Deep Learning in Automatic Fingerprint Identification

Abstract

The development of fingerprint identification as a computer application technology is closely related to the new technology of computer science. The artificial intelligence technology, especially the image technology based on deep learning, has opened a new mode of fingerprint identification algorithm. In this paper, we divide the development of artificial intelligence in fingerprint field into three stages, and analyze the development trend of the second stage. The fingerprint identification technology based on deep learning uses image features instead of traditional minutiae feature, which changes the cognition of fingerprint recognition in the field of forensic science. This study investigate the application mode and typical methods of deep learning technology in fingerprint identification, give the technical schemes based on deep learning, and put forward several key technologies such as image processing and dimensionality reduction. The existing DNN models that can be used for fingerprint identification are introduced, such as convolutional neural network and auto-encoder network. The results show that the performance of artificial intelligence fingerprint identification algorithm is better than the traditional algorithm in many indicators.