Fingerprint templates combination

The commercially available touch type fingerprint sensor, especially the solid state sensor, usually has a small capture area that is not sufficient to cover the entire finger. This causes the false rejection experienced by the user to be higher if the region of the finger scanned by the sensor is not significantly similar to the region presented when the user is enrolled in the fingerprint recognition system. This paper introduces an approach to reduce the high false rejection rate of the small touch sensor. We propose a fingerprint template combination approach which combines the fingerprint template obtained from multiple independent image capture to form an image that would represent the template obtained using a large fingerprint sensor. The combination process takes into account the deformation of the finger. Main advantages of this approach over existing image mosaicing approach include low memory storage requirement and low computational complexity. Moreover, a new region of the finger can be registered into the template at any time without having to store the entire fingerprint image and the overhead needed to search for the matching fingerprint can be reduced due to the reduction in the data redundancy. Extensive experiments were conducted to determine the best transformation suitable for minutiae alignment. Among the three transformations studied, quasi-affine transformation is found to be most suitable. The proposed combination approach is experimentally shown to improve the false rejection rate due to the user using different fingerprint regions while touching the sensor for recognition.