American Journal of Engineering Research (AJER) 2016 American Journal of Engineering Research (AJER) e-ISSN: 2320-0847 p-ISSN : 2320-0936 Volume-5, Issue-2, pp-248-253 www.ajer.org Research Paper Open Access www.ajer.org Page 248 The performance evaluation of an algorithm for fingerprint biometric recognition Virtyt Lesha Department of electronics and telecommunications/Polytechnic University of Tirana/Albania ABSTRACT: In this article we have treated the performance of an algorithm designed to achieve the biometric identification through fingerprints. The paper consists in providing the design steps of this algorithm consisting of transformations that are currently being introduced in the system images. The algorithm used is in itself a database that holds a certain number of fingerprint images which are chosen to be included in the database in question. After the introduction of the database, the system is ready to perform the fingerprint identification. This process comprises a series of sub-processes that make up this algorithm. One of the performance parameters of the algorithm is the execution time needed to carry out the identification of various fingerprints. We have used 200 different traces of fingers and we have entered them in the database. Then, for each finger traces we have calculated the time of the execution of milliseconds needed to recognize the fingerprints. This execution time is set in relation to an arbitrary identification number ranging from 1 to 200 and so we have reflected the exponential regression trend-line together with the coefficient of determination. This coefficient has also led to the identification of the limits of this paper. Keywords: algorithm, fingerprint, execution, biometric, regression I. INTRODUCTION The personal identification consists in accompanying a particular individual with a corresponding identity. He plays an important role in our society, [1] where questions relating to the identity of an individual are performed millions of times a day by financial services organizations, health care, in electronic commerce, in telecommunications, government agencies, etc. With a more rapid development of information technology, people are becoming more and more connected with electronics. As a result, the ability to achieve accurate identification of individuals has become critical [2]. A diversity of varieties of systems requires personal and sustainable authentication schemes in order to confirm and identify the identity of persons seeking certain services. The aim of these schemes or systems is to ensure that services are accessible to legitimate users. Examples of these systems include secure access among different buildings, computer systems, mobile phones, ATM, etc [3]. In the lack of robust authentication systems, these systems are imminent. Consequently, the use of biometric parameters was born as a need to confront these threats. II. METHODOLOGY The methodology used in this study consists of two aspects. The first is the fingerprint identification algorithm and the second is the performance of this algorithm discussed in terms of execution time of the algorithm in conjunction with different fingerprints that are part of the database. When a trace finger is added to the database of the application, the algorithm is executed 2 times: the first time for image input and a second time for the image rotated at an suitable angle (22.5 / 2 grade) so that the process becomes as varied in rotation. The rotation of the image is realized through using Matlab program using ―imrotate‖ function. [4] When an image of the fingerprint is added to the database, there is only one core point. On the other hand, when an input image is selected to perform the compliance of the fingerprint, then it will activate a series of fingerprint core and align points shall be carried out for each of them. Finally, it will be taken into consideration only the candidate with the smallest distance.[5] For example, we have 3 images in the database, img1, img2 and img3 where, each of them is characterized only by a core point and therefore will have 3 points each of their core is associated with an image present in the database. If we select an image for fingerprint compliance (should be ―ImgNew‖) we will find a number of core points (let it be their number N). For each of these N