A fingerprint recognition method and device based on deep learning
A technology of fingerprint recognition and deep learning
Active Publication Date: 2022-02-18
NORTH CHINA UNIVERSITY OF TECHNOLOGY
View PDF3 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
[0006] Aiming at the problem that the existing fingerprint recognition algorithm based on deep learning does not make sufficient use of neural network information, the present invention simultaneously uses the cross-entropy loss function and the contrastive loss function to train the neural network, and utilizes these two kinds of losses simultaneously in the fingerprint recognition stage The information provided by the function to achieve a better fingerprint recognition effect
Method used
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment Construction
[0030] Below through embodiment and accompanying drawing, the present invention is described in detail.
[0031] Step 1: Design and build as figure 1 The deep neural network shown. The network model uses a residual network, which consists of multiple residual units in series. The residual unit helps to speed up the training process of the neural network. The output part of the neural network adopts a parallel structure as a cross-entropy loss function and comparison Input to the loss function.
[0032] The residual network has cross-layer data flow between different layers, such as figure 2 As shown, where Relu is an activation function, F(X) is the output of parameter layer 1, and H(X) is the output of parameter layer 2.
[0033] Step 2: Input the fingerprint image registered by the user into the neural network, and use the cross-entropy loss function and the contrastive loss function to train the deep neural network.
[0034] For the cross entropy loss function L cross...
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More PUM
Login to View More Abstract
The present invention relates to a fingerprint recognition method and device based on deep learning. First, a deep neural network is built, and the fingerprint image set registered by the user is input into the deep neural network, and the cross-entropy loss function and the contrastive loss function are used for training. After the training is completed, input the registered user image used for training into the deep neural network again to obtain the input feature vector of the contrast loss function, and use the clustering algorithm to cluster the fingerprint image of each registered user to obtain its specific number of clusters. The class center acts as a local feature library. In the fingerprint verification stage, input the fingerprint image to be identified into the trained deep neural network, obtain its cross-entropy loss value and contrast loss function to directly train the output vector of the node, and use the judgment function to judge similar fingerprints according to the preset threshold , to complete the fingerprint verification process. The present invention can obtain a lower false recognition rate under the premise of ensuring a high recognition rate, and has higher security.
Description
technical field [0001] The invention belongs to the application of deep learning in the field of fingerprint recognition, in particular to a method and device for feature extraction and recognition of small-area fingerprint images using a deep neural network. Background technique [0002] Fingerprint, as a biometric feature with high uniqueness, high stability and high anti-fraud, has been successfully used in many fields, such as mobile fingerprint payment on mobile phones, fingerprint check-in in work units, etc. Fingerprint identification technology extracts the features of the fingerprint image registered by the user, obtains the texture, key points and other detailed information of the fingerprint image, and identifies the user's fingerprint. Traditional fingerprint recognition algorithms can achieve better fingerprint recognition results on the premise of obtaining a complete user fingerprint image. However, in recent years, fingerprint recognition technology has been...
Claims
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More Application Information
Patent Timeline
Login to View More Patent Type & Authority Patents(China)
IPC IPC(8): G06N3/04G06N3/08G06K9/00G06K9/62
CPCG06N3/08G06V40/12G06N3/045G06F18/23213
Inventor 曾凡锋胡胜达肖珂
Owner NORTH CHINA UNIVERSITY OF TECHNOLOGY