The Designand Implement Of Fingerprint Access Controlsystem Based On ARM And Linux

With the rapid development of society, the security is facing new problems and the pastaccess control system can not meet the requirement of people,on the other hand the biometricauthentication methods and network is to be the direction of the access control system.Fingerprint identification technology is used more widely than others because of its convenienceand efficience. Moreover the ARM+Linux embedded platform has a high efficiency of workand can develop in a simple way, meanwhile, it has a good network performance, which can beused well in this system.Firstly, this paper discusse the development status of access control system, and then theproposed system will be based on ARM+Linux embedded platform, the fingerprint as theauthentication methods. After accomplishing the design of system, we set up the platform fordevelpoment. Based on these things, we design the circuit using the USB function and internalROM of Fps200(Fingerprint-gathering chip) and of course the USB driver programme,application program, which use the Linux’s asynchronous notification mechanism.Finally, wecan use this module to get the fingerprint.As the core of this access control system, and considering the characteristics of the system,we should finish the image preprocessing of fingerprint, include cutting, enhacement,binarization and thinning of fingerprint. Then, we can extract the minutiae on the skeletonfingerprint image. On the other hand, we must eliminate the false minutiae, because it impactseriously to matching work, and then save the information of minutiae get two fingrprinttemplate. And now, we can match the minutiae in order to tell that the two fingerprints comefrom the same finger. The core that is used in fingerprint’s point mode matching generally haveproblems, meanwhile the opportunity of false minutiae come from bifurcation is much less thanendpoint, so this paper will chose five bifurcations which have smallest distance to the centroidin the fingerprint and decide a couple of bifurcations in the two fingerprint template which havehighest match degree among these bifurcations, and then these two bifurcations is used as thematching reference points, by which the fingerprint will rotate and move. At last, we can matchthese two fingerprint on the boundaries box of point mode matching algorithms. At end of this paper, we test the main performance parameters of this system. The resultshow that the fingerprint acquisition module can reach the requirement of this system, on theother hand, the matching algorithm based on bifurcation in centroid domain increase theperformance partly. At last, we summarize the work of this paper andgive a outlook for actual use in the future.