CN107016372A - Face identification method based on neutral net - Google Patents
Face identification method based on neutral net Download PDFInfo
- Publication number
- CN107016372A CN107016372A CN201710234331.5A CN201710234331A CN107016372A CN 107016372 A CN107016372 A CN 107016372A CN 201710234331 A CN201710234331 A CN 201710234331A CN 107016372 A CN107016372 A CN 107016372A
- Authority
- CN
- China
- Prior art keywords
- training
- face
- neutral net
- identification method
- sample
- 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.)
- Withdrawn
Links
Classifications
-
- 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
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
-
- 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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Molecular Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Human Computer Interaction (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
The invention provides a kind of face identification method based on neutral net, including training set image collection is set up, training set face bitmap is stored, and read bitmap data；Training sample in original input space carries out feature extraction, forms training set sample set；A RBF neural is trained using the training set sample set after training set bitmap data and feature extraction；Test set image collection is set up, test set face bitmap is stored, and read bitmap data；Test set bitmap data is inputted into the RBF neural that training is completed, test set sample point set is obtained；Using grader, Classification and Identification is carried out to test set image.Compared with prior art, the face identification method based on neutral net that the present invention is provided, velocity of approch is fast, and discrimination is higher, and under conditions of certain recognition correct rate is met, recognition time is short.
Description
Technical field
The present invention relates to a kind of face identification method based on neutral net, belong to field of face identification.
Background technology
Recognition of face is to reach that identity differentiates a kind of computer technology of purpose by analyzing human face's visual signature.
Academia gives to recognition of face and is specifically defined of both broad sense and narrow sense.The recognition of face of broad sense includes Face datection
（face detection）, face characterize（face representation）, face detection（ face identification
）, Expression analysis（ face expression analysis ）And physical classification（physical classification）
Etc. a series of correlation techniques；And the recognition of face of narrow sense is then defined as a kind of technology or system, this technology or system can
Identity validation, identity are carried out by the feature of face to compare and identity finder.
At present, because face recognition technology can pass through organism（Typically refer in particular to people）The biological characteristic of itself is individual to distinguish
Body, improves the precision of organism identification, therefore, the technology gets the attention and praised highly, and the field is also become raw
Focus in thing identification feature research.By taking the mankind as an example, biological characteristic mostlys come from following aspect：Face, retina, iris,
Palm line, fingerprint, voice, the bodily form, custom etc., thus based on the above, research has then been focused on identification face, view
In the Computer Recognition Technologies of individual features such as film, iris, palm line, fingerprint, voice, the bodily form, keyboard are tapped, signature, and take
Obtained significant achievement.
The characteristics of advantage of recognition of face is its naturality and friendly.So-called naturality, refers to that the mankind are also in itself
Distinguished by observing and comparing human face's feature and confirm other side's identity, such as speech recognition, the bodily form are recognized similarly
Feature with naturality, and the mankind or other biological generally do not distinguish individual, therefore above-mentioned spy by features such as fingerprint, irises
Levy the feature that identification does not just have naturality.
So-called friendly, refers to that the recognition methods does not increase the psychological burden for being authenticated people because of special treatment, and
Therefore it is easier to obtain direct and real characteristic information.Fingerprint or iris recognition need using electronic pressure transmitter or
The special technique such as infrared ray means gather information, and above-mentioned special acquisition technique is easily found by people, is considerably increased and is authenticated people
Hide the possibility of identity discriminating, reduce the efficiency of identity discriminating.
However, recognition of face but can directly obtain the face information for being authenticated people by simple image or video technique,
This information gathering mode is not easy to be therefore easily perceived by humans, and adds the authenticity and reliability of information.
Although face recognition technology has above-mentioned advantage, the realization of the technology is not easy to.The main life by face
Thing characteristic is limited, and is in particular in：
First, because the structure of same type of face all has higher similitude.The feature can be used for Face detection, but
It is but to considerably increase the difficulty for differentiating individual using face characteristic.
Second, limited by factors such as age, mood, temperature light condition, overcovers, the profile of face is very unstable,
Even in different viewing angles, the characteristics of image of face adds answering for face recognition technology application there is also significant difference
Polygamy.
To make field needed for face recognition technology is preferably served, then need to seek above-mentioned two limitation progress researchs
Break through.
Face recognition technology based on kernel space is one of technology for being most widely used in field of face identification.One
General kernel space face recognition algorithms flow is as shown in figure 1, because the expression of base in kernel space will use all instructions
Practice sample, therefore increasing with training sample number, the projection speed of test sample slows down, and then has had a strong impact on face knowledge
Other speed, what this drawback embodied especially in real-time system and on-line system becomes apparent.
The content of the invention
In view of in place of above-mentioned the deficiencies in the prior art, it is an object of the invention to provide a kind of face based on neutral net
Recognition methods, it is desirable to set up a kind of face recognition algorithms model, can about be subtracted to the basis representation of core proper subspace.
In face recognition process, test sample is set to avoid the base to the proper subspace being made up of whole training samples from being projected,
But to the approximate subspace projection about subtracted, recognition of face speed is improved with this.
In order to achieve the above object, this invention takes following technical scheme：
The invention provides a kind of face identification method based on neutral net, comprise the following steps：
Step 1: setting up training set image collection, training set face bitmap is stored, and read bitmap data；
Step 2: the training sample in original input space carries out feature extraction, training set sample set is formed；
Step 3: training a RBF neural using the training set sample set after training set bitmap data and feature extraction；
Step 4: setting up test set image collection, test set face bitmap is stored, and read bitmap data；
Step 5: test set bitmap data to be inputted to the RBF neural trained and completed, test set sample point set is obtained；
Step 6: using grader, Classification and Identification is carried out to test set image.
It is preferred that, above-mentioned steps two, which carry out feature extraction, to be carried out by KPCA methods.
It is preferred that, the Training RBF Neural Network of above-mentioned steps three includes unsupervised and two steps of Training.
It is preferred that, above-mentioned training comprises the following steps：
The first step, the method calculating Basis Function Center by cluster；
Second step, calculating variance；
3rd step, acquisition hidden layer neuron to output layer neuron connection weight.
It is preferred that, the above-mentioned first step includes adjustment cluster centre, that is, obtains different cluster set vpMiddle training sample average,
I.e. brand-new cluster centre ci, judge whether brand-new cluster centre changes, if the constant c so obtainediIt is exactly final base letter
Number center, continues to adjust cluster centre if changing, carries out next round solution.
Compared with prior art, the face identification method based on neutral net that the present invention is provided, velocity of approch is fast, and identification
Rate is higher, and under conditions of certain recognition correct rate is met, recognition time is short.
Brief description of the drawings
Fig. 1 is kernel space face identification method schematic flow sheet general in the prior art；
Fig. 2 is the face identification method schematic flow sheet of the invention based on neutral net.
Embodiment
The present invention provides a kind of face identification method based on neutral net, for make the purpose of the present invention, technical scheme and
Effect is clearer, clear and definite, and the present invention is described in more detail for the embodiment that develops simultaneously referring to the drawings.It should be appreciated that herein
Described specific embodiment only to explain the present invention, is not intended to limit the present invention.
First, RBF neural is determined using Self-organizing Selection Center method by training.It is divided into unsupervised and has supervision
Training.The idiographic flow of training is as follows：
The first step：Basis Function Center c is calculated by the method for cluster.
（1）Netinit：H training sample is randomly selected as cluster centre ci（i=1,2,…,h）, input is instructed
Practice sample set to be grouped with Nearest Neighbor Method：By xpAccording to xpWith center ciBetween Euclidean distance distribute to different poly- of input sample
Class set vp(p=1,2 ... P) in.
（2）Adjust cluster centre：Obtain different cluster set vpMiddle training sample average, i.e., brand-new cluster centre ci, judge
Whether brand-new cluster centre changes, if the constant c so obtainediBe exactly final Basis Function Center, if change if after
Continuous adjustment cluster centre, carries out next round solution.
Second step：Calculate variances sigmai。
Using Gaussian function as neutral net basic function, if between Selection Center distance maximum be Cmax, then variances sigmai
Calculation expression is as follows:
3rd step：Hidden layer neuron is obtained to output layer neuron connection weight
Hidden layer neuron and the connection weight of output layer neuron can be calculated with least square method, its expression formula is as follows：
As shown in Fig. 2 the face identification method based on neutral net that the present embodiment is provided, specifically includes following steps：
（1）Training set image collection is set up, training set face bitmap is stored, and read bitmap data.
（2）Feature extraction is carried out to the training sample in original input space using KPCA methods, training set sample is formed
Set.
（3）A RBF neural is trained using the training set sample set after training set bitmap data and feature extraction.
（4）Test set image collection is set up, test set face bitmap is stored, and read bitmap data.
（5）Test set bitmap data is inputted into the RBF neutral nets that training is completed, test set sample point set is obtained.
（6）Using grader, Classification and Identification is carried out to test set image.
Compared with prior art, the face identification method based on neutral net that the present invention is provided, velocity of approch is fast, and identification
Rate is higher, and under conditions of certain recognition correct rate is met, recognition time is short.
It is understood that for those of ordinary skills, can be with technique according to the invention scheme and its invention structure
Think of is subject to equivalent substitution or change, and all these changes or replacement should all belong to the protection model of appended claims of the invention
Enclose.
Claims (5)
1. a kind of face identification method based on neutral net, it is characterised in that：The recognition methods comprises the following steps：
Step 1: setting up training set image collection, training set face bitmap is stored, and read bitmap data；
Step 2: the training sample in original input space carries out feature extraction, training set sample set is formed；
Step 3: training a RBF neural using the training set sample set after training set bitmap data and feature extraction；
Step 4: setting up test set image collection, test set face bitmap is stored, and read bitmap data；
Step 5: test set bitmap data to be inputted to the RBF neural trained and completed, test set sample point set is obtained；
Step 6: using grader, Classification and Identification is carried out to test set image.
2. the face identification method as claimed in claim 1 based on neutral net, it is characterised in that：The step 2 carries out special
Levying extraction is carried out by KPCA methods.
3. the face identification method as claimed in claim 1 based on neutral net, it is characterised in that：The step 3 training
RBF neural includes unsupervised and two steps of Training.
4. the face identification method as claimed in claim 3 based on neutral net, it is characterised in that：The training includes following
Step：
The first step, the method calculating Basis Function Center by cluster；
Second step, calculating variance；
3rd step, acquisition hidden layer neuron to output layer neuron connection weight.
5. the face identification method as claimed in claim 4 based on neutral net, it is characterised in that：The first step includes adjusting
Whole cluster centre, that is, obtain different cluster set vpMiddle training sample average, i.e., brand-new cluster centre ci, judge in brand-new cluster
Whether the heart changes, if the constant c so obtainediIt is exactly final Basis Function Center, continues to adjust cluster if changing
Center, carries out next round solution.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710234331.5A CN107016372A (en) | 2017-04-12 | 2017-04-12 | Face identification method based on neutral net |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710234331.5A CN107016372A (en) | 2017-04-12 | 2017-04-12 | Face identification method based on neutral net |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107016372A true CN107016372A (en) | 2017-08-04 |
Family
ID=59445239
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710234331.5A Withdrawn CN107016372A (en) | 2017-04-12 | 2017-04-12 | Face identification method based on neutral net |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107016372A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109558774A (en) * | 2017-09-27 | 2019-04-02 | 中国海洋大学 | Object automatic recognition system based on depth residual error network and support vector machines |
| CN110348898A (en) * | 2019-06-28 | 2019-10-18 | 广东奥园奥买家电子商务有限公司 | A kind of information-pushing method and device based on human bioequivalence |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102831396A (en) * | 2012-07-23 | 2012-12-19 | 常州蓝城信息科技有限公司 | Computer face recognition method |
| CN104700076A (en) * | 2015-02-13 | 2015-06-10 | 电子科技大学 | Face image virtual sample generating method |
| CN104915680A (en) * | 2015-06-04 | 2015-09-16 | 河海大学 | Improved RBF neural network-based multi-label metamorphic relationship prediction method |
| CN105975959A (en) * | 2016-06-14 | 2016-09-28 | 广州视源电子科技股份有限公司 | Face feature extraction modeling and face recognition method and device based on neural network |
-
2017
- 2017-04-12 CN CN201710234331.5A patent/CN107016372A/en not_active Withdrawn
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102831396A (en) * | 2012-07-23 | 2012-12-19 | 常州蓝城信息科技有限公司 | Computer face recognition method |
| CN104700076A (en) * | 2015-02-13 | 2015-06-10 | 电子科技大学 | Face image virtual sample generating method |
| CN104915680A (en) * | 2015-06-04 | 2015-09-16 | 河海大学 | Improved RBF neural network-based multi-label metamorphic relationship prediction method |
| CN105975959A (en) * | 2016-06-14 | 2016-09-28 | 广州视源电子科技股份有限公司 | Face feature extraction modeling and face recognition method and device based on neural network |
Non-Patent Citations (1)
| Title |
|---|
| 王坚等: "基于神经网络的快速核子空间人脸识别算法研究", 《计算机科学》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109558774A (en) * | 2017-09-27 | 2019-04-02 | 中国海洋大学 | Object automatic recognition system based on depth residual error network and support vector machines |
| CN110348898A (en) * | 2019-06-28 | 2019-10-18 | 广东奥园奥买家电子商务有限公司 | A kind of information-pushing method and device based on human bioequivalence |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Yuan et al. | Fingerprint liveness detection using an improved CNN with image scale equalization | |
| Rigas et al. | Biometric recognition via eye movements: Saccadic vigor and acceleration cues | |
| CN106203497B (en) | Finger vena area-of-interest method for screening images based on image quality evaluation | |
| WO2016090376A1 (en) | Eye tracking via patterned contact lenses | |
| CN105559802A (en) | Tristimania diagnosis system and method based on attention and emotion information fusion | |
| Puengdang et al. | EEG-based person authentication method using deep learning with visual stimulation | |
| EP3449409B1 (en) | Biometric method and device for identifying a person through an electrocardiogram (ecg) waveform | |
| Zhang et al. | On biometrics with eye movements | |
| Wang et al. | Continuous authentication using EEG and face images for trusted autonomous systems | |
| WO2018187953A1 (en) | Facial recognition method based on neural network | |
| CN110309813A (en) | A kind of model training method, detection method, device, mobile end equipment and the server of the human eye state detection based on deep learning | |
| CN106446849A (en) | Fatigue driving detection method | |
| CN113221655A (en) | Face spoofing detection method based on feature space constraint | |
| CN106874867A (en) | A kind of face self-adapting detecting and tracking for merging the colour of skin and profile screening | |
| Zhang et al. | On biometric verification of a user by means of eye movement data mining | |
| Gale et al. | Evolution of performance analysis of iris recognition system by using hybrid methods of feature extraction and matching by hybrid classifier for iris recognition system | |
| Daza et al. | Alebk: Feasibility study of attention level estimation via blink detection applied to e-learning | |
| CN116348042A (en) | Method and system for quantifying attention | |
| Bhamare et al. | Deep neural networks for lie detection with attention on bio-signals | |
| CN107016372A (en) | Face identification method based on neutral net | |
| CN113343198B (en) | Video-based random gesture authentication method and system | |
| CN111046978A (en) | Skin mirror image identification method based on StyleGANs and decision fusion | |
| Chevallier et al. | Riemannian classification for SSVEP-based BCI: Offline versus online implementations | |
| Mohammad et al. | Towards ethnicity detection using learning based classifiers | |
| Tovarek et al. | Human abnormal behavior impact on speaker verification systems |
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 | ||
| WW01 | Invention patent application withdrawn after publication | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20170804 |