Touchless Fingerprint 3d Reconstruction Based On Binocular Vision

Fingerprint is currently one of the most widely used biometric technologies. Fingerprint technologies have been researched and developed several decades which have been focused on touching fingerprint recognition.For touching fingerprint recognition, a finger touches or rolls onto sensor surface, the elastic skin is easily deformed which caused the different quality of fingerprint images and directions by the different pressure of users. The skin conditions of different moisture also caused inconsistent, irreproducible fingerprint images. Hence, the representation of the same fingerprint changes every time the finger is placed on the sensor platen, increasing the complexity of fingerprint matching and representing a negative influence on system performance.Recently, a new approach to capture fingerprints has been proposed. This approach, referred to as touchless or contactless fingerprint, tries to overcome the above-cited problems. Because of the lack of contact between the finger and sensor surface, the skin does not deform during the capture. So touchless fingerprint could get better recognition result, but the shape of fingers are curved, one camera only captures a part of the finger which decrease the information of fingerprint, so generally speaking, touchless fingerprint devices use more than two cameras, its aim to capture the whole and clear fingerprint which increase the expense of fingerprint devices. In order to decrease the expense of fingerprint devices, we propose to use two cameras for capture and reconstruct fingerprint 3D information to increase information of fingerprint. Meanwhile, 3D information reconstruction based on binocular vision needed a lot of matching pairs, but the number of fingerprint minutiae could not qualify for this condition, so we propose to extract extreme fingerprint features to increase number of matching pairs which is named Scale Invariant Feature Transform. Experiments show that it is very helpful for 3D reconstruction by increasing extreme fingerprint features.Meanwhile, SIFT matching algorithm for fingerprint minutiae is tested on the Hong Kong Polytechnic University of Touchless Fingerprint Database and compared to my laboratory results, experiments show that SIFT matching algorithm could get more number of minutiae and improve correct recognition rate.