New IJECT Vo l . 6, Issu E 1, spl -1 Jan - Mar C h 2015 Secure and in … · 2016. 4. 1. · IJECT Vo l. 6, Issu E 1, spl - 1 Jan - Mar C h 2015 ISSN : 2230-7109 (Online) | ISSN :

IJECT Vol. 6, IssuE 1, spl-1 Jan - MarCh 2015

w w w . i j e c t . o r g InternatIonal Journal of electronIcs & communIcatIon technology 143

ISSN : 2230-7109 (Online) | ISSN : 2230-9543 (Print)

Secure and in-vivo Biometric Fingerprint Detection Using Optical Coherence Tomography

1Vishal Srivastava, 2Krishna Dalal, 2Devjyoti Dalal, 2Elanchezhiyan Devarajan1Dept. of ECE, Amity University, Noida, UP, India

2Dept. of Biophysics, All India Institute of Medical Sciences, New Delhi, India

AbstractFingerprint distinctive patterns are very helpful in biometric identification but are very vulnerable to spoofing such as dummy or fake fingers. In this context, we propose a method for identifying fingerprints based on the subcutaneous structure of fingers using swept source-optical coherence tomography. The proposed method has a potential in finding the artificial material on the surface of the fingers. The results indicated the accuracy and reliability of fingerprint identification.

KeywordsFingerprint, Optical Coherence Tomography and Biometric Identification

I. IntroductionIn modern electronically connected world, accuracy in personal identification is very important. To increase accountability, biometric identification based on physical or behavioral characteristics has been tremendously increased in last two decades. Amongst the existing different biometric attributes such as, DNA, face, signature, voice, iris, retina, fingerprints, facial thermogram recognition etc., fingerprint based identification received the most attention because of its extensive use in forensic [1 -4]. Skin of human finger consists of the series of ridges and valleys and form patterns which are distinct in nature [2]. These alternate ridge and valley patterns are called fingerprints. Injuries like cuts and burns temporally damage the fingerprints and once it fully healed up, it regains its original structure [2]. Due to its distinctive patterns, it is very helpful in biometric identification especially in forensic to solve the crimes [1 - 7]. Optical Coherence Tomography (OCT) is a non-contact, non-invasive and high resolution 3D optical imaging technology based on low coherence interferometry. OCT takes the advantage of low temporal coherence property of light source for ranging and high resolution optical sectioning. It measures the amplitude of backscattered light of the sample being imaged. Owing to the advantages of non-contact and high resolution, OCT has become a powerful technique for biomedical application [8]. The principle of OCT is very similar to ultrasound except that it uses light in place of sound [9-10]. It can be broadly classified in two categories: time-domain OCT (TD-OCT) and Fourier-domain OCT (FD-OCT). Further, FD-OCT can be divided into two parts: Spectral-domain OCT (SD-OCT) and swept source OCT (SS-OCT). The mechanical depth scans (A-scan) of TD-OCT is either replaced by spectrogram (SD-OCT) or by swept-source (SD-OCT) in FD-OCT. In SD-OCT broad band source is used at the input of interferometer while the output is connected to a spectrometer and line scan camera. In SS-OCT system broadband light is tuned with the help of tunable filter to achieve longer depth range and sensitivity [9]. Amongst the biomedical applications, OCT has attracted great attention in the field of forensic science specially, in fingerprint identification. There are two types of fingerprints: exemplar and latent. In the former one we directly get the fingerprints through specific fluid or scanner while in the later one when finger comes in contact with any surface it gets imprinted and through some contrast

enhancement we get fingerprints. In both the cases the fingerprint detection is based on the detection of the features of the surface of the human skin so that with the help of artificial dummy or mold, one can easily deceive the biometric system and attack the security system. FD-OCT is used to find out the internal structures of the tissues [9 - 10]. Further full-field OCT is used to find out the latent finger prints (based on surface imaging) [11]. The accuracy of the fingerprint detection based on the human surface skin is less in the case of artificial dummy. In this letter, we report the detection of artificial layers by using optical properties of the SS-OCT images. We anticipate that the high precision of this proposed method may be helpful to improve the system resistance to spoofing.

II. Experimental DetailsFig. 1 shows the schematic diagram of SS-OCT system. A swept laser source with 1325 nm central wavelength, spectral bandwidth 100 nm, 6 mm coherence length and output power of 10 mW was used.

Fig. 1: Schematic diagram of SS-OCT system. The swept laser source (SS), fiber coupler (FC), polarization controller (PC), circulator (CIR), collimator (C), adjustable pinhole variable attenuator (AP), data acquisition board (DAQ) and mirror (M).

IJECT Vol. 6, IssuE 1, spl- 1 Jan - MarCh 2015 ISSN : 2230-7109 (Online) | ISSN : 2230-9543 (Print)

w w w . i j e c t . o r g 144 InternatIonal Journal of electronIcs & communIcatIon technology

Fig. 2 (a) Typical OCT B-scan image of real finger, (b) corresponding OCT signal of (a), (c) OCT B-scan image of real finger with glycerin dummy layer (d) corresponding OCT signal of (c), (e) OCT B-scan image of real finger with soap dummy layer and (f) corresponding OCT signal of (e).

The volume image size acquired by the setup is 1024x512x512. The axial scan (A- scan) line rate (~16 kHz) is controlled by the sweeping frequency, while the transverse scan (B-scan) acquired by the galvo scanning mirrors. To provide frequency clock to laser, Mach-Zehnder Interferometer with the swept source was used. The output of the laser is coupled to the input of the fiber-based Michelson interferometer, which divides the input beam into the reference and sample beam using a broadband 50/50 fiber coupler. The reference beam gets back reflected into the fiber by a mirror. The sample beam pass through the handheld probe with XY galvo scanning mirrors towards the sample. The visible light scattered by the sample, fall onto the CCD camera by inserting a dichroic mirror along the beam path. The interference pattern is only observed when the reference arm and the sample arm is within the coherence length. The sample is placed on the X-Y and rotational translation stage to capture the data. To have the sample’s conventional microscopic view, there is an integrated CCD camera in the probe. For generating an artificial layer on the surface of the human finger skin soap and glycerin were used.

Fig. 3: Flow Diagram of the Study Design

III. Results and DiscussionMost of fingerprint identification methods are based on surface imaging of the human skin which can produce false results in the case artificial layers. OCT generates both the surface and the subsurface images of the human skin. We conducted an experiment on 18 individuals of both genders of age group 30 to 60 years to detect fingerprints. The finger was positioned on a mark such that the core of the fingerprint was placed at the center of the acquired image. In these 18 objects, 6 were reference ones which represented the data records without applying any artificial material on the surface of the finger-skin, 6 were coated with thick layer of glycerin and rest 6 were coated with thick layer of soap. Fig. 2 (a), (c) and (e) shows the B-scan images of the fingerprints, where the brightest region was at the top and it is due to refractive index mismatch at the surface of the skin and air and Fig. 2 (b), (d) and (f) show the corresponding OCT signal of real, coated with glycerin and soap respectively obtained from the thumb skin. To obtain this, we found out A-scan line from the B-scan OCT images. The OCT signal shown in Fig. 2 (b) depicts the highly scattering stratum corneum as the first layer and below that epidermis and dermis are distinctly visible. As can be clearly seen from Fig. 2(b) that there is no peak before the stratum corneum peak while when we put a glycerin on thumb skin there is another peak at 0.20 mm just before the stratum corneum as shown in Fig. 2 (d). Similarly, in the case of thumb coated with soap, we observed the peak just before the stratum corneum peak at 0.30 mm as shown in Fig. 2 (f). The whole procedure/algorithm for finger identification is explained in Fig. 3. We also reconstructed the 2D images (X-Y) of fingers as shown in Fig. 4 and used correlation algorithm to identify the fingerprint. In Fig 4 (c) the yellow arrows show that the thickness of the ridges was increased due to soap and would generate false result. In correlation algorithm we took two fingerprints: one was reference one and another was sample. and normalizing both the images we found out the correlation. If the maximum correlation value was within the range of [0 1], the object was declared to be true, else false. The main advantage of the proposed method is summarized in Table 1.

Fig. 4 (a): Typical OCT 2D X-Y image of real finger, (b) OCT 2D X-Y image of real finger with glycerin dummy layer and (c) OCT 2D X-Y image of real finger with soap dummy layer.

IJECT Vol. 6, IssuE 1, spl-1 Jan - MarCh 2015

w w w . i j e c t . o r g InternatIonal Journal of electronIcs & communIcatIon technology 145

ISSN : 2230-7109 (Online) | ISSN : 2230-9543 (Print)

Table 1: Advantages of the Proposed Method for Fingerprint Detection

These results demonstrated the capability the OCT to detect the fake finger print by imaging multilayer objects with high resolution and from 2D and 3D images. This multilayer based fingerprint detection technology is more robust as compare to the conventional biometric systems. We propose the flowchart (Fig. 5), which may be adopted for detecting the in-vivo and secured real and fake fingerprints on the spot.

Start

Select the test image

Check surface pattern through

scanner

Check OCT A-scan signal

End

Wrong Person

Wrong Person

Y

Y

N

N

Right Person

Fig. 5: Proposed Algorithm for Fingerprint Identification.

VI. ConclusionIn this study, we presented the OCT as a more powerful technique for biometric (fingerprint) identification. Combination of 1D and 2D OCT images increased the accuracy of fingerprint detection. We could successfully demonstrate the detection of artificial materials that were put on the fingers to deceive the biometric system.

References[1] Lee, H. C., Gaensslen, R. E. Eds.,“Advances in Fingerprint

Technology”, 2nd ed., CRC Press, 2001.[2] Maltoni, D., Maio, D., Jain, A. K., Prabhakar, S., “Handbook

of Fingerprint Recognition”, Springer, 2003.[3] Jain, K., Ross, A., Prabhakar, S.,“An introduction to biometric

recognition”, IEEE Transactions on Circuits and Systems for Video Technology, Vol. 14, 2004, pp. 4–20,

[4] Munir, M. U., Javed, M. Y.,“Fingerprint matching using ridge patterns”, Proceeding of IEEE-ICICT, 2005, pp. 116-20.

[5] Prabhakar, S., Pankanti, S., Jain, A. K.,“Biometric recognition: Security and privacy concerns”, IEEE Security Privacy, Vol. 1, 2003, pp. 33–42.

[6] Chang, J. H., Fan, K. C.,“Fingerprint ridge allocation in direct gray-scale domain”, Pattern Recognition, Vol. 34, 2004, pp. 1907–1925.

[7] Jain, A. K., Bolle, R., Pankanti, S., Eds.,“Biometrics: Personal identification in networked Society”, Kluwer Academic Publishers, 1999.

[8] Huang, D., Swanson, E., Lin, C., Schuman, J., Stinson, W. , Chang, W., Hee, M., Flotte, T., Gregory, K., Puliafito C., Fujimoto, J., “Optical coherence tomography”, Science, Vol. 254, 1991, pp. 1178–1181.

[9] Bouma, B. E., Tearney, G. J.,“Handbook of Optical Coherence Tomography”, Marcel Dekker, Inc. 2002.

[10].Brenzinski, M. E., “Optical coherence tomography-principles and applications”, Elsevier Inc. 2006.

[11] Dubey, S. K., Anna, T., Shakher, C., Mehta, D. S., “Fingerprint detection using full-field swept-source optical coherence tomography”, Applied Physics Letter, Vol. 91, 2009, pp: 181106 – 181108.