Study On Automatic Fingerprint Identification Matching Technique

With the rapid development of electronic information, computer and network technology, biometric authentication technique, which identifies human from the physiological or behavioral characteristics, is widely recognized by more and more people. Compared with the traditional identification techniques, biometric authentication can avoid being fabricated and stolen, so it has excellent security. At the same time, the biometrics need not carry, so it's convenient in identification technique. Fingerprint, as one of the biological characteristics, not only has the same superiority as the above biological characteristics have, but also has many other advantages:informative, easy to identify, not change over time. So, fingerprint identification has become one of the most popular, the most convenient and the most reliable personal identification techniques.Automatic fingerprint identification system includes fingerprint image acquisition, image preprocessing, minutia extraction and minutia matching. This paper focuses on two algorithms about fingerprint identification. One is fingerprint matching algorithm based on genetic algorithm. The other one is fingerprint matching algorithm based on regional feature points. Genetic algorithm can globally search optimal or near optimal solution, which is frequently used in fingerprint verification. Firstly, we use the local feature information of fingerprint feature points to generate the initial population of genetic algorithm, then, we use this population which has local alignment information to search globally and compute the fitness of every individual to search for the global optimal individual. We compare its fitness with the threshold to decide whether the two fingerprints are from the same finger. Because the fingerprint feature points around the center have a small interference from noise, using them for fingerprint matching can achieve more reliable result. So we suggest a proposed fingerprint matching algorithm based on regional feature points. Using core points and minutiae in the same region around the core points of the template and the fingerprint to match, we search for the translation and rotation parameters. Experimental results show that this algorithm has a small computation but the ability to resist noise is relatively inferior and the matching accuracy is slightly. Relatively, the fingerprint matching algorithm based on genetic algorithm has a better ability to resist noise.