Fingerprint Authentication System Using A Low-priced Webcam

International Conference on Data Management (ICDM2008), IMT Ghaziabad, India, Feb. 25-26, 2008

Fingerprint Authentication System using a low-priced Webcam Md. Rajibul Islam, Md. Shohel Sayeed, Andrews Samraj Faculty of Information Science and Technology (FIST) Multimedia University, Jalan Ayer Keroh lama, 75450 Melaka, Malaysia E-mail: {md.rajibul.islam05, shohel.sayeed, andrews.samraj}@mmu.edu.my Abstract A number of biometric techniques have been proposed for personal identification in the past. Among the vision-based ones, we can point out fingerprint, face, palm, ear, iris and retina recognition. Voice recognition or signature verification are the most widely known among the non-vision based ones. Signature verification requires the use of electronic tablets or digitizers for on-line capturing and optical scanners for on-line adaptation. These interfaces have some negative aspect that they are large and convoluted to use, increasing the intricacy of the whole identification system. On the other hand, scanners and cameras are much smaller and easy to handle, and are becoming all over in the current computer atmosphere. Lots of vision-based biometric techniques have been projected in the past for personal identification. We present a novel one based on visual capturing of fingerprints using a Webcam. However, there is an open issue to use webcam in stead of any scanner because of the low price, available in the market and easy to adjust anywhere. In this paper, we describe our implementation of the fingerprint authentication system using webcam having Pentium IV CPU, 256 RAM and a piece of transparent glass and a light source. Also, we describe a preprocessing technique based on gamma manipulation and gamma correction that can be executed to adjust lightness and intensities of the fingerprint image before fingerprint image enhancement and feature extraction. Key words: webcam, fingerprint, gamma manipulation, gamma correction, fingerprint authentication. 1.0 INTRODUCTION Because of ridge direction and minutiae such as ridge endings and ridge bifurcations are used for matching, so the performance of automatic fingerprint matching systems depends on local ridge characteristics. The ridges can be easily detected and minutiae can be correctly extracted in an ideal fingerprint thinned image. However, the quality of many fingerprints is often poor due to the injured part on the skin and the atmosphere in which it was taken. Also the quality of fingerprints is very poor which are captured by a low priced webcam. The ridge formations in these poor-quality fingerprint images are not well defined and minutiae cannot be correctly detected. Therefore, a discrete ridge structure is necessary to assurance robust minutiae detection in spite of image quality. As such, the goal of this research is to improve the clarity of ridge structures of a poor fingerprint image captured by webcam to assist the correct extraction of minutiae. Enhancing fingerprint images facilitate matching is a problem that has been much studied [1] [2] [3]. Minutiae extraction from fingerprint images is one of the most important steps in automatic fingerprint identification and classification. Minutiae are local discontinuities in the fingerprint pattern, mainly terminations and bifurcations [1]. Fingerprint image quality is an important factor in the performance of Automatic Fingerprint Identification Systems (AFIS). It is used to evaluate the system performance, assess enrollment acceptability, and evaluate fingerprint sensors. [2]

689

International Conference on Data Management (ICDM2008), IMT Ghaziabad, India, Feb. 25-26, 2008

A number of techniques to enhance fingerprint images have been proposed, which take advantage of ridge characteristics such as directionality. In this paper, we have presented a preprocessing system using gamma manipulation and gamma correction methods which adjust lightness and intensities of the poor quality fingerprint image. The rest of the paper is prearranged as, in section 2 we present an overview of the proposed system and also our contribution in this paper specially gamma manipulation and gamma correction. In section 3 we show the experiments, results and discussion and finally section 4 concludes the paper. 2.0 OVERVIEW OF OUR PROPOSED SYSTEM Fig. 1 shows our proposed block diagram of the whole authentication system and the experimental setup. The preprocessing stage performs the initialization of the algorithm, i.e. it captures a colorful low resolution fingerprint image and convert it to grayscale image and performs the gamma manipulation and gamma correction to adjust lightness and intensities of the fingerprint image. Then feeds it into the next block. The fingerprint enhancement block has the task of enhance the fingerprint on each impression of each user by using the code loosely follows the approach presented by P.D. Kovesi [4]. Just before feature extraction a thinning process needs to be performed as indicated in [5]. In which two tests are run one after the other until none of them discover pixels that need to be removed. However, this method did not meet the requirements imposed to a thinning algorithm because it still left some spurious structures that did not permit a single point inside a line to have only two neighbors, a ridge-end only one, and a bifurcation three. The conformance to the established criteria was obtained by the creation of a third test to be run once after the former two are passed to test for certain conditions in matrices of 3x3 pixels that indicate a spurious structure that shall be properly modified. The minutiae extraction process, defined in [6], uses matrices of 3x3 pixels to search for typical minutiae, that is: ridge endings and ridge bifurcations. After extraction minutiae the extracted data stores to the system database. Finally for the matching process, the live extracted data is to be compared with the extracted data stored in the system database. (a) Enrolment Process (b) Verification Process

Glass

System Database

Feature extraction

Webcam

Fingerprint enhancement

Preprocessing stage Yes/ No

Matching

Fig. 1: (a) Block Diagram of the Enrolment system. The webcam captures a fingerprint impression to the preprocessing stage. This block initializes the algorithm and selects the grey scale fingerprint to perform the enhancement. The enhanced fingerprint obtains the position of the minutiae in the feature extraction block and

690

International Conference on Data Management (ICDM2008), IMT Ghaziabad, India, Feb. 25-26, 2008

finally stored the extracted data in system database block. (b) Block Diagram of the verification system. Like the enrolment system after successfully satisfied all the blocks until Feature extraction, live feature data will verify with the feature data stored with the system database. 2.1 Our Contribution Our proposed scheme is the most automatic and unobtrusive security device. We use a webcam attached to the computer and applies fingerprint recognition to analyze the fingerprint it sees, matching it against the minutiae of the fingerprint belonging to the authorized user. When the proposed protocol ready to capture fingerprint using USBconnected webcam, we tried both, a relatively high end Creative unit and a low end Logitech QuickCam and didn't notice any difference in performance. It seems that the fingerprint recognition process does not depend on a particularly high resolution image to identify the necessary features which we have presented below in this section. On the other hand, we did notice a difference when changing lighting conditions. When the light source is behind the fingerprint, our proposed scheme has more difficulty identifying the features than when the lighting is in front or to the side. Another problem is to get the whole minutiae as well as ridges and valleys of the fingerprint as shown in Fig 2.

(a)

(b)

Fig. 2: (a) Fingerprint captured by webcam without using transparent glass. (b) Enhanced fingerprint using the code loosely follows the approach presented by Peter Kovesi [4]. To overcome this problem we have used a piece of clear transparent glass. The user has to press his finger on the glass and from the opposite side the fingerprint will be captured. And between the webcam and glass we have used light source lighting to the side because if the lighting is in front of the glass it reflects and captured a shadow with the fingerprint image. The lightness of the fingerprint images captured by low-priced webcam is so bright and blur, and for that it doesn’t give better result to enhance the ridges of the fingerprint image. That’s why we’ve used a preprocessing stage in our system to perform gamma manipulation and correction to adjust lightness and intensities of the fingerprint image. The outcomes are shown in Fig. 5 and Fig. 6. 2.2 Gamma Manipulation The fingerprint images captured by webcam are not gamma corrected. During the preprocessing stage when the image processing operations are performed on color fingerprint

691

International Conference on Data Management (ICDM2008), IMT Ghaziabad, India, Feb. 25-26, 2008

images, it is normal that the production of out-of-gamut pixels is not prevented. The gamut mapping may reduce the effect of the image processing algorithm [8]. In this paper we propose a standard method that allows lightness processing on grey image without exceeding the limits of the gamut of the technique. In such exercise it will be asked to display gray scale images in Matlab. In that case if P is an image that takes on the values [0,1,……..,255], then it may be displayed by using the following commands. image(P+1); axis('image'); graymap = [0:255; 0:255; 0:255]'/255; colormap(graymap); In this technique the lightness processing is a function of the color of the pixel to change the desired lightness and a maximum and minimum lightness per pixel. This maximum and minimum depend on the position of the pixel in the gamut of the fingerprint image and the relation between the lightness change and the chroma change. The hue of all pixels is kept constant. The proposed method can be used to apply a variety of grey value algorithms on color fingerprint images. In this section we show the results for contrast improvement using gamma manipulation. The effect of a gamma manipulation is that the lightness values are distributed nonlinearly over the range that is used. This may increase the contrast in one or more regions of the lightness range, at the cost of decreasing the contrast in other regions. In fig. 3 the most common form of gamma manipulation is shown, mathematically described by: * out

L

* min

=L

* max

+ (L

* min

−L

 L* − L*min ) *  * *  Lmax − Lmin

γ

  …………………….(1) 

L* and L*out are the input and output lightness and L*min and L*max are the minimum and maximum of the lightness range. When this manipulation is used with γ >1, the higher (lighter) lightness range gains more contrast, at the expense of the contrast of the darker colors. At the same time the mean lightness is decreased, i.e. all new colors are darker than the original colors. When γ < 1, the opposite occurs (more contrast in the darker colors, less contrast in the lighter colors, and mean lightness increases).

692

International Conference on Data Management (ICDM2008), IMT Ghaziabad, India, Feb. 25-26, 2008

Fig. 3: Gamma manipulation. The result of equation 1 is given for three different gamma values. It can be seen that for γ < 1 the lightness of the image ( L*out ) is always higher than for the original image ( L*in ), and that darker colors have more contrast. For γ > 1 the opposite holds true. 2.3 Gamma Correction We’ve used gamma correction to compensate for the nonlinear behavior of a displayed fingerprint image. Most often images are already encoded in gamma corrected form when anyone using a high quality Digital camera, and will appear fine when displayed on most video monitors but for the fingerprint image captured by webcam has to encoded in gamma corrected form before using the enhancement algorithm to obtain better result. However, if a fingerprint image is stored with a linear scaling it becomes necessary to correct the image. If the value of gamma for the webcam is known, then the correction process consists of applying the inverse of equation (2). γ

 n  m = 255  .............................(2)  255  Where n is the original pixel value and m is the pixel intensity as it appears on the display. This relationship is illustrated in Fig. 4. The fingerprint images captured by webcam, especially during the gamma manipulation they are not corrected for the nonlinear relationship between pixel value and displayed intensity that is typical for a webcam. This nonlinear relationship is roughly a power function, i.e. displayed_intensity = pixel_value^gamma.

693

International Conference on Data Management (ICDM2008), IMT Ghaziabad, India, Feb. 25-26, 2008

Fig. 4: approximate curve to show the intensity response over pixel value This is an approximated curve to show how the intensity response of a fingerprint image captured by webcam is non-linear. Bright colours tend to be displayed too bright. This can be corrected. The process of adjusting the intensities to look correct is known as Gamma Correction. The amount of Gamma Correction we shall call G is usually greater than 1. The range of displayable intensities, P, is between 0 and 1. The formula is thus: pixel = P ^ (1/G) ………………(3) A G value of 1 gives no Gamma Correction. Higher values give more correction. Because values of P must be between 0 and 1, it will have to divide the intensity by the maximum displayable intensity, perform the Gamma Correction, and then multiply up again. pixel = ((p / MaxIntensity) ^ (1/G)) * MaxIntensity ………………(4)

(a) (b) (c) (d) Fig. 5: (a) Image captured by webcam, (b) Grayscale conversion, (c) Shown the results of gamma manipulation and inverting the gamma correction, (d) Enhance part of fingerprint-c.

694

International Conference on Data Management (ICDM2008), IMT Ghaziabad, India, Feb. 25-26, 2008

(a) (b) Fig. 6: (a) Image enhancement before preprocessing, (b) Image enhancement after preprocessing

3.0 EXPERIMENTS AND DISCUSSIONS We conduct experiments with data of fingerprint verification competitions, to demonstrate the advantages of our proposed approach to fingerprint verification using low-priced webcam.

3.1 Datasets We used an improved fingerprint matching approach using TSVM [7] in order to prove the influence of different image qualities and image amount to our proposed approach, we have collected 5 datasets and within these four datasets from FVC2004 (The Second International Fingerprint Verification Competition) and one dataset which is obtained using our proposed scheme. The information of each dataset is shown in Table 1. Each fingerprint image allows a rotation angle that belongs to [−π/4, π/4] (compared with the vertical line). Every two images from one finger have an overlap of common region. But there may be no delta points or core points in some fingerprint images.

3.2 Experiments Setup We posed 2 experiments. For each experiment, we compared the FAR and FRR of our webcam database with the rest of 4 database which are taken from FVC2004 using TSVM. Both the experiments are done by the method of 5-folder cross validation, but have differences in the size of test sets and training sets.

Experiment 1. For database 1 to database 4, 880 images are divided into 5 parts, each of which has 176 images. The algorithm TSVM runs five times. For each time, four of the five parts are used as training sets (our approach only), and the other one part is used as test set. The averaged verification result will be reported over these 5 times. Experiment 2. For database 5, 1200 images are divided into 5 parts, each of which has 240 images. Then the algorithm TSVM runs again five times. For each time, one of the five parts

695

International Conference on Data Management (ICDM2008), IMT Ghaziabad, India, Feb. 25-26, 2008

is used as training set (our approach only), and the other four parts are used as test sets. The averaged verification result will be reported over these 5 times. 3.3 Measures The performance of our proposed fingerprint authentication system can be measured by FRR (False Rejection Rate: each sample in the subset A is matched against the remaining samples of the same finger), FAR (False Acceptance Rate: the first sample of each finger in the subset A is matched against the first sample of the remaining fingers in A). The configuration of running computer is Pentium IV CPU, 256 RAM.

3.4 Results

1st DB 2nd DB 3rd DB 4th DB 5th DB

Table 1. Experimental results of db1 to db5 using TSVM The source of Sensors Image size Resolution FAR the datasets FVC2004 DB1 Optical sensor 640 x 480 500 dpi 0.064% FVC2004 DB2 Optical sensor 328 x 364 500 dpi 0.094% FVC2004 DB3 Thermal 300 x 480 512 dpi 0.057% sweeping sensor FVC2004 DB4 SFinGe v3.0 288 x 384 500 dpi 0.059% Collected using Webcam 640 x 480 450 dpi 0.042% proposed approach

FRR 1.98% 8.87% 6.18%

5.77% 3.92%

The experimental results of db1 to db5 are shown in Table 1. We see that our proposed authentication system using a low priced webcam really can achieve much better accuracy which we compared by the FAR and FRR rate using the matching approach TSVM. As shown in Table 1, fingerprints of the five datasets are captured by sensors of different types. So the images have different qualities. This strongly suggests that our proposed scheme as well as preprocessing system, image enhancement, feature extraction and TSVM methods capture well the information needed for fingerprint verification, and have a low influence by fingerprint image quality. We see in experiment 2, that although the proportion of training sets is reduced, and the number of test members is increased in db5, our proposed approach using a low priced webcam as a sensor still works better than the authentication system using expensive sensors which are available in the market. This implies that this approach has a low influence by fingerprint image amount. Comparing the experimental results of all the other datasets with webcam datasets using our approach, it turns out that the transductive learning technique can provide some help to fingerprint authentication system. We think this is because our preprocessing phase of the proposed scheme makes effective use of the matching vectors to enhance classification and to derive a threshold selection range for better accuracy. The experimental results of our webcam datasets obtained by the proposed approach are compared in Table 1. It is clear that the accuracy of the webcam datasets using our proposed approach outperforms with the other datasets consistently and significantly.

4.0 CONCLUSION In this paper, a fingerprint preprocessing approach using a webcam was proposed. Since the characteristics of fingerprint images acquired with a webcam are quite different from those

696

International Conference on Data Management (ICDM2008), IMT Ghaziabad, India, Feb. 25-26, 2008

acquired by conventional touch-based sensors, a new fingerprint preprocessing algorithm was used. The main contributions of this paper are: First, we introduced a fingerprint acquisition model incorporating gamma manipulation and gamma correction during preprocessing that can be executed to adjust lightness and intensities of the fingerprint image before fingerprint image enhancement and feature extraction. We have found that this clearly outperforms defining the features from a very low quality fingerprint captured by a low priced webcam. Second, we used a fast TSVM matching algorithm to estimate the feature locations, which will evaluate the performance of our proposed approach by the FAR and FRR rate of the webcam datasets with the datasets obtained by other sensors.

REFERENCE [1] Greenberg, S.; Aladjem, M.; Kogan, D.; Dimitrov, I. “Fingerprint image enhancement using filtering techniques”, Proceedings. 15th International Conference on Pattern Recognition. Vol.3, pp.322 – 325, 2000. [2] Chaohong Wu, Sergey Tulyakov and Venu Govindaraju, “Image Quality Measures for Fingerprint Image Enhancement”, Lecture Notes in Computer Science. Springer Berlin / Heidelberg. Vol. 4105, pp. 215-222, 2006 [3] O'Gorman, L. Nickerson, J.V. “Matched filter design for fingerprint image enhancement” International Conference on Acoustics, Speech, and Signal Processing, ICASSP-88., vol.2, pp.916-919, 1988. [4] P.D. Kovesi. Matlab functions for computer vision and image analysis. http://www.csse.uwa.edu.au/~pk/Research/MatlabFns/index.html. [5] T. Y. Zhang , C. Y. Suen, “A fast parallel algorithm for thinning digital patterns”, Communications of the ACM, vol.27 n.3, p.236-239, March 1984 [6] C.Arcelli and G.S.D.Baja, “A Width Independent Fast Thinning Algorithm,” IEEE Trans. Pattern Analysis Machine Intelligence, vol. 7, no. 4, pp. 463-474, 1984. [7] Jia Jia, Lianhong Cai, “A TSVM-Based Minutiae Matching Approach for Fingerprint Verification.” Lecture Notes in Computer Science. Springer Berlin / Heidelberg. IWBRS 2005, vol. 3781, pp. 85-94, 2005. [8] J. Dijk and P.W. Verbeek, “Lightness Filtering in Color Images with Respect to the Gamut”, CGIV 2006, Proc. Third European Conference on Colour in Graphics, Imaging, and Vision (University of Leeds, UK, June 19-22), 2006, pp. 330-335.

697