DESIGN A DEEP LEARNING MODEL FOR AN ENHANCED FINGERPRINT IDENTIFICATION SCHEME
In this research work, we analyze a new way to improve fingerprints when there is a lot of background noise. First, the image of the fingerprint is enhanced in the frequency domain, and then a binary fingerprint is made. After the second step wrongly boosts these regions, they are reassembled using a classification deep Convolution neural network with orientation selection. Both the traditional frequency domain enhancement method and the deep learning method can work together in the framework of the proposed method. To design a deep learning model for enhanced fingerprint identification, the results of the tests show that the suggested method works better than other methods.
Keywords: Fingerprint, Deep Learning, Restricted Boltzmann machine, Convolution neural network.