CN103605962A - Between-class and within-class distance based human face verification method - Google Patents
Between-class and within-class distance based human face verification method Download PDFInfo
- Publication number
- CN103605962A CN103605962A CN201310589074.9A CN201310589074A CN103605962A CN 103605962 A CN103605962 A CN 103605962A CN 201310589074 A CN201310589074 A CN 201310589074A CN 103605962 A CN103605962 A CN 103605962A
- Authority
- CN
- China
- Prior art keywords
- facial image
- class
- difference
- human face
- people
- 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.)
- Pending
Links
Abstract
The invention relates to a human face identification technology, and provides a between-class and within-class distance based human face verification method, for solving the problems of quite low recognition rate and reject rate in conventional human face verification. The technical scheme can be summarized as follows: first of all, performing human face image feature extraction on a given human face image sample, forming a between-class space and a within-class space, and marking positive samples and negative samples; then performing characteristic selection by use of Adaboost to obtain a characteristic sequence; performing feature extraction on all the positive samples and the negative samples according to the characteristic sequence to obtain positive negative samples; training the positive negative samples by use of an SVM to obtain two types of classifiers; during use, performing human face image feature extraction and feature extraction on two human face image samples to be verified, and performing classification by use of the two types of classifiers; and if a classification result is a positive sample, it is determined that verification succeeds, otherwise, it is determined the verification fails. The advantages are as follows: the method provided by the invention improves the correctness of human face verification, thereby being applied to the human face verification.
Description
Technical field
The present invention relates to face recognition technology, particularly utilize the technology that in class, between class distance carries out the checking of people's face.
Background technology
At present, face recognition technology is widely used in each field, under complex illumination, attitude, background, require real-time and reliable accuracy, people's face verification technique is a great problem in recognition of face field always, most of algorithm of present stage is all devoted to improve the accuracy rate of recognition of face, and in a lot of application scenarioss, not only requires system can correctly identify the people's face in system, also require stranger to refuse i.e. people's face checking.What the checking of people's face will realize is exactly to judge whether two given facial images are same people, traditional people's face verification method is the feature of extracting respectively two facial images, then by distance function, measure, adopt the method for threshold value to provide result of determination, adopt the method wide usage of threshold value poor, under applicable cases complicated and changeable, a given threshold value is difficult to guarantee correct recognition rata and correct rejection ratio.
Adaboost algorithm and SVM(Support Vector Machine) algorithm is all the method for classical machine learning, Adaboost has been widely used in the realization that people's face detects, by in a large number targetedly positive negative sample train, obtain one group of strong classifier, thereby new sample introduction is originally classified; SVM algorithm is a kind of trainable machine learning method, is widely used in recognition of face.Yet Adaboost and SVM algorithm application are not but significantly embodied in the validity of people's face checking.
Summary of the invention
The object of the invention is to overcome the lower shortcoming of discrimination and reject rate in current people's face checking, a kind of people's face verification method based between class distance in class is provided.
The present invention solves its technical matters, and the technical scheme of employing is that the people's face verification method based between class distance in class, is characterized in that, comprises the following steps:
Step 1, given facial image sample is carried out to facial image feature extraction, be designated as I
i,j, representing that i people's j opens facial image feature, the intrinsic dimensionality of I is designated as N;
The difference of step 2, calculating different people face characteristics of image forms difference space, is designated as H(I
i,j-I
m,k), wherein, the difference of the different people face characteristics of image of same person forms space in class, and when i=m, space between the difference of the facial image feature of different people formation class, during i ≠ m;
The difference of the facial image feature in step 3, all classes of mark in space is positive sample, and the difference of the facial image feature between all classes in space is negative sample;
Step 4, employing Adaboost do feature selecting, obtain the characteristic sequence of an X dimension, described X<N;
Step 5, all positive samples and negative sample are carried out to feature extraction according to the characteristic sequence of X dimension, obtain the positive negative sample of X dimension;
Step 6, the positive negative sample that X is tieed up adopt SVM to train, and obtain binary classifier;
Step 7, two people's face image patterns that needs are verified carry out facial image feature extraction, and calculate the poor of its facial image, the difference of this facial image is carried out to feature extraction according to the characteristic sequence of X dimension, after feature extraction, use binary classifier to classify, if classification results is positive sample, think and be verified, if classification results is negative sample, think authentication failed.
Concrete, in step 1, described facial image is characterized as gabor facial image feature or LBP(Local Binary Patterns) facial image feature.
Further, in step 2, the difference of described facial image feature be calculated as Euclidean distance or manhatton distance.
The invention has the beneficial effects as follows, by above-mentioned people's face verification method based between class distance in class, can obviously improve the correctness of people's face checking, be difficult for makeing mistakes, and spent system resource is not huge.
Embodiment
Below in conjunction with embodiment, describe technical scheme of the present invention in detail.
People's face verification method based between class distance in class of the present invention is: first given facial image sample is carried out to facial image feature extraction, be designated as I
i,j, representing that i people's j opens facial image feature, the intrinsic dimensionality of I is designated as N, and the difference of then calculating different people face characteristics of image forms difference space, is designated as H(I
i,j-I
m,k), wherein, the difference of the different people face characteristics of image of same person forms space in class, when i=m, the difference of the facial image feature of different people forms space between class, while being i ≠ m, in all classes of mark, the difference of the facial image feature in space is positive sample again, the difference of the facial image feature between all classes in space is negative sample, and then adopt Adaboost to do feature selecting, obtain the characteristic sequence of an X dimension, described X<N, and all positive samples and negative sample are carried out to feature extraction according to the characteristic sequence of X dimension, obtain the positive negative sample of X dimension, to the positive negative sample of X dimension, adopt SVM to train again, obtain binary classifier, during use, two people's face image patterns to needs checking carry out facial image feature extraction, and calculate the poor of its facial image, the difference of this facial image is carried out to feature extraction according to the characteristic sequence of X dimension, after feature extraction, use binary classifier to classify, if classification results is positive sample, think and be verified, if classification results is negative sample, think authentication failed.
Embodiment
People's face verification method based between class distance in class in this example, its concrete steps are as follows:
Step 1, given facial image sample is carried out to facial image feature extraction, be designated as I
i,j, representing that i people's j opens facial image feature, the intrinsic dimensionality of I is designated as N.
In this step, facial image feature can be gabor facial image feature or LBP facial image feature etc.
The difference of step 2, calculating different people face characteristics of image forms difference space, is designated as H(I
i,j-I
m,k), wherein, the difference of the different people face characteristics of image of same person forms space in class, and when i=m, space between the difference of the facial image feature of different people formation class, during i ≠ m.
In this step, the calculating of the difference of facial image feature can be Euclidean distance or manhatton distance etc.
The difference of the facial image feature in step 3, all classes of mark in space is positive sample, and the difference of the facial image feature between all classes in space is negative sample.
Step 4, employing Adaboost do feature selecting, obtain the characteristic sequence of an X dimension, described X<N.
In this step, the Size-dependent of X in the selection of facial image sample, the selection of the computing method of feature, positive sample and negative sample etc., the characteristic sequence of X dimension sorts according to the differentiation intensity that aligns sample and negative sample, and more forward, the decision-making ability that aligns sample and negative sample is stronger.
Step 5, all positive samples and negative sample are carried out to feature extraction according to the characteristic sequence of X dimension, obtain the positive negative sample of X dimension.
In this step, all positive samples and negative sample are carried out to feature extraction according to the characteristic sequence of X dimension, all positive samples and negative sample are lowered to X dimension, thereby greatly reduce intrinsic dimensionality when meeting the maximum separability of positive sample and negative sample.
Step 6, the positive negative sample that X is tieed up adopt SVM to train, and obtain binary classifier.
Step 7, two facial image samples that needs are verified carry out facial image feature extraction, and calculate the poor of its facial image, the difference of this facial image is carried out to feature extraction according to the characteristic sequence of X dimension, after feature extraction, use binary classifier to classify, if classification results is positive sample, think and be verified, if classification results is negative sample, think authentication failed.
In this step, be verified that to need two facial image samples of checking be same people, otherwise belong to different people, in people's face checking, be judged to be refusal.
Claims (3)
1. the people's face verification method based between class distance in class, is characterized in that, comprises the following steps:
Step 1, given facial image sample is carried out to facial image feature extraction, be designated as I
i,j, representing that i people's j opens facial image feature, the intrinsic dimensionality of I is designated as N;
The difference of step 2, calculating different people face characteristics of image forms difference space, is designated as H(I
i,j-I
m,k), wherein, the difference of the different people face characteristics of image of same person forms space in class, and when i=m, space between the difference of the facial image feature of different people formation class, during i ≠ m;
The difference of the facial image feature in step 3, all classes of mark in space is positive sample, and the difference of the facial image feature between all classes in space is negative sample;
Step 4, employing Adaboost do feature selecting, obtain the characteristic sequence of an X dimension, described X<N;
Step 5, all positive samples and negative sample are carried out to feature extraction according to the characteristic sequence of X dimension, obtain the positive negative sample of X dimension;
Step 6, the positive negative sample that X is tieed up adopt SVM to train, and obtain binary classifier;
Step 7, two people's face image patterns that needs are verified carry out facial image feature extraction, and calculate the poor of its facial image, the difference of this facial image is carried out to feature extraction according to the characteristic sequence of X dimension, after feature extraction, use binary classifier to classify, if classification results is positive sample, think and be verified, if classification results is negative sample, think authentication failed.
2. the people's face verification method based between class distance in class according to claim 1, is characterized in that, in step 1, described facial image is characterized as gabor facial image feature or LBP facial image feature.
3. according to the people's face verification method based between class distance in class described in claim 1 or 2, it is characterized in that, in step 2, the difference of described facial image feature be calculated as Euclidean distance or manhatton distance.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310589074.9A CN103605962A (en) | 2013-11-19 | 2013-11-19 | Between-class and within-class distance based human face verification method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310589074.9A CN103605962A (en) | 2013-11-19 | 2013-11-19 | Between-class and within-class distance based human face verification method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN103605962A true CN103605962A (en) | 2014-02-26 |
Family
ID=50124182
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310589074.9A Pending CN103605962A (en) | 2013-11-19 | 2013-11-19 | Between-class and within-class distance based human face verification method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103605962A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104239862A (en) * | 2014-09-11 | 2014-12-24 | 中国电子科技集团公司第二十九研究所 | Face recognition method |
| CN105224957A (en) * | 2015-10-23 | 2016-01-06 | 苏州大学 | A kind of method and system of the image recognition based on single sample |
| CN107886102A (en) * | 2016-09-29 | 2018-04-06 | 北京君正集成电路股份有限公司 | Adaboost classifier training method and system |
| CN109977803A (en) * | 2019-03-07 | 2019-07-05 | 北京超维度计算科技有限公司 | A kind of face identification method based on Kmeans supervised learning |
| CN110263755A (en) * | 2019-06-28 | 2019-09-20 | 上海鹰瞳医疗科技有限公司 | Eye fundus image identification model training method, eye fundus image recognition methods and equipment |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101699470A (en) * | 2009-10-30 | 2010-04-28 | 华南理工大学 | Extracting method for smiling face identification on picture of human face |
| US20130163829A1 (en) * | 2011-12-21 | 2013-06-27 | Electronics And Telecommunications Research Institute | System for recognizing disguised face using gabor feature and svm classifier and method thereof |
-
2013
- 2013-11-19 CN CN201310589074.9A patent/CN103605962A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101699470A (en) * | 2009-10-30 | 2010-04-28 | 华南理工大学 | Extracting method for smiling face identification on picture of human face |
| US20130163829A1 (en) * | 2011-12-21 | 2013-06-27 | Electronics And Telecommunications Research Institute | System for recognizing disguised face using gabor feature and svm classifier and method thereof |
Non-Patent Citations (2)
| Title |
|---|
| 李靖: "基于AdaBoost算法的人脸识别研究", 《中国优秀硕士学位论文全文数据库·信息科技辑》 * |
| 殷争光: "基于Log-Gabor和KD-LDA的人脸识别算法研究和实现", 《中国优秀硕士学位论文全文数据库·信息科技辑》 * |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104239862A (en) * | 2014-09-11 | 2014-12-24 | 中国电子科技集团公司第二十九研究所 | Face recognition method |
| CN104239862B (en) * | 2014-09-11 | 2018-06-19 | 中国电子科技集团公司第二十九研究所 | A kind of face identification method |
| CN105224957A (en) * | 2015-10-23 | 2016-01-06 | 苏州大学 | A kind of method and system of the image recognition based on single sample |
| CN105224957B (en) * | 2015-10-23 | 2019-03-08 | 苏州大学 | A kind of method and system of the image recognition based on single sample |
| CN107886102A (en) * | 2016-09-29 | 2018-04-06 | 北京君正集成电路股份有限公司 | Adaboost classifier training method and system |
| CN107886102B (en) * | 2016-09-29 | 2020-04-07 | 北京君正集成电路股份有限公司 | Adaboost classifier training method and system |
| CN109977803A (en) * | 2019-03-07 | 2019-07-05 | 北京超维度计算科技有限公司 | A kind of face identification method based on Kmeans supervised learning |
| CN110263755A (en) * | 2019-06-28 | 2019-09-20 | 上海鹰瞳医疗科技有限公司 | Eye fundus image identification model training method, eye fundus image recognition methods and equipment |
| CN110263755B (en) * | 2019-06-28 | 2021-04-27 | 上海鹰瞳医疗科技有限公司 | Eye ground image recognition model training method, eye ground image recognition method and eye ground image recognition device |
| US11893831B2 (en) | 2019-06-28 | 2024-02-06 | Shanghai Eaglevision Medical Technology Co., Ltd. | Identity information processing method and device based on fundus image |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105550658A (en) | Face comparison method based on high-dimensional LBP (Local Binary Patterns) and convolutional neural network feature fusion | |
| CN102663413B (en) | Multi-gesture and cross-age oriented face image authentication method | |
| CN102194131B (en) | Fast human face recognition method based on geometric proportion characteristic of five sense organs | |
| CN101739555B (en) | Method and system for detecting false face, and method and system for training false face model | |
| CN106355138A (en) | Face recognition method based on deep learning and key features extraction | |
| CN101980242B (en) | Human face discrimination method and system and public safety system | |
| CN107506702A (en) | Human face recognition model training and test system and method based on multi-angle | |
| US9563821B2 (en) | Method, apparatus and computer readable recording medium for detecting a location of a face feature point using an Adaboost learning algorithm | |
| CN103605962A (en) | Between-class and within-class distance based human face verification method | |
| CN103778409A (en) | Human face identification method based on human face characteristic data mining and device | |
| CN103824091B (en) | A kind of licence plate recognition method for intelligent transportation system | |
| CN105512638A (en) | Fused featured-based face detection and alignment method | |
| CN103729625A (en) | Face identification method | |
| CN103679158A (en) | Face authentication method and device | |
| CN105760858A (en) | Pedestrian detection method and apparatus based on Haar-like intermediate layer filtering features | |
| CN103413119A (en) | Single sample face recognition method based on face sparse descriptors | |
| CN103473571A (en) | Human detection method | |
| CN104504383A (en) | Human face detecting method based on skin colors and AdaBoost algorithm | |
| CN104143091B (en) | Based on the single sample face recognition method for improving mLBP | |
| CN105160331A (en) | Hidden Markov model based face geometrical feature identification method | |
| CN104680154A (en) | Identity recognition method based on fusion of face characteristic and palm print characteristic | |
| CN103077378A (en) | Non-contact human face identifying algorithm based on expanded eight-domain local texture features and attendance system | |
| Dehshibi et al. | Persian vehicle license plate recognition using multiclass Adaboost | |
| CN103927539A (en) | Efficient feature extraction method for off-line recognition of Uyghur handwritten signature | |
| CN105320948A (en) | Image based gender identification method, apparatus and system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20140226 |
|
| RJ01 | Rejection of invention patent application after publication |