NATIONAL LEVEL TECHNICAL SYMPOSIM REGENCY INSTITUTE OF TECHNOLOGY YANAM

S.R.K.R.ENGINEERING COLLEGE BHIMAVARAM ANDHRA PRADESH

DOOR OPENING SYSTM WITH LOW COST FINGERPRINT SCANNER AND PC USING FINGER PRINT VERIFICATION

AUTHOR : J.M.B.Seshu Kumar, E.C.E, 3rd Year, B.E E-mail:seshukumarjmb@yahoo.co.in

Door Opening System with Low cost Fingerprint Scanner and PC Using Fingerprint Verification

ABSTRACT Biometrics offer improved security for access control. Fingerprints are used for identification to improve the error rates obtained with hand geometry biometrics and costs with iris recognition system. The first step is to obtain an image of the users fingerprint. After this, a pre-processing algorithm is applied, which enables feature extraction to obtain the location and type of all minutiae, i.e., the discontinuities in the ridges and valleys of the fingerprint. Due to the fact that it is not possible to always catch the same number of minutiae, a special matching algorithm should be applied to compare the feature vector with the corresponding users template. Several biometric devices can be adapted to our special requirements and some criterion is needed to choose the best solution ,a standalone embedded system or a low-cost PC-based system can be better fitted to our special requirements, by means of special purpose software can be a good alternative. This paper describes one of these systems for door-opening with the users fingerprint. Special emphasis is given on the design, advantages, and inconveniences with respect to embedded systems.

INTRODUCTION Among all biometric techniques known today, one of the most widely used is the one based on the users fingerprints. Modern fingerprint identification, was established by Sir Francis Galton and Sir Edward Henry in the late nineteenth century. Following these works, fingerprints are considered as a set of ridges separated by valleys. The ridge flow pattern can present local discontinuities, such as terminations and / or bifurcations. These discontinuities are called minutiae, and they represent their type, location, and orientation, from the features that make each fingerprint different.

Figure 1

Sample Fingerprint with both the minutiae

Even, the ten fingerprints, corresponding to each of the ten fingers of the same user, are totally different from each other.Figure1 shows a sample fingerprint with both types of minutiae. Studying those minutiae, on can determine the identify of the subject. New approaches have appeared in order to identify fingerprints without using minutiae information, such as multiresolution analysis or location of pores throughout the surface of the fingertip.

As in any other biometric technique, the system is composed of four different blocks : Capture of the fingerprint Pre-processing , in order to eliminate redundant information and to adapt the sample to the next block requirements. Feature Extraction , where minutiae ( as shown here ), pores or any other information related with the unicity of the fingerprint is obtained. Matching of the features obtained with the template previously computed in the enrollment phase. This matching will provide a percentage of similarity, that will be used to determine whether this user is the same as the enrolled. In this paper, a design for door-opening system is presented. Once a decision has been taken to adopt a biometric solution instead of ( or in combination with ) a handheld token or a knowledge-based system, the next step is to adopt a standalone embedded system or a PC-based system. So far, the former has seemed to be more professional, compact, and low-cost. However, nowadays, the cost of personal computers is considerably low, and there are small low-cost cards available ( PC-on- chip module ). All together is an alternative to be taken into account before making a decision. Table 1 summarizes the main advantages and drawbacks of both solutions. FINGERPRINT CAPTURE Due to the fact that this biometric technique is one of the oldest, a lot of work is being applied to develop sensors that can capture a detailed image of the fingerprint in a comfortable way for the user. Years ago, all sensors were mainly based on optics, even using infrared video cameras. Although reaching high quality images ( over 500 dpi ),

their cost, size, and maintenance needs have pushed industry to look for solutions that can fit into other applications. Therefore, semiconductor sensors based on the capacitance of the ridges in the fingerprint have appeared. This kind of sensor can provide a 500 dpi image, having a size of 15 x 15 mm and a very low power consumption. All this allows sensors to be placed in any kind of portable device. In this work, both kinds of sensors is semiconductor based sensor and optical based sensors have been used without noticing much of a difference in the performance of the biometric algorithms. PRE-PROCESSING AND FEATURE EXTRACTION After capturing an image of the fingerprint, image processing techniques should be applied in order to obtain the unique features of such fingerprint. Here, have used minutiae as these feature. To extract those minutiae, the Ridge Following Algorithm developed by D.Maio and D.Maltoni .

Figure 2 Basics of the Ridge Fallowing Algorithm Before using Maios work, some pre-processing steps should be performed. In

fact, an equalization of the image is applied to take profit of all the diversity of gray tones present. After that, the orientation field, i.e., the low-scale gradient of ridges and valleys, computed, using it to locate the fingerprint inside the image. Then, the maximum rectangle containing only ridges and valleys is extracted and used to detect the minutiae. Finally, this rectangle is stretched in intensity and enters Maios algorithm. The Ridge Following Algorithm is mainly based on an iterative search of ridges inside the image. Once a ridge is found, i.e., the darkest pixel in a local zone, it is followed using the direction given by the orientation field, previously computed, and the darkest pixel in a short line, perpendicular to the orientation given in the first pixel, and translated a determined number of pixels ( ) in that orientation. This algorithm is illustrated in Figure 2. As the ridge is followed, an auxiliary image T is created, substituting the ridge by a polygonal image as shown in figure3..

Figure 3 If the ridge is found to be finished, or of it collides with a previously found ridge, a minutiae is detected, being a termination or bifurcation.

The feature vector extracted is composed of each of the minutiae found, i.e., their location, orientation, type, and for some needs of the matching algorithm, a small segment of the ridge involved. With this new method, many image processing steps involved in typical minutiae extraction algorithms are avoided. The feature vectors obtained are 10 bytes/minutiae. The number of minutiae found depends on the finger scanned and the part of the finger taken after segmentation. MATCHING ALGORITHM The feature vector obtained from the fingerprint sample captured must be compared with the template stored previously in the enrollment phase. As in all other biometric techniques, tolerances and probabilities should be used for such comparison. A direct matching of the features found is not possible due to the variations in the sample taken, the position of the finger, etc. Also, they elasticity of the finger introduces new variations,, such as the distance between ridges, quality of the image due to differences in pressure, etc. Therefore, in fingerprint verification, an elastic matching algorithm should be applied. First a reference minutiae is located in both the template and the sample, and the rest of the minutiae are converted to polar coordinates in respect to that reference minutiae. To select the reference minutiae in both feature vectors, the Euclidean distance of every pair of minutiae in location and orientation is measured. The best match, if it doesnt go above a certain threshold, is taken as the reference minutiae. After that, both feature vectors are aligned, and the distances among all the minutiae are measure, using bounding boxes to adjust the elasticity of the fingers.

DOOR-OPENING SYSTEM BASED ON A PERSONAL COMPUTER Although our actual version is based on a standard PC, the application can be easily transferred to a low-cost PC on chip module, and some of the advantages of the embedded system can be achieved. An example of these systems is the PC-on-chip module. Table 2 summarizes the main features of one of these hardware systems. Anyway, the proposed PC-based system seems to be more suitable for testing the device, and a further step would be to transfer the application to this module.

Table 2. Features of a PC-on-chip module Feature Price CPU RAM Disk Power consumption Weight Size Operating systems Specifications Below 50 USD ( basic configuration ) Up to 400 MHz, 32 + 32 KB cache Up to 64 Mbyte 1-128 Mbyte flash disk Below 1 W 25 grams 66 x 44 x 7 mm Linux, Vx Works and Windows CE

Low-cost fingerprint scanners are designed for a PC environment ,there were no problems due to the sensor, problems may appear in computer ,operating in continuous mode for several days. Taking into account that the door-opening system , install a program, like iCPU shutdown ,which also reduces the power consumption. This program turns on and off the computer at predefined hours.

Table 1. Comparison Between Embedded and Personal Computer Based Systems for Fingerprint Access Control Aspect Embedded system Personal computer + software Sensor * It usually is a solid-state sensor * It can be freely chosen and easily in order to reduce the devices replaced if it goes old or is destroyed size. Thus, it should be fixed by due to vandalism. a specialized repairing service. * It is connected by a standard USB, parallel etc. port, so no soldering process is needed. Cost * It requires a special hardware, * PCs are massive sold, so their price so its cost depends on the can be quite competitive. number of products sold by the * Low-cost cards that implement the manufacturer. main functionalities of a computer are available. Size * It is quite compact, and devices * Its size is bigger, but low-cost and about 20 cm x 15 cm x 10 cm small-size cards exist that have the exist. required PC functionalities. Consumption * They just have the strictly * Unless a PC-on-chip module is used, necessary electronics, so their their power consumption is higher power consumption is usually than an embedded system because a low. PC is a general purpose device with more electronic components than needed to open a door. Vulnerability * Main vulnerability points are * The PC should be inside the door. the sensor and decision levels Thus, the unique vulnerability point ( the whole system can be pulled is the sensor. out from the wall and the * The PC should be disconnected from control wires for opening the internet, if remote attacks want to be door manipulated ). avoided. Adaptability to * The companies offer a general * It is just needed to modify the special product that cannot normally be software, which can be C program. requirements adapted to special requirements. Thus, it is easy to include or eliminate some functionality. Computational * In order to reduce the cost, size * State of the art computers include a power and consumption, these powerful microprocessor that lets to systems have a microprocessor. operate on an identification mode and * Due to the processor limitations make hundreds of comparisons per it is restricted to verification second. mode. Malfunctioning * It is a closed system, so it is * It is quite easy to debug the problems solving quite difficult to arrange, so a and replace the damaged portions, special repairing service must which are quite standard ( the be called. hardware consists of a personal computer ).

Figure 4

The basic structure of the Implementation

Figure 4 presents the main constitutive blocks of a PC-based door opening system. Mainly they are : 1. Low-cost inkless fingerprint sensor : An optical sensor has been chosen. In r implementation, this is the unique part outside the door. The other parts are inside, although they are visible through a big glass.

2. Personal computer : The core of the system is a standard Pentium III at 800 MHz. No special considerations required , an old one can be used.

3. Screen : It displays the result of the identification ( if the user is recognized ) and the entrance time. For future versions it can be replaced by a simpler display ( for instance, a red and green light ).

4. Keyboard : The system works on identification mode. The user does not need to provide any information except for his/ her fingerprint. Thus, the keyboard is just used to typewrite the usernames for registering new users, and is also used during normal operation mode. 5. Rele card: Although it is quite easy to output a control signal through the serial or parallel port, a rele card is included which able to manage up to 16 signals. This will be used for future versions with one computer controlling several doors at a time.

6. Actuator ( Door-opening device ) : The computer just takes the decision to open or not open the door. A standard device is needed, similar to the manual remote dooropening systems in our houses ( entry phone ).

7. Software : A special application in visual C++ has been programmed taking the above procedure This program controls the door, but it also produces a Microsoft Access file with the registers of the user accesses ( name and access time). Interestingly, it is worth observing that we also record the attempts to access of unauthorized people. CONCLUSION In spite of conceptual complexity, the method proposed performs better both in terms of efficiency and robustness. Future work will try to improve this result, optimizing the code involved. It will also try to eliminate the need to store a certain part of the ridge of any of the minutiae, reducing storage needs and the computer power required.

The door-opening system can be used to control access to the laboratories and other confidential areas such that they do not need to have a key nor do they have to wait for someone to open the door. It lets us rationalize the entry , update the authorized persons inside the database without collecting or distributing physical keys, etc.The dooropening system with a proper configuration, it is able to work during the entire working day, and it also enables us to study the number of accesses and users that it supports. It can be entered to any number of users, the PC provides in the enough memory. BIBLOGRAPHY [1] L.C.Jain, U.Halici, I.Hayashi, S.B.Lee and S.Tsutsui, 1999, Intelligent Biometric Techniques in Fingerprint and Face Recognition. [2] D. Maio and D. Mmaltoni, January 19967, Direct Gray- Scale Minutiae Detection in Fingerprints, IEEE Trans, on Pattern Analysis and Machine Intelligence, Vol. 19, No. 1, January 1997 [3] A.K. Jain , L. Hong and R. Bolle, April 1997, On-Line Fingerprint Verification, IEEE Trans, on Pattern Analysis and Machine Intelligence, Vol. 19. No.4, April 1997. [4] A.K. Jain , L. Hong , S. Pankanti , and R. Bolle, September, 1997, An IdentifyAuthentication System Using Fingerprints, Proceedings of the IEEE, Vol. 85, No. 9, September 1997. [5] M.FaundezZanurry, Vulnerability of Biometric security systems.Proceedings of IEEE Vol.36,No 10, December 2003. [6] David Maltoni,Dario Maio, Anil K Jain and Salil Prabhakar,Hand book of Finger print Recognition ,Springer professional computing ,2003.