Fingerprint reconstruction and recognition based on the three-dimensional structure of fingertip skin

Abstract

Among the biometric identification methods, fingerprint identification is one of the most widely researched and applied biometric identification technologies. However, the traditional fingerprint identification system is vulnerable to attacks with the use of fake fingerprints, causing security problems. At the same time, when the skin of the finger is worn, wet, stained the efficiency of fingerprint identification will suffer. Optical Coherence Tomography is a non-invasive high resolution imaging technology that can image the subcutaneous depth of 1mm. Therefore, OCT can be used to obtain fingerprints inside the finger to effectively solve the security problem of fingerprint recognition, and at the same time solve the problem of the reduction in the recognition performance when the finger epidermis is damaged by external factors. In this research, OCT technology is used to collect the data of the three-dimensional structure of the fingertip by the aid of the deep learning U-net, SIFT and FLANN algorithm to ensure the reconstruction and recognition of internal fingerprints. The results show that U-net can extract the contour of the subcutaneous papilla layer and reconstruct the 2D internal fingerprint. Then we use Sift algorithm to match and splice the feature points of the internal fingerprints collected by multiple overlapping and establish a large area of internal finger template library. Finally, the FLANN algorithm library is used to extract the minutiae of the tested internal fingerprint and match the fingerprint template to achieve identity recognition. Compared with the traditional algorithm, this method is difficult to imitate and has high security.