Research Of Fingerprint Recognition Algorithms Based On Minutia

With the characteristics of uniqueness, reliability and stability, fingerprint is widely used in identification and verification fields, and is becoming a more and more mature business technique nowadays. Extracting high quality fingerprint template and studying fingerprint matching algorithms based on minutia with high accuracy and robustness, are preconditions to implement Challenge/Response verification based on biometrics. Meanwhile, in order to implement Challenge/Response verification based on biometrics, we need to get rid of commercial fingerprint verification system with protected patents that belong to commercial companies. Therefore, fingerprint recognition is still a far reaching research field.Based on mounts of studies on the fingerprint home and abroad, the paper does deeply researches on how to extract high quality fingerprint template and fingerprint matching algorithms based on minutia. The paper consists of several characteristics as follows:1. High quality feature extraction. Firstly, the preprocessing technology is used for the fingerprint image enhancement, including normalization, segmentation, orientation field estimation, frequency estimation, filter enhancement based on Gabor, binarization algorithms and so on. Especially, the paper proposed an improved segmentation algorithm based on Harris corner, which not only effectively eliminates false minutiae from the fingerprint background region, but also reduces the calculation of fingerprint image pre-processing. Secondly, with thinning, feature extraction and reducing false minutiae from false features based on ridge tracking algorithm, high quality fingerprint template will be obtained.2. Fingerprint matching algorithm based on minutia. Through lots of researches on fingerprint matching algorithms, this paper proposed an improved fingerprint matching algorithm based on similar vector triangle, which is independent of the global features of fingerprint, such as the core points and ridge structure. Furthermore, the algorithm is conducted to solve the key problem of fingerprint matching based on minutia: reference point orientation. And the improved fingerprint matching algorithm based on minutia, not only makes the fingerprint template size meet the requirements of Challenge/Response Verification based on biometrics, but also obtains relatively high accuracy and robustness to the fingerprint translation, rotation and deformation.In this paper, all algorithms are tested on the platform of Microsoft Visual C++ 6.0, with the production of an AFIS prototype. Finally statistic experimental results show that the AFIS prototype has achieved our predicted goals and meet the research's requirements.