Abstract— For providing the secure facilities and services to the user, the accurate identification is necessary. Multimodal Biometric System using multiple sources of information for establishing the identity has been widely recognized.
Multibiometric recognition systems, which aggregate information from multiple biometric sources, are able to overcome limitations such as non-universality, noisy sensor data and susceptibility. Multibiometric systems promise significant improvements as higher accuracy and increased resistance to spoofing over the single biometric systems. In this paper multimodal biometric image such as fingerprint and iris are extracted individually and are fused together using a score level fusion mechanism.
Index Terms— Biometric, Fingerprint recognition, fusion, Iris recognition, Multimodal Biometric
I. INTRODUCTION
Biometric technology deals with recognizing the identity of individuals based on their unique physical or behavioral characteristics. Physical characteristics such as fingerprint, palm print, hand geometry and iris patterns or behavioral characteristics such as typing pattern and hand-written signature present unique information about a person and can be used in authentication applications. Biometrics can be divided into two types. Unimodal and multimodal. Many unimodal biometrics systems suffer from limitations such as inability to tolerate deformed data due to noise, deformed data from the sensor device, distorted signal from environmental noise and variability of an individual’s physical appearance and pattern over time. Multimodal biometrics are able to solve some of these limitations by combining information from multiple biometric sources, e.g.
Palmprint and fingerprint, face and iris etc. Among the many current biometric technologies, fingerprint identification is the oldest and the most popular one. Fingerprint technology has low cost comparing to others and high user acceptance. It is the method of identification using the impressions made by the minute ridge formats or patterns found on the fingertips.
For every individual the ridge patterns will be different throughout the life. Fingerprints will offer an infallible means of personal authentication. Other personal characteristics may change, but fingerprints do not. However some people do not have clear fingerprints because of their physical work or problematic skin. Iris and retina recognition system provide very high accuracy [30] but input devices cost high.
A typical multimodal biometric authentication system consists of five parts. Image capture, pre-processing, feature extraction,
fusion and matching. Special biometric scanners are used for image capturing. It may vary depending on the type of biometric traits used [1, 2]. At the pre-processing stage the image is enhanced to remove noise and unwanted areas.
Feature extraction gets effective features from the pre-processed biometric trait. After feature extraction fusion is carried out to combine different features and stored in the database as templates. A matching algorithm is used to compare it with the stored one in the database. Fig 1 gives the basic block diagram of a biometric system.
II- LITERATURES SURVEY/RELATED WORK A number of studies have been done on multimodal biometrics and these works show that multi- biometric has more advantage than single- biometric. A variety of articles can be found, which propose different approaches for unimodal and multimodal biometric systems. Multimodal biometric systems are based on different biometric features and/or introduce different fusion algorithms for these features.
Many researchers have demonstrated that the fusion process is effective, because fused scores provide much better discrimination than individual scores. Such results have been achieved using a variety of fusion techniques. Toh et al. [5]
combined hand geometry, fingerprint and voice by using global and local learning decision as fusion approach. The accuracy performance is 85% to 95%. Viriri and Tapamo [6]
introduced a multimodal approach including iris and signature biometrics at score level fusion with False Reject Rate (FRR) 0.008% on a False Accept Rate (FAR) of 0.01%.
Fierrez-Aguilar and Ortega-Garcia [7] proposed a multimodal approach including face, a minutiae-based fingerprint and online signature with fusion at the matching score level. The fusion approach obtained Equal Error Rate (EER) of 0.5. Luca et al. [8] used fingerprint and face to be fused at the match score level. PCA and LDA are used for the feature extraction and classification. Mean rule, product rule and Bayesian rule are used as the fusion techniques with FAR of 0% and FRR of 0.6% to 1.6%. Meraoumia et al. [9]
presented a multimodal biometric system using hand images and by integrating two different biometric traits palmprint and finger-knuckle-print (FKP) with EER = 0.003 %.
Rodriguez et al. [10] used signature with iris by using sum rule and product rule as the fusion techniques. Neural Network is used as the classification technique with EER below than 2.0%. Kartik et al. [11] combined speech and signature by using sum rule as fusion technique after the min max normalization is applied. Euclidean distance is used as the classification technique with 81.25% accuracy performance rate. Aggithaya et al. [12] proposed a personal
Performance of Identification System Based on Score Level Fusion
Subhash V.Thul, Anurag Rishishwar, Manish Trivedi
simultaneously exploits 2D and 3D Palmprint features. The sum rule classifier achieves the best EER of 0.002. Feng et al.
[13] combined face and palmprint at feature level by concatenating the features extracted by using PCA and ICA with the nearest neighbor classifier and support vector machine as the classifier. Kisku et al. [14] proposed a multibiometric system including face and Palm print biometrics at feature level fusion. The system attained 98.75% recognition rate with 0% FAR. Kazi and Rody [15]
presented a multimodal biometric system using face and signature with score level fusion. The results showed that face and signature based bimodal biometric system can improve the accuracy rate about 10%, higher than single face/signature based biometric system.
III PROPOSED MULTIMODAL SYSTEM For combining two or more uni-modal biometrics and making a multi-biometric system, two or more acceptance results must be combined as fusion. Fusion strategies can be divided into two main categories: premapping fusion (before the matching phase) and postmapping fusion (after the matching phase). The first strategy deals with the sensor level fusion, feature level fusion. Usually, these techniques are not used because they result in many implementation problems.
The second strategy is realized through fusion at the decision level, based on some algorithms, which combine single decisions for each component of the system. Furthermore, the second strategy is also based on the matching-score level, which combines the matching scores of each component system.
Fig.-1:Schematic diagram of multibiometric system
A generic biometric system has 4 important modules: (a) the sensor module which captures the trait in the form of raw biometric data; (b) the feature extraction module which processes the data to extract a feature set that is a compact representation of the trait; (c) the matching module which employs a classifier to compare the extracted feature set with the stored templates to generate matching scores; (d) the decision module which
uses the matching scores to either determine an identity or validate a claimed identity. Figure (i) is the representation of a conventional biometric system. The main operations that the system can perform are enrolment and testing. During enrolment biometric information of an individual are stored, during test biometric information are detected and compared with the stored ones. The first block (sensor) is the interface between the real world and our system; it has to acquire all the necessary data. Most of the times it is an image acquisition system, but it can change according to the characteristics we want to consider. The second block performs all the necessary preprocessing: it has to remove artifacts from the sensor, to enhance the input (e.g. removing some noise), to use some kind of normalization, etc. In the third block we have to extract the features we need. This step is really important: we have to choose which features to extract and how to do it, with certain efficiency to create a template. After that, we are matching the input pattern and the Data base pattern using pattern matching technique.
Finally Authentication occurs based on pattern matching.
System is divided into three sub-systems:- Fingerprint recognition, Iris recognition, Fusion techniques:
3.1 Fingerprint recognition
Fingerprint recognition involves the following four steps:
The pre-processing of the image involves taking the image and applying various processes on the image so that it can easily be processed to find out the ridge endings and bifurcation points. The two major steps in the pre-processing are:
1) Binarizing: In this step the colors of the image are binaries so that the output image consists of only two colors, black and white.
2) Thinning: After the fingerprint image is converted to binary form, submitted to the thinning algorithm which reduces the ridge thickness to one pixel wide, demonstrates that the global thresholding technique is effective in separating the ridges (black pixels) from the valleys (white pixels). The results of thinning show that the connectivity of the ridge structures is well preserved, and that the skeleton is eight-connected throughout the image.
3) Minutiae extraction: The most commonly employed method of minutiae extraction is the Crossing number (CN) concept. This method involves the use of the skeleton image where the ridge flow pattern is eight-connected. The minutiae are extracted by scanning the local neighborhood of each ridge pixel in the image using a 3*3 window. The CN value is then computed, which is defined as half the sum of the differences between pairs of adjacent pixels in the eight-neighborhood. Using the properties of the CN as, the ridge pixel can then be classified as a ridge ending, bifurcation or non-minutiae point.
4) Matching: The algorithm that we have applied to match two fingerprints involves calculating hamming distance between each ridge end and all other ridge ends and similarly between all bifurcations and then taking
the averages for both and at last add the results for achieving high accuracy and precision level. This process is applied on both the images and the results of the two images are compared to give the percentage match between the two images.
Fig.-2: (a) Input image (b)Binarized image (c)Thinned image (d)Ridge end+Bifurcation (e)Common Region of ROI and Image (f)Final Minutiae
3.2 Iris recognition
1) Iris segmentation: It is a significant module in iris recognition. It comprises of two steps 1) canny edge detection technique 2) The parabolic Hough transform. The iris image is first fed as input to the canny edge detection algorithm that produces the edge map of the iris image for boundary estimation. The exact boundary of pupil and iris is located from the detected edge map using the Hough transform. Hough transform has been enhanced to find positions of arbitrary shapes, usually circles or ellipses. For the parameters of circles passing through every edge point, votes are being casted in Hough space, from the obtained edge map. These parameters are the centre coordinates x and y, and the radius are capable to describe the circle in accordance with this equation:
……….. (1) 2) Iris normalization: Once the segmentation module has estimated the iris’s boundary, the normalization module uses image registration technique to transform the iris texture from Cartesian to polar coordinates. Daugman`s Rubber Sheet Model is utilized for the transformation process.
3) Feature encoding: Feature encoding extracts the underlying information in an iris pattern and generates the binary iris template that is used in matching. The normalized 2D form image is disintegrated up into 1D signal, and these signals are made use to convolve with 1D Gabor wavelets.
The frequency response of a Log-Gabor filter is as follows,
……….. (2) Where fo indicates the centre frequency and σ provides bandwidth of the filter. The Log-Gabor filter generates the biometric feature (texture properties) of the iris.
4) Matching: we are using Hamming distances (HD) to calculate the matching scores between two iris templates. It is
patterns. Using the Hamming Distance of two bit patterns, a decision is made as to whether the two patterns were generated from different Irises or from the same one.
Fig.-3: Steps involved in iris feature set extraction
3.3 Range normalization
The scores generated by a biometric system can be either similarity scores or distance scores, one need to convert these scores into a same nature. Normalization maps the raw matching scores to interval [0, 1] and retains the original distribution of matching scores except for a scaling factor.
Given that max(X) and min(X) are the maximum and minimum values of the raw matching scores, respectively, the normalized score is calculated as
………..……….….
(3)
3.4 Sum rule based score level fusion
The procedure for sum rule-based fusion is stated as following.
After we get a set of normalized scores(x1, x2,……., xm) from a particular person (here the index i=1,….,m indicates the biometric matcher), the fused score fs is evaluated using the formula,
fs =w1x1+ . . .
+wmxm…………..………...(4) The notation wistands for the weight which is assigned to the matcher–i, for i=1,…, m. There are many choices of how to calculate these weights based on some preliminary results. In the next step, the fused score fs will be compared to a pre- specified threshold t. If fs≥ t,
then a person declares as to be genuine otherwise, declare as an impostor.
4. EXPERIMENTAL OUTPUT
Fig-4: Processing of fingerprint recognition
Fig.-5: Processing of iris recognition
Fig.-6: Graph of FAR and FRR for Fingerprint
Fig.-7: Graph of FAR and FRR for iris
Fig.-8: Graph of FAR and FRR for fusion
Fig.-9: Graph of FRR for fusion, fingerprint and iris
Fig.-10: Graph of FAR for fusion, fingerprint and iris
5. CONCLUSIONS
Biometric features are unique to each individual and remain unaltered during a person’s lifetime. An efficient algorithm using the phase-based image matching is particularly effective for verifying low-quality fingerprint images that could not be identified correctly by conventional techniques.
Log-Gabor filter is effective method than any other technique to extract feature from iris image capture. Fusion can be applied to enhance the performance of system and security level. Fusion increases accuracy, but it generally increases template sizes, computation costs.
REFERENCES
[1] C. Sanderson and K. K. Paliwal, Information Fusion and Person Verification Using Speech and Face, Information.
Research Paper IDIAP-RR 02-33, IDIAP, September 2002.
[2] A. Ross, K. Nandakumar, and A. K. Jain, Handbook of Multibiometrics, New York: Springer, 2006.
[3] A. Ross and R. Govindarajan, Feature Level Fusion Using Hand and Face Biometrics, In Proceedings of SPIE Conference on Biometric Technology for Human Identification II, volume 5779, pages 196–204, Orlando, USA, March 2005.
[4] A.K. Jain, A. Ross, Multibiometric systems, Communications of the ACM, Special Issue on Multimodal Interfaces, Vol. 47, January 2004, 34-40.
[5] Toh.K.A, J. Xudong and Y. Wei-Yun, “Exploiting global and local decisions for multimodal biometrics verification,”
Signal Processing, IEEE Transactions on Signal Processing, vol. 52, pp. 3059-3072, 2004.
[6] S. Viriri and R. Tapamo, “Integrating Iris and Signature Traits for Personal Authentication using User-Specific Weighting”, 2009.
[7] J. Fierrez-Aguilar, J. Ortega-Garcia, D. Garcia-Romero, and J. Gonzalez Rodriguez, “ A comparative evaluation of
Proc. 4th Int, Conf,Audio-video-based Biometric Person Authentication , J. Kittler and M. Nixon, Eds., vol. LNCS 2688, pp. 830–837, 2003.
[8] Gian Luca Marcialis and Fabio Roli, “Serial Fusion of Fingerprint and Face Matchers”, M. Haindl, MCS 2007, LNCS volume 4472, pp. 151-160, © Springer-Verlag Berlin Heidelberg 2007
[9] A. Meraoumia, S. Chitroub and A. Bouridane, “Fusion of Finger-Knuckle-Print and Palmprint for an Efficient Multi-biometric System of Person Recognition”, IEEE ICC 2011.
[10] Rodriguez.L.P, Crespo.A.G, Lara.M and Mezcua.M.R,
“Study of Different Fusion Techniques for Multimodal Biometric Authentication,” in Networking and Communications. IEEE International Conference on Wireless and Mobile Computing, 2008
[11] Kartik.P, S.R. Mahadeva Prasanna and Vara.R.P,
“Multimodal biometric person authentication system using speech and signature features,” in TENCON 2008 - 2008 IEEE Region 10 Conference, pp. 1-6, Ed, 2008.
[12] V. Aggithaya, D. Zhang and N. Luo “A Multimodal biometric authentication system based on 2D and 3D palmprint features”, Proc. of SPIE Vol. 6944 69440C-1- 2012.
[13] G. Feng, K. Dong, D. Hu and D. Zhang, “When Faces Are Combined with Palmprints: A Novel Biometric Fusion Strategy,” in Biometric Authentication. vol. 307, 2004.
[14] D. Kisku, P. Gupta and J. Sing, “Multibiometrics Feature Level Fusion by Graph Clustering”, International Journal of Security and Its Applications Vol. 5 No. 2, April, 2011
[15] M. Kazi and Y. Rode, “multimodal biometric system using face and signature: a score level fusion approach” ,Advancies in Computational Research, Vol. 4, No.
1, 2012.
[16] R. Wildes, J. Asmuth, G. Green, S. Hsu, and S. Mcbride.
“A System for Automated Iris Recognition”, Proceedings IEEE Workshop on Applications of Computer Vision, Sarasota, FL, USA, 1994.
[17] K. Dmitry, “Iris Recognition: Unwrapping the Iris”, The Connexions Project and Licensed Under the Creative Commons Attribution License, Version 1.3. (2004).
[18] A. K. Jain, K. Nandakumar, & A. Ross, “Score Normalization in multimodal biometric systems”, The Journal of Pattern Recognition Society, 38(12), 2005, 2270-2285.
[19] Bhupesh Gour, T. K. Bandopadhyaya and Sudhir Sharma, “Fingerprint Feature Extraction usingMidpoint Ridge Contour Method and Neural Network”, International Journal of Computer Science and Network Security, vol. 8, no, 7, pp. 99-109, (2008).
[20] G. Aguilar, G. Sanchez, K. Toscano, M. Nakano, and H.
Perez, Mul-timodal biometric system using fingerprint,ǁ in Proc. Int. Conf. Intell. Adv. Syst. 2007, pp. 145–150. DOI:
10.1109/ICIAS.2007.4658364
[21] F. Besbes, H. Trichili, and B. Solaiman, “Multimodal biometric system based on fingerprint identification and Iris recognition,” in Proc. 3rd Int.IEEE Conf. Inf. Commun.
Technol.: From Theory to Applications (ICTTA2008), pp.
1–5. DOI: 10.1109/ICTTA. 2008. 4530129.
[22] Prince,Manvjeet K., “Fingerprint matching system using level 3 feature”,Intrrnational journal of engg. science and tech Vol.2(6),2010, 2258- 2262.
[23] R. Snelick, U. Uludag, A. Mink, M. Indovina, A. Jain, Large-scale evaluation of multimodal biometric
authentication using state-of-the-art systems, IEEE Trans- actions on Pattern Analysis and Machine Intelligence 27 (3) (2005) 450–455.
[24] M. Vatsa, R. Singh, and A. Noore, “Improving iris recognition performance using segmentation, quality enhancement, match score fusion, and indexing,” IEEE Trans. Syst., Man, Cybern. B, Cybern., vol. 38, no. 4, pp.
1021–1035, Aug. 2008.
[25] A. Nagar, S. Rane, and A. Vetro, “Privacy and security of features extracted from minutiae aggregates,” in Proc. IEEE Int. Conf. Acoustics, Speech and Signal Processing, Dallas, TX, Mar. 2010, pp. 524–531.
Subhash V. Thul PG Scholar, Department of Electronics
& Communication, RKDF Institute of Science & Technology Bhopal, Madhya Pradesh, INDIA.
Anurag Rishishwar Asst. Professor, Department of Electronics & Communication, RKDF Institute of Science &
Technology Bhopal, Madhya Pradesh, INDIA.
Manish Trivedi Asst. Professor, Department of Electronics & Communication, RKDF Institute of Science &
Technology Bhopal, Madhya Pradesh, INDIA.
