融合多尺度特征的多监督人脸活体检测算法

doi:10.19678/j.issn.1000-3428.0063424

摘要/Abstract

摘要： 目前的多数活体检测算法忽略了特征挖掘，导致判别性信息提取不足。提出一种融合梯度纹理和群感受野的活体检测算法。使用中心差分卷积计算感受野周围点与中心点的差值，提取图像的梯度纹理特征，设计群感受野模块，采用不同尺寸的卷积核结合空洞卷积组成多分支结构，在使用较少参数量的情况下获得更大的感受野和多尺度特征，并将两种特征融合输入到残差结构中。此外，在使用深度图进行监督的同时，增加二值掩模进行辅助监督，使得网络将学习的中心放到人脸部位，进一步提升模型的鲁棒性。在此基础上，综合深度图生成器和掩模生成器的输出结果，计算预测得分，实现端到端的活体检测。实验结果表明，该算法在公开数据集OULU-NPU 4个协议上的平均分类错误率分别为0.9%、1.9%、1.6%±2.0%和2.7%±1.8%，在数据集CASIA-MFSD和Replay-Attack上可实现无误差活体检测，并且模型参数量仅为1.1 MB，与Auxiliary和STASN等活体检测算法相比，检测精度更高，具有更好的鲁棒性。

关键词: 卷积神经网络, 人脸活体检测, 纹理特征, 感受野, 多监督

Abstract: To address the problem that most face liveness detection algorithms ignore feature mining and cannot extract enough discriminative information, this paper proposes a novel algorithm that combines gradient texture and group receptive fields. On the one hand, central difference convolution is used to calculate the difference between the points around the receptive field and the center point.In this way, the gradient texture feature of the image can be fully extracted.On the other hand, a group receptive field module is proposed to obtain larger receptive fields and multi-scale features with fewer parameters.It uses convolution kernels of different sizes combined with dilated convolution to form multiple small branch structures.Then, two types of feature are combined and input into the residual structure.In addition, while a depth map is used for the main supervision, binary mask supervision is added to assist with supervision. As a result, the network focuses on the face part, further improving the robustness of the model.Finally, the outputs of the depth map generator and mask generator are combined to calculate the prediction score to realize end-to-end detection.Experimental results show that the average classification error rates of the proposed algorithm on the four protocols of the OULU-NPU database are 0.9%, 1.9%, 1.6%±2.0%, and 2.7%±1.8%.Moreover, error-free liveness detection was achieved on the CASIA-MFSD and Replay-Attack datasets.In addition, the size of the model is only 1.1 MB.It has higher detection accuracy and better robustness than auxiliary and the Spatio-Temporal Anti-Spoof Network (STASN) liveness detection algorithms.

Key words: Convolutional Neural Network(CNN), face liveness detection, texture feature, receptive field, multi-supervision

中图分类号:

TP391.4

陈苏阳, 宋晓宁. 融合多尺度特征的多监督人脸活体检测算法[J]. 计算机工程, 2022, 48(12): 79-85,94.

CHEN Suyang, SONG Xiaoning. Multi-supervision Face Liveness Detection Algorithm Fused with Multi-scale Feature[J]. Computer Engineering, 2022, 48(12): 79-85,94.

http://www.ecice06.com/CN/Y2022/V48/I12/79

图/表 9

20230112182725

20230112182728

20230112182732

20230112182736

20230112182739

20230112182743

20230112182746

20230112182750

20230112182753

参考文献

[1] 马钰锡, 谭励, 董旭, 等.面向VTM的交互式活体检测算法[J].计算机工程, 2019, 45(3):256-261. MA Y X, TAN L, DONG X, et al.Interactive liveness detection algorithm for VTM[J].Computer Engineering, 2019, 45(3):256-261.(in Chinese)
[2] 胡斐, 文畅, 谢凯, 等.基于微调策略的多线索融合人脸活体检测[J].计算机工程, 2019, 45(5):256-260. HU F, WEN C, XIE K, et al.Multi-cue fusion face liveness detection based on fine-tuning strategy[J].Computer Engineering, 2019, 45(5):256-260.(in Chinese)
[3] LIU Y J, JOURABLOO A, LIU X M.Learning deep models for face anti-spoofing:binary or auxiliary supervision[C]//Proceedings of 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition.Washington D.C., USA:IEEE Press, 2018:389-398.
[4] JOURABLOO A, LIU Y J, LIU X M.Face de-spoofing:anti-spoofing via noise modeling[C]//Proceedings of European Conference on Computer Vision.Berlin, Germany:Springer, 2018:290-306.
[5] YANG X, LUO W H, BAO L C, et al.Face anti-spoofing:model matters, so does data[C]//Proceedings of 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition.Washington D.C., USA:IEEE Press, 2019:3502-3511.
[6] 林云, 孙晓刚, 姜尧岗, 等.基于语义分割的活体检测算法[J].吉林大学学报(工学版), 2020, 50(3):1040-1046. LIN Y, SUN X G, JIANG X G, et al.Face anti-spoofing algorithm based on semantic segmentation[J].Journal of Jilin University (Engineering and Technology Edition), 2020, 50(3):1040-1046.(in Chinese)
[7] YU Z T, ZHAO C X, WANG Z Z, et al.Searching central difference convolutional networks for face anti-spoofing[C]//Proceedings of 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition.Washington D.C., USA:IEEE Press, 2020:5294-5304.
[8] SZEGEDY C, LIU W, JIA Y, et al.Going deeper with convolutions[C]//Proceedings of IEEE Conference on Computer Vision and Pattern Recognition.Washington D.C., USA:IEEE Press, 2015:1-9.
[9] BOULKENAFET Z, KOMULAINEN J, LI L, et al.Oulu-npu:a mobile face presentation attack database with real-world variations[C]//Proceedings of the 12th IEEE International Conference on Automatic Face & Gesture Recognition.Washington D.C., USA:IEEE Press, 2017:612-618.
[10] ZHANG Z W, YAN J J, LIU S F, et al.A face antispoofing database with diverse attacks[C]//Proceedings of the 5th IAPR International Conference on Biometrics.Washington D.C., USA:IEEE Press, 2012:26-31.
[11] CHINGOVSKA I, ANJOS A, MARCEL S.On the effectiveness of local binary patterns in face anti-spoofing[C]//Proceedings of International Conference on Biometrics Special Interest Group.Washington D.C., USA:IEEE Press, 2012:1-7.
[12] BENGIO S, MARIETHOZ J.A statistical significance test for person authentication[C]//Proceedings of IEEE Conference on Speaker and Language Recognition.Washington D.C., USA:IEEE Press, 2004:158-167.
[13] ZHANG K P, ZHANG Z P, LI Z F, et al.Joint face detection and alignment using multitask cascaded convolutional networks[J].IEEE Signal Processing Letters, 2016, 23(10):1499-1503.
[14] KINGMA D P, BA J.Adam:a method for stochastic optimization[EB/OL].[2021-11-01].https://arxiv.org/abs/1412.6980.
[15] CAI R Z, LI H L, WANG S Q, et al.DRL-FAS:a novel framework based on deep reinforcement learning for face anti-spoofing[J].IEEE Transactions on Information Forensics and Security, 2021, 16(4):937-951.
[16] KIM T, KIM Y, KIM I, et al.BASN:enriching feature representation using bipartite auxiliary supervisions for face anti-spoofing[C]//Proceedings of 2019 IEEE/CVF International Conference on Computer Vision.Washington D.C., USA:IEEE Press, 2019:494-503.
[17] ZHANG K Y, YAO T P, ZHANG J, et al.Face anti-spoofing via disentangled representation learning[C]//Proceedings of European Conference on Computer Vision.Berlin, Germany:Springer, 2020:641-657.
[18] LI L, FENG X Y, BOULKENAFET Z, et al.An original face anti-spoofing approach using partial convolutional neural network[C]//Proceedings of the 6th International Conference on Image Processing Theory, Tools and Applications.Washington D.C., USA:IEEE Press, 2020:1-6.
[19] YANG J W, LEI Z, LI S Z.Learn convolutional neural network for face anti-spoofing[EB/OL].[2021-11-01].https://arxiv.org/abs/1408.5601.
[20] XU Y W, WU L F, JIAN M, et al.Identity-constrained noise modeling with metric learning for face anti-spoofing[J].Neurocomputing, 2021, 434:149-164.
[21] CHEN H N, HU G S, LEI Z, et al.Attention-based two-stream convolutional networks for face spoofing detection[J].IEEE Transactions on Information Forensics and Security, 2020, 15(5):578-593.
[22] BOULKENAFET Z, KOMULAINEN J, HADID A.Face antispoofing using speeded-up robust features and fisher vector encoding[J].IEEE Signal Processing Letters, 2017, 24(2):141-145.
[23] AGARWAL A, SINGH R, VATSA M.Face anti-spoofing using Haralick features[C]//Proceedings of the 8th IEEE International Conference on Biometrics Theory, Applications and Systems.Washington D.C., USA:IEEE Press, 2016:1-6.
[24] BHARADWAJ S, DHAMECHA T I, VATSA M, et al.Computationally efficient face spoofing detection with motion magnification[C]//Proceedings of 2013 IEEE Conference on Computer Vision and Pattern Recognition.Washington D.C., USA:IEEE Press, 2013:105-110.
[25] DE FREITAS PEREIRA T, ANJOS A, DE MARTINO J M, et al.Can face anti-spoofing countermeasures work in a real world scenario?[C]//Proceedings of 2013 International Conference on Biometrics.Washington D.C., USA:IEEE Press, 2013:1-8.
[26] PINTO A, PEDRINI H, SCHWARTZ W R, et al.Face spoofing detection through visual codebooks of spectral temporal cubes[J].IEEE Transactions on Image Processing, 2015, 24(12):4726-4740.
[27] VARETO R H, DINIZ M A, SCHWARTZ W R.Face spoofing detection on low-power devices using embeddings with spatial and frequency-based descriptors[C]//Proceedings of International Conference on Image Analysis, Computer Vision, and Applications.Berlin, Germany:Springer, 2019:187-197.
[28] SANDLER M, HOWARD A, ZHU M L, et al.MobileNetV2:inverted residuals and linear bottlenecks[C]//Proceedings of 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition.Washington D.C., USA:IEEE Press, 2018:4510-4520.

选择文件类型/文献管理软件名称

选择包含的内容