Fingerprint Enhancement and its features purification

1 downloads 0 Views 2MB Size Report
B.G. Sherlock, D.M. Monro, and K.Millard, ―Fingerprint enhancement by ... L.Hong, A.K Jain, S Pankanti, and Ruud Bolle, ―Fingerprint enhancement‖,.
JOURNAL OF COMPUTING, VOLUME 4, ISSUE 5, MAY 2012, ISSN 2151-9617 https://sites.google.com/site/journalofcomputing WWW.JOURNALOFCOMPUTING.ORG

78

Fingerprint Enhancement and its features purification Raju Rajkumar, Th. Churjit Meetei, Shahin Ara Begum, K Hemachandra Abstract- It is difficult to extract only genuine minutiae from fingerprints. Enhancement techniques are being used as preprocessing methods for minutiae extraction. The fingerprint enhancement is a challenging process. To overcome the adverse effect caused by spurious minutiae in fingerprint matching, a new method for fingerprint enhancement is proposed. Since some of the spurious minutiae in the boundary region cannot be remo ve in the minutiae purification, we draw a region of interest in the fingerprint image to remove the remaining boundary minutiae which exists as a ridge ending. These boundary minutiae affect the accuracy in fingerprint matching. Experimental result shows that the proposed method can eliminate the effect caused by spurious minutiae. Index Term- Boundary minutiae, genuine minutiae, spurious minutiae, fingerprint enhancement, FAR, FRR.

——————————  ——————————

1 INTRODUCTION The fingerprint has many particular properties such as uniqueness, stableness, and inseparability from the host. It has been used for personal verification for more than ten decades [1], and is the most widely used biometrics in day today life. It can be represented by a large number of features, including the overall ridge flow pattern, ridge frequency, location and position of singular points, location of minutia points, ridge counts between pairs of minutiae, and location of pores [2]. All these features contribute to fingerprint individuality. In this study, minutia base representation of the fingerprints is chosen because it has been utilized by forensic experts, and has been adopted by most of the commercially available Automated Fingerprint Identification Systems (AFIS). There are two types of minutiae which are commonly used in AFIS, ridge ending and ridge bifurcation as shown in Fig 1.

a) Ridge bifurcation b) Ridge Ending Fig 1. The most common type of Fingerprint feature

These minutiae are totally depend on the input image. Sometimes the input images are very noisy. It is necessary to

————————————————

   

Raju Rajkumar is with the Computer Science Department, Assam university, Silchar, India. Th. Churjit Meetei is with the Computer Science Department, Assam University, Silcha, India. Shahin Ara Begum is with the Computer Science Department, Assam University, Silchar, India. K Hemachandran is with the Computer Science Department, Assam University, Silchar, India.

enhance the input fingerprint image inorder to extract reliable minutiae. A number of enhancement algorithms are available in the literature. O’ Gorman and Nickerson [3] have proposed different filters for the fingerprint image enhancement and the k X k mask coefficients are generated, based on the local ridge orientation. Only three orientation directions are used. The four model parameters derived from ridge width (Wmax, Wmin), valley width (Ẃmax, Ẃmin), and the minimum radius of curvature are used to describe a fingerprint. It is assumed that the Wmax + Wmin = Ẃmax +Ẃmin . The mask is convoled with the input image. The enhanced image is binarized and post processed. Mehtre [4] computes the directional image, representing the local ridge direction, in a block of size 16 X 16 pixels. For this purpose, local gray level intensity variances along eight different directions are computed. The direction with the least variance is the desired least direction. A set of eight 7 X 7 convolution masks is applied to the input image for ridge enhancement. The fingerprint area is segmented from the background before applying standard locally adaptive thresholding and thinning operators. Features are obtained based on the computation of the connection number (CN) described in [3]. Sherlock et. al. [5] have studied the enhanced fingerprint images by using a directional Fourier filter. The direction of the filtering is decided by the local ridge orientation. A 32 X 32 window is used to obtain a projection of the pattern in 16 directions. The projection with the maximum variance is the desired ridge direction for the window. The result of the enhancement is compared with feature extraction techniques used in a system by the UK Home office. Hong et. al. [6] introduced a new fingerprint enhancement algorithm which decomposes the input fingerprint image into a set of filtered images. A set of band pass filters can efficiently remove the undesired noise and preserve the true ridge/valley structure. Gabor filters [7] have both frequency –selective and orientation-selective properties and have optimal joint resolution in both spatial and frequency domains. Therefore, it is beneficial to use Gabor filters as bandpass filters to remove the noise and preserve true ridge/valley structure. One of the heuristics to detect the

JOURNAL OF COMPUTING, VOLUME 4, ISSUE 5, MAY 2012, ISSN 2151-9617 https://sites.google.com/site/journalofcomputing WWW.JOURNALOFCOMPUTING.ORG

spurious minutiae resulting from these crack, is based on the observation that these minutiae are anti-aligned and the region between them is brighter than the average brightness of the foreground region. Hong et. al.[8] improved the computationally expensive enhancement method [6], which makes it unsuitable for an online verification system. The authors have presented a fast enhancement algorithm to enhance adaptively, the ridge and furrow structures using both the local ridge orientation and local frequency information. Yang et. al.[9] modified the method proposed by Hong et. al. [8] by discarding the inaccurate prior assumption of sinusoidal plane wave, and making the parameter selection process independent of fingerprint image. Rajkumar and Hemachandran [15] used both FFT and Gaussian filter for fingerprint enhancement in preprocessing stage. FFT have 32 X 32 pixel block size and Gaussian filter gives little blur by giving σ =0.5 which used in removing the hairy structure when it is thinning.

2 PROPOSED ALGORITHM An automatic fingerprint identification system consists of various processing stages. The overall process of the proposed algorithm can be divided into four main operations: 2.1 Image enhancement 2.2 Binarization and thinning 2.3 Feature extraction and its purification 2.4 Demarcation of boundary feature

2.1. Fingerprint Enhancement In this study two methods are used for Image enhancement process: (i) Normalization and (ii) Fast Fourier Transformation. Fingerprint image

Image enhancement

Binarization and Thinning

Feature extraction & purification

Taking ROI

79

𝑀𝑂 +

VAR O I i,j −M 2

𝑖𝑓 𝐼 𝑖, 𝑗 .

𝐺 𝑖, 𝑗 = 𝑀𝑂 −

Fig 2. Proposed algorithm for Fingerprint Identification System

(i) Normalization Normalization is done so that the gray level values lies within a given set of values. The fingerprint image is normalized to have a predefined mean and variance. The normalization is required as the fingerprint image usually has distorted levels of gray values among the ridges and furrows. This is a pixelwise operation although it does not change the ridge and furrows structure. Let I(i,j) denotes the gray-level value at pixel (i,j), M and VAR denote the estimated mean and variance of I, respectively, and G(i,j) denote the normalized gray-level value at pixel (i,j).The normalized image is defined as follows [ 9]:

>𝑀 (1)

𝑉𝐴𝑅 𝑂 𝐼 𝑖,𝑗 −𝑀 2 𝑉𝐴𝑅

𝑖𝑓 𝐼 𝑖, 𝑗

,

≤𝑀

where MO and VARO are the desired mean and variance values, respectively. Normalization is achieved by histogram equalization. It increases the local contrast in an image. Thus the intensities can be distributed on the histogram. This allows for areas of lower local contrast to gain a higher contrast without affecting the global contrast. Histogram equalization accomplishes this by effectively spreading out the intensity values. (ii) Fast Fourier Transformation The primary enhancement is done through Fourier transformation. Initially the fingerprint image is divided into small processing blocks (32 X 32 pixels) and the Fourier transformation is performed according to the following equation: 𝑀−1 𝑁−1

𝐹 𝑢, 𝑣 =

.

. 𝑓 𝑥, 𝑦 ∗ exp −𝑗2𝜋 ∗

𝑥 =0 𝑦=0

𝑢𝑥 𝑣𝑦 + 𝑀 𝑁

(2)

for u = 0,1,2,….,31 and v = 0,1,2,….,31. In order to enhance a specific block by its dominant frequencies, the FFT of the block was multiplied by its magnitude a set of times. Where the magnitude of the original FFT = |F(u,v)|. The enhanced block is obtained according to g x, y = F −1 F u, v ∗ |F u, v |K (3) where F-1 (F(u,v)) is done by:

1 𝑓 𝑥, 𝑦 = 𝑀𝑁

𝑀−1 𝑁−1

𝐹 𝑢, 𝑣 𝑥=0

𝑦=0

𝑢𝑥 𝑣𝑦 + 𝑀 𝑁 for x = 0, 1, 2, ..., 31 and y = 0, 1, 2, ..., 31. ∗ exp 𝑗2𝜋 ∗

Fingerprint feature

,

VAR

(4)

The k in (3) is an experimentally determined constant, as k = 0.45. While having a higher "k" value improves the appearance of the ridges, by filling up small holes in ridges and having a too high "k" value can result in false joining of ridges. Thus a termination might become a bifurcation.

2.2. Binarization and Thinning This process is used to convert the gray scale image to two bit image (black and white) and after the ridge pixel is eliminate to one pixel wide. The details of this process are described below: (i) Binarization The Fingerprint image binarization is to transform the 8-bit gray fingerprint image to a 1-bit image with 0 value for ridges and 1 value for furrows. After the operation, ridges in

JOURNAL OF COMPUTING, VOLUME 4, ISSUE 5, MAY 2012, ISSN 2151-9617 https://sites.google.com/site/journalofcomputing WWW.JOURNALOFCOMPUTING.ORG

the fingerprint are highlighted with black color while furrows with white. We implement a threshold value 100 for transforming a pixel value to 1 when the value is larger than the threshold and 0 for vice versa. (ii) Thinning Ridge thinning is to eliminate the redundant pixels of the ridges till the ridge is just one pixel wide. No further removal of the pixels should be possible after completion of thinning process. The proposed algorithm used the built-in Morphological thinning function in MATLAB environment The objective of a good thinning algorithm are: 1. To obtain a thinned fingerprint image with a single pixel width and no discontinuities. 2. To eliminate the noise and singular pixels.

2.3. Feature extraction and its purification There are two types of feature in the fingerprint identification system: Minutiae based approach and Image based approach. Minutiae based approach is the most common method used in fingerprint identification system. Many number of spurious minutiae are present while extraction these minutiae. These spurious minutiae are needed to remove for higher accuracy of matching. The method of minutiae extraction and remove spurious minutiae are explained below: (i) Minutiae Minutiae are extracted by identifying a pixel value in the ridge orientation flow. If the central is 1 and has only 1 one-value neighbor, then the central pixel is a termination and if the central is 1 and has 3 one-value neighbor, then the central pixel is a bifurcation. If the central is 1 and has 2 one-value neighbor, then the central pixel is a usual pixel. (ii) Remove spurious minutiae The minutiae extracted in the above section includes many spurious minutiae. These spurious minutiae will significantly affect the accuracy of matching if they are simply regarded as genuine minutiae. A threshold D is used to eliminate the spurious minutiae under the following conditions: 1. if the distance between a termination and a bifurcation is smaller than D. 2. if the distance between two bifurcations is smaller than D 3. if the distance between two terminations is smaller than D The experimental value of D in the proposed algorithm is 6. 2.4. Demarcation of boundary feature Some spurious minutiae may still available in the boundary of the ridge orientation flow as shown in fig. 3 (a). These spurious minutiae (i.e. boundary minutiae) cannot be eliminated by using threshold D. So it is required to draw a region of interest (within the ridge orientation) for eliminating boundary minutiae. We use the ROI tools of MATLAB, to draw the ROI. Because of the different shape and size of the fingerprints, it is difficult to have a suitable robust algorithm for the region of interest. The region of interest as shown in Fig 3(b) is drawn manually. Once we defined the ROI, we can eliminate the boundary minutiae which are located outside the ROI. The image after elimination of boundary minutiae is shown in Fig 3 (c).

80

(a)

(b)

(c) Fig 3 a) before suppression of external minutiae b) ROI of the given image c)After suppression of external minutiae.

3 EXPERIMENTAL RESULT The algorithm which is described in the previous section has been implemented and tested in MATLAM 7.3.0. The fingerprint images, used in this study are taken from the standard Fingerprint Verification Competition (FCV) 2004 database[14] which contains hundreds of fingerprint image. The execution time of each process is shown in Table 1. Image no 1_4 3_3 4_5 5_4 6_3 7_6 9_3 10_6 11_1 13_2

Fingerprint Enhancement

Binarization & thinning

0.333 0.448 0.392 0.416 0.393 0.397 0.389 0.334 0.414 0.388

0.625 0.574 0.771 0.779 0.742 0.774 0.903 0.757 0.859 0.645

Feature extraction & purification 2.732 3.119 3.176 2.905 2.583 2.698 3.012 3.552 3.18 2.986

Table 1: Execution time (in seconds) of the proposed algorithm.

The calculation of execution time does not include the drawing of region of interest as it is drawn manually. The average execution time is 3.9 seconds. The performance of verification system is tested using False Acceptance Rate (FAR) and False Reject Rate (FRR). FAR is the percentage of imposter fingerprints accepted by the system, and FRR is

JOURNAL OF COMPUTING, VOLUME 4, ISSUE 5, MAY 2012, ISSN 2151-9617 https://sites.google.com/site/journalofcomputing WWW.JOURNALOFCOMPUTING.ORG

a)

Input image

c) Fast Fourier transformation

e) Thinning

g) elimination of spurious minutiae

81

b) Normalization

d) Binarization

f) Minutiae extraction

h) elimination of boundary minutiae

Fig 4. The results of different steps of the proposed algorithm

JOURNAL OF COMPUTING, VOLUME 4, ISSUE 5, MAY 2012, ISSN 2151-9617 https://sites.google.com/site/journalofcomputing WWW.JOURNALOFCOMPUTING.ORG

Image no 1-4 3-3 4-5 5-4 6-3 7-6 9-3 10-6 11-1 13-2

Total Minutiae 466 805 1083 682 384 325 772 1539 911 901

Purified Minutiae 64 71 56 62 59 42 85 78 67 72

Missed Minutiae 9 20 11 12 7 13 9 11 11 8

82

Successful Fraud Minutiae 15 22 23 19 14 10 16 33 11 38

FAR(n)

FRR(n)

0.03 0.03 0.02 0.03 0.04 0.03 0.02 0.02 0.01 0.04

0.019313 0.024845 0.010157 0.017595 0.018229 0.04 0.011658 0.007147 0.012075 0.008879

Table 2: FAR and FRR for different 10 fingerprint image.

the percentage of genuine fingerprints rejected by the system. The performance evaluation of minutiae extraction for the proposed algorithm of different twenty fingerprint images are shown in Table2. This table contains total minutiae, purified minutiae, missed minutiae and successful fraud minutiae. The average ratio of False Acceptance Rate (FAR) and False Rejection Rate(FRR) are 1.3% and 2.8% respectively. The equation to calculation of False Acceptance Rate and False Rejection Rate are shown below: FAR = FRR =

i FAR (i)

[2]

[3]

[5]

i FRR (i)

(6)

N

Where N is the total no of images. FAR(i) =

Sucessful Fraud Minutia Total no of minutia

FRR(i) =

Missing Minutia Total no of minutia

[6] [7] [8]

The comparison of different algorithm with respect to FAR and FRR is shown in Table 3. The experimental results of different processing stage are shown in fig4. Method

[1]

[4]

(5)

N

References

FAR

Wuzhili’s algorithm [10] 2.5 Zhu’s algorithm[16] 1.8 Proposed algorithm 1.3 Table 3: Comparison of

FRR 3.5 2.9 2.8 algorithm

Execution time(Sec) 1.9 2.7 3.9

4 CONCLUSION The traditional minutiae purification in fingerprint enhancement concentrate only in h-breaks and spur, while the elimination of h-break and spur are not sufficient for minutiae purification. We found another spurious minutiae i.e. boundary minutiae(ridge ending) which effect the accuracy of matching. The experimental results shows that the proposed algorithm gives more accuracy in minutiae purification.

[9]

[10] [11]

[12] [13] [14] [15]

[16]

B.V. Subramanyam Modies. ―Medical jurisprudence and Toxilology‖, 22nd Edition. Rashtriya Printer 27/487, Zulfe Bengal, Dishad Garden, 1938. N. Ratha, S. Chen, and A. K. Jam, "Adaptive flow orientation based feature extraction in fingerprint images" Pattern Recognition, vol. 28(11), 1995, pp. 1657-1672. O’Gorman and J.V. Nickerson, ―An approach to fingerprint filter design‖, Pattern Recognition, vol. 22, No. 1, 1989, pp. 29-38. B.M. Mehtre, ―Fingerprint image analysis for automatic identification‖, Machine Vision and Applications, vol. 6, 1993, pp. 124-139. B.G. Sherlock, D.M. Monro, and K.Millard, ―Fingerprint enhancement by directional Fourier filtering‖, IEE Proc. Vis Image Signal Processing, vol. 141(2), April 1994, pp. 87-94. L.Hong, A.K Jain, S Pankanti, and Ruud Bolle, ―Fingerprint enhancement‖, IEEE vol. 5,1996, pp. 202-207. D. Gabor, ―Theory of communication‖, Journal of IEE, vol. 92, 1946, pp. 429-457. L. Hong, Y. Wan, and A. K. Jain, ―Fingerprint image enhancement: algorithm and performance evaluation", IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 20(8), 1998, pp. 777-789. J. Yang, L. Liu, T. Jiang and Y. Fan, ―A modified Gabor filter design method for fingerprint image enhancement‖, Pattern Recognition letter, vol. 24, August 2003, pp. 1805-1817. Wu zhili, ―Fingerprint Recognition‖, Thesis, Hong Kong, Baptist University, 2002. L.C. Jain, U. Halici, I. Hayashi, S.B. Lee, and S. Tsutsui, ―Intelligent biometric techniques in fingerprint and face recognition‖, The CRC Press, 1999. D. Maltoni, D. Maio, A. K. Jain, and S. Prabhakar, ―Handbook of Fingerprint Recognition‖, Springer, New York, 2003. R.C Gonzales & R.E. Woods, ―Digital Image Processing‖. New York, USA: Addison-Wesley, 1993. FVC2004 http://bias.csr.unibo.it/fvc2004 R. Rajkumar, K. Hemachandran, ―A Secondary Fingerprint Enhancement For Identification System ―, Journal Of Computer Science And Engineering, Vol 6(1), March 2011, pp 9-17. N Zhu, R H Shi, K H Wu, ―A New Fingerprint Matching Method of Minutiae‖, Journal of Computer Engineering and Applications, vol 5, 2006, pp. 74-76