基于双分支深度图卷积网络的指静脉识别研究

doi:10.19678/j.issn.1000-3428.0070022

摘要/Abstract

摘要：

基于图卷积神经网络(GCNN)的指静脉识别方法不仅可以解决传统指静脉识别方法识别率较低的问题, 还可以解决其计算量大的问题。针对目前指静脉图模型结构不稳定和匹配效率因模型增大而下降的问题, 采用SLIC(Simple Linear Iterative Clustering)超像素分割算法来构建加权图并改变GCNN提取加权图的图级特征。为了有效抓取图数据中的高阶特征并避免过平滑, 建立一种双分支多交互的深度图卷积网络(GCN), 旨在提升节点对高阶特征的掌握能力。首先根据节点特征对图结构进行调整; 然后结合原始和重构后的图结构, 构建了双分支网络架构以充分挖掘高阶特征; 最后设计一种通道信息互动机制, 以促进不同分支间的信息交流, 从而提高特征的多样性。实验结果显示, 在多个标准数据集上进行指静脉识别任务时, 该网络能减少单张图片识别时间, 提高识别效率, 并有效减轻过平滑现象, 相较于单分支的GCN, 在识别精度上平均取得了超过1.5百分点的性能提升。

关键词: 指静脉识别, 图像分割算法, 图卷积神经网络, 交叉熵函数, 通道信息交互

Abstract:

This study presents a finger-vein recognition method based on a Graph Convolutional Neural Network (GCNN) to overcome the low recognition rates and high computational cost of traditional methods. The study aims to address issues of graph structure instability and degraded matching efficiency in current finger-vein graph models. For this purpose, a Simple Linear Iterative Clustering (SLIC) superpixel segmentation algorithm is utilized to construct a weighted graph, based on which the GCNN is adapted for graph-level feature extraction. A dual-branch multi-interaction deep Graph Convolutional Network (GCN) is proposed to enhance the node's capability to represent higher-order features, to effectively capture these features in the graph data while avoiding oversmoothing. This study first adjusts the graph structure based on node features. Subsequently, by integrating the original and reconstructed graph structures, a dual-branch network architecture is built to fully explore higher-order features. Furthermore, a feature channel interaction mechanism is designed to facilitate information exchange between different branches, thereby improving feature diversity. Experimental results on multiple standard datasets for finger-vein recognition show that the proposed network reduces recognition time per image, improves efficiency, and effectively alleviates oversmoothing. Compared with the single-branch GCN, it improves recognition accuracy by an average of over 1.5 percentage points.

Key words: finger-vein recognition, image segmentation algorithm, Graph Convolutional Neural Network (GCNN), cross-entropy function, channel information interaction

程俊军, 王明文. 基于双分支深度图卷积网络的指静脉识别研究[J]. 计算机工程, 2026, 52(3): 152-160.

CHENG Junjun, WANG Mingwen. Research on Finger-vein Recognition Based on Deep Graph Convolutional Network with Dual-Branch[J]. Computer Engineering, 2026, 52(3): 152-160.

https://www.ecice06.com/CN/Y2026/V52/I3/152

图/表 12

图1 节点集构建

Fig.1 Construction of the node set

图2 边集构建

Fig.2 Construction of the edge set

图3 指静脉识别网络框架

Fig.3 Finger-vein recognition network framework

图4 部分指静脉数据集图像展示

Fig.4 Partial finger-vein dataset image presentation

图5 α、β不同时本文网络的识别准确率

Fig.5 Recognition accuracy of the proposed network when α and β are different

图6 不同超像素数下的指静脉分割示例

Fig.6 Examples of finger-vein segmentation with different number of hyperpixels

图7 不同指静脉识别网络的ROC对比

Fig.7 ROC comparison of different finger-vein recognition networks

参考文献 25

1	郑方, 艾斯卡尔·肉孜, 王仁宇, 等. 生物特征识别技术综述. 信息安全研究, 2016, 2(1): 12- 26.
	ZHENG F, Askar Rozi, WANG R Y, et al. Overview of biometric recognition technology. Journal of Information Security Research, 2016, 2(1): 12- 26.
2	杨金锋, 石滨萌. 手指双模态特征图像感兴趣区域稳定定位方法研究. 信号处理, 2019, 35(7): 1191- 1200.
	YANG J F, SHI B M. Reliably locating ROIs of dual-modal biometric images of human fingers. Journal of Signal Processing, 2019, 35(7): 1191- 1200.
3	MERAOUMIA A, KORICHI M, CHITROUB S, et al. Finger-knuckle-print identification based on histogram of oriented gradients and SVM classifier[C]//Proceedings of the 1st International Conference on New Technologies of Information and Communication (NTIC). Washington D.C., USA: IEEE Press, 2015: 1-6.
4	SHI J B, MALIK J. Normalized cuts and image segmentation. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2000, 22(8): 888- 905. doi: 10.1109/34.868688
5	陶志勇, 高亚静, 王萌, 等. 轻量级Transformer的双向交互近红外手指静脉图像识别. 兰州大学学报(自然科学版), 2024, 60(5): 621- 628.
	TAO Z Y, GAO Y J, WANG M, et al. Lightweight Transformer with bidirectional interaction for near-infrared finger vein image recognition. Journal of Lanzhou University (Natural Sciences), 2024, 60(5): 621- 628.
6	叶子云, 杨金锋. 一种基于加权图模型的手指静脉识别方法. 山东大学学报(工学版), 2018, 48(3): 103- 109.
	YE Z Y, YANG J F. A finger-vein recognition method based on weighted graph model. Journal of Shandong University (Engineering Science), 2018, 48(3): 103- 109.
7	DAS R, PICIUCCO E, MAIORANA E, et al. Convolutional neural network for finger-vein-based biometric identification. IEEE Transactions on Information Forensics and Security, 2019, 14(2): 360- 373. doi: 10.1109/TIFS.2018.2850320
8	BRUNA J. Spectral networks and locally connected networks on graphs[EB/OL]. [2024-05-04]. https://arxiv.org/abs/1312.6203.
9	BRUNA J, ZAREMBA W, SZLAM A, et al. Spectral networks and locally connected networks on graphs[EB/OL]. [2024-05-04]. https://arxiv.org/abs/1312.6203.
10	LI Q M, HAN Z C, WU X M. Deeper insights into graph convolutional networks for semi-supervised learning[C]//Proceedings of the AAAI Conference on Artificial Intelligence. Palo Alto, USA: AAAI Press, 2018: 3538-3545.
11	HE K M, ZHANG X Y, REN S Q, et al. Deep residual learning for image recognition[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Washington D.C., USA: IEEE Press, 2016: 770-778.
12	KIPF T N, WELLING M. Semi-supervised classification with graph convolutional networks[EB/OL]. [2024-05-04]. https://openreview.net/pdf?id=SJU4ayYgl.
13	李冉, 苏志刚, 张海刚, 等. 改进GCNs在指静脉特征表达中的应用. 信号处理, 2020, 36(4): 550- 561.
	LI R, SU Z G, ZHANG H G, et al. Application of improved GCNs in feature representation of finger-vein. Journal of Signal Processing, 2020, 36(4): 550- 561.
14	江波, 汤进, 罗斌. 计算机视觉中的图匹配方法研究综述. 安徽大学学报(自然科学版), 2017, 41(1): 29- 36.
	JIANG B, TANG J, LUO B. A survey on graph matching algorithms in computer vision. Journal of Anhui University (Natural Science Edition), 2017, 41(1): 29- 36.
15	ZHANG S, TONG H H, XU J J, et al. Graph convolutional networks: a comprehensive review. Computational Social Networks, 2019, 6(1): 11. doi: 10.1186/s40649-019-0069-y
16	CHEN M, WEI Z W, HUANG ZF, et al. Simple and Deep Graph Convolutional Networks[C]//Proceedigns of the 37th International Conference on Machine Learning. New York, USA: ACM Press, 2020: 1725-1735.
17	楼嘉琪, 叶海良, 杨冰, 等. 双分支多交互的深度图卷积网络. 模式识别与人工智能, 2022, 35(8): 754- 763.
	LOU J Q, YE H L, YANG B, et al. Deep graph convolutional network with dual-branch and multi-interaction. Pattern Recognition and Artificial Intelligence, 2022, 35(8): 754- 763.
18	PASZKE A, GROSS S, MASSA F, et al. PyTorch: an imperative style, high-performance deep learning library[C]//Proceedigns of the 33rd International Conference on Neural Information Processing Systems. Cambridge, USA: MIT Press, 2019: 8026-8037.
19	KINGMA D P, BA J L. Adam: a method for stochastic optimization[EB/OL]. [2024-05-04]. https://arxiv.org/pdf/1412.6980.pdf.
20	YIN Y L, LIU L L, SUN X W. SDUMLA-HMT: a multimodal biometric database[M]//BERTINO E, GAO W, STEFFEN B. Biometric recognition. Berlin, Germany: Springer, 2011: 260-268.
21	RONG Y, HUANG W B, XU T Y, et al. DropEdge: towards deep graph convolutional networks on node classification[EB/OL]. [2024-05-04]. https://openreview.net/pdf?id=Hkx1qkrKPr.
22	ASAARI M S M, SUANDI S A, ROSDI B A. Fusion of band limited phase only correlation and width centroid contour distance for finger based biometrics. Expert Systems with Applications, 2014, 41(7): 3367- 3382. doi: 10.1016/j.eswa.2013.11.033
23	KUMAR A, ZHOU Y B. Human identification using finger images. IEEE Transactions on Image Processing, 2012, 21(4): 2228- 2244. doi: 10.1109/TIP.2011.2171697
24	KRIZHEVSKY A, SUTSKEVER I, HINTON G E. ImageNet classification with deep convolutional neural networks. Communications of the ACM, 2015, 60, 84- 90.
25	孙旭, 胡伟, 李瑞瑞. 基于Lighten CNN的自动人脸分类模型的优化方法. 计算机应用, 2018, 38(S1): 32-35, 82.
	SUN X, HU W, LI R R. Optimization method for automatic facial classification model based on Lighten CNN. Journal of Computer Applications, 2018, 38(S1): 32-35, 82.

[1]	蒋杰平, 王明文. 基于时空置换注意力机制的残差行为识别模型[J]. 计算机工程, 2025, 51(4): 119-128.
[2]	胡书林, 张华军, 邓小涛, 王征华. 结合依存图卷积的中文文本相似度计算研究[J]. 计算机工程, 2025, 51(3): 76-85.
[3]	张洋, 刘畅, 李少青. 基于可控制性度量的图神经网络门级硬件木马检测方法[J]. 计算机工程, 2024, 50(7): 164-173.
[4]	李政学, 李枝名, 彭德中, 陈杰. 基于特征对比学习和图卷积的社交网络用户分类[J]. 计算机工程, 2024, 50(4): 258-266.
[5]	代祖华, 刘园园, 狄世龙. 语义增强的图神经网络方面级文本情感分析[J]. 计算机工程, 2023, 49(6): 71-80.
[6]	衡红军, 苗菁. 语义与句法信息加强的二元标记实体关系联合抽取[J]. 计算机工程, 2023, 49(4): 77-84.
[7]	徐康, 李霏, 姬东鸿. 结合依存图卷积与文本片段搜索的方面情感三元组抽取[J]. 计算机工程, 2023, 49(4): 61-67.
[8]	陈治旭, 靳雁霞, 芦烨, 杨晶, 刘亚变, 史志儒. 基于子图卷积神经网络的多精度服装建模方法[J]. 计算机工程, 2023, 49(4): 174-181.
[9]	翟社平, 张宇航, 柏晓夏. 融合实体邻域信息的知识图谱嵌入负采样方法[J]. 计算机工程, 2023, 49(3): 95-104.
[10]	HAYTHEM Alhag, 杨军. 基于隐式描述符的三维模型对应关系计算[J]. 计算机工程, 2022, 48(5): 229-234.
[11]	唐鑫, 徐彦彦, 潘少明. 基于图卷积神经网络的智能路由算法[J]. 计算机工程, 2022, 48(3): 38-45.
[12]	施荣华, 金鑫, 胡超. 基于图注意力网络的方面级别文本情感分析[J]. 计算机工程, 2022, 48(2): 34-39.
[13]	孙亚茹, 杨莹, 王永剑. 基于知信图卷积神经网络的开放域知识图谱自动构建模型[J]. 计算机工程, 2022, 48(10): 116-122.
[14]	段大高, 白宸宇, 韩忠明, 熊海涛. 基于多传递影响力的社交媒体谣言检测方法[J]. 计算机工程, 2022, 48(10): 138-145,157.
[15]	赖河蒗, 李玲俐, 胡婉玲, 颜学明. 一种基于层次化R-GCN的会话情绪识别方法[J]. 计算机工程, 2022, 48(1): 85-92.

选择文件类型/文献管理软件名称

选择包含的内容