Fingerprint recognition technology plays a critical role in modern security and information protection. Traditional 2D fingerprint recognition methods are still limited due to an imbalance between growing security demands and inefficiency of encoding detailed information. Although various 3D fingerprint technologies have been introduced recently, their practical applications are restricted by complex sampling procedures and bulky equipment. This paper proposes a new 3D fingerprint fragments reconstruction method based on the condensation of microdroplet clusters, resulting in efficiently extracting detailed structural information from fingerprint patterns. By identifying the unique topological features of fingerprint valleys, a micrometer-scale vapor transport model is developed. A differential approach is used to divide the microdroplet clusters formed when a finger is pressed on a cold surface into discrete units. In each unit, the diffusion distance and mass transfer in the condensation process are calculated. Nonlinear regression techniques are then utilized to reconstruct the 3D fingerprint fragments. Furthermore, the experimental validation shows excellent consistency with premeasured fingerprint data, with a reconstruction error of less than 9.3%. It has made a significant improvement in capturing high-density fingerprint data in a short period of time, completing the data acquisition in less than 1 second. Compared with ultrasound imaging techniques, this method significantly shortens the acquisition time, which typically involve complex procedures. Additionally, it offers a more efficient alternative to deep learning methods, which require extensive data training and computational processes. This 3D fingerprint reconstruction method provides an efficient, low-cost and easy-to-operate solution. It holds the potential to significantly enhance personal identification and information protection systems, contributing to the advancement of 3D fingerprint recognition technology in practical applications.
DownLoad:
