基于局部感知图重构的点云异常检测网络

doi:10.16451/j.cnki.issn1003-6059.202511001

摘要
图/表
参考文献(0)
相关文章 (15)

全文: PDF (2302 KB) HTML (1 KB)
输出: BibTeX | EndNote (RIS)

摘要

点云异常检测旨在从整体数据分布中识别缺陷样本,并进一步定位其在空间中偏离预期模式的异常区域.针对现有全局匹配策略难以有效捕捉集中于局部细微几何结构异常的问题,文中提出基于局部感知图重构的点云异常检测网络.首先,将点云建模为图结构,通过局部敏感的边卷积操作挖掘点云的局部结构特征,提升对局部细微异常的识别能力.然后,基于子图结构对齐策略设计局部对齐重建损失,从结构层面放大正常样本与异常样本间的差异性.此外,引入局部异常生成策略,构建异常-正常样本对,约束网络学习异常样本向正常样本的模式映射.最后,采用局部匹配检测算法计算异常样本与期望正常样本之间的差异,实现对点云异常区域的检测.实验表明,文中网络显著增强对局部细节的感知能力,在多个类别的点云异常检测任务中均取得较优性能.

	服务

	把本文推荐给朋友
	加入我的书架
	加入引用管理器
	E-mail Alert
	RSS
	作者相关文章
	高程玲
	黄昌勤
	郑忠龙
	蒋云良

关键词 ：图神经网络, 无监督学习, 异常检测, 点云

Abstract：

Point cloud anomaly detection aims to identify defective samples from the overall data distribution and further locate the abnormal regions deviating from the expected pattern in space. Existing global matching strategies struggle to effectively capture anomalies concentrated in local subtle geometric structures. To address this issue, a point cloud anomaly detection network based on local perception graph reconstruction is proposed. First, the point cloud is modeled as a graph structure, and local sensitive edge convolution operations are utilized to mine local structural features to enhance the identification ability for local subtle anomalies. Second, a local alignment reconstruction loss is designed based on a subgraph structure alignment strategy to amplify the differences between normal and abnormal samples at the structural level. Furthermore, a local anomaly simulation strategy is introduced to construct an anomaly-normal sample pair. Through this strategy, the model is constrained to learn the pattern mapping from abnormal samples to normal samples. Finally, a local matching algorithm is applied to calculate the differences between abnormal samples and expected normal samples to achieve the detection of point cloud abnormal regions. Experimental results show that the proposed method significantly enhances the perception ability for local details and achieves excellent performance on multiple categories of point cloud anomaly detection tasks.

Key words： Graph Neural Network Unsupervised Learning Anomaly Detection Point Cloud

收稿日期: 2025-10-20

ZTFLH:

TP391

基金资助:

科技创新2030——“新一代人工智能”重大项目(No.2022Z D0117104)、国家自然科学基金项目(No.62337001)资助

通讯作者: 黄昌勤,博士,教授,主要研究方向为机器学习、教育大数据、智能教育.E-mail:cqhuang@zju.edu.cn.

作者简介: 高程玲,博士研究生,主要研究方向为图表示学习、图异常检测.E-mail:chl_gao@zjnu.edu.cn.
郑忠龙,博士,教授,主要研究方向为机器学习、计算机视觉、智联网与区块链.E-mail:zhonglong@zjnu.edu.cn.
蒋云良,博士,教授,主要研究方向为智能信息处理、地理信息系统.E-mail:jyl2022@zjnu.cn.

引用本文:

高程玲, 黄昌勤, 郑忠龙, 蒋云良. 基于局部感知图重构的点云异常检测网络[J]. 模式识别与人工智能, 2025, 38(11): 961-973. GAO Chengling, HUANG Changqin, ZHENG Zhonglong, JIANG Yunliang. Point Cloud Anomaly Detection Network Based on Local Perception Graph Reconstruction. Pattern Recognition and Artificial Intelligence, 2025, 38(11): 961-973.

链接本文:

http://manu46.magtech.com.cn/Jweb_prai/CN/10.16451/j.cnki.issn1003-6059.202511001 或 http://manu46.magtech.com.cn/Jweb_prai/CN/Y2025/V38/I11/961

[1] QI C R, SU H, MO K, et al. PointNet: Deep Learning on Point Sets for 3D Classification and Segmentation // Proc of the IEEE Conference on Computer Vision and Pattern Recognition. Washington, USA: IEEE, 2017: 77-85.
[2] GUO Y L, WANG H Y, HU Q Y, et al. Deep Learning for 3D Point Clouds: A Survey. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2021, 43(12): 4338-4364.
[3] MA X X, WU J, XUE S, et al. A Comprehensive Survey on Graph Anomaly Detection with Deep Learning. IEEE Transactions on Knowledge and Data Engineering, 2023, 35(12): 12012-12038.
[4] BERGMANN P, JIN X, SATTLEGGER D, et al. The MVTec 3D-AD Dataset for Unsupervised 3D Anomaly Detection and Localization[C/OL].[2025-09-07]. https://arxiv.org/pdf/2112.09045v1.
[5] BERGMANN P, SATTLEGGER D.Anomaly Detection in 3D Point Clouds Using Deep Geometric Descriptors // Proc of the IEEE/CVF Winter Conference on Applications of Computer Vision. Washington, USA: IEEE, 2023: 2612-2622.
[6] ROTH K, PEMULA L, ZEPEDA J, et al. Towards Total Recall in Industrial Anomaly Detection // Proc of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington, USA: IEEE, 2022: 14298-14308.
[7] HE H Y, ZHANG J N, CHEN H X, et al. A Diffusion-Based Framework for Multi-class Anomaly Detection. Proc of the AAAI Conference on Artificial Intelligence, 2024, 38(8): 8472-8480.
[8] HORWITZ E, HOSHEN Y.Back to the Feature: Classical 3D Features Are(Almost) All You Need for 3D Anomaly Detection // Proc of the IEEE/CVF Conference on Computer Vision and Pattern Re-cognition. Washington, USA: IEEE, 2023: 2968-2977.
[9] WANG Y, PENG J L, ZHANG J N, et al. Multimodal Industrial Anomaly Detection via Hybrid Fusion // Proc of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington, USA: IEEE, 2023: 8032-8041.
[10] CAO Y K, XU X H, SHEN W M.Complementary Pseudo Multimodal Feature for Point Cloud Anomaly Detection. Pattern Recognition, 2024, 156. DOI: 10.1016/j.patcog.2024.110761.
[11] RUDOLPH M, WEHRBEIN T, ROSENHAHN B, et al. Asymme-tric Student-Teacher Networks for Industrial Anomaly Detection // Proc of the IEEE/CVF Winter Conference on Applications of Computer Vision. Washington, USA: IEEE, 2023: 2591-2601.
[12] LIU J Q, XIE G Y, CHEN R T, et al. Real3D-AD: A Dataset of Point Cloud Anomaly Detection // Proc of the 37th International Conference on Neural Information Processing Systems. Cambridge, USA: MIT Press, 2023: 30402-30415.
[13] HOLLES R C, FISCHLER M A.A RANSAC-Based Approach to Model Fitting and Its Application to Finding Cylinders in Range Data // Proc of the 7th International Joint Conference on Artificial Intelligence. San Francisco, USA: IJCAI, 1981: 637-643.
[14] HUANG C Q, XU Q W, WANG Y F, et al. Self-Supervised Mas-king for Unsupervised Anomaly Detection and Localization. IEEE Transactions on Multimedia, 2022, 25: 4426-4438.
[15] PIRNAY J, CHAI K.Inpainting Transformer for Anomaly Detection // Proc of the 21st International Conference on Image Analysis and Processing. Berlin, Germany: Springer, 2022: 394-406.
[16] YAN X D, ZHANG H D, XU X M, et al. Learning Semantic Context from Normal Samples for Unsupervised Anomaly Detection. Proc of the AAAI Conference on Artificial Intelligence, 2021, 35(4): 3110-3118.
[17] LI W X, XU X H, GU Y, et al. Towards Scalable 3D Anomaly Detection and Localization: A Benchmark via 3D Anomaly Synthesis and a Self-Supervised Learning Network // Proc of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington, USA: IEEE, 2024: 22207-22216.
[18] ZHOU Z Y, WANG L, FANG N Y, et al. R3D-AD: Reconstruction via Diffusion for 3D Anomaly Detection // Proc of the 18th European Conference on Computer Vision. Berlin, Germany: Springer, 2024: 91-107.
[19] DOHMATOB E, FENG Y Z, KEMPE J. Model Collapse Demystified: The Case of Regression // Proc of the 38th International Conference on Neural Information Processing Systems. Cambridge, USA: MIT Press, 2024: 46979-47013.
[20] LI W Q, ZHENG B Z, XU X H, et al. Multi-sensor Object Ano-maly Detection: Unifying Appearance, Geometry, and Internal Properties // Proc of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington, USA: IEEE, 2025: 9984-9993.
[21] ZHOU Q H, YAN J T, HE S B, et al. PointAD: Comprehending 3D Anomalies from Points and Pixels for Zero-Shot 3D Anomaly Detection // Proc of the 38th International Conference on Neural Information Processing Systems. Cambridge, USA: MIT Press, 2024: 84866-84896.
[22] SHI Y K, ZHAN X, LI Y Q, et al. Cycle-CFM: An Unsupervised Framework for Robust Multimodal Anomaly Detection in Industrial Settings. Expert Systems with Applications, 2026, 298(C). DOI: 10.1016/j.eswa.2025.129745.
[23] ZHU W B, WANG L D, ZHOU Z Q, et al. Real-IAD D³: A Real-World 2D/Pseudo-3D/3D Dataset for Industrial Anomaly Detection // Proc of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington, USA: IEEE, 2025: 15214-15223.
[24] CHEN C F, QIAN S S, FANG Q, et al. HAPGN: Hierarchical Attentive Pooling Graph Network for Point Cloud Segmentation. IEEE Transactions on Multimedia, 2020, 23: 2335-2346.
[25] SIMONOVSKY M, KOMODAKIS N.Dynamic Edge-Conditioned Filters in Convolutional Neural Networks on Graphs // Proc of the IEEE Conference on Computer Vision and Pattern Recognition. Washington, USA: IEEE, 2017: 29-38.
[26] WANG Y, SUN Y B, LIU Z W, et al. Dynamic Graph CNN for Learning on Point Clouds. ACM Transactions on Graphics , 2019, 38(5). DOI: 10.1145/3326362.
[27] QI C R, YI L, SU H, et al. PointNet++: Deep Hierarchical Feature Learning on Point Sets in a Metric Space // Proc of the 31st International Conference on Neural Information Processing Systems. Cambridge, USA: MIT Press, 2017: 5105-5114.
[28] LIN Z H, HUANG S Y, WANG Y F.Learning of 3D Graph Convolution Networks for Point Cloud Analysis. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2022, 44(8): 4212-4224.
[29] HU T, ZHANG J N, YI R, et al. AnomalyDiffusion: Few-Shot Anomaly Image Generation with Diffusion Model. Proc of the AAAI Conference on Artificial Intelligence, 2024, 38(8): 8526-8534.
[30] LI C L, SOHN K, YOON J, et al. CutPaste: Self-Supervised Lear-ning for Anomaly Detection and Localization // Proc of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington, USA: IEEE, 2021: 9659-9669.
[31] YE J N, ZHAO W G, YANG X, et al. PO3AD: Predicting Point Offsets Toward Better 3D Point Cloud Anomaly Detection // Proc of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington, USA: IEEE, 2025: 1353-1362.
[32] CHANG A X, FUNKHOUSER T, GUIBAS L, et al. ShapeNet: An Information-Rich 3D Model Repository[C/OL].[2025-09-07]. https://arxiv.org/pdf/1512.03012.
[33] PASZKE A, GROSS S, MASSA F, et al. PyTorch: An Imperative Style, High-Performance Deep Learning Library // Proc of the 33th International Conference on Neural Information Processing Systems. Cambridge, USA: MIT Press, 2019: 8026-8037.
[34] KINGMA D P, BA J L. Adam: A Method for Stochastic Optimiza-tion[C/OL]. [2025-09-07].https://arxiv.org/pdf/1412.6980.
[35] ZHU H Z, XIE G Y, HOU C B, et al. Towards High-Resolution 3D Anomaly Detection via Group-Level Feature Contrastive Lear-ning // Proc of the 32nd ACM International Conference on Multimedia. New York, USA: ACM, 2024: 4680-4689.
[36] WANG J X, NIU Y C, HUANG B Q.Fusion-Restoration Model for Industrial Multimodal Anomaly Detection. Neurocomputing, 2025, 637. DOI: 10.1016/j.neucom.2025.130073.
[37] RUSU R B, BLODOW N, BEETZ M.Fast Point Feature Histograms(FPFH) for 3D Registration // Proc of the IEEE Internatio-nal Conference on Robotics and Automation. Washington, USA: IEEE, 2009: 3212-3217.
[38] PANG Y J, WANG W X, TAY F E H, et al. Masked Autoenco-ders for Point Cloud Self-Supervised Learning // Proc of the 17th European Conference on Computer Vision. Berlin, Germany: Springer, 2022: 604-621.