Co-occurrence Matrix Features for Fingerprint Verification Mohammed S. Khalil, Muhammad Khurram Khan, Muhammad Imran Razzak Center of Excellence in Information Assurance, King Saud University, Saudi Arabia sayimkhalil@ksu.edu.sa, mkhurram@ksu.edu.sa, mirpak@gmail.com Abstract—In this paper, an enhanced image-based fingerprint verification algorithm is presented that improves matching accuracy by overcoming the shortcomings of previous methods due to poor image quality. It reduces multi-spectral noise by enhancing a fingerprint image to accurately and reliably determine a reference point and then extracts a 129 X 129 block, making the reference point its center. From the 12 co-occurrence matrices, four statistical descriptors are computed. Experimental results show that the proposed method has more accurate and performance than other methods the average false acceptance rate (FAR) is 0.48% and the average false rejection rate (FRR) is 0.18%. Keywords: Fingerprint verification, Security, Biometrics. I. INTRODUCTION Biometric-fingerprints are probably the most widely used personal identification tool, as they have been used for many centuries due to their individuality, uniqueness and reliability. A distinctive feature of fingerprints lies in the high degree of difficulty in terms of forgery, along with the fact that fingerprints are unique to each person; this means that fingerprints provide an excellent source of entropy, which makes fingerprinting an excellent candidate for security applications. Users cannot pass their fingerprint characteristics to other users as easily as they do with their cards or passwords [1, 2]. The pattern of the valleys and ridges on human fingertips forms the fingerprint image. Analyzing this pattern at different level reveals different types of features. Methods to extract and match fingerprint features can be classified into three categories: minutiae-based, correlation-based, and hybrid [3]. Minutiae-based techniques attempt to align two sets of minutiae points from two fingerprints and count the total number of matched minutia [4, 5]; the performance of minutiae-based techniques relies on the accurate detection of minutiae points as well as the use of sophisticated matching techniques to compare two minutiae fields that undergo non-rigid transformations. In the correlation-based approach, global patterns of ridges and furrows are compared to determine whether two fingerprints align [6, 7]; the performance of correlation-based techniques is affected by non-linear distortions and noise present in the image. Finally, in hybrid methods, local orientation and frequency, ridge shape, and texture information are used to extract fingerprint features [8, 9]; the robustness of hybrid methods is affected by the difficulty of detecting all minutiae. Moreover, the computational requirements are very high. Recently, there has been research on using co-occurrence matrices with fingerprinting. Arivazhagan et al. [10] proposed a fingerprint verification method using Gabor wavelets and co- occurrence matrices to obtain a fingercode. Yazdi et al. [11] proposed a fingerprint classification based on co-occurrence matrices. This paper proposes a new method to verify an enhanced fingerprint image using four descriptors to characterize a co- occurrence matrix. The paper is organized as follows. In section 2, the proposed method is presented. It includes an algorithm for enhancing an image by using the estimated ridge frequency and applying the Gabor filter, presented in section II.A. In section II.B, the procedure for locating the reference point is discussed; the extraction of the reference point block is explained in section II.C. The Gray-Level Co-occurrence Matrix is described in section II.D. Feature extraction is mentioned in section II.E. Experimental results and conclusions are discussed in sections III and IV. II. PROPOSED METHOD A. Fingerprint Image Enhancement Fingerprint images are not always good quality; in real life, skin condition, sensor noise, and incorrect finger pressure produce low-quality images. In order to have a good quality image, enhancement is used to improve the contrast between ridges and valleys in the fingerprint images Figure 1 shows original images; Figure 2 shows enhanced images. The enhancement method [12] consists of the following steps: 1) Normalization Normalization is a pixel-wise operation. It does not change the structure of the ridges and valleys. The main purpose of normalization is to reduce the variations in gray-level values along ridges and valleys, which facilitates the subsequent processing steps. The normalized image is defined as follows: (, ) = ⎩ ⎨ ⎧ 0 +  0 ((, )−) 2   (, ) >   0 −  0 ((, )−) 2  ℎ  (1) where M0 and VAR0 are the desired mean and variance values, respectively. 2) Estimation of Local Ridge Orientation This paper makes use of the method proposed in [13] to estimate local ridge orientation. The algorithm proceeds as follows.  Divide the normalized image into blocks of size 16 X 16.  Compute the gradients x(i,j) and y(i,j) at each pixel of the smoothed block using 3 X 3 sobel masks.  Compute the dominant ridge direction in the 16 X 16 block using the following equation: ___________________________________ 978-1-61284-632-3/11/$26.00 ©2011 IEEE