CN101584575B - Age assessment method based on face recognition technology - Google Patents
Age assessment method based on face recognition technology Download PDFInfo
- Publication number
- CN101584575B CN101584575B CN2009100327563A CN200910032756A CN101584575B CN 101584575 B CN101584575 B CN 101584575B CN 2009100327563 A CN2009100327563 A CN 2009100327563A CN 200910032756 A CN200910032756 A CN 200910032756A CN 101584575 B CN101584575 B CN 101584575B
- Authority
- CN
- China
- Prior art keywords
- age
- facial image
- face
- characteristic vector
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 27
- 238000005516 engineering process Methods 0.000 title claims abstract description 22
- 230000001815 facial effect Effects 0.000 claims abstract description 64
- 239000013598 vector Substances 0.000 claims description 40
- 238000011156 evaluation Methods 0.000 claims description 14
- 238000001514 detection method Methods 0.000 claims description 6
- 230000008569 process Effects 0.000 claims description 6
- 239000000284 extract Substances 0.000 claims description 4
- 230000008878 coupling Effects 0.000 claims description 3
- 238000010168 coupling process Methods 0.000 claims description 3
- 238000005859 coupling reaction Methods 0.000 claims description 3
- 238000000605 extraction Methods 0.000 claims description 3
- 230000006870 function Effects 0.000 description 2
- 241000208125 Nicotiana Species 0.000 description 1
- 235000002637 Nicotiana tabacum Nutrition 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000002596 correlated effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000010195 expression analysis Methods 0.000 description 1
- 230000008921 facial expression Effects 0.000 description 1
- 230000008676 import Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000003909 pattern recognition Methods 0.000 description 1
- 241000894007 species Species 0.000 description 1
- 230000037303 wrinkles Effects 0.000 description 1
Images
Landscapes
- Image Analysis (AREA)
Abstract
An age assessment method based on face recognition technology is a method capable of considering affects on age estimation caused by race, sex, latitude (region), rural-urban difference. The method ofusing face recognition technology to estimate age has vast applicability and improves accuracy of age estimation observably. The method is based on an effective hypothesis, that is people with simila r appearance has similar facial features at different ages, when estimating ages, just comparing the face image to be estimated with face image of different ages with higher appearance similarity, finding numbers of face images at different ages has highest appearance similarity with the object being estimated, proceeding weighted average to infer age of the object being estimated.
Description
Technical field
The present invention is a kind of age assessment method based on face recognition technology, belongs to the technical field that image recognition and pattern recognition are used.
Background technology
In the last few years, at the relevant computer vision field of facial image research, along with people's face detects and the obtained remarkable break-throughs of face recognition technology, people's research focus had turned to more advanced research topics such as human face expression analysis and age assessment gradually.In these problems, the age is evaluated at the application of " for the people of all ages and classes section provides different services " aspect, and huge market potential is arranged.Such as, whether the web browser with age evaluation function can the user be limited to visit some webpages, has the automatic vending machine of age evaluation function, can refuse to sell tobacco and wine etc. to the minor.
At present common age assessment method is the facial image data that import the people of a large amount of all ages and classes sections earlier to the data base, and the machine learning method in the application mode identification field then extracts and write down the facial image characteristic vector of each age bracket.Need carry out when assessment at age as the target facial image, compare, find the highest age bracket of characteristic vector similarity, as the age bracket assessment result of target image with the face characteristic vector of all ages and classes section of lane database storage.
Because at present common appraisal procedure has two critical defects: the firstth, do not consider the difference of people's face portion feature that factors such as race, sex, living environment cause, thereby the limitation that has caused the age assessment, can only guarantee that promptly evaluated facial image when all consistent, just has higher accuracy at aspects such as race, sexs.When if evaluated facial image data exist than big-difference at aspects such as race, sexs with the facial image data among the data base, tend to the assessment result that makes mistake.Such as, if having only Asian facial image characteristic vector in the feature database, and evaluated facial image is African facial image, then the age assessment result may differ more than 10 years old with actual age.The secondth, even having collected huger data base, current technology carries out computing, but owing to race, living environment differ and make people's eigenvalue of same age bracket greatest differences be arranged and can't set up effective characteristic vector.Such as the people who is 40 years old equally, the metropolitan resident of China from differing in person more than 20 years old, can't set up unified characteristic vector with the resident in rural area according to features such as the colour of skin, wrinkle, the corners of the mouth, lower jaws.
Summary of the invention
Technical problem: the purpose of this invention is to provide a kind of age assessment method, extensive applicability is arranged, also significantly improved the accuracy of age assessment based on face recognition technology.
Technical scheme: the age assessment method based on face recognition technology of the present invention, based on a kind of available hypothesis, be that the similar people of appearance also is similar at the facial characteristics of all ages and classes section, carrying out when assessment at age, only need to follow the facial image with the higher people's of its appearance similarity all ages and classes section to compare evaluated facial image, find several portraits with the highest age bracket of evaluation object similarity, by weighted average, thus the age of inferring evaluation object.
Described appraisal procedure comprises off-line training and two stages of matching inquiry,
Off-line training step: the main task in this stage is integrated application face recognition technology and ordering learning algorithm, sets up with artificial grouping unit, and the facial image feature database that everyone facial image feature is arranged from small to large according to age bracket,
The matching inquiry stage: the main task in this stage is in the facial image feature database, inquiry and the higher facial image of evaluated facial image similarity, and utilization ordering learning algorithm, calculate respectively evaluation object with the characteristic sequence of the higher facial image of its similarity in sorting position, thereby infer its age, and resulting age inferred results weighting averaged, as final age assessment result output.
The main task of described off-line training is to set up the facial image property data base, and the flow process of setting up feature database is as follows:
A. use human face detection tech, the plurality of pictures of being collected carried out facial image detect, reject the picture that detects less than people's face, suppose that here all pictures have all successfully passed through people's face and detected,
B. will divide into groups with artificial unit by the picture that people's face detects, and everyone picture arranged from small to large according to age bracket, just obtain many group people face pictures,
C. order travels through the every pictures in every group of human face photo, should choose face with other technology, extraction comprises the facial image characteristic vector that the information such as position, size and shape of the eyes, nose, mouth, chin of people's face etc. are formed, and the characteristic vector of several photos of everyone each age bracket is weighted averages, characteristic vector as this age bracket
D. still divide into groups with artificial unit, the characteristic vector of storage facial image, thus set up the feature database that comprises many eigenvectors sequence,
E. use the ordering learning algorithm, the characteristic vector in every stack features sequence sorted from small to large again according to age bracket,
F. so far, the feature database foundation that comprises many group face characteristic sequence vectors finishes, and every eigenvectors has all been finished ordering according to age bracket.
The main task in matching inquiry stage is that the facial image of importing is carried out the age assessment, and the flow process of input picture being carried out the age assessment is as follows:
A. use human face detection tech, the facial image of importing is carried out effectiveness detect, hypothesis detects successfully and passes through here,
B. use face recognition algorithms, extract the characteristic vector of the facial image of input, and characteristic vector and the characteristic vector in the face characteristic storehouse extracted are carried out the similarity coupling, inquire the characteristic vector of n the facial image similar to input picture,
C. use the ordering learning algorithm, respectively the characteristic vector of input picture searched on position according to age bracket in n relevant characteristic sequence,
The result that the D.C step is carried out obtains n age assessment result,
E. the assessment result that the D step is drawn is used weighting evaluation algorithm, calculates final assessment result.
F. export final assessment result.
Beneficial effect: method of the present invention is a kind of influence that can take all factors into consideration factors such as race, sex, latitude (region), town and country difference to the age assessment, use the method that face recognition technology carries out the age assessment, extensive applicability is arranged, also significantly improved the accuracy of age assessment.
Description of drawings
Fig. 1 is " setting up face characteristic storehouse flow chart " of the present invention,
Fig. 2 is " a matching inquiry flow chart " of the present invention,
Fig. 3 is " functional module structure figure " of the present invention.
The specific embodiment
Method of the present invention is based on a kind of available hypothesis, and promptly the similar people of appearance also is similar at the facial characteristics of all ages and classes section.Carrying out when assessment at age, only need to follow the facial image with the higher people's of its appearance similarity all ages and classes section to compare evaluated facial image, find several portraits with the highest age bracket of evaluation object similarity, by weighted average, thus the age of inferring evaluation object.
The realization of the inventive method comprises off-line training and two stages of matching inquiry.
Off-line training step, the main task in this stage is integrated application face recognition technology and ordering learning algorithm, sets up with artificial grouping unit the facial image feature database that everyone facial image feature is arranged from small to large according to age bracket.
The matching inquiry stage, the main task in this stage, in the facial image feature database, inquiry and the higher facial image of evaluated facial image similarity, and utilization ordering learning algorithm, calculate respectively evaluation object with the characteristic sequence of the higher facial image of its similarity in sorting position, thereby infer its age, and resulting age inferred results weighting averaged, as final age assessment result output.
Technology among the present invention realizes being divided into off-line training and two stages of matching inquiry.Come the details of detailed description technology realization below in conjunction with the accompanying drawing of this description with example.
1. off-line training: the main task in this stage is to set up the facial image property data base.
As shown in Figure 1, it is as follows to set up the flow process of feature database:
Suppose now to have collected each facial image data base of 10000 of man/woman's face image in Asia, America, Europe, Africa.Related people's age bracket was from 2 years old to 90 years old, and everyone has 20 age brackets, 5 photos of each age bracket.
A. use human face detection tech, 40000 pictures of being collected are carried out facial image detect, reject the picture that detects less than people's face, suppose that here all pictures have all successfully passed through people's face and detected.
B. will divide into groups with artificial unit by the picture that people's face detects, and everyone picture is arranged from small to large according to age bracket.After this step is finished, just obtain 400 groups of people's face pictures.
C. order travels through the every pictures in every group of human face photo, use face recognition technology, extraction comprises the facial image characteristic vector that the information such as position, size and shape of the eyes, nose, mouth, chin of people's face etc. are formed, and the characteristic vector of 5 photos of everyone each age bracket is weighted averages, as the characteristic vector of this age bracket.
D. still divide into groups with artificial unit, the characteristic vector of storage facial image, thus set up the feature database that comprises 400 eigenvectors sequences.
E. use the ordering learning algorithm, the characteristic vector in every stack features sequence is sorted from small to large again according to age bracket.
F. so far, the feature database foundation that comprises 400 groups of face characteristic sequence vectors finishes, and every eigenvectors has all been finished ordering according to age bracket.
2. matching inquiry stage: the main task in this stage is that the facial image to input carries out the age assessment.
Suppose that the owner X that will carry out the facial image of age assessment is the Asia male of a youth.
As shown in Figure 2, it is as follows input picture to be carried out the flow process of age assessment:
A. use human face detection tech, the facial image of importing is carried out effectiveness detect, hypothesis detects successfully and passes through here.
B. use face recognition algorithms, extract the characteristic vector of the facial image of input, and characteristic vector and the characteristic vector in the face characteristic storehouse extracted are carried out similarity coupling (similarity of characteristic vector representative image in calculating input image and the feature database).Here hypothesis inquires the characteristic vector of 3 facial images similar to input picture (A, B, C).
C. use the ordering learning algorithm, respectively the characteristic vector of input picture is searched on position according to age bracket in the characteristic sequence that A, B, C three are correlated with.
The result that the D.C step is carried out obtains three age assessment results, supposes to be respectively 20.x, 24.y, 22.z.
E. the assessment result that the D step is drawn is used weighting evaluation algorithm, calculates final assessment result.
F. export final assessment result.
The advantage of the present embodiment is not rely on single eigenvalue similarity, effectively reduces the error rate of age identification; And solved the problem that eigenvalue can't be unified under town and country differences, latitude areal variation, people's species diversity, the difference between male condition.
Claims (3)
1. age assessment method based on face recognition technology, it is characterized in that this method is based on a kind of available hypothesis, be that the similar people of appearance also is similar at the facial characteristics of all ages and classes section, carrying out when assessment at age, only need to follow the facial image with the higher people's of its appearance similarity all ages and classes section to compare evaluated facial image, find several portraits with the highest age bracket of evaluation object similarity, by weighted average, thus the age of inferring evaluation object;
Described appraisal procedure comprises off-line training and two stages of matching inquiry,
Off-line training step: the main task in this stage is integrated application face recognition technology and ordering learning algorithm, sets up with artificial grouping unit, and the facial image feature database that everyone facial image feature is arranged from small to large according to age bracket,
The matching inquiry stage: the main task in this stage is in the facial image feature database, inquiry and the higher facial image of evaluated facial image similarity, and utilization ordering learning algorithm, calculate respectively evaluation object with the characteristic sequence of the higher facial image of its similarity in sorting position, thereby infer its age, and resulting age inferred results weighting averaged, as final age assessment result output.
2. the age assessment method based on face recognition technology according to claim 1, the main task that it is characterized in that described off-line training are to set up the facial image property data base, and the flow process of setting up feature database is as follows:
A. use human face detection tech, the plurality of pictures of being collected carried out facial image detect, reject the picture that detects less than people's face, suppose that here all pictures have all successfully passed through people's face and detected,
B. will divide into groups with artificial unit by the picture that people's face detects, and everyone picture arranged from small to large according to age bracket, just obtain many group people face pictures,
C. order travels through the every pictures in every group of human face photo, use face recognition technology, extraction comprises the facial image characteristic vector that position, size and the shape information of eyes, nose, mouth, the chin of people's face are formed, and the characteristic vector of several photos of everyone each age bracket is weighted averages, characteristic vector as this age bracket
D. still divide into groups with artificial unit, the characteristic vector of storage facial image, thus set up the feature database that comprises many eigenvectors sequence,
E. use the ordering learning algorithm, the characteristic vector in every stack features sequence sorted from small to large again according to age bracket,
F. so far, the feature database foundation that comprises many group face characteristic sequence vectors finishes, and every eigenvectors has all been finished ordering according to age bracket.
3. the age assessment method based on face recognition technology according to claim 1, the main task that it is characterized in that the matching inquiry stage are that the facial image of importing is carried out the age assessment, and the flow process of input picture being carried out the age assessment is as follows:
A. use human face detection tech, the facial image of importing is carried out effectiveness detect, hypothesis detects successfully and passes through here,
B. use face recognition algorithms, extract the characteristic vector of the facial image of input, and characteristic vector and the characteristic vector in the face characteristic storehouse extracted are carried out the similarity coupling, inquire the characteristic vector of n the facial image similar to input picture,
C. use the ordering learning algorithm, respectively the characteristic vector of input picture searched on position according to age bracket in n relevant characteristic sequence,
The result that the D.C step is carried out obtains n age assessment result,
E. the assessment result that the D step is drawn is used weighting evaluation algorithm, calculates final assessment result,
F. export final assessment result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2009100327563A CN101584575B (en) | 2009-06-19 | 2009-06-19 | Age assessment method based on face recognition technology |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2009100327563A CN101584575B (en) | 2009-06-19 | 2009-06-19 | Age assessment method based on face recognition technology |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN101584575A CN101584575A (en) | 2009-11-25 |
| CN101584575B true CN101584575B (en) | 2011-05-04 |
Family
ID=41369159
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2009100327563A Active CN101584575B (en) | 2009-06-19 | 2009-06-19 | Age assessment method based on face recognition technology |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101584575B (en) |
Families Citing this family (31)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101877054A (en) * | 2009-11-23 | 2010-11-03 | 北京中星微电子有限公司 | Method and device for determining age of face image |
| CN102214286A (en) * | 2010-04-08 | 2011-10-12 | 鸿富锦精密工业（深圳）有限公司 | Network content restricted browsing control system and method |
| US9189679B2 (en) | 2010-06-21 | 2015-11-17 | Pola Chemical Industries, Inc. | Age estimation method and sex determination method |
| JP4924742B2 (en) * | 2010-06-30 | 2012-04-25 | カシオ計算機株式会社 | Image processing apparatus and method, and program |
| CN102508606A (en) * | 2011-11-10 | 2012-06-20 | 广东步步高电子工业有限公司 | Method and system for subdividing user groups by face recognition and setting corresponding functions of mobile handheld devices |
| TWI467498B (en) | 2011-12-19 | 2015-01-01 | Ind Tech Res Inst | Method and system for recognizing images |
| US9626597B2 (en) | 2013-05-09 | 2017-04-18 | Tencent Technology (Shenzhen) Company Limited | Systems and methods for facial age identification |
| CN104143079B (en) | 2013-05-10 | 2016-08-17 | 腾讯科技（深圳）有限公司 | The method and system of face character identification |
| CN103886284B (en) * | 2014-03-03 | 2017-03-01 | 小米科技有限责任公司 | Character attribute information identifying method, device and electronic equipment |
| CN103984948B (en) * | 2014-06-03 | 2017-04-05 | 南京大学 | A kind of soft double-deck age estimation method based on facial image fusion feature |
| CN104036247A (en) * | 2014-06-11 | 2014-09-10 | 杭州巨峰科技有限公司 | Facial feature based face racial classification method |
| CN104616002A (en) * | 2015-03-06 | 2015-05-13 | 李志刚 | Facial recognition equipment used for judging age groups |
| CN105117207B (en) * | 2015-07-27 | 2018-09-18 | 小米科技有限责任公司 | Photograph album creation method and device |
| CN105469064B (en) * | 2015-12-03 | 2018-12-28 | 广州市皓品信息科技有限公司 | A kind of acquisition methods and device of skin aging trend |
| CN105678253B (en) * | 2016-01-04 | 2019-01-18 | 东南大学 | Semi-supervised face age estimation device and semi-supervised face age estimation method |
| CN106295499B (en) * | 2016-07-21 | 2019-10-11 | 北京小米移动软件有限公司 | Age estimation method and device |
| CN106200432A (en) * | 2016-07-29 | 2016-12-07 | 深圳前海弘稼科技有限公司 | Control method and device, intelligence plantation equipment for intelligence plantation equipment |
| CN107133576A (en) * | 2017-04-17 | 2017-09-05 | 北京小米移动软件有限公司 | Age of user recognition methods and device |
| CN107680278A (en) * | 2017-09-22 | 2018-02-09 | 珠海格力电器股份有限公司 | Control method and device of purifier |
| CN108021863B (en) * | 2017-11-01 | 2022-05-06 | 平安科技（深圳）有限公司 | Electronic device, age classification method based on image and storage medium |
| CN108108668B (en) * | 2017-12-01 | 2021-04-27 | 北京小米移动软件有限公司 | Age prediction method and device based on image |
| KR102543656B1 (en) * | 2018-03-16 | 2023-06-15 | 삼성전자주식회사 | Screen controlling method and electronic device supporting the same |
| CN108564055A (en) * | 2018-04-24 | 2018-09-21 | 上海思依暄机器人科技股份有限公司 | A kind of offline face identification method and system |
| CN109730638A (en) * | 2018-12-29 | 2019-05-10 | 缤刻普达（北京）科技有限责任公司 | A kind of ethnic group measurement pattern switching method and device applied in Human fat balance |
| CN109948545A (en) * | 2019-03-20 | 2019-06-28 | 百度在线网络技术（北京）有限公司 | A kind of reminding method of user behavior, device, electronic equipment and storage medium |
| CN110046571B (en) * | 2019-04-15 | 2020-05-15 | 北京字节跳动网络技术有限公司 | Method and device for identifying age |
| CN110163170A (en) * | 2019-05-27 | 2019-08-23 | 北京字节跳动网络技术有限公司 | The method and apparatus at age for identification |
| CN110189179A (en) * | 2019-05-31 | 2019-08-30 | 广州软盈科技有限公司 | A kind of precision advertisement broadcast method |
| CN110334643B (en) * | 2019-06-28 | 2023-05-23 | 知鱼智联科技股份有限公司 | Feature evaluation method and device based on face recognition |
| CN110442294A (en) * | 2019-07-10 | 2019-11-12 | 杭州鸿雁智能科技有限公司 | Interface display method, device, system and the storage medium of operation panel |
| CN113780220A (en) | 2021-09-17 | 2021-12-10 | 东胜神州旅游管理有限公司 | Child sitting posture detection method and system based on child face recognition |
-
2009
- 2009-06-19 CN CN2009100327563A patent/CN101584575B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN101584575A (en) | 2009-11-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101584575B (en) | Age assessment method based on face recognition technology | |
| CN108520216B (en) | Gait image-based identity recognition method | |
| US7881524B2 (en) | Information processing apparatus and information processing method | |
| CN103824052B (en) | Multilevel semantic feature-based face feature extraction method and recognition method | |
| CN102831404B (en) | Gesture detecting method and system | |
| CN103810490B (en) | A kind of method and apparatus for the attribute for determining facial image | |
| CN102122354B (en) | Adaptive characteristic block selection-based gait identification method | |
| CN111989689A (en) | Method for recognizing objects in images and mobile device for performing the method | |
| CN108520226A (en) | A Pedestrian Re-Identification Method Based on Body Decomposition and Saliency Detection | |
| CN109784270B (en) | A Processing Method for Improving the Integrity of Face Image Recognition | |
| CN102508606A (en) | Method and system for subdividing user groups by face recognition and setting corresponding functions of mobile handheld devices | |
| CN102945366A (en) | Method and device for face recognition | |
| CN103530659B (en) | With reference to the face identification method and attendance checking system of original and symmetrical face facial image | |
| CN102902959A (en) | Face recognition method and system for storing identification photo based on second-generation identity card | |
| CN107958217A (en) | A kind of fingerprint classification identifying system and method based on deep learning | |
| CN107610177B (en) | The method and apparatus of characteristic point is determined in a kind of synchronous superposition | |
| CN102682309A (en) | Face feature registering method and device based on template learning | |
| CN107480586B (en) | Detection method of biometric photo counterfeiting attack based on facial feature point displacement | |
| CN105956570B (en) | Smile recognition method based on lip features and deep learning | |
| CN106529377A (en) | Age estimating method, age estimating device and age estimating system based on image | |
| Tong et al. | Cross-view gait recognition based on a restrictive triplet network | |
| Sun et al. | Human recognition for following robots with a Kinect sensor | |
| CN106709508A (en) | Typical weight correlation analysis method utilizing characteristic information | |
| CN106407966B (en) | A face recognition method applied to attendance | |
| CN107392105A (en) | A kind of expression recognition method based on reverse collaboration marking area feature |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| DD01 | Delivery of document by public notice |
Addressee: Li Qing Document name: Notification of Approving Refund |