEE Press, 2024: 5701-5711., articleTitle=A hybrid ANN-SNN architecture for low-power and low-latency visual perception, refAbstract=null), Reference(id=1263550878112407863, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, doi=null, pmid=null, pmcid=null, year=2024, volume=null, issue=null, pageStart=924, pageEnd=927, url=null, language=null, rfNumber=[20], rfOrder=24, authorNames=FAN Q D, journalName=2024 IEEE 2nd International Conference on Sensors, Electronics and Computer Engineering, refType=null, unstructuredReference=FAN Q D. Research on the dynamic object segmentation based on computer vision technology[C]// 2024 IEEE 2nd International Conference on Sensors, Electronics and Computer Engineering. New York: IEEE Press, 2024: 924-927., articleTitle=Research on the dynamic object segmentation based on computer vision technology, refAbstract=null), Reference(id=1263550878477312318, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, doi=null, pmid=null, pmcid=null, year=2023, volume=null, issue=null, pageStart=512, pageEnd=520, url=null, language=null, rfNumber=[21], rfOrder=25, authorNames=HE X, ZHAO D C, LI Y, journalName=The 38th AAAI Conference on Artificial Intelligence, refType=null, unstructuredReference=HE X, ZHAO D C, LI Y, et al. An efficient knowledge transfer strategy for spiking neural networks from static to event domain[C]// The 38th AAAI Conference on Artificial Intelligence. Washington: AAAI Press, 2023: 512-520., articleTitle=An efficient knowledge transfer strategy for spiking neural networks from static to event domain, refAbstract=null), Reference(id=1263550878779302209, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, doi=null, pmid=null, pmcid=null, year=2023, volume=null, issue=null, pageStart=6532, pageEnd=6542, url=null, language=null, rfNumber=[22], rfOrder=26, authorNames=SU Q Y, CHOU Y H, HU Y F, journalName=2023 IEEE/CVF International Conference on Computer Vision, refType=null, unstructuredReference=SU Q Y, CHOU Y H, HU Y F, et al. Deep directly-trained spiking neural networks for object detection[C]// 2023 IEEE/CVF International Conference on Computer Vision. New York: IEEE Press, 2023: 6532-6542., articleTitle=Deep directly-trained spiking neural networks for object detection, refAbstract=null), Reference(id=1263550879026766150, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://arxiv.org/abs/2502.12524.pdf, language=null, rfNumber=[23], rfOrder=27, authorNames=TIAN Y J, YE Q X, DOERMANN D, journalName=null, refType=null, unstructuredReference=TIAN Y J, YE Q X, DOERMANN D. YOLOv12:attention- centric real-time object detectors[EB/OL]. [2025-02-18]. https://arxiv.org/abs/2502.12524.pdf., articleTitle=YOLOv12:attention- centric real-time object detectors, refAbstract=null), Reference(id=1263550879366504778, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, doi=null, pmid=null, pmcid=null, year=2021, volume=null, issue=null, pageStart=2126, pageEnd=2136, url=null, language=null, rfNumber=[24], rfOrder=28, authorNames=KIM J, BAE J, PARK G, journalName=2021 IEEE/CVF International Conference on Computer Vision, refType=null, unstructuredReference=KIM J, BAE J, PARK G, et al. N-ImageNet: towards robust, fine-grained object recognition with event cameras[C]// 2021 IEEE/CVF International Conference on Computer Vision. New York: IEEE Press, 2021: 2126-2136., articleTitle=N-ImageNet: towards robust, fine-grained object recognition with event cameras, refAbstract=null), Reference(id=1263550879790129487, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, doi=null, pmid=null, pmcid=null, year=2024, volume=null, issue=null, pageStart=5694, pageEnd=5703, url=null, language=null, rfNumber=[25], rfOrder=29, authorNames=MA X, DAI X Y, BAI Y, journalName=2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition, refType=null, unstructuredReference=MA X, DAI X Y, BAI Y, et al. Rewrite the star[C]// 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition. New York: IEEE Press, 2024: 5694-5703., articleTitle=Rewrite the star, refAbstract=null), Reference(id=1263550880150839635, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, doi=null, pmid=null, pmcid=null, year=2023, volume=10, issue=22, pageStart=140, pageEnd=145, url=null, language=null, rfNumber=[26], rfOrder=30, authorNames=GUO A J, MA J P, DIAN R W, journalName=National Science Review, refType=null, unstructuredReference=GUO A J, MA J P, DIAN R W, et al. Surface defect detection competition with a bio-inspired vision sensor[J]. National Science Review, 2023, 10(22): 140-145., articleTitle=Surface defect detection competition with a bio-inspired vision sensor, refAbstract=null)], funds=[Fund(id=1263550865516912810, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, awardId=2022YFB3303804, language=EN, fundingSource=National Key Research and Development Program of China under Grant(2022YFB3303804), fundOrder=null, country=null), Fund(id=1263550865953120431, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, awardId=2022YFB3303804, language=CN, fundingSource=国家重点研究发展计划项目(2022YFB3303804), fundOrder=null, country=null), Fund(id=1263550866758426806, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, awardId=62473127, language=EN, fundingSource=National Natural Science Foundation of China under Grant(62473127), fundOrder=null, country=null), Fund(id=1263550867404349631, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, awardId=62473127, language=CN, fundingSource=国家自然科学基金(62473127), fundOrder=null, country=null)], companyList=[AuthorCompany(id=1263550824567923537, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, xref=null, ext=[AuthorCompanyExt(id=1263550824584700756, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, companyId=1263550824567923537, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=School of Artificial Intelligence and Data Science, Hebei University of Technology, Tianjin 300401, China), AuthorCompanyExt(id=1263550824626643797, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, companyId=1263550824567923537, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=河北工业大学人工智能与数据科学学院, 天津 300401)])], figs=[ArticleFig(id=1263550843551343604, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, language=EN, label=Fig. 1, caption=Schematic diagram of the event generation principle, figureFileSmall=5evi4Ky7zjJHNvZV2/cb+A==, figureFileBig=iuLcfTyBuySAqEJ1hHFpVg==, tableContent=null), ArticleFig(id=1263550844838994942, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, language=CN, label=图1, caption=事件产生原理图, figureFileSmall=5evi4Ky7zjJHNvZV2/cb+A==, figureFileBig=iuLcfTyBuySAqEJ1hHFpVg==, tableContent=null), ArticleFig(id=1263550847078752263, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, language=EN, label=Fig. 2, caption=Schematic diagram of defect imaging ((a) Optical path with a stationary workpiece;(b) The optical path during relative displacement between the workpiece and the DVS), figureFileSmall=KUItEPMWDI7qDyUMn8UkHg==, figureFileBig=F1iACfDO/Gz8czo6D+s2dQ==, tableContent=null), ArticleFig(id=1263550847233941516, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, language=CN, label=图2, caption=缺陷成像示意图((a) 工件位置未发生变化时的光路;(b) 工件与DVS发生相对位移时的光路), figureFileSmall=KUItEPMWDI7qDyUMn8UkHg==, figureFileBig=F1iACfDO/Gz8czo6D+s2dQ==, tableContent=null), ArticleFig(id=1263550847686926354, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, language=EN, label=Fig. 3, caption=Experimental platform, figureFileSmall=T+ovgx1SajCGthP0mkNfMA==, figureFileBig=MW/aJlpsDC4nh5LHXK32eA==, tableContent=null), ArticleFig(id=1263550848160882708, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, language=CN, label=图3, caption=实验平台, figureFileSmall=T+ovgx1SajCGthP0mkNfMA==, figureFileBig=MW/aJlpsDC4nh5LHXK32eA==, tableContent=null), ArticleFig(id=1263550848907468827, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, language=EN, label=Fig. 4, caption=Comparison of imaging of weak defects (surface scratch defects in red box) ((a) DVS imaging effect; (b) Traditional camera imaging effect), figureFileSmall=GzGMJUVJosF/f2eWciqerQ==, figureFileBig=4WWnkZ8kjxdQ137MiiR3nA==, tableContent=null), ArticleFig(id=1263550849335287843, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, language=CN, label=图4, caption=弱小缺陷成像对比(红框中为表面划痕缺陷)((a) DVS成像效果;(b) 传统相机成像效果), figureFileSmall=GzGMJUVJosF/f2eWciqerQ==, figureFileBig=4WWnkZ8kjxdQ137MiiR3nA==, tableContent=null), ArticleFig(id=1263550849800855592, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, language=EN, label=Fig. 5, caption=High dynamic range imaging comparison (surface scratch defects and point scratch defects in the red box) ((a) DVS imaging effect; (b) Traditional camera imaging effect), figureFileSmall=gXVC0ccFC2GSP1yXd2SrWw==, figureFileBig=t/Eh8fHObvojNecmPGwp7A==, tableContent=null), ArticleFig(id=1263550851386302508, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, language=CN, label=图5, caption=高动态范围成像对比(红框中为表面划痕缺陷与点痕缺陷) ((a) DVS成像效果;(b) 传统相机成像效果), figureFileSmall=gXVC0ccFC2GSP1yXd2SrWw==, figureFileBig=t/Eh8fHObvojNecmPGwp7A==, tableContent=null), ArticleFig(id=1263550851872841781, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, language=EN, label=Fig. 6, caption=Visualization images of event flows for different types of defects (the scratches are in the red boxes, the dot marks are in the red circles, and the edge damages are in the red triangles), figureFileSmall=c2GQdZMMAPYP3y7OVl1NxA==, figureFileBig=qL5MBc9TCAckQJcQz9gGGw==, tableContent=null), ArticleFig(id=1263550852355186743, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, language=CN, label=图6, caption=不同类型缺陷的事件流可视化图像(红色方框中为划痕,红色圆圈中为点痕,红色三角形中为边缘损伤), figureFileSmall=c2GQdZMMAPYP3y7OVl1NxA==, figureFileBig=qL5MBc9TCAckQJcQz9gGGw==, tableContent=null), ArticleFig(id=1263550852770422844, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, language=EN, label=Fig. 7, caption=EDD-AB Characteristics Statistics of Dataset ((a) Area distribution of bounding boxes for seratches; (b) Area distribution of bounding boxes for point marks; (c) Area distribution of bounding boxes for edge damage; (d) Area distribution of bounding boxes for all defect targets), figureFileSmall=B0qq6JJ77dy1ndmiU6557w==, figureFileBig=3CfRJitmWUegBGF2A3hfNQ==, tableContent=null), ArticleFig(id=1263550853533786184, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, language=CN, label=图7, caption=数据集 EDD-AB 特性统计((a) 划痕损伤目标边界框宽高比的分布;(b) 点痕损伤目标边界框宽高比的分布;(c) 边缘损伤目标边界框宽高比的分布;(d) 所有缺陷目标边界框的面积分布), figureFileSmall=B0qq6JJ77dy1ndmiU6557w==, figureFileBig=3CfRJitmWUegBGF2A3hfNQ==, tableContent=null), ArticleFig(id=1263550854393618508, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, language=EN, label=Fig. 8, caption=Small target detection head, figureFileSmall=yW8PgPCcXfOSjPup7R+IGQ==, figureFileBig=EePU/MfmTJUSAJQBqj6DLQ==, tableContent=null), ArticleFig(id=1263550856100700242, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, language=CN, label=图8, caption=小目标检测头, figureFileSmall=yW8PgPCcXfOSjPup7R+IGQ==, figureFileBig=EePU/MfmTJUSAJQBqj6DLQ==, tableContent=null), ArticleFig(id=1263550856260083798, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, language=EN, label=Fig. 9, caption=Schematic diagram of the SimAM attention structure, figureFileSmall=Hmj8Sqf90qCU5UaRbL0AdQ==, figureFileBig=GB4WDM65FNWE6k5tjO4Fjg==, tableContent=null), ArticleFig(id=1263550856880840796, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, language=CN, label=图9, caption=SimAM注意力结构示意图, figureFileSmall=Hmj8Sqf90qCU5UaRbL0AdQ==, figureFileBig=GB4WDM65FNWE6k5tjO4Fjg==, tableContent=null), ArticleFig(id=1263550857518375011, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, language=EN, label=Fig. 10, caption=Schematic diagram of star-shaped convolution operation(* is element-wise multiplication), figureFileSmall=g7HlP8zHfvSZdFAMedjqsg==, figureFileBig=sdQpcqmHyyGdiO5AucEVWw==, tableContent=null), ArticleFig(id=1263550857749061733, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, language=CN, label=图10, caption=星形卷积操作示意图(*为星形运算), figureFileSmall=g7HlP8zHfvSZdFAMedjqsg==, figureFileBig=sdQpcqmHyyGdiO5AucEVWw==, tableContent=null), ArticleFig(id=1263550858369818733, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, language=EN, label=Fig. 11, caption=Comparison of hatemap detection ((a) Original image; (b) DETR; (c) Yolov10; (d) TIFF-EDD; (e) AEAF-ABDD), figureFileSmall=zCzcQRzCc5fX8SkVL87UCw==, figureFileBig=da2LsD0GawjCD0XMI8B9fw==, tableContent=null), ArticleFig(id=1263550860236284018, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, language=CN, label=图11, caption=检测热力图对比, figureFileSmall=zCzcQRzCc5fX8SkVL87UCw==, figureFileBig=da2LsD0GawjCD0XMI8B9fw==, tableContent=null), ArticleFig(id=1263550860731211897, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, language=EN, label=Table 1, caption=

Experimental environment configuration

, figureFileSmall=null, figureFileBig=null, tableContent=
配置 参数
操作环境 Windows11
深度学习框架 Pytorch1.12
CUDA 11.3
Python版本 Python-3.12
CPU Intel(R) Core(TM) i7-14700HX 2.10 GHz
GPU NVIDIA GeForce RTX 4070Laptop 8 GB
), ArticleFig(id=1263550861087727742, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, language=CN, label=表1, caption=

实验环境配置

, figureFileSmall=null, figureFileBig=null, tableContent=
配置 参数
操作环境 Windows11
深度学习框架 Pytorch1.12
CUDA 11.3
Python版本 Python-3.12
CPU Intel(R) Core(TM) i7-14700HX 2.10 GHz
GPU NVIDIA GeForce RTX 4070Laptop 8 GB
), ArticleFig(id=1263550861465215111, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, language=EN, label=Table 2, caption=

Experimental parameter settings

, figureFileSmall=null, figureFileBig=null, tableContent=
超参数 参数值 超参数 参数值
Images size 640×640 Optimize SGD
Epochs 300 Momentum 0.973
Batch size 32 Learning rate 0.01
), ArticleFig(id=1263550861901422732, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, language=CN, label=表2, caption=

实验参数设置

, figureFileSmall=null, figureFileBig=null, tableContent=
超参数 参数值 超参数 参数值
Images size 640×640 Optimize SGD
Epochs 300 Momentum 0.973
Batch size 32 Learning rate 0.01
), ArticleFig(id=1263550862417322131, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, language=EN, label=Table 3, caption=

Comparative experiments with other algorithms on the EDD-AB dataset

, figureFileSmall=null, figureFileBig=null, tableContent=
Method P/% R/% Parms/106 GFLOPs FPS mAP/%
综合 划痕 点痕 边缘损伤
Faster-RCNN 76.6 77.6 165.00 199.0 66 80.2 87.7 66.2 86.6
SSD 63.0 65.3 24.50 87.9 72 71.0 77.6 55.3 79.6
DETR 77.5 78.0 9.49 16.8 88 87.0 91.3 77.5 91.8
Yolov10 89.6 87.5 2.59 6.4 96 92.8 97.1 85.7 96.0
Yolov12 85.3 88.7 2.52 6.0 93 91.8 94.8 89.1 92.2
EMS-YOLO[18] 82.3 84.5 14.40 6.8 90 88.8 90.2 83.3 91.9
TIFF-EDD[13] 90.6 91.1 3.06 28.4 82 94.1 95.8 89.9 96.5
AEAF-ABDD 93.2 93.9 3.00 12.8 105 97.7 98.6 95.9 97.9
), ArticleFig(id=1263550862631231641, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, language=CN, label=表3, caption=

与其他算法在EDD-AB数据集上的对比实验

, figureFileSmall=null, figureFileBig=null, tableContent=
Method P/% R/% Parms/106 GFLOPs FPS mAP/%
综合 划痕 点痕 边缘损伤
Faster-RCNN 76.6 77.6 165.00 199.0 66 80.2 87.7 66.2 86.6
SSD 63.0 65.3 24.50 87.9 72 71.0 77.6 55.3 79.6
DETR 77.5 78.0 9.49 16.8 88 87.0 91.3 77.5 91.8
Yolov10 89.6 87.5 2.59 6.4 96 92.8 97.1 85.7 96.0
Yolov12 85.3 88.7 2.52 6.0 93 91.8 94.8 89.1 92.2
EMS-YOLO[18] 82.3 84.5 14.40 6.8 90 88.8 90.2 83.3 91.9
TIFF-EDD[13] 90.6 91.1 3.06 28.4 82 94.1 95.8 89.9 96.5
AEAF-ABDD 93.2 93.9 3.00 12.8 105 97.7 98.6 95.9 97.9
), ArticleFig(id=1263550862979358876, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, language=EN, label=Table 4, caption=

Ablation experiments on the EDD-AB dataset

, figureFileSmall=null, figureFileBig=null, tableContent=
Base MSAFPN SimAM StarNet P/% R/% Parms/106 GFLOPs FPS mAP/%
综合 划痕 点痕 边缘损伤
87.1 89.6 3.00 8.2 99 92.9 96.4 87.8 94.7
92.6 93.0 2.93 12.4 100 94.0 96.4 87.6 96.5
89.6 90.8 3.01 8.2 98 93.6 95.4 90.9 97.0
88.6 90.5 3.08 8.5 89 94.1 96.9 89.2 95.8
91.9 92.3 2.93 12.4 101 95.5 97.6 93.2 96.5
90.3 90.8 3.08 8.5 96 94.2 96.3 88.6 96.6
91.6 92.5 3.00 12.8 93 95.2 95.8 89.7 97.3
93.2 93.9 3.00 12.8 105 97.7 98.6 95.9 97.9
), ArticleFig(id=1263550864938098851, tenantId=1146029695717560320, journalId=1263187241531621409, articleId=1263514355325387636, language=CN, label=表4, caption=

在EDD-AB数据集上的消融实验

, figureFileSmall=null, figureFileBig=null, tableContent=
Base MSAFPN SimAM StarNet P/% R/% Parms/106 GFLOPs FPS mAP/%
综合 划痕 点痕 边缘损伤
87.1 89.6 3.00 8.2 99 92.9 96.4 87.8 94.7
92.6 93.0 2.93 12.4 100 94.0 96.4 87.6 96.5
89.6 90.8 3.01 8.2 98 93.6 95.4 90.9 97.0
88.6 90.5 3.08 8.5 89 94.1 96.9 89.2 95.8
91.9 92.3 2.93 12.4 101 95.5 97.6 93.2 96.5
90.3 90.8 3.08 8.5 96 94.2 96.3 88.6 96.6
91.6 92.5 3.00 12.8 93 95.2 95.8 89.7 97.3
93.2 93.9 3.00 12.8 105 97.7 98.6 95.9 97.9
)], attaches=null, journal=Journal(id=1263185177418154013, delFlag=0, nameCn=图学学报, nameEn=Journal of Graphics, nameHistory1=null, nameHistory2=null, issn=2095-302X, eissn=null, cn=10-1034/T, coden=null, periodic=1, language=CN, oaType=null, 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=o2S94GfCLjDYYJqsdQm5Mw==, journalPrice=null, startedYear=null, abbrevIsoEn=Journal of Graphics, journalRemark=null, publicationField=null, createdTime=1779096415144, updatedTime=1779097332356, createdBy=18614031015, updatedBy=13701087609, firstLetterCn=J, firstLetterEn=J, subjectCode=Natural Sciences, subjectName=null, subjectCodeEn=Natural Sciences, subjectNameEn=null, picCn=o2S94GfCLjDYYJqsdQm5Mw==, picEn=f7IBTa1ZXKPoIhMUwL1eVw==, jcr=null, cjcr=null, exts=[JournalExt(id=1263189024693469540, language=CN, name=图学学报, nameHistory1=null, nameHistory2=null, managedBy=, sponsoredBy=, publishedBy=, editorOffice=, officeProv=null, officeCity=null, officeAddr=, officeZip=, editDirector=, officeDirector=null, officePhone=null, coverPicUrl=null, journalRemark=, submitArticleUrl=null, websiteUrl=, createdTime=1779097332404, updatedTime=1779097332404, createdBy=13701087609, updatedBy=13701087609, submissionGuidelinesUrl=, submissionAuthorUrl=http://www.txxb.com.cn/Journalx_txxb/authorLogOn.action, submissionEditorUrl=http://www.txxb.com.cn/Journalx_txxb/editorLogOn.action, submissionReviewUrl=http://www.txxb.com.cn/Journalx_txxb/expertLogOn.action, submissionCeEditorUrl=, submissionAeEditorUrl=, option={"copyright":""}), JournalExt(id=1263189024764772709, language=EN, name=Journal of Graphics, nameHistory1=null, nameHistory2=null, managedBy=, sponsoredBy=, publishedBy=, editorOffice=, officeProv=null, officeCity=null, officeAddr=, officeZip=, editDirector=, officeDirector=null, officePhone=null, coverPicUrl=null, journalRemark=, submitArticleUrl=null, websiteUrl=, createdTime=1779097332421, updatedTime=1779097332421, createdBy=13701087609, updatedBy=13701087609, submissionGuidelinesUrl=, submissionAuthorUrl=http://www.txxb.com.cn/Journalx_txxb/authorLogOn.action, submissionEditorUrl=http://www.txxb.com.cn/Journalx_txxb/editorLogOn.action, submissionReviewUrl=http://www.txxb.com.cn/Journalx_txxb/expertLogOn.action, submissionCeEditorUrl=, submissionAeEditorUrl=, option={"copyright":""})], databaseList=null, tenantJournalId=1263187241531621409, websiteList=[Website(id=1263188159811207473, webName=null, webTitle=null, webDomain=null, webCopyrigh=null, webIpcNo=null, seoTitle=null, seoKeywords=null, seoDescription=null, tenantJournalId=null, journalId=1263187241531621409, 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/txxb/CN, language=CN, createTime=1779097126200, createBy=18614031015, updateTime=1779097561850, updateBy=18614031015, name=图学学报-中文, tplId=1146099689490845704, title=图学学报, delFlag=0, indexPage=/home, props=[WebsiteProps(id=1263194793929204110, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1263188159811207473, code=articleTextType, value=kx, createTime=1779098707897, updateTime=1779098707897, creator=18614031015, updator=18614031015), WebsiteProps(id=1263194793853706635, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1263188159811207473, code=banner, value=null, createTime=1779098707879, updateTime=1779098707879, creator=18614031015, updator=18614031015), WebsiteProps(id=1263194793966952849, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1263188159811207473, code=grayFlag, value=0, createTime=1779098707906, updateTime=1779098707906, creator=18614031015, updator=18614031015), WebsiteProps(id=1263194793824346506, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1263188159811207473, code=logo, value=https://castjournals.cast.org.cn/joweb/txxb/CN/file/pic?fileId=lmN7m2FoR6FhgnIimGeZBg==, createTime=1779098707872, updateTime=1779098707872, creator=18614031015, updator=18614031015), WebsiteProps(id=1263194793992118675, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1263188159811207473, code=minRunFlag, value=0, createTime=1779098707912, updateTime=1779098707912, creator=18614031015, updator=18614031015), WebsiteProps(id=1263194793891455373, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1263188159811207473, code=picServerUrl, value=https://castjournals.cast.org.cn/joweb/txxb/CN/file/pic, createTime=1779098707888, updateTime=1779098707888, creator=18614031015, updator=18614031015), WebsiteProps(id=1263194793979535762, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1263188159811207473, code=silenceFlag, value=0, createTime=1779098707909, updateTime=1779098707909, creator=18614031015, updator=18614031015), WebsiteProps(id=1263194793883066764, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1263188159811207473, code=staticResourcePath, value=https://castjournals.cast.org.cn/joweb/cast_kjdb_cn_619/, createTime=1779098707886, updateTime=1779098707886, creator=18614031015, updator=18614031015), WebsiteProps(id=1263194793933398415, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1263188159811207473, code=themeColor, value=null, createTime=1779098707898, updateTime=1779098707898, creator=18614031015, updator=18614031015), WebsiteProps(id=1263194793941787024, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1263188159811207473, code=themeStyle, value=null, createTime=1779098707900, updateTime=1779098707900, creator=18614031015, updator=18614031015)]), Website(id=1263188160184500536, webName=null, webTitle=null, webDomain=null, webCopyrigh=null, webIpcNo=null, seoTitle=null, seoKeywords=null, seoDescription=null, tenantJournalId=null, journalId=1263187241531621409, 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/txxb/EN, language=EN, createTime=1779097126289, createBy=18614031015, updateTime=1779097557941, updateBy=18614031015, name=图学学报-英文, tplId=1146101810881728533, title=Journal of Graphics, delFlag=0, indexPage=/home, props=[WebsiteProps(id=1263194821443838360, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1263188160184500536, code=articleTextType, value=kx, createTime=1779098714457, updateTime=1779098714457, creator=18614031015, updator=18614031015), WebsiteProps(id=1263194821393506709, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1263188160184500536, code=banner, value=null, createTime=1779098714445, updateTime=1779098714445, creator=18614031015, updator=18614031015), WebsiteProps(id=1263194822144287131, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1263188160184500536, code=grayFlag, value=0, createTime=1779098714624, updateTime=1779098714624, creator=18614031015, updator=18614031015), WebsiteProps(id=1263194821334786452, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1263188160184500536, code=logo, value=https://castjournals.cast.org.cn/joweb/txxb/EN/file/pic?fileId=lmN7m2FoR6FhgnIimGeZBg==, createTime=1779098714431, updateTime=1779098714431, creator=18614031015, updator=18614031015), WebsiteProps(id=1263194822316253597, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1263188160184500536, code=minRunFlag, value=0, createTime=1779098714665, updateTime=1779098714665, creator=18614031015, updator=18614031015), WebsiteProps(id=1263194821414478231, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1263188160184500536, code=picServerUrl, value=https://castjournals.cast.org.cn/joweb/txxb/EN/file/pic, createTime=1779098714450, updateTime=1779098714450, creator=18614031015, updator=18614031015), WebsiteProps(id=1263194822249144732, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1263188160184500536, code=silenceFlag, value=0, createTime=1779098714649, updateTime=1779098714649, creator=18614031015, updator=18614031015), WebsiteProps(id=1263194821401895318, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1263188160184500536, code=staticResourcePath, value=https://castjournals.cast.org.cn/joweb/cast_kjdb_en_623/, createTime=1779098714448, updateTime=1779098714448, creator=18614031015, updator=18614031015), WebsiteProps(id=1263194821716468121, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1263188160184500536, code=themeColor, value=null, createTime=1779098714522, updateTime=1779098714522, creator=18614031015, updator=18614031015), WebsiteProps(id=1263194821850685850, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1263188160184500536, code=themeStyle, value=null, createTime=1779098714554, updateTime=1779098714554, creator=18614031015, updator=18614031015)])], journalTitle=图学学报, weixinUrl=null, journalUrl=http://www.txxb.com.cn/, iacademicId=null, status=1, seqNo=null, journalTitleEn=Journal of Graphics, journalPhotoCn=o2S94GfCLjDYYJqsdQm5Mw==, journalPhotoEn=f7IBTa1ZXKPoIhMUwL1eVw==, journalFirstLetter=J, 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=, provinceCode=null, provinceName=null, collectFlag=false), detailUrlCn=https://castjournals.cast.org.cn/joweb/txxb/CN/10.11996/JG.j.2095-302X.2026010120, detailUrlEn=https://castjournals.cast.org.cn/joweb/txxb/EN/10.11996/JG.j.2095-302X.2026010120, pdfUrlCn=https://castjournals.cast.org.cn/joweb/txxb/CN/PDF/10.11996/JG.j.2095-302X.2026010120, pdfUrlEn=https://castjournals.cast.org.cn/joweb/txxb/EN/PDF/10.11996/JG.j.2095-302X.2026010120, aliStartDate=null, aliEndDate=null, collectionFlag=false, citedCount=null, citedUrl=null, reference=null)
收藏切换
基于动态视觉传感器的航发叶片缺陷检测
收藏切换
PDF下载
张行顺 , 陈海永
图学学报 | 图像处理与计算机视觉 2026,47(1): 120-130
收起
收藏切换
图学学报 | 图像处理与计算机视觉 2026, 47(1): 120-130
基于动态视觉传感器的航发叶片缺陷检测
全屏
张行顺, 陈海永
作者信息
  • 河北工业大学人工智能与数据科学学院, 天津 300401

通讯作者:

陈海永,E-mail:
Defect detection of aero-engine blades based on dynamic vision sensors
Xingshun ZHANG, Haiyong CHEN
Affiliations
  • School of Artificial Intelligence and Data Science, Hebei University of Technology, Tianjin 300401, China
出版时间: 2026-02-28 doi: 10.11996/JG.j.2095-302X.2026010120
文章导航
收藏切换

航空发动机叶片作为发动机核心零部件,其表面微小缺陷可能导致严重安全事故,传统视觉检测技术受限于运动模糊、动态范围低及背景冗余等问题。针对上述挑战,提出一种基于动态视觉传感器(DVS)的航发叶片缺陷检测方法。动态视觉传感器数据格式为异步事件流,故也被称作事件相机,具有动态范围大、高帧率和微小目标捕捉能力强等优势。首先搭建基于DVS的缺陷检测平台,探索总结了其成像特点及优势。在此基础上,构建首个基于DVS的航发叶片缺陷检测数据集(EDD-AB),涵盖划痕、点痕、边缘损伤3类缺陷近6 000张图像,精细标注近1.2万个目标标签,数据集已开源(链接: https://github.com/NiBieZhouMei5520/EDD-AB.git)。进一步提出基于异步事件流帧聚合的多尺度缺陷检测算法(AEAF-ABDD):通过固定时间窗的帧聚合技术实现事件流可视化;构建多分辨率自适应特征金字塔网络(MRAFPN)增强多尺度缺陷特征提取能力;引入轻量级SimAM注意力机制强化关键区域聚焦;融合星形卷积模块(StarNet)提升高维非线性特征映射效率,实现复杂曲面工件多尺度缺陷的精准检测。实验表明,AEAF-ABDD在EDD-AB数据集上的平均精度均值(mAP)达97.7%,检测速度达105帧/秒,显著优于主流算法,为高反光曲面工件的自动化质检提供了高效解决方案,推动了DVS在工业检测领域的应用。

动态视觉传感器  /  航空发动机叶片  /  缺陷检测  /  异步事件流  /  多尺度特征融合

Aeroengine blades are core components of engines; tiny surface defects can lead to serious safety accidents. Traditional vision detection technology is limited by motion blur, low dynamic range, background redundancy, and so forth. To address these challenges, a method of aeroengine blade defect detection based on Dynamic Vision Sensor (DVS) was proposed. Dynamic vision sensor produced data in an asynchronous event-stream format, and were therefore referred to as event camera, which exhibited the advantages of large dynamic range, high frame rate, and strong ability to capture small targets. Firstly, a defect detection platform based on DVS was built, and its imaging characteristics and advantages were explored. On this basis, the first Event-based Defect Detection Dataset of Aeroengine Blade (EDD-AB) dataset based on DVS was constructed, covering nearly 6 000 images of scratches, point marks and edge damage, with approximately 12 000 finely annotated target labels. The dataset was released as open source (link: https://github. com/NiBieZhouMei5520/EDD-AB.git). Furthermore, a multi-scale defect-detection algorithm based on asynchronous event-stream frame aggregation (AEAF-ABDD) was proposed: event streams were visualized through frame aggregation technology using a fixed time window; a Multi-Resolution Adaptive Feature Pyramid Network (MRAFPN) was developed to enhance multi-scale defect feature extraction capability; a lightweight SimAM attention mechanism was incorporated to strengthen focus on key regions; a star-convolution module (StarNet) was fused to improve the efficiency of high-dimensional nonlinear feature mapping, enabling accurate detection of multi-scale defects on complex curved workpieces. Experiments demonstrated that AEAF-ABDD achieved a mean Average Precision (mAP) of 97.7% on the EDD-AB dataset and a detection speed of 105 frames per second, substantially outperforming mainstream algorithms. An efficient solution for automated quality inspection of highly reflective curved workpieces was thereby provided, promoting the application of DVS in the field of industrial inspection.

dynamic vision sensor  /  aeroengine blades  /  defect detection  /  asynchronous event stream  /  multi-scale feature
张行顺, 陈海永. 基于动态视觉传感器的航发叶片缺陷检测. 图学学报, 2026 , 47 (1) : 120 -130 . DOI: 10.11996/JG.j.2095-302X.2026010120
Xingshun ZHANG, Haiyong CHEN. Defect detection of aero-engine blades based on dynamic vision sensors[J]. Journal of Graphics, 2026 , 47 (1) : 120 -130 . DOI: 10.11996/JG.j.2095-302X.2026010120
航空发动机叶片作为航空发动机涡轮和压气机的核心零部件,需要在高转速、高温、高压力的严苛环境中长期工作,任何微小的缺陷或损伤(如裂纹、腐蚀、磨损、疲劳损伤等)都可能导致叶片失效,进而引发发动机故障。故早期发现和及时修复潜在缺陷至关重要,不仅可以避免因叶片失效造成的重大安全事故,也可以延长叶片的使用寿命,减少提前更换的频率,保持发动机在最佳性能状态,提高飞机整体运行效率,降低航空公司的维护成本。
因此,国内外在发动机服役期间,均需要对叶片进行定期检修。目前主流的检测方法分为:人工目视检测、利用介质无损探伤和机器视觉检测3类方法。
传统的人工目视检测依赖于操作人员的经验和技能,主观性较强、准确性和检测精度有限,且耗时长、效率低,不适应大规模工业生产应用。利用介质的无损探伤对于复杂几何形状的叶片适应性较差,并且航发叶片使用的高强度合金材料与表面涂层也影响了检测效果。随着机器视觉的发展[1],视觉缺陷检测技术[2-3]凭借其高精度、高速度、自动化和智能化等优势,成为当前的主流方法,广泛应用于3C电子、汽车制造、金属加工和印刷纺织等行业。目前也已经有成熟可行的缺陷检测算法,如以SSD (Single Shot Multibox Detector)[4]系列和YOLO (You Only Look Once)[5]系列等为代表的单步检测算法,及以R-CNN (Region-Based Convolutional Neural Networks)[6]系列为代表的两步检测算法。
然而,视觉缺陷检测技术依赖于相机提供的信息,在工业品表面缺陷检测中,会受到传统相机以下缺陷的限制:
1)传统视觉传感器以固定帧率采集图像,存在延迟问题和中间帧信息丢失问题,在针对钢管、叶片等曲面工件需要360°全周成像的高速成像场景时尤为突出。此外,由于较慢的成像速度,在曝光时间内拍摄流水线中快速运动目标时,易产生运动模糊。
2)对前景和背景赋予相同权重成像,易造成感兴趣目标不够显著、信息冗余等问题,进而影响后续识别算法的精度与检测效率。
3)传统视觉相机成像动态范围小,在过曝或过暗的环境中无法有效成像[7-9],尤其在对金属工件、曲面工件等具有类镜面反射特性的物体补光拍摄时,易产生过曝现象,使画面高光区域部分信息丢失。此外,由物体表面反射产生的类缺陷伪影也会降低检测系统的可靠性。
上述问题导致依靠传统视觉的缺陷检测系统,难以对金属、曲面等类镜面反射物体进行快速成像和精准缺陷检测。由传统相机的硬件限制造成的问题,难以从软件方面去避免。
针对航空发动机叶片等高反光材质表面的微小缺陷快速检测需求,本文提出一种基于动态视觉传感器(Dynamic Vision Sensor, DVS)的缺陷检测方法。DVS具有与传统相机不同的成像原理,其只对动态变化的光强作出快速响应,具有低延迟、高动态范围和低功耗等优点,可以适应复杂环境和强曝光干扰等挑战性场景[10-12]。目前已有若干将DVS运用到缺陷检测中的成功案例,如马居坡等[13]提出的基于DVS的铝基盘片表面缺陷检测方法(Temporal Irregular Feature Aggregation Framework for Event-Based Defect Detection, TIFF-EDD);吕和平和洪流[14]利用最大池卷积脉冲神经网络(Convolutional Spiking Neural Network Max-pooling,CSNN Max-pooling)处理事件相机数据,并进行铝基盘片划痕检测研究;邓坚和万小芳[15]提出了基于事件相机的工件缺陷检测方法及装置专利。
然而,上述研究大多集中于针对铝基盘片等平面光滑物体,不同于本文中曲面形态的航发叶片工件。航发叶片的高曲率结构导致表面缺陷形态更加多变,反光效应也更加强烈。
此外,铝基盘片等平面工件仅需进行平面旋转就可实现单面检测,而航空发动机叶片需进行360°高速旋转以实现全面、高效的缺陷检测。并且由于其旋转运动,导致同一缺陷在不同角度和光照条件下的成像各异,由此对缺陷检测提出了更高挑战。
本文根据航发叶片物理特性和DVS的成像原理,构建了一套仿人眼DVS缺陷检测实验平台,选用实际服役中因损伤而报废替换的航空发动机叶片进行一系列的验证试验,分析DVS的成像特点与应用于缺陷检测的优势。实验结果表明,DVS可以在强曝光与高速运动的情况下稳定、清晰检出缺陷目标,且可去掉背景冗余信息和环境倒影信息,有效检出航发叶片的损伤。
随后,本文构建了首个基于DVS的航发叶片缺陷检测数据集(Event-Based Defect Detection Dataset of Aeroengine Blade, EDD-AB),包含划痕、点痕和边缘损伤3类缺陷,并完成近1.2万个样本的标注工作。
最后,本文提出了一种基于异步事件流帧聚合的航发叶片多尺度缺陷检测算法(Asynchronous Event Stream Frame Aggregation-based Multi-Scale Defect Detection Algorithm for Aeroengine Blades, AEAF-ABDD)。选择CSPDarkNet53,PAFPN和DeCouple Head组成基础模型。针对事件流数据异步、空间稀疏的特点,提出一种基于异步事件流帧聚合技术的数据提取及可视化方法;针对叶片缺陷损伤微小,与背景相似度高,设计一种多分辨率自适应特征金字塔网络(Multi-Resolution Adaptive Feature Pyramid Network, MRAFPN),实现对不同种类、不同大小缺陷的精准提取与判别;为抑制图片中不重要信息,集中处理重要信息,提高模型的鲁棒性和泛化能力,加入一个轻量级自注意机制(Similarity-Based Attention Mechanism, SimAM)[16]模块;为了应对缺陷形态多样,缺陷总体较小,设计一种星操作的高维非线性特征映射卷积StarNet。在EDD-AB数据集上进行了一系列的验证实验,结果表明,本文算法能够实现对航发叶片缺陷的有效检测。主要贡献如下:
1) 提出了基于DVS的航发叶片表面缺陷检测方法,搭建了基于DVS的实验平台并开展缺陷成像实验,分析了DVS缺陷成像的特性优势。
2) 构建首个基于DVS的航发叶片表面缺陷检测数据集,包含214段事件流,5 964张事件聚合帧,对3类常见缺陷类型完成近1.2万个样本的标注工作。目前数据集已经开源(数据集链接: https://github.com/NiBieZhouMei5520/EDD-AB.git)。
3) 针对事件流数据异步、空间稀疏、缺陷形态与大小差异大的特点,提出 AEAF-ABDD算法,并在EDD-AB数据集上进行一系列实验评估,具备显著有效性。
DVS也称作事件相机,是根据生物视网膜系统特性研制的仿生相机。相机芯片上的每个感光单元独立地对光强变化作出响应,当变化量超过阈值,生成非结构化的异步事件流[17]。每一个事件event由一个四元组表示为$e=\left({x}_{e},{y}_{e},{t}_{e},{p}_{e}\right)$。其中${x}_{e}$${y}_{e}$为该事件发生的坐标;${t}_{e}$为该事件发生的时间戳;${p}_{e}$为该事件的极性,表征光强变化的趋势。对于一个时间间隔内产生的n个事件可表示为
$E=\left\{{e}_{1},{e}_{2},\cdots,{e}_{n}\right\}={\left\{\left({x}_{i},{y}_{i},{t}_{i},{p}_{i}\right)\right\}}_{i=1}^{n}$
感光单元输出事件的数学表达式为
$\left|\Delta \mathrm{log}I\right|=\left|\mathrm{log}I\left(u,t\right)-\mathrm{log}I\left(u,t-\Delta t\right)\right|\ge \left|{p}_{k}C\right|$
式中:$u=\left(x,y\right)$表示感光单元坐标位置;I表示光强;t表示当前事件的发生时刻;C表示事件产生阈值。当光强变化超过阈值C时,则产生事件信号。${p}_{k}\in \{+1,-1\}$表示产生的事件极性,其中${p}_{k}=+1$表示发生光强变大超过阈值的“ON”事件;${p}_{k}=-1$表示发生光强变小超过阈值“OFF”事件。图1展示了DVS感光单元由于光强变化产生事件的工作原理图。
图1可知,单位时间内光强变化越大,产生事件越多。若要相机感受到的光强发生变化,可令环境的亮度发生变化,或令相机与待测工件发生相对位移。当环境亮度变化越剧烈时,相机或待测工件移动越快,单位时间内产生的事件就越多。
图2中,Light表示光源,Camera表示相机,箭头表示光线传播路径;从图2(a)图2(b),光源与相机均固定,只有工件移动。在拍摄规则平滑工件时,如果工件表面存在缺陷区域,如细微划痕或凹坑,当工件移动时,可见凹坑缺陷处光线的反射发生了变化,导致相机中的感光单元感受光强发生了变化,从而产生记录缺陷信息的事件,而正常区域对光线的反射未变化,则不会产生事件信息。由图1可知,单位时间内光强变化越快,产生事件信息越多,故可令工件移动速度变快,加剧缺陷部位的光强变化,从而记录更多的缺陷部位事件信息。
本文以航空发动机叶片为检测目标,模拟真实工厂实际质检场景,构建了一套基于DVS的缺陷检测实验平台,并展开了一系列的成像实验。
航空发动机叶片是航空发动机涡轮和压气机的核心零部件,任何微小的缺陷或损伤都可能降低涡轮机组的工作效率和使用寿命,甚至造成重大安全事故,故需要定期对叶片进行检修。一组航空涡轮发动机叶片数量有3 000 ~ 4 000件,甚至更多,但目前主流的方法还是使用人工目检,效率低下,亟需一种可靠高效的自动化检测方法。
图3为本文搭建的实验平台,主要由DVS、转台、工业光源、相机支架及航发叶片夹具组成。动态视觉传感器选择DVSLume,搭载PROPHESEE芯片,空间分辨率为1280 × 720,像元尺寸为4.86 μm × 4.86 μm,等效时间分辨率 >10 k FPS,动态范围 > 110 dB,相机通过支架固定,动态视觉传感器距航发叶片约130 mm,视野范围为230 mm × 150 mm;转台选用Microvision维视图像的,最高转速可达60°/s,航发叶片通过夹具固定于转台中央,经调试实验选择以60°/s的转速匀速运动。条形灯带选用海康威视白色工业光源对叶片进行补光,顶光光源功率8 W,背光光源功率3 W,确保叶片表面能获得充分且均匀的光照,提高成像效果。
基于该实验平台展开了一系列的测试,对DVS缺陷成像效果进行了实验分析。其特性如下:
1) 弱小缺陷成像明显。DVS可以精准捕获由于缺陷导致的光线强度发生变化。图4展示了DVS与传统相机对于深度约为1 μm、长度约为5 mm的划痕成像效果。如图4(a)中红框所示,DVS对微小划痕成像明显;如图4(b)中红框所示,传统相机难以对曲面金属物体上的微小缺陷进行成像。
2) 强曝光下稳定成像。在成像过程中常需借助额外光源进行补光,使其清晰稳定成像,实现目标特征与背景的有效分离并增强信噪比。但对于类镜面反射物体,强烈的反光会导致传统相机过曝而无法成像。图5展示了DVS与传统相机在补光条件下的成像效果。如图5(a)中红框所示,DVS凭借其高动态范围可以克服强光干扰,依然稳定有效成像;如图5(b)中红框所示,传统相机在强光条件下时,无法对缺陷成像。
3) 可以只保留感兴趣目标,去除背景信息干扰。传统相机对前景和背景进行同权重成像,造成信息冗余。而DVS只对光强变化作出响应,即只对运动物体成像。故在拍摄时只令叶片进行旋转,便可以去掉背景冗余信息,并进行成像,以提高检测速度。
4) 响应速度快,能有效捕捉快速移动物体。因为叶片呈现曲面结构,同一缺陷在不同角度缺陷成像存在差异,要实现缺陷的精准检测,需要尽可能对每个角度的叶片进行拍摄成像。传统高精度相机成像速度慢,无法拍摄高速运动物体,叶片只能慢速旋转,检测效率低下。DVS可以捕捉高速运动物体,允许叶片在高速旋转状态下完成全周检测,显著提升了检测效率。
基于2.1节中的实验平台,本文采集了大量数据,并构建了首个基于DVS的航空发动机叶片数据集EDD-AB。为保证缺陷真实可靠,所有叶片均来自检修厂实际因缺陷报废的航发叶片。根据检修厂技术负责人的反馈,最常见的3类高危害性缺陷及其特征为:划痕长度一般在1~25 mm、深度在0.5~500.0 μm,点痕直径在0.1~2.0 mm,以上2种缺陷通常导致产生更严重的裂纹和疲劳损伤,且由其造成的表面涂层划伤,常成为腐蚀的起点。导致边缘不再平整光滑,且边缘损伤直接影响其几何完整性,会对涡轮机组空气动力性能和结构强度构成更严重的威胁。
本文共对67张叶片进行了数据采集,获取了214段不同光照强度与不同叶片姿态下的叶片数据,收集了以上3种常见并危害最大的缺陷类型,包括划痕、点痕和边缘损伤,如图6所示。
针对事件流数据异步、空间稀疏带来的可视化与标注困难等问题,本文创新性提出一种基于异步事件流帧聚合技术的数据处理方法。具体而言,采用固定时间窗宽∆T = 10 ms的时序划分策略,通过三维时空事件向二维图像平面的投影转换,实现异步事件流的高效可视化表征。基于此方法构建包含5 964张叶片缺陷图像的数据集,并借助LabelImg专业图像标注工具完成近12 000个目标的精细标注工作。
为了进一步研究DVS的成像特性与缺陷的特性,针对性的设计缺陷检测算法,本文统计了3种缺陷的宽高比特征与像素尺度分布,如图7所示。
对数据集中各种缺陷进行统计可知,划痕数量为5 210个,宽高比覆盖0 ~ 6的宽泛区间,主要集中在0 ~ 1之间,说明纵向划痕占比较大,部分3 ~ 4分布,说明横向划痕占比较小;点痕数量为3 553个,边缘损伤数量为3 068个,点痕与边缘损伤的宽高比集中在0.4 ~ 1.4之间,说明2种缺陷呈现局部点状损伤的形态特征。图7(d)揭示了缺陷目标的像素尺度分布特征:约70%的样本面积集中在2 000像素以下,同时存在1 000 ~ 10 000像素的中大型缺陷(占27%),以及少量超10 000像素的特例样本(占3%)分析可知,大尺度样本主要是斜形的划痕导致目标框变大。由图7可知,缺陷样本在宽高比和尺寸范围上均呈显著差异性,对缺陷检测算法的多尺度特征提取与目标检测能力提出了挑战。
动态视觉传感器的输出与传统相机的输出有着根本的不同:事件是异步的,空间上很稀疏;而图像是同步的,很密集。因此,为图像序列设计的基于帧的视觉算法并不直接适用于事件数据。
针对DVS特殊的数据格式,许多学者做了大量研究,主要可以分为2类:①直接处理数据流的算法;②基于事件流与传统图像帧结合的算法。CHEN等[18]提出了基于帧的卷积神经网络(Convolutional Neural Network, CNN)来训练脉冲神经网络(Spiking Neural Network, SNN),并在现场可编程门阵列(Field-Programmable Gate Array, FPGA)中进行实现。 AYDIN等[19]采用混合ANN-SNN架构,通过低频运行的辅助ANN初始化SNN状态,解决了SNN状态瞬变和状态衰减的问题。FAN[20]提出一种针对伪帧的目标检测算法,首先将事件信息转化为伪帧,以YOLO网络为基础,优化网络的各部分结构,在较小规模的深度学习模型下获得很好的检测效果。文献[13]则基于时序不规则特征聚合框架的DVS缺陷检测算法TIFF-EDD,实现对缺陷目标的有效检测。
对比以上方法,SNN[21-22]和事件流数据更加契合,但存在训练难度大和数据需求高的问题。综合考虑实际任务需求与硬件环境,平衡提升检测效率和保留事件流数据特性。本文基于异步事件流帧聚合的方法设计一种多分辨率自适应特征金字塔网络,提出更高效便捷的解决方案。参照YOLO系列的网络结构,选择CSPDarkNet53[23],PAFPN和DeCouple Head组成基础模型,结合DVS的数据时空关联特性和缺陷形状多变,缺陷尺寸不一,提出基于异步事件流帧聚合的航发叶片多尺度缺陷检测算法,以实现对航发叶片更高效的缺陷检测。
事件仅对场景中的亮度变化进行非冗余编码,通过对事件流进行聚合,获取亮度变化数据,以表征运动和叶片表面缺陷信息。凭借事件固有的超高时间分辨率,这种聚合方法能够以极高的帧率重建出累积的亮度图像。
选取一个时间窗口∆T = 10 ms,统计该时间段内事件流$E={\{({x}_{i},{y}_{i},{t}_{i},{p}_{i})\}}_{i=1}^{n}$每个像素点发生的事件情况,并按式(3)进行处理,对聚合帧图像F[24]的每一个像素点赋值,即
$F(x,y)=1$,if $x={x}_{e}$ and $y={y}_{e}$
在该事件窗口∆T内,只要某像素点产生事件,则该像素点取1,否则取0。通过该方法,可将一段事件流转换为事件帧序列${\displaystyle \sum {F}_{k}}$
参照先进高效的YOLO检测模型,设计了一种多尺度自适应特征金字塔网络(Multi-Scale Adaptive Feature Pyramid Network, MSAFPN)。针对传统特征金字塔因网络深度增加导致的图像信息丢失与特征图分辨率衰减问题,本文通过FPN+ PAN的双向特征融合机制,系统整合低分辨率强语义特征与高分辨率弱语义特征。值得注意的是,随着卷积层堆叠和重复下采样操作,小目标特征易被高层语义信息淹没的痛点问题尤为突出。为此,本文对特征融合模块进行多维拓展:首先通过设计四尺度自适应检测头,在浅层至深层网络分别部署大、中、小、微小目标专用检测分支,形成缺陷特征解耦增强架构;其次通过强化局部特征提取能力,充分利用不同层级特征图的细粒度信息。该架构创新性地实现了从宏观表面划痕到亚像素级点痕缺陷的多维覆盖,其浅层检测头可捕获微小缺陷的局部突变特征,深层检测头则聚焦大尺度缺陷的全局表征。图8为添加的小目标检测头。
SimAM是一种基于能量函数的轻量级无参的空间-通道协同优化方法,其核心在于通过空间-通道交互建模实现特征自适应增强,并同时关注特征图的空间维度、通道维度和实例维度,其结构图如图9所示。尤其是在小目标的检测中,小目标通常体现在图像的局部区域,特征较弱,但SimAM通过对空间、通道和实例的全面关注,能够在不增加参数的情况下放大其微小特征,提升模型对小目标的检测能力。
SimAM注意力模块基于神经科学理论来优化能量函数,通过评估每个神经元的重要性为其分配注意力权重。单个神经元的最小能量函数为
${e}_{t}^{*}=\frac{4\left({\widehat{\sigma }}^{2}+\lambda \right)}{{\left(t-\widehat{\mu }\right)}^{2}+2{\widehat{\sigma }}^{2}+2\lambda }$
$\widehat{\mu }=\frac{1}{M}{\displaystyle \sum _{i=1}^{M}{x}_{i}}$
${\widehat{\sigma }}^{2}=\frac{1}{M}{\displaystyle \sum _{i=1}^{M}{\left({x}_{i}-\widehat{\mu }\right)}^{2}}$
式中:${e}_{t}^{\ast }$表示每个神经元的最小能量函数;t表示输入特征图中每个神经元的值;$\lambda $表示正则化系数;M表示该通道的神经元总数;${x}_{i}$表示该通道中其他神经元;$\widehat{\mu }$${\widehat{\sigma }}^{2}$分别表示该通道上所有神经元的均值和方差。
由式(4)可知,能量越低,神经元t与其他神经元的区别度越大,表明其重要性越高,将能量函数取倒数$1/{e}_{t}^{*}$作为权重系数,并使用Sigmoid函数进行增强处理限制其大小,即
$\tilde{X}=\text{Sigmoid}\left(\frac{1}{E}\right)\odot X$
式中:$\tilde{X}$表示输出的增强特征;X表示输入特征;⊙表示点乘符号;E表示每个通道上的能量函数。
SimAM通过赋予不同区域不同的重要性权重,进一步提升模型目标的检测能力。SimAM注意力机制加入模型,将捕获的多尺度空间信息进一步通过注意力机制提炼出更加精确和有区分性的特征,从而提高模型的整体性能。
为了应对缺陷形态多样,缺陷总体较小,设计一种基于星形运算的高维非线性特征映射卷积。星形运算具有将特征投影到一个极高维隐式特征空间的能力,类似于多项式核函数[25]。其与传统神经网络的增加宽度或通道数的方法有所不同,星形运算是在不同的通道上逐元素相乘融合不同的子空间特征,通过隐式高维特征映射显著提升网络表征能力。元素级乘法本身具有非线性特性,可以结合通道间交叉乘积,且无需额外激活函数即可捕获复杂模式,从而降低计算复杂度。
图10所示,星形卷积Star Blocks主要由以下几个关键操作组成:
1) 深度可分离卷积(DepthWise Convolution, DW-Conv)。模块中使用DW-Conv层进行初步特征提取。其核大小(ks)为7,步长(Stride)为1。
2) 批量归一化(Batch Normalization, BN)和激活函数(ReLU)。该操作用于归一化数据和引入非线性。
3) 全连接层(Fully Connected layer, FC)。通过1×1卷积实现通道级的线性变换。
4) 星型运算(Element-wise multiplication),即星型乘法。这是Star Blocks模块的关键操作。星型运算将不同层的特征进行逐元素乘法,从而在不增加网络宽度的情况下,将输入映射到高维非线性特征空间,实现高效的特征表示。
星形卷积操作一般可以写作为$\left({W}_{1}^{\text{T}}X+{B}_{1}\right)\ast $$\left({W}_{2}^{\text{T}}X+{B}_{2}\right)$,其通过逐元素乘法可融合2个线性变换的特征。以一个输出通道转换和一个单元素输入场景为例,${w}_{1},{w}_{2},x\in {ℝ}^{\left(d+1\right)\times 1}$,其中d为通道数,${ℝ}^{(d+1)\times 1}$表示(d+1)×1维的实数向量空间;扩展到多个输出通道${W}_{1},{W}_{2},{ℝ}^{\left(d+1\right)\times 1}$,并处理多个特征元素,其中$X\in {ℝ}^{d+1}\times n$。一般表达为
$\begin{array}{l}{w}_{1}^{\text{T}}x\ast {w}_{2}^{\text{T}}x=\left({\displaystyle \sum _{i=1}^{d+1}{w}_{1}^{i}{x}^{i}}\right)\ast \left({\displaystyle \sum _{j=1}^{d+1}{w}_{2}^{j}{x}^{j}}\right)={\displaystyle \sum _{i=1}^{d+1}{\displaystyle \sum _{j=1}^{d+1}{w}_{1}^{i}{w}_{2}^{j}{x}^{i}{x}^{j}}}\\ =\underset{\left(d+2\right)\left(d+1\right)/2}{\underbrace{{\alpha }_{\left(1,1\right)}{x}^{1}{x}^{1}+\cdots +{\alpha }_{\left(4,5\right)}{x}^{4}{x}^{5}+\cdots +{\alpha }_{\left(d+1,d+1\right)}{x}^{d+1}{x}^{d+1}}}\end{array}$
式中,*表示星形运算,使用ij来索引通道,$\alpha $表示每个项的系数,即
${\alpha }_{\left(i,j\right)}\left\{\begin{array}{l}{w}_{1}^{i}{w}_{2}^{j},\text{if}i==j\\ {w}_{1}^{i}{w}_{2}^{j}+{w}_{1}^{j}{w}_{2}^{i},\text{if}i!==j\end{array}\right.$
在重新定义式(8)后,所描述的星操作可展开为$\left(d+2\right)\left(d+1\right)/2$个不同的项。除了${\alpha }_{\left(d+1,:\right)}{x}^{d+1}x$外,每个项与x都表现出非线性关联,表明其是独立且隐式的维度。因此可以在d维空间中进行运算,并在$\left(d+2\right)\left(d+1\right)/2\approx {\left(d/\sqrt{2}\right)}^{2}$(考虑到$d\ge 2$)的隐式维度空间中表示,显著增加了特征维度。
使用第2节构建的基于DVS的航发叶片缺陷检测数据集(EDD-AB),对本文提出的基于异步事件流帧聚合的航发叶片多尺度缺陷检测算法进行了一系列实验验证。
本文所有实验均在同一实验环境下进行,实验环境的软硬件配置见表1,实验参数见表2
本文检测精度指标使用精确率(Precision, P)、召回率(Reacll, R)、平均精度均值(Mean Average Precision, mAP),模型复杂度使用参数量(Params)、计算量(Giga Floating-Point Operations Per Second, GFLOPs),检测速度指标使用每秒帧数(Frames Per Second, FPS)。精确率、平均精度均值与召回率分别为
$P=\frac{TP}{TP+FP}$
$mAP=\frac{{\displaystyle \sum \left(AP\right)}}{n}$
$R=\frac{TP}{TP+FN}$
式中:TP表示模型正确预测为正类的样本数;FP表示模型错误预测为正类的样本数;FN表示模型错误预测为负类的样本数;AP表示单个类别的准确率;n表示类别数。
本文采用经典的10倍交叉验证来评估准确性[26]。将训练集分割为10个子集,每个子集用作测试集一次,其余9个子集用于训练集,用10次的平均结果来评估模型的性能。同时为了全面评估模型性能,保证模型可靠性,本文使用3个交并比(Intersection over Union, IoU)阈值(0.4,0.5,0.6),然后将每个IoU值对应的mAP取平均值作为最终结果。
为了验证本文提出的基于异步事件流帧聚合的航发叶片多尺度缺陷检测算法的优越性,在相同的实验环境下,与目前主流的性能优良的目标检测算法进行对比测试,对比结果见表3。并将检测结果热力图进行了对比展示,如图11所示。热力图可以提供模型预测结果的置信度的可视化表示,红色区域是模型关注的重点目标特征,用以衡量模型的检测性能。
表3实验结果可知,本文算法相较于大部分经典的算法,在精确度、召回率、mAP都有显著提升。2个经典算法SSD与Faster R-CNN在检测精度和检测速度方面较其他模型均无优势。DETR作为基于Transformer架构的检测算法,虽然在精度和检测速度均有提升,但相较于YOLO系列模型,在计算量、参数量和检测速度仍有较大差距;基于ANN-SNN混合架构的EMS-YOLO模型,在缺陷尺度差异更大、缺陷微小的ABDD-AB数据集中,表现并不如YOLO原始模型。相较于YOLO系列算法,本文算法虽然Parms和GFLOPs有增加,但是mAP提升了4.9%,实时检测帧率提高了39。在本文的检测任务中,AEAF-ABDD在参数量与计算量均低于TIFF-EDD的情况下,检测精度与效率更高,在实际应用场景和价值上更具优势。
图11所示,划痕缺陷特征明显,各项算法均能检出,点痕与边缘损伤因尺寸较小且特征微弱,致使检测难度增大,导致DETR和yolov10存在小目标漏检的情况,并且缺陷检测热力图均不如本算法AEAF-ABDD效果良好。TIFF-EDD虽然能检出所有缺陷类型,但其针对划痕和边缘损伤生成的热力图中,存在高响应区域(红色区域)与真实缺陷轮廓贴合度不足的问题,这表明在应对航发叶片缺陷形态分布广、尺寸差异大的复杂场景时,TIFF-EDD的定位精度和检测置信度均低于本文的AEAF-ABDD算法。
综上表明,AEAF-ABDD在目标检测领域实现了突破性的性能跨越,为高精度实时检测任务提供了更优解决方案。
为了评估本文方法对于航发叶片表面微小缺陷检测的有效性,在使用CSPDarkNet53,PAFPN和DeCoupleHead组成Base模型上设置多组消融实验,以评估种改进方式对于模型性能的贡献,结果见表4
表4可知,在Base模型中使用MSAFPN优化PAFPN,虽然增加了计算量,但模型参数量有稍许降低,且mAP取得1.0%的提升;加入SimAM注意力机制,在参数量和计算量几乎不变的情况下,mAP提高了0.7%,验证了注意力机制对特征聚焦的有效性,强化了局部缺陷区域的响应,尤其对不规则划痕检测提升显著;使用StarNet优化Base模型中的CSPDarkNet53,在适度增加3.2%计算量的前提下mAP提升1.2%,说明其层级特征融合策略对复杂场景检测具有显著增益;将改进模块两两加入Base,对mAP分别有2.6%,1.3%和2.3%的提升;本文模型将3个改进结合,计算量相较于基础模型有一定增加,但是检测精度显著提升,检测速度略有提升,mAP为97.7%,P为93.2%,R为93.9%,FPS为105帧/秒,相较于原始模型分别提升了4.8%,6.1%,4.3%和1.9%。说明本方法在针对航发叶片表面缺陷微小、形态多变,缺陷尺寸差异大等挑战时表现优异,可以实现对缺陷目标的准确、快速检测。
本文针对航空发动机叶片表面缺陷检测中传统视觉技术面临易过曝、缺陷微小多变等困难,提出了一种基于动态视觉传感器的缺陷检测新方法。首先搭建了一套基于DVS的航发叶片质检实验平台,探究DVS的成像机理与优势,构建了首个基于DVS的航发叶片缺陷检测数据(EDD-AB)。最后,结合异步事件流帧聚合技术与多尺度特征融合策略,设计了基于异步事件流帧聚合的航发叶片多尺度缺陷检测算法(AEAF-ABDD)。实验结果表明,该方法在面对复杂曲面工件及强曝光、高速运动情况下,能够显著提升微小缺陷(如划痕、点痕)的检测精度与鲁棒性,其综合性能优于传统检测算法。本研究不仅为航空发动机叶片的自动化质检提供了高效解决方案,还为DVS在工业高反光表面缺陷检测领域的应用提供了范例,未来有望进一步推广至其他复杂工况下的缺陷检测任务。
但目前算法计算复杂度相较于部分模型(如YOLOv10)有所增加,并且数据集覆盖范围有限,只包括3种缺陷数据。考虑到异步时间帧聚合可以去除背景冗余信息,未来工作将会继续对模型进行轻量化研究、减少模型参数量和冗余通道数,提升在嵌入式平台和工业边缘设备上的实时推理能力,并继续采集更多种类的缺陷数据、丰富数据集。
  • 国家重点研究发展计划项目(2022YFB3303804)
  • 国家自然科学基金(62473127)
参考文献 引证文献
排序方式:
[1]
陶显, 侯伟, 徐德. 基于深度学习的表面缺陷检测方法综述[J]. 自动化学报, 2021, 47(5): 1017-1034.
TAO X, HOU W, XU D. A survey of surface defect detection methods based on deep learning[J]. Acta Automatica Sinica, 2021, 47(5): 1017-1034 (in Chinese).
[2]
CHEN Y Q, PAN J W, LEI J Y, et al. EEE-Net: efficient edge enhanced network for surface defect detection of glass[J]. IEEE Transactions on Instrumentation and Measurement, 2023, 72: 5029013.
[3]
JIANG W B, LIU M, PENG Y N, et al. HDCB-Net: a neural network with the hybrid dilated convolution for pixel-level crack detection on concrete bridges[J]. IEEE Transactions on Industrial Informatics, 2021, 17(8): 5485-5494.
[4]
LIU W, ANGUELOV D, ERHAN D, et al. SSD: single shot MultiBox detector[C]// The 14th European Conference on Computer Vision. Cham: Springer, 2016: 21-37.
[5]
REDMON J, DIVVALA S, GIRSHICK R, et al. You only look once: unified, real-time object detection[C]// 2016 IEEE Conference on Computer Vision and Pattern Recognition. New York: IEEE Press, 2016: 779-788.
[6]
GIRSHICK R, DONAHUE J, DARRELL T, et al. Rich feature hierarchies for accurate object detection and semantic segmentation[C]// 2014 IEEE Conference on Computer Vision and Pattern Recognition. New York: IEEE Press, 2014: 580-587.
[7]
ZHENG Y J, ZHENG L X, YU Z F, et al. High-speed image reconstruction through short-term plasticity for spiking cameras[C]// 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition. New York: IEEE Press, 2021: 6354-6363.
[8]
XU X Y, SUN D Q, PAN J S, et al. Learning to super-resolve blurry face and text images[C]// 2017 IEEE International Conference on Computer Vision. New York: IEEE Press, 2017: 251-260.
[9]
李家宁, 田永鸿. 神经形态视觉传感器的研究进展及应用综述[J]. 计算机学报, 2021, 44(6): 1258-1286.
LI J N, TIAN Y H. Recent advances in neuromorphic vision sensors: a survey[J]. Chinese Journal of Computers, 2021, 44(6): 1258-1286 (in Chinese).
[10]
LICHTSTEINER P, POSCH C, DELBRUCK T. A 128 × 128 120db 30mw asynchronous vision sensor that responds to relative intensity change[C]// 2006 IEEE International Solid State Circuits Conference-Digest of Technical Papers. New York: IEEE Press, 2006: 2060-2069.
[11]
CHEN S S, GUO M H. Live demonstration: CeleX-V: a 1M pixel multi-mode event-based sensor[C]// 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops. New York: IEEE Press, 2019: 1682-1683.
[12]
BRANDLI C, MULLER L, DELBRUCK T. Real-time, high-speed video decompression using a frame- and event-based DAVIS sensor[C]// 2014 IEEE International Symposium on Circuits and Systems. New York: IEEE Press, 2014: 686-689.
[13]
马居坡, 陈周熠, 吴金建. 基于动态视觉传感器的铝基盘片表面缺陷检测[J]. 自动化学报, 2024, 50(12): 2407-2419.
MA J P, CHEN Z Y, WU J J. Dynamic vision sensor based defect detection for the surface of aluminum disk[J]. Acta Automatica Sinica, 2024, 50(12): 2407-2419 (in Chinese).
[14]
吕和平, 洪流. 基于事件相机的铝基片划痕检测研究[J]. 数字自造科学, 2024, 22(2): 140-145.
LV H P, HONG L. Research on scratch detection of aluminum substrate based on event camera[J]. Digital Manufacture Science, 2024, 2(22): 140-145 (in Chinese).
[15]
邓坚, 万小芳. 一种基于事件相机的缺陷检测装置: 中国, 217212305U[P]. 2022-08-16.
DENG J, WAN X F. Defect detection device based on event camera: China, 217212305U[P]. 2022-08-16 (in Chinese).
[16]
YANG L X, ZHANG R Y, LI L D, et al. SimAM: a simple, parameter-free attention module for convolutional neural networks[EB/OL]. [2025-03-30]. https://proceedings.mlr.press/v139/yang21o.html. https://proceedings.mlr.press/v139/yang21o.html
[17]
GALLEGO G, DELBRUCK T, ORCHARD G, et al. Event-based vision: a survey[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2022, 44(1): 154-180.
[18]
CHEN S S, MARTINI B, CULURCIELLO E. A bio-inspired event-based size and position invariant human posture recognition algorithm[C]// 2009 IEEE International Symposium on Circuits and Systems. New York: IEEE Press, 2009: 775-778.
[19]
AYDIN A, GEHRIG M, GEHRIG D, et al. A hybrid ANN-SNN architecture for low-power and low-latency visual perception[C]// 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops. New York: IEEE Press, 2024: 5701-5711.
[20]
FAN Q D. Research on the dynamic object segmentation based on computer vision technology[C]// 2024 IEEE 2nd International Conference on Sensors, Electronics and Computer Engineering. New York: IEEE Press, 2024: 924-927.
[21]
HE X, ZHAO D C, LI Y, et al. An efficient knowledge transfer strategy for spiking neural networks from static to event domain[C]// The 38th AAAI Conference on Artificial Intelligence. Washington: AAAI Press, 2023: 512-520.
[22]
SU Q Y, CHOU Y H, HU Y F, et al. Deep directly-trained spiking neural networks for object detection[C]// 2023 IEEE/CVF International Conference on Computer Vision. New York: IEEE Press, 2023: 6532-6542.
[23]
TIAN Y J, YE Q X, DOERMANN D. YOLOv12:attention- centric real-time object detectors[EB/OL]. [2025-02-18]. https://arxiv.org/abs/2502.12524.pdf. https://arxiv.org/abs/2502.12524.pdf
[24]
KIM J, BAE J, PARK G, et al. N-ImageNet: towards robust, fine-grained object recognition with event cameras[C]// 2021 IEEE/CVF International Conference on Computer Vision. New York: IEEE Press, 2021: 2126-2136.
[25]
MA X, DAI X Y, BAI Y, et al. Rewrite the star[C]// 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition. New York: IEEE Press, 2024: 5694-5703.
[26]
GUO A J, MA J P, DIAN R W, et al. Surface defect detection competition with a bio-inspired vision sensor[J]. National Science Review, 2023, 10(22): 140-145.
2026年第47卷第1期
PDF下载
25
10
引用本文
BibTeX
文章信息
doi: 10.11996/JG.j.2095-302X.2026010120
  • 接收时间:2025-05-30
  • 首发时间:2026-05-19
  • 出版时间:2026-02-28
补充材料
相关文章
文章信息
作者
出版历史
  • 收稿日期:2025-05-30
  • 录用日期:2025-08-28
基金
National Key Research and Development Program of China under Grant(2022YFB3303804)
国家重点研究发展计划项目(2022YFB3303804)
National Natural Science Foundation of China under Grant(62473127)
国家自然科学基金(62473127)
作者信息
    河北工业大学人工智能与数据科学学院, 天津 300401

通讯作者:

陈海永,E-mail:
参考文献
分享链接
https://castjournals.cast.org.cn/joweb/txxb/CN/10.11996/JG.j.2095-302X.2026010120
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