A new approach to identify fingerprint using support vector machine

Fingerprint identification is an important biometric technique which has been used and studied a lot in the past. The most popular traditional approach for identifying fingerprint basing on extracting minutiae and then matching them together to authenticate has some limitations in practice. For a considerable fraction of population, it is very difficult to automatically extract minutiae in a noise image. Further, minutiae-based matching has difficulties in quickly matching two fingerprint images. So, a new approach to overcome these limitations for automatically identifying fingerprint is necessary. In this paper, we describe a new method to identify fingerprint by combining an extreme classifying method, support vector machine, and filter bank-based technique. This approach is a learning method whose identifying time is so fast while training time is acceptable.