Presenting a method for identifying individuals using biometric fingerprint location

Background and Aim: Fingerprint recognition is one of the most popular biometric methods for authentication. For this reason, research literature widely supports fingerprints for its high reliability, low cost and ease of measurement, however. In recent years, attacks and threats have jeopardized the security of these systems, and the widespread use of fake fingerprint identification has compromised this authentication method, so the purpose of this research is to provide a suitable, fast, reliable and low cost method for identifying people with fingerprints. This research is of a type of developmental study with a purposeful purpose, and on The basis of the design of a SVM-based model and the Learn ++ NC algorithm, in addition to fingerprint identification, also attempted to discover the falsification of fingerprints, for this purpose the Blonnia University s databases such as FVC2000, FVC2002 and FVC2004 are used. The findings of this study show that in all three datasets, the average of the performance error is less than the three methods of existing algorithms. As there are relatively equal functionalities in the six datasets, the dataset in Learning Learn ++ NC is an exception and results in slightly improved error rates, and the learning outcomes and the recognition of the 32x32 input image compared to the images 64x64 suggests improvements to the proposed algorithms. Experiments performed on databases have shown that the proposed algorithm from other existing algorithms has better performance in terms of error and detection and compliance.

کلیدواژه ها:

fingerprint ، fake fingerprint ، support vector machine ، and incremental learning algorithm