基于U-Net的掌纹图像增强与ROI提取

陆展鸿; 单鲁斌; 苏立循; 焦雨欣; 王家骅; 王海霞

doi:10.13700/j.bh.1001-5965.2020.0309

基于U-Net的掌纹图像增强与ROI提取

doi: 10.13700/j.bh.1001-5965.2020.0309

1.
杭州海康威视数字技术股份有限公司, 杭州 310051
2.
浙江工业大学信息工程学院, 杭州 310023
3.
浙江工业大学计算机科学与技术学院, 杭州 310023

基金项目:

浙江省重点研发计划 2019C01007

详细信息

作者简介:
陆展鸿男, 学士。主要研究方向:图像识别技术

王海霞女, 博士, 副教授, 博士生导师。主要研究方向:精密测量与图像处理

通讯作者:
王海霞, E-mail:hxwang@zjut.edu.cn

中图分类号: TP391.41
计量
- 文章访问数: 1574
- HTML全文浏览量: 367
- PDF下载量: 222
- 被引次数: 0
出版历程
- 收稿日期: 2020-07-01
- 录用日期: 2020-07-22
- 网络出版日期: 2020-09-20

Palmprint enhancement and ROI extraction based on U-Net

1.
Hangzhou Hikvision Digital Technology Co., Ltd., Hangzhou 310051, China
2.
College of Information Engineering, Zhejiang University of Technology, Hangzhou 310023, China
3.
College of Computer Science and Technology, Zhejiang University of Technology, Hangzhou 310023, China

Funds:

Zhejiang Povincial Key Research and Development Program of China 2019C01007

More Information

Corresponding author: WANG Haixia, E-mail:hxwang@zjut.edu.cn

摘要

摘要:
掌纹因稳定性、唯一性、难复制性及易获取等特点，是极具应用潜力的生物识别特征。针对掌纹识别中，获取掌纹感兴趣区域（ROI）与增强掌纹信息普遍存在的时间成本大，方法之间有依赖关系等问题，提出了一种基于U-Net神经网络结构的掌纹图像预处理方法。通过香港理工大学掌纹库进行实验对比，结果表明，所提方法能消除预处理方法之间的相互影响，实现对掌纹图像的去噪与增强处理，并能快速、高精度提取感兴趣区域。
- U-Net /
- 感兴趣区域(ROI)增强 /
- 深度学习 /
- 感兴趣区域(ROI)提取 /
- 掌纹
Abstract:
Palmprint is a biometric feature with great application potential due to its stability, uniqueness, hardness in copying, and easy access.In the palmprint recognition, palmprint Region of Interest(ROI) acquisition and palmprint enhancement usually have the problems of high time cost and high dependency between methods.This paper proposes a palmprint preprocessing method based on U-Net neural network structure.The experiments are carried out on palmprint database from Hong Kong Polytechnic University.The results show that the proposed method can eliminate the mutual influence between the preprocessing methods.Both denoising and enhancement of palmprint images are realized, and the region of interest can be extracted quickly and accurately; plamprint
- U-Net /
- Region of Interest (ROI) enhancement /
- deep learning /
- Region of Interest (ROI) extraction /
- plamprint

HTML全文

图 1 本文方法流程图

Figure 1. Proposed method flow chart

下载: 全尺寸图片幻灯片

图 2 预处理步骤

Figure 2. Pre-processingsteps

下载: 全尺寸图片幻灯片

图 3 U-Net神经网络结构

Figure 3. U-Net neural network structure

下载: 全尺寸图片幻灯片

图 4 激活函数图形

Figure 4. Graph of activation functions

下载: 全尺寸图片幻灯片

图 5 关键点及掌纹对应谷点

Figure 5. Key points and correspondingly valley points on plamprint

下载: 全尺寸图片幻灯片

图 6 ROI提取处理

Figure 6. ROI extraction processing diagram

下载: 全尺寸图片幻灯片

图 7 不同掌纹图像的增强结果

Figure 7. Enhancement results of different palmprint images

下载: 全尺寸图片幻灯片

图 8 BM3D去噪+Gabor小波增强后为真值的掌纹增强对比

Figure 8. Comparison of palmprint enhancement after BM3D denoising +Gabor wavelet enhancement as truth value

下载: 全尺寸图片幻灯片

图 9 OBLNM增强后为真值的掌纹增强对比

Figure 9. Comparison of palmprint enhancement after OBLNM enhancement as truth value

下载: 全尺寸图片幻灯片

表 1 U-Net神经网络模型参数

Table 1. U-Net neural network model parameters

层类型	输入大小	滤波器大小	重复次数
卷积模块	1×384×280	3×3	1
卷积模块	64×384×280	3×3	1
池化层	64×384×280	2×2	1
卷积模块	64×192×140	3×3	2
池化层	128×192×140	2×2	1
卷积模块	128×96×70	3×3	2
池化层	256×96×70	2×2	1
卷积模块	256×48×35	3×3	2
上采样层	512×48×35	3×3	1
卷积模块	512×96×70	3×3	2
上采样层	256×96×70	3×3	1
卷积模块	256×192×140	3×3	2
上采样层	128×192×140	3×3	1
卷积模块	128×384×280	3×3	1
卷积模块	64×384×280	3×3	1
卷积模块	64×384×280	1×1	1

下载: 导出CSV

表 2 掌纹增强效果比较结果

Table 2. Comparison results of palmprint enhancement

方法	BM3D去噪+Gabor小波增强后为真值			OBLNM增强后为真值
方法	PSNR	Entropy	AMBE	PSNR	Entropy	AMBE
BM3D去噪+Gabor小波增强	24.727 6	6.435 6	3.163 0	24.049 3	6.531 6	4.168 1
OBLNM	23.171 6	6.916 0	12.419 8	26.334 8	6.877 0	5.930 9
RKTFILT	23.499 4	6.820 0	15.239 3	28.022 2	6.753 7	6.806 8
SRAD	26.925 8	6.005 4	7.346 4	27.037 8	6.154 6	5.653 2
本文方法	27.120 8	7.247 6	3.165 3	27.916 9	6.920 7	3.210 1

下载: 导出CSV

表 3 掌纹感兴趣区域重合度分析

Table 3. Analysis of coincidence degree of palmprint ROI

方法	面积重合率/%
方法1	91.18
方法2	88.00
本文方法	94.00

下载: 导出CSV

表 4 掌纹分割图像相似度对比

Table 4. Similarity comparison of palmprint segmentation images

方法	Cosin相似度/%	哈希相似度/%
方法1	99.32	73.17
方法2	99.13	71.61
本文方法	99.49	76.47

下载: 导出CSV

表 5 每张掌纹图像处理时间

Table 5. Processing time of each palmprint image

方法	时间/ms
方法	增强	分割	增强和分割
方法1	1 355.73	32.23	1 387.96
方法2	1 355.73	14.49	1 370.22
本文方法			90.49

下载: 导出CSV

参考文献(39)

[1]	KUMAR A, KWONG C.Towards contactless, low-cost and accurate 3D fingerprint identification[J].IEEE Transactions on Pattern Analysis and Machine Intelligence, 2015, 37(3):681-696. doi: 10.1109/TPAMI.2014.2339818
[2]	LU J W, LIONG V E, ZHOU X Z, et al.Learning compact binary face descriptor for face recognition[J].IEEE Transactions on Pattern Analysis and Machine Intelligence, 2015, 37(10):2041-2056. doi: 10.1109/TPAMI.2015.2408359
[3]	NGUYEN K, FOOKES C, JILLELA R, et al.Long range iris recognition:A survey[J].Pattern Recognition, 2017, 72(12):123-143.
[4]	XIANG B H, YU Z X, QU H B.An improved algorithm of principal line extraction and registration based on MFRAT and ICP[J].Pattern Recognition and Artificial Intelligence, 2015, 28(3):260-265.
[5]	WONG M, ZHANG D, KONG W K, et al.Real-time palmprint acquisition system design[J].IEE Proceedings Vision, Image and Signal Processing, 2005, 152(5):527-534. doi: 10.1049/ip-vis:20049040
[6]	李艳, 吴贵芳, 戴高乐, 等.一种新的掌纹轮廓特征点提取算法[J].微电子学与计算机, 2010, 27(5):90-94. LI Y, WU G F, DAI G L, et al.A new algorithm to extract contour feature points of palmprint[J].Microelectronics and Computer, 2010, 27(5):90-94(in Chinese).
[7]	卢妍辉.智能交通系统中车辆路径识别方法[J].交通世界, 2018(35):15-16. LU Y H.Vehicle route recognition method in intelligent transportation system[J].Transpoworld, 2018(35):15-16(in Chinese).
[8]	张馨心, 姚爱琴, 孙运强, 等.基于深度学习的步态识别算法优化研究[J].自动化与仪表, 2020, 35(4):70-74. ZHANG X X, YAO A Q, SUN Y Q, et al.Research on optimization of gait recognition algorithm based on deep learning[J].Automation & Instrumentation, 2020, 35(4):70-74(in Chinese).
[9]	RONNEBERGER O, FISCHER P, BROX T.U-Net: Convolutional networks for biomedical image segmentation[EB/OL].(2015-05-18)[2020-06-28].
[10]	YUE F, ZUO W M, ZHANG D, et al.Orientation selection using modified FCM for competitive code-based palmprint recognition[J].Pattern Recognition, 2009, 42(11):2841-2849. doi: 10.1016/j.patcog.2009.03.015
[11]	SHEN L L, WU S P, ZHENG S H, et al. Embedded palmprint recognition system using OMAP 3530[J].Sensors, 2012, 12(2):1482-1493. doi: 10.3390/s120201482
[12]	LI C, LIU F, ZHANG Y Z.A principal palm-line extraction method for palmprint images based on diversity and contrast[C]//Proceedings of the 3rd International Congress on Image and Signal Processing.Berlin: Springer, 2010: 1772-1777.
[13]	ALI M M H, YANAWAR P, GAIKWAD A T.Study of edge detection methods based on palmprint lines[C]//Proceeding of the International Conference on Electrical, Electronics, and Optimization Techniques.Berlin: Springer, 2016: 1344-1350.
[14]	许学斌, 张德运, 张新曼, 等.基于离散曲波变换和支持向量机的掌纹识别方法[J].红外与毫米波学报, 2009, 28(6):456-460. XU X B, ZHANG D Y, ZHANG X M, et al.Palmprint recognition method based on discrete curvelet transform and support vector machine[J].Journal of Infrared and Millimeter Waves, 2009, 28(6):456-460(in Chinese).
[15]	XU X B, LU L B, ZHANG X M, et al.Multispectral palmprint recognition using multiclass projection extreme learning machine and digital shearlet transform[J].Neural Computing and Applications, 2016, 27(1):143-153.
[16]	吴婕, 任江涛.基于子空间特征融合的两级掌纹识别算法[J].计算机工程与应用, 2015, 51(11):175-178. WU J, REN J T.Two level palmprint recognition algorithm based on subspace features[J].Computer Engineering and Applications, 2015, 51(11):175-178(in Chinese).
[17]	李新春, 马红艳, 林森.基于子空间与纹理特征融合的掌纹识别[J].激光与光电子学进展, 2019, 56(7):147-155. LI X C, MA H Y, LI S.Palmprint recognition based on subspace and texture feature fusion[J].Laser & Optoelectronics Progress, 2019, 56(7):147-155(in Chinese).
[18]	IZADPANAHKAKHK M, RAZAVI S, TAGHIPOUR-GORJIKOL-AIE M, et al. Deep region of interest and feature extraction models for palmprint verification using convolutional neural networks transfer learning[J].Applied Sciences, 2018, 8(7):1210. doi: 10.3390/app8071210
[19]	GAO X J, LUO X L, PAN X, et al.Palmprint recognition based on deep learning[C]//Proceedings of the International Conference on Wireless Mobile and Multi-media.Berlin: Springer, 2015: 212-214.
[20]	HUANG D S, JIA W, ZHANG D.Palmprint verification based on principal lines[J].Pattern Recognition, 2008, 41(4):1316-1328.
[21]	WU X, ZHANG D, WANG K.Palm line extraction and matching for personal authentication[J].IEEE Transactions on Systems Man & Cybernetics Part A:Systems & Humans, 2006, 36(5):978-987. doi: 10.1109/TSMCA.2006.871797
[22]	谢杰成, 张大力, 徐文立.小波图象去噪综述[J].中国图象图形学报, 2002, 7(3):3-11. XIE J C, ZHANG D L, XU W L.Overview on wavelet image denoising[J].Journal of Image and Graphics, 2002, 7(3):3-11(in Chinese).
[23]	孔平.基于偏微分方程的图像去噪与增强研究[D].南京: 南京理工大学, 2012. KONG P.Research on image denoising and enhancement based on partial differential equation[D].Nanjing: Nanjing University of Science and Technology, 2012(in Chinese).
[24]	DANIELYAN A, KATKOVNIK V, EGIAZARIAN K.BM3D frames and variational image deblurring[J].IEEE Transactions on Image Processing, 2012, 21(4):1715-1728.
[25]	张秀峰, 张真林, 谢红.掌纹ROI分割算法的研究与实现[J].计算机科学, 2016, 43(S2):170-173. ZHANG X F, ZHANG Z L, XIE H.Research and realization of palmprint ROI segmentation algorithm[J].Computer Science, 2016, 43(S2):170-173(in Chinese).
[26]	MENG Z, FAN Z Y, ZHAO Z C, et al.ENS-Unet: End-to-end noise suppression U-Net for brain tumor segmentation[C]//Proceedings of Annual International Conference of the IEEE Engineering in Medicine and Biology Society.Piscataway: IEEE Press, 2018: 5886-5889.
[27]	QIAN P, LI A J, LIU M H.Latent fingerprint enhancement based on denseUNet[C]//2019 International Conference on Biometrics (ICB).Piscataway: IEEE Press, 2019: 1-6.
[28]	LIU D, SUN D M.Contactless palmprint recognition based on convolutional neural network[C]//Proceedings of the IEEE International Conference on Signal Processing.Piscataway: IEEE Press, 2016: 1363-1367.
[29]	SUN Q L, ZHANG J X, YANG A Q, et al.Palmprint recognition with deep convolutional features[J]//Proceedings of the Chinese Conference on Image and Graphics Technologies.Berlin:Springer, 2017:12-19.
[30]	LIN Z, CHENG Z X, YING S, et al.Palmprint and palmvein recognition based on DCNN and a new large-scale contactless palmvein dataset[J].Symmetry, 2018, 10(4):78.
[31]	王海霞, 杨熙丞, 梁荣华, 等.基于轻量级U-Net神经网络提取OCT内指纹和外指纹[J].高技术通讯, 2019(11):1063-1072. WANG H X, YANG X C, LIANG R H, et al.OCT internal and external fingerprints generation based on a lightweight U-Net neural network[J].Chinese High Technology Letters, 2019(11):1063-1072(in Chinese).
[32]	周瞳, 黄晨昕, 胡锐.基于深度卷积网络的主掌线纹区域的提取[J].工业控制计算机, 2018, 31(7):97-98. ZHOU T, HUANG C X, HU R.Extracting main lines area on palm based on deep convolutional network[J].Industrial Control Computer, 2018, 31(7):97-98(in Chinese).
[33]	PolyU Palmprint Palmprint Database[EB/OL].(2020-08-06)[2020-06-28].
[34]	COUPÉ P, HELLIER P, KERVRANN C.Nonlocal means-based speckle filtering for ultrasound images[J].IEEE Transactions on Image Processing, 2009, 18(10):2221-2229. doi: 10.1109/TIP.2009.2024064
[35]	ROGOWSKA J, BRYANT C M, BREZINSKI M E.Cartilage thickness measurements from optical coherence tomography[J].Journal of the Optical Society of America A:Optics, Image Science, and Vision, 2003, 20(2):357-367. doi: 10.1364/JOSAA.20.000357
[36]	YU Y J, ACTON S T.Speckle reducing anisotropic diffusion[J].IEEE Transactions on Image Processing, 2002, 11(11):1260-1270. doi: 10.1109/TIP.2002.804276
[37]	刘佳雯, 陈勇.关于掌纹图像基于形态学的角点检测的定位分割方法[J].现代计算机, 2016(24):35-38. LIU J W, CHEN Y.Research on go chess game image recognition algorithm[J].Modern Computer, 2016(24):35-38(in Chinese).
[38]	ZHANG D, KONG W K, YOU J, et al.Online palmprint identification[J].IEEE Transactions on Pattern Analysis and Machine Intelligence, 2003, 25(9):1041-1050. doi: 10.1109/TPAMI.2003.1227981
[39]	VENKATESAN R, KOON S M, JAKUBOWSKI M H, et al.Robust image hashing[C]//IEEE International Conference on Image Processing.Piscataway: IEEE Press, 2000: 664-666.