Three-dimensional Fingerprintrecognition By Using Deep Learning

In recent years,with the development of science and technology,people urgently need new recognition technology for more stable authentication.Because of its reliability,high recognition,and easy acquisition,fingerprint recognition technology has become a hot research field in biometrics fields.A complete fingerprinting system includes the following two main steps: fingerprint data acquisition and fingerprint matching recognition,in which matching recognition of fingerprint data is the key.The speed and precision of matching directly affect the final recognition rate of the system.The traditional two-dimensional(2D)fingerprint recognition method mostly depends on the matching of feature points and has a high recognition rate.However,this method based on feature point matching not only takes long time,but also loses the depth information of the fingerprint.As the fingerprint rotates and scales,the robustness will be seriously reduced.To solve these problems,threedimensional(3D)fingerprint recognition technology appeared.As it is an emerging research field,there are still many challenging unsolved issues in the 3D fingerprint recognition technology.Traditional methods find fingerprint feature points and match feature points.For the long time and high complexity of the algorithm,a 3D fingerprint recognition algorithm by using deep learning is proposed in this paper.By introducing a convolutional neural network,the step of finding fingerprint feature points is reduced,the complexity of the recognition algorithm is reduced,and the depth information of the fingerprint is also effectively preserved.The specific steps of the proposed algorithm are as follows: firstly,the fingerprint depth image and the 2D fingerprint image are input to different convolutional neural networks respectively to obtain the fingerprint depth feature and the 2D fingerprint feature,and then the feature fusion is performed through another neural network,and finally the merged features are used for 3D fingerprinting.Experimental results show that the algorithm not only improves the recognition rate by introducing 2D fingerprint features,but also solves the matching problem between the 3D fingerprint recognition algorithm and the existing 2D fingerprint database.Compared with other machine learning algorithms,the recognition rate of this algorithm has been effectively improved and it is feasible.