A fingerprint verification system using minutiae and wavelet based features

Minutiae based approach is most widely used for fingerprint matching. Minutiae can be extracted either directly from gray-scaled image or from a thinned image. During matching, finding an exact match depends on the best matched minutiae pairs from both images. For matching stage, different kinds of features are extracted from extracted minutiae. The structure of some features allows us to have rotation and translation invariance. Minutiae based approach also has some drawbacks because it requires very lengthy preprocessing operations for minutiae extraction and still can result in false minutiae. Previously, to overcome this problem some kind of post-processing is used, which also eliminates valid minutiae along with false ones. So eventually, we can say that the strength of matching algorithm depends on the strength of extracted features from fingerprint. In our research, we have presented a new approach which uses wavelet based features which are fused with minutiae based features for matching purpose. In particular, we find that among the algorithms we studied, our proposed work have significant effects on overall performance. Experiment results show that using these features have made the matching process much more accurate even in the presence of false minutiae.