CN110472501A - A kind of fingerprint pore coding specification method neural network based - Google Patents
A kind of fingerprint pore coding specification method neural network based Download PDFInfo
- Publication number
- CN110472501A CN110472501A CN201910618569.7A CN201910618569A CN110472501A CN 110472501 A CN110472501 A CN 110472501A CN 201910618569 A CN201910618569 A CN 201910618569A CN 110472501 A CN110472501 A CN 110472501A
- Authority
- CN
- China
- Prior art keywords
- fingerprint
- neural network
- pore
- layer
- hidden layer
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
- G06V10/267—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion by performing operations on regions, e.g. growing, shrinking or watersheds
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/12—Fingerprints or palmprints
- G06V40/1365—Matching; Classification
- G06V40/1371—Matching features related to minutiae or pores
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Molecular Biology (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Multimedia (AREA)
- Human Computer Interaction (AREA)
- Collating Specific Patterns (AREA)
Abstract
The invention proposes a kind of fingerprint pore coding specification methods neural network based, the following steps are included: step 1) extracts high-resolution fingerprint image, by segmentation, method for normalizing pre-processes fingerprint image, obtains the position feature of fingerprint pore by Gabor filter;Step 2 is trained the position feature using neural network, obtains classification based training collection;This method facilitates direct, reduces external environment influence, reduces the complexity of recognizer, improve the robustness of system, saves recognition time, be with a wide range of applications in security system, advanced door lock access control system and criminal investigation field.
Description
Technical field
The present invention relates to a kind of fingerprint pore classification methods neural network based, and specifically one kind is by pore
Position carry out BP neural network training, belong to technical field.
Background technique
Currently, traditional fingerprint recognition system based on fingerprint minutiae is more perfect, but its accuracy still needs
It improves, it is more accurate using the identifying system based on fingerprint pore for there is the place of the requirement of high security.Jian A K etc.
People proposes improved iteration closest approach algorithmic match pore, has good anti-noise effect;Zhao Q et al. proposes pore-valley line
Descriptor solves the translation and rotational invariance of pore feature, improves accuracy.However fingerprint pore is usually more, at present
Algorithm it is low to its treatment effeciency, complexity is very high, time-consuming very long.
Summary of the invention
The method that encode the object of the present invention is to provide a kind of pair of fingerprint pore and then classify, to reduce fingerprint pore
Matching algorithm complexity improves system robustness, saves the time.
The object of the present invention is achieved like this: a kind of fingerprint pore coding specification method neural network based, including
Following steps:
Step 1) extracts high-resolution fingerprint image, and by segmentation, method for normalizing pre-processes fingerprint image,
The position feature of fingerprint pore is obtained by Gabor filter;
Step 2) is trained the position feature using neural network, obtains classification based training collection.
It is further limited as of the invention, step 1) specifically includes: fingerprint pore is obtained by Gabor filter filtering
Model:
Wherein, δiIndicate the pore size perpendicular to crestal line direction, δjIndicate the pore size along crestal line direction, θ is indicated
Crestal line direction can show that Rot is to P by calculating the field of direction0The operation for carrying out rotation θ angle, passes through the above filtering operation
The position feature of you can get it fingerprint pore.
It is further limited as of the invention, step 2) specifically includes: carrying out off-line training using BP neural network, obtain
The function model of fingerprint pore identification, for the identification to fingerprint database;The function model is as follows:
Y=purelin (W2×tan sig(W1×Xn+θ1)+θ2) wherein, XnFor the input vector of BP neural network;W1For
Weight between input layer and hidden layer;W2Weight between hidden layer and output layer;θ1Between input layer and hidden layer
Threshold value;θ2Threshold value between hidden layer and output layer;Y is the output vector of BP neural network;Tan sig () be input layer and
Tanh S function between hidden layer;Linear transfer function of the purelin () between hidden layer and output layer.
It is further limited as of the invention, for BP neural network input layer, sets the input points of BP neural network
It is 500, this 500 input values are the local detail feature of pore of taking the fingerprint;Utilize formula
Calculate hidden layer and output layer number of nodes, nIFor the number of nodes of input layer, nOFor the number of nodes of output layer, ncIt is
Constant sets constant nc=8, the number of nodes of output layer is set to 200, and hidden layer number of nodes is then n=34;
For output layer, 200 binary codings are carried out in advance to each fingerprint image, coding range is 0~
2200-1;Each binary numeral is classified as one, a corresponding people;
In training module, using three layers of BP neural network of input layer, hidden layer and output layer, BP nerve net is constructed
Network model;The learning rate γ that the model is arranged is 0.8, and momentum coefficient α is 0.9, maximum cycle 5000, and target error is
1e-5;Fingerprint recognition function is obtained after off-line training, is used for subsequent recognition operation;
In identification module, it is stored in a feature vector, X, connects from 500 characteristic values that fingerprint image extracts
By vector X input BP neural network model in, the model has trained and has obtained corresponding weight value W and threshold θ at this time, thus
Output Y is obtained, output Y is a binary coding, it is corresponding with the fingerprint pore number in database, the element of vector Y
Value is 0 or 1, and the classification to fingerprint pore can be realized in this way, can the other fingerprint of knowledge due to the corresponding people of each classification
Which people image is derived from.
The invention adopts the above technical scheme compared with prior art, has following technical effect that and is instructed by neural network
Practice, so that each fingerprint image coding is unique in fingerprint base, recognition accuracy is improved;In addition by pretreatment after,
Fingerprint recognition function is brought directly to link to be encoded rapidly, match complexity is reduced, and time-consuming shortens, and system robustness becomes
It is high.
Detailed description of the invention
Fig. 1 is neural network model figure of the invention.
Specific embodiment
Technical solution of the present invention is described in further detail with reference to the accompanying drawing:
The present embodiment proposes a kind of coding specification method of fingerprint pore neural network based, and process is as follows:
1. extracting the higher fingerprint image of resolution ratio, by segmentation, method for normalizing pre-processes fingerprint image.It is logical
It crosses Gabor filtering and obtains fingerprint pore model:
Wherein, δxIndicate scale factor along the x-axis direction, δyIndicate scale factor along the y-axis direction.
Wherein, δiIndicate the pore size perpendicular to crestal line direction, δjIndicate the pore size along crestal line direction, θ is indicated
Crestal line direction can show that Rot is to P by calculating the field of direction0Carry out the operation of rotation θ angle.
Pass through the position feature of the above filtering operation you can get it fingerprint pore.
2. the position for each pore carries out BP neural network training, the pore of fingerprint image is subjected to 0 and 1 coding.
Fingerprint pore sorting algorithm based on BP error Feedback Neural Network, this method only need to train enough training
Sample can make BP network adjust weight and the threshold value of itself to adapt to the various features of fingerprint pore, the mould of BP neural network
Type is shown in attached drawing 1, and formula is as follows
Y=purelin (W2×tan sig(W1×Xn+θ1)+θ2) (3)
Wherein, XnFor the input vector of BP neural network;W1Weight between input layer and hidden layer;W2For hidden layer
Weight between output layer;θ1Threshold value between input layer and hidden layer;θ2Threshold value between hidden layer and output layer;Y
For the output vector of BP neural network;Tanh S function of the tan sig () between input layer and hidden layer;purelin
The linear transfer function of () between hidden layer and output layer.
One width complete finger print image pore quantity is about 500, and the input points that can set BP neural network are 500,
This 500 input values are the local detail feature to take the fingerprint.According to formula
Hidden layer and output layer number of nodes, n can be calculatedIFor the number of nodes of input layer, nOFor the number of nodes of output layer, nc
It is constant, sets constant nc=8, the number of nodes of output layer is set to 200, and hidden layer number of nodes is then n=34.
In training module, using three layers of BP neural network of input layer, hidden layer and output layer, BP nerve net is constructed
Network model, the learning rate γ for being provided with the model is 0.8, and momentum coefficient α is 0.9, and maximum cycle 5000, target is missed
Difference is 1 × 10-5。
In identification module, it is stored in a feature vector, X, connects from 500 characteristic values that fingerprint image extracts
By vector X input BP neural network model in, the model has trained and has obtained corresponding weight value W and threshold θ at this time, thus
Obtain output Y.Exporting Y is a binary coding, it is corresponding with the fingerprint pore number in database, the element of vector Y
Value is 0 or 1, can be classified in this way to fingerprint pore.
The above, the only specific embodiment in the present invention, but scope of protection of the present invention is not limited thereto, appoints
What is familiar with the people of the technology within the technical scope disclosed by the invention, it will be appreciated that expects transforms or replaces, and should all cover
Within scope of the invention, therefore, the scope of protection of the invention shall be subject to the scope of protection specified in the patent claim.
Claims (4)
1. a kind of fingerprint pore coding specification method neural network based, which comprises the following steps:
Step 1) extracts high-resolution fingerprint image, and by segmentation, method for normalizing pre-processes fingerprint image, passes through
Gabor filter obtains the position feature of fingerprint pore;
Step 2) is trained the position feature using neural network, obtains classification based training collection.
2. a kind of fingerprint pore coding specification method neural network based according to claim 1, which is characterized in that step
It is rapid 1) to specifically include: fingerprint pore model is obtained by Gabor filter filtering:
Wherein, δiIndicate the pore size perpendicular to crestal line direction, δjIndicate the pore size along crestal line direction, θ indicates crestal line
Direction can show that Rot is to P by calculating the field of direction0The operation for carrying out rotation θ angle, passes through the above filtering operation
Obtain the position feature of fingerprint pore.
3. a kind of fingerprint pore coding specification method neural network based according to claim 2, which is characterized in that step
It is rapid 2) to specifically include: to carry out off-line training using BP neural network, obtain the function model of fingerprint pore identification, for fingerprint
The identification of database;The function model is as follows:
Y=purelin (W2×tan sig(W1×Xn+θ1)+θ2)
Wherein, XnFor the input vector of BP neural network;W1Weight between input layer and hidden layer;W2For hidden layer and defeated
Weight between layer out;θ1Threshold value between input layer and hidden layer;θ2Threshold value between hidden layer and output layer;Y is BP
The output vector of neural network;Tanh S function of the tan sig () between input layer and hidden layer;Purelin () is
Linear transfer function between hidden layer and output layer.
4. a kind of fingerprint pore coding specification method neural network based according to claim 3, which is characterized in that right
In BP neural network input layer, the input points for setting BP neural network are 500, this 500 input values are to take the fingerprint
The local detail feature of pore;Utilize formula
Calculate hidden layer and output layer number of nodes, nIFor the number of nodes of input layer, nOFor the number of nodes of output layer, ncIt is constant,
Set constant nc=8, the number of nodes of output layer is set to 200, and hidden layer number of nodes is then n=34;
For output layer, 200 binary codings are carried out in advance to each fingerprint image, coding range is 0~2200-1;
Each binary numeral is classified as one, a corresponding people;
In training module, using three layers of BP neural network of input layer, hidden layer and output layer, BP neural network mould is constructed
Type;The learning rate γ that the model is arranged is 0.8, and momentum coefficient α is 0.9, maximum cycle 5000, target error 1e-
5;Fingerprint recognition function is obtained after off-line training, is used for subsequent recognition operation;
It in identification module, is stored in a feature vector, X from 500 characteristic values that fingerprint image extracts, then will
Vector X is inputted in BP neural network model, and the model has trained and obtained corresponding weight value W and threshold θ at this time, to obtain
Y is exported, output Y is a binary coding, it is corresponding with the fingerprint pore number in database, and the element value of vector Y is 0
Or 1, the classification to fingerprint pore can be realized in this way, it, can the other fingerprint image of knowledge due to the corresponding people of each classification
Which people be derived from.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910618569.7A CN110472501B (en) | 2019-07-10 | 2019-07-10 | Neural network-based fingerprint sweat pore coding classification method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910618569.7A CN110472501B (en) | 2019-07-10 | 2019-07-10 | Neural network-based fingerprint sweat pore coding classification method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110472501A true CN110472501A (en) | 2019-11-19 |
CN110472501B CN110472501B (en) | 2022-08-30 |
Family
ID=68507469
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910618569.7A Active CN110472501B (en) | 2019-07-10 | 2019-07-10 | Neural network-based fingerprint sweat pore coding classification method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110472501B (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112668595A (en) * | 2021-01-25 | 2021-04-16 | 数网金融有限公司 | Image processing method, device, equipment and storage medium |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107480649A (en) * | 2017-08-24 | 2017-12-15 | 浙江工业大学 | Fingerprint sweat pore extraction method based on full convolution neural network |
CN108449295A (en) * | 2018-02-05 | 2018-08-24 | 西安电子科技大学昆山创新研究院 | Combined modulation recognition methods based on RBM networks and BP neural network |
CN108959833A (en) * | 2018-09-26 | 2018-12-07 | 北京工业大学 | Tool wear prediction technique based on improved BP neural network |
CN109547431A (en) * | 2018-11-19 | 2019-03-29 | 国网河南省电力公司信息通信公司 | A kind of network security situation evaluating method based on CS and improved BP |
-
2019
- 2019-07-10 CN CN201910618569.7A patent/CN110472501B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107480649A (en) * | 2017-08-24 | 2017-12-15 | 浙江工业大学 | Fingerprint sweat pore extraction method based on full convolution neural network |
CN108449295A (en) * | 2018-02-05 | 2018-08-24 | 西安电子科技大学昆山创新研究院 | Combined modulation recognition methods based on RBM networks and BP neural network |
CN108959833A (en) * | 2018-09-26 | 2018-12-07 | 北京工业大学 | Tool wear prediction technique based on improved BP neural network |
CN109547431A (en) * | 2018-11-19 | 2019-03-29 | 国网河南省电力公司信息通信公司 | A kind of network security situation evaluating method based on CS and improved BP |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112668595A (en) * | 2021-01-25 | 2021-04-16 | 数网金融有限公司 | Image processing method, device, equipment and storage medium |
Also Published As
Publication number | Publication date |
---|---|
CN110472501B (en) | 2022-08-30 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111524525B (en) | Voiceprint recognition method, device, equipment and storage medium of original voice | |
Fahmy | Online handwritten signature verification system based on DWT features extraction and neural network classification | |
Kashi et al. | A Hidden Markov Model approach to online handwritten signature verification | |
CN107967695B (en) | A kind of moving target detecting method based on depth light stream and morphological method | |
CN109241995B (en) | Image identification method based on improved ArcFace loss function | |
Ng et al. | Iris recognition using rapid Haar wavelet decomposition | |
CN107832747B (en) | Face recognition method based on low-rank dictionary learning algorithm | |
CN108564040B (en) | Fingerprint activity detection method based on deep convolution characteristics | |
CN108681698B (en) | Large-scale iris recognition method with privacy protection function | |
CN113312989B (en) | Finger vein feature extraction network based on aggregated descriptors and attention | |
CN113779643B (en) | Signature handwriting recognition system and method based on pre-training technology and storage medium | |
CN107403153A (en) | A kind of palmprint image recognition methods encoded based on convolutional neural networks and Hash | |
CN105118509A (en) | Security authentication method based on voiceprint two-dimensional code | |
CN107122725B (en) | Face recognition method and system based on joint sparse discriminant analysis | |
CN110956082A (en) | Face key point detection method and detection system based on deep learning | |
Saponara et al. | Recreating fingerprint images by convolutional neural network autoencoder architecture | |
Öztürk et al. | Minnet: Minutia patch embedding network for automated latent fingerprint recognition | |
CN110472501A (en) | A kind of fingerprint pore coding specification method neural network based | |
Saffar et al. | Online signature verification using deep representation: a new descriptor | |
CN115984906A (en) | Palm print image recognition method, device, equipment and storage medium | |
CN113076930B (en) | Face recognition and expression analysis method based on shared backbone network | |
CN111950333B (en) | Electronic handwritten signature recognition method based on neural network | |
CN108537213A (en) | Enhance the system and method for iris recognition precision | |
CN114155554A (en) | Transformer-based camera domain pedestrian re-recognition method | |
CN108197573A (en) | The face identification method that LRC and CRC deviations based on mirror image combine |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |