=Paper=
{{Paper
|id=Vol-2603/short1
|storemode=property
|title=User Identification based on the Vein Pattern in Biometric Immobilizer
|pdfUrl=https://ceur-ws.org/Vol-2603/short1.pdf
|volume=Vol-2603
|authors=Michael Basarab,Tatyana Buldakova,Kristina Smolyaninova,Michael Sokolov
}}
==User Identification based on the Vein Pattern in Biometric Immobilizer==
https://ceur-ws.org/Vol-2603/short1.pdf
Copyright © 2019 for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0)
User Identification based on the Vein Pattern in
Biometric Immobilizer
Michael A. Basarab1, Tatyana I. Buldakova2, Kristina A. Smolyaninova3, Michael N. Sokolov4
Faculty of Informatics and Control Systems
Bauman Moscow State Technical University
Moscow, Russia
1
bmic@mail.ru, 2buldakova@bmstu.ru
3
kriszzztina@yandex.ru, 4mike.sv@mail.ru
Abstract—Article deals with the problem of vehicles However, the disadvantage of existing biometric immobilizer
protection against theft by using biometric immobilizers. The use systems is the use of the fingerprint authentication method [5-
of a vascular authentication method based on the vein pattern of 7]. This method is not safe, since various methods of imitating
the driver's finger is suggested. The flowcharts of algorithms for a fingerprint are known.
the image preprocessing and the formation of the biometric
pattern are given. The article proposes to use the vascular authentication
method, which provides high recognition accuracy and
Keywords—biometry; pattern recognition; authentication characteristics concealment. The pattern of the veins is only
algorithm; vein pattern; biometric immobilizer; vehicle visible in the infrared spectrum, so it cannot be falsified [8, 9].
One of the key tasks of constructing biometric
I. INTRODUCTION authentication systems based on finger veins is preliminary
Nowadays we are facing the problem of vehicles protection image processing, which cuts unnecessary areas and prepares
from theft. More electronic devices are added to cars, and a image for extraction of the biometric features [10].
growing number of vulnerabilities allow attackers to hack and
steal the transport. At the same time, the ways of hacking II. FUNCTIONAL MODEL OF THE BIOMETRIC AUTHENTICATION
become more sophisticated, therefore, the methods of MODULE
protection should be more effective.
The technology of biometric authentication by vein pattern
Traditional means of protecting vehicles are mainly related is based on optical visualization of human veins and their
to scaring, tracing and attracting attention. Along with this, further recognition [11, 12]. Since hemoglobin in blood
there are regular means of blocking the car, but intruders absorbs infrared radiation and other tissues reflect it, the veins
overcome them by prescribing counterfeit keys to the central appear darker in the image than other tissues. This allows us to
control unit [1, 2]. use them for further user authentication.
The solution of this problem is the use of biometric The main element of the biometric immobilizer is the
immobilizers, which allow the vehicle's engine to be blocked image capture and recognition module, as it provides the
by breaking critical electrical circuits [3]. Biometric signs of a system with user identification functions. In Fig. 1 you can see
person are unique, which allows them to be successfully the block diagram of the biometric module of user
applied in security systems for identification of users [4]. authentication.
1
Fig. 1. Block diagram of the biometric module of user authentication
The input of the module receives an image from the For this purpose, the sum of the multiplications of values of
scanner, after that the image is pre-processed, which includes: pixels by their position in the horizontal and vertical
1) extraction of the area of interest; 2) resizing the image; 3) coordinates is calculated, i.e.
improving the quality of the image.
The input image contains an unwanted background, so the aw Heigth i1 pij xi
W idth
first step is to filter image and select the area of interest. j 1
Filtering allows you to distinguish significant areas of finger
veins, reduce areas of noise and glare. Then the original image and
is converted to a binary code using the Otsu method [13].
pij y j
Binarization allows you to determine the center of the finger
and crop the image based on the selected center point. ah Wi1idthHeigth
j 1
To reduce the algorithm computation time and to further
reduce noise, the size of the cropped image is scaled (its respectively. Here pij – value of pixel with coordinates xi and yj,
resolution is reduced). Since the resulting image, basically, has Width – image width, Heigth – image heigth.
a low contrast level, it needs to be increased with a modified
Gaussian high-pass filter. Barycenter coordinates is calculated as Bx aw W and
This filter allows you to extract low-frequency components, B y ah W , where W Wi1idthHeigth
j 1 pij . Barycenter
such as the borders and veins of the finger. A flowchart of the coordinates Bx and By is used as a reference point for
algorithm for pre-processing of the user's venous finger image determining the coordinates of the singular points of the vein
is shown in Fig. 2. Let us consider the main steps of the pattern.
algorithm and their features. Based on the calculated point, the image is cropped to
480x160, and thus the area of interest is distinguished. In the
III. SEARCH OF INTERESTED AREA future, to increase the speed of the algorithm, as well as to get
First, the input image is binarized using the Otsu method rid of pixel noise, the cropped image is scaled to a resolution of
(Figure 3), and the barycenter of the white region (the 192x64 pixels.
conditional finger center) is determined in the binarized image.
2
Begin
Begin
Image
Image
Histogram of image color
intensity calculation;
Histogram sum calculation;
Otsu Normalized histogram calculation;
method Threshold calculation
Centroid calculation; No Do we have not
Image cropping by processed pixel?
centroid to 480x160;
Yes
Image resizing to 192x64;
Yes No
Pixel value >
Threshold?
Image
filtering
Pixel value = white Pixel value = black
Return
preprocessed image
Return binarized
image
End End
Fig. 2. Block diagram of the image preprocessing algorithm Fig. 3. Block diagram of the Otsu method
Thus, binarization allows us to determine the center of the Different approaches to texture analysis are possible. In this
area of interest (the conditional center of the finger) and to case, the features of the venous finger pattern of vehicle user
obtain a scaled image that will be used to highlight the pattern are detected using the algorithm LLBP (Local Line Binary
of veins. Pattern). To extract the biometric image, this technique uses a
new texture descriptor [15]. One of the advantages of the
IV. IMAGE QUALITY IMPROVEMENT AND BIOMETRIC FEATURES LLBP algorithm is that it can emphasize the change in image
EXTRACTION
intensity (for example, in the area of bifurcation (separation) of
vessels, in the areas of the end or bend of the vessel).
Next, it is necessary to create the biometric image that
contains the features of user’s finger veins as a digital code. The operator used for texture analysis consists of two
But since the input image has a low contrast value, it needs to components: horizontal (LLBPh) and vertical (LLBPv)
be increased. This procedure can be performed, for example, components. The LLBP value can be obtained by computing
using a modified Gaussian high-pass filter [14]. The filter is the binary string codes for both components. These
calculated by the formula: components are calculated by the following formulas:
H x, y a 1 e D x, y 2 D0 b
2 2
1, x 0,
sx
0, x 0,
where Dx, y x x0 2 y y0 2 – distance from a point
(x, y) to a cutoff frequency locus with coordinates (x0, y0); D0 – LLBPh x, y cn11 shn hc 2c n 1 nNc 1 shn hc 2n c 1
cutoff frequency locus distance from origin of coordinates; a
and b – correcting variables for changing the amplitude and the
initial level of the filter mask signal. An example of the
filtering result is shown in Fig. 4. LLBPv x, y cn11 svn vc 2c n 1 nNc 1 svn vc 2n c 1
After filtering, the image has a sufficient level of contrast to
extract the biometric code, which describes the features of the
venous pattern in the form of texture descriptors.
3
LLBP LLBPh2 LLBPv2 Fig. 5 shows the graphical result of the LLBP encoding for
the vertical, horizontal, and final component.
The further authentication process is based on comparing
where s(x) – threshold function; N – length of pixel line;
the extracted features with template features stored in the
c N 2 – central pixel position (hc, vc). database. When identifying, the similarity between them can be
measured using, for example, the Hamming distance.
Fig. 4. Image filtering result: a – original image, b – filtered image
Fig. 5. Result of encoding by the LLBP method
Since the images of one user, calculated at different time Depending on the mode of operation of the biometric
intervals with different positions of the finger, can differ from immobilizer module, the extracted image is either recorded in
each other, the recognition algorithm shall consider these the database or compared with the existing etalon images in the
differences. Therefore, the comparison is performed using a database. In the second case, a decision about the degree of
threshold value, the change of which will influence the images coincidence is made.
recognition accuracy and the magnitude of the errors of the
first and second kind. In the presence of these errors there is a REFERENCES
successful authentication of the user who is absent in the
[1] S. Tillich, M. Wójcik “Security Analysis of an Open Car Immobilizer
database, or access to a legal user is denied, respectively [16]. Protocol Stack,” in Proceedings of 4th International Conference on
The threshold value should be selected based on statistical Trusted Systems. Lecture Notes in Computer Science. C.J. Mitchell and
A. Tomlinson, Eds. 2012. Vol. 7711. Pp. 83-94.
error rates for different threshold values [17]. At the choice of https://doi.org/10.1007/978-3-642-35371-0_8.
value it is required to minimize indicators of errors of the first [2] J. Wei, Y Matsubara and H. Takada, “HAZOP-Based Security Analysis
and second kind. for Embedded Systems: Case Study of Open Source Immobilizer
Protocol Stack,” In Recent Advances in Systems Safety and Security.
Studies in Systems, Decision and Control. E. Pricop and G. Stamatescu,
V. CONCLUSION Eds. 2016. Vol. 62. Pp. 79-96.
The biometric image, extracted from the image of the blood [3] J.C. van Ours, B. Vollaard, “The Engine Immobiliser: A Non‐ starter for
vessels and defining the specific points of the veins, makes it Car Thieves,” The Economic Journal. 2016. Vol. 126, No. 593. Pp.
possible to accurately identify a specific user. Patterns of blood 1264–1291.
vessels are unique for each person and, unlike fingerprints, they [4] S. Prabhakar, S. Pankanti and A.K. Jain, “Biometric recognition:
security and privacy concerns,” IEEE Security & Privacy,vol. 99, Issue
can not be faked, so it is not necessary to re-register users 2, pp. 33-42, Mar-Apr 2003.
through certain periods. [5] C.-N. Liang, H.-B. Huang and B.-C. Chen, “Fingerprint Identification
In this research, a functional model of the work of the Keyless Entry System,” International Journal of Electronics and
Communication Engineering, vol. 2, No. 8, pp. 1554-1559, 2008.
biometric authentication module based on finger veins was
[6] A. Kumar, Y. Zhou, “Human identification using finger images,” IEEE
developed; algorithms for image pre-processing and biometric Transactions on Image Processing, 21 (4), pp. 2228-2244, 2012.
features extracting were considered.
4
[7] S.C. Draper, A. Khisti, E. Martinian, A. Vetro and J.S. Yedidia, “Using [12] C. Wilson, Vein Pattern Recognition: A Privacy-Enhancing Biometric,
distributed source coding to secure fingerprint biometrics,” IEEE CRC Press, 2017, 307 p.
International Conference on Acoustics, Speech and Signal Processing, [13] N. Otsu, “A threshold selection method from gray-level histograms,”
Hawaii, pp. 129–132, 2007. IEEE Transaction Systems, Man, and Cybernetics, vol. 9, No. 1, pp. 62-
[8] Y. Zhou and A. Kumar, “Human identification using palm-vein images,” 66, 1979.
IEEE Transactions on Information Forensics and Security. 6 (4), pp. [14] E.C. Lee, H. Jung and D. Kim, “New finger biometric method using
1259-1274, 2011. near infrared imaging,” Sensors, 11(3), pp. 2319–2333, 2011.
[9] S. Liu and Sh. Song, “An embedded real-time finger-vein recognition [15] A.R. Bakhtiar, W.S. Chai and A.S. Shahrel, “Finger Vein Recognition
system for mobile devices,” IEEE Transactions on Consumer Using Local Line Binary Pattern,” Sensors, 11(12), pp. 11357-11371,
Electronics, vol. 58, Issue 2, pp. 522-527, May 2012. 2011. doi:10.3390/s111211357
DOI: 10.1109/TCE.2012.6227456
[16] K.W. Ko, J. Lee, M. Ahmadi and S. Lee, “Development of Human
[10] S.I. Suyatinov, S.V. Kolentev and T.I. Bouldakova, “Criteria of Identification System Based on Simple Finger-Vein Pattern-Matching
identification of the medical images,” Proceedings of SPIE - The Method for Embedded Environments,” International Journal of Security
International Society for Optical Engineering, vol. 5067, pp. 148-153, and Its Applications, vol. 9, No. 5, pp. 297-306, 2015. URL:
2002. http://dx.doi.org/10.14257/ijsia.2015.9.5.29
[11] N. Kaur and P. Chopra, “Vein Pattern Recognition: A secured way [17] R. Chopra and S.Kaur, “Finger print and finger vein recognition using
of authentication,” International Journal of Engineering And Computer repeated line tracking and minutiae,” International Journal of Advanced
Science, vol. 5, Issue 10, pp. 18377-18383, Oct. 2016. DOI: Science and Research, vol. 2, Issue 2,pp. 13-22, 2017.
10.18535/ijecs/v5i10.26
5