CN104036236A - Human face gender recognition method based on multi-parameter index weighting - Google Patents
Human face gender recognition method based on multi-parameter index weighting Download PDFInfo
- Publication number
- CN104036236A CN104036236A CN201410230388.4A CN201410230388A CN104036236A CN 104036236 A CN104036236 A CN 104036236A CN 201410230388 A CN201410230388 A CN 201410230388A CN 104036236 A CN104036236 A CN 104036236A
- Authority
- CN
- China
- Prior art keywords
- face
- people
- gender
- human face
- sex
- 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
Landscapes
- Image Analysis (AREA)
Abstract
The invention provides a human face gender recognition method based on multi-parameter index weighting. According to the method, through tracking the human face of the same person in a video sequence, a plurality of human face images of the same person are collected; the human face image quality analysis is carried out on the collected human face images; the human face image quality analysis results are used as human face gender recognition accuracy weights for calculating human face gender confidence values subjected to image analysis correction; in addition, the human face gender confidence values of the human face images of the same person are subjected to index weighting operation; and the human face gender is finally judged out according to the calculation results. The human face gender recognition method has the advantages that the gender of a certain person is judged by using a plurality of human faces in the video sequence, so the problem that when a single static picture is taken for human face gender recognition, the interference by in-site environment is greater is solved; and in addition, the problem of recognition accuracy interference due to in-site complicated environment is solved through human face image quality analysis and index weight operation.
Description
Technical field
The present invention relates to a kind of face gender identification method based on multiparameter exponential weighting.
Background technology
The method of gender classification is mainly divided into two classes both at home and abroad at present: the method based on feature and the method based on statistics.
Method based on signature analysis refers to by the observation to men and women's image, utilizes hair length, eyebrow thickness, chin width, whether has the low-level features of the facial images such as beard as the foundation of Sex Discrimination.These class methods be mainly that distance by tolerance visible features realizes the sex of facial image is identified.
Method based on statistics is regarded sex identification as two classification problems.By the study of a large amount of training samples being set up to one, can realize to the face gender on image the sorter of correct identification, then by sorter, the facial image in test set be carried out to Gender Classification.Typical classifier methods comprises artificial neural network, adaboost, support vector machine etc.
Current gender classification is mainly to analyze for still image.In video scene, traditional method is by the mode of external trigger, judges when having personnel to pass through, and automatically captures photo site, then the people's face in candid photograph photo is carried out to gender analysis.And in video monitoring scene, site environment more complicated, camera acquisition to people's face angle and illumination all cannot guarantee, therefore capture that people's face of arriving is may angle of arrival deflection large, focal length, to situations such as, uneven illumination are even, do not have a strong impact on age-sex's effect.Therefore, current many gender classification algorithms cannot be applied in traditional video monitoring scene.
Summary of the invention
The object of the invention is to propose a kind of face gender identification method based on multiparameter exponential weighting, can in unrestricted video monitoring scene, keep the precision of gender classification algorithm.
A kind of face gender identification method based on multiparameter exponential weighting of the present invention, by the people's face to same person in video sequence, follow the tracks of, gather a plurality of facial images of same person, a plurality of facial images that collect are carried out to quality of human face image analysis, using quality of human face image analysis result as gender classification accuracy weights, calculate the face gender the value of the confidence after graphical analysis is corrected, and the face gender the value of the confidence of a plurality of facial images of same person is carried out to exponential weighting computing, according to result of calculation, finally judge face gender.
Specifically comprise the steps:
Step 1, by the people's face to same person in video sequence, follow the tracks of, gather a plurality of facial images that obtain same person in video monitoring scene:
By RTSP stream media protocol, connecting camera head obtains stream medium data and decodes, complete after decoding, end user's face detection algorithm detects the people's face in video sequence, when present frame detects new people's face, using the interval time of the coordinate position of people's face and interframe as input, use kalman wave filter predict this people's face in next frame by the coordinate range there will be, then next frame image is carried out to the detection of people's face, if the position that people's face occurs is positioned at the position range of kalman filtering threshold, think that people's face of working as forefathers' face and previous frame belongs to same person, according to above-mentioned judgment mode, from a plurality of images of Real-time Collection, obtain a plurality of facial images of same person,
Step 2, quality of human face image analysis, calculate the mass parameter that belongs to a plurality of facial images of same person in above-mentioned video sequence:
For a plurality of facial images that belong to same person in above-mentioned video sequence, first calculate three mass parameters of the fog-level of the size of people's face, the angle of people's face and people's face, then these three mass parameters are weighted to summation, obtain final quality of human face image value c;
Step 3, employing LBP feature are described people's face, use support vector machine to carry out face gender classification as face characteristic sorter:
First, ready multiple people's face training samples that completed Sex-linked marker are carried out to LBP feature extraction, recycling face characteristic sorter is trained these features, obtains sex disaggregated model, comprises male sex faceform and women faceform in this Gender Classification model; When carrying out face gender classification, first load this Gender Classification model, people's face to new collection carries out modeling, again the LBP characteristic information of a plurality of facial images of same person in the video monitoring scene gathering in step 1 is input to respectively in face characteristic sorter and is calculated, obtain the sex the value of the confidence x of every facial image, when sex the value of the confidence is greater than 0.5, be the male sex, it is women that sex the value of the confidence is less than at 0.5 o'clock, completes the face gender classification to every facial image;
Step 4, according to index weights, people's face is carried out to gender analysis:
First the N that is ready to belong to same person opens facial image, calculates the mass calibration sex the value of the confidence h of each facial image, and computing method are as follows:
Wherein c represents quality of human face image value, and x is illustrated in the face gender the value of the confidence of calculating acquisition in step 3;
Obtain after mass calibration sex the value of the confidence h, then carry out exponential weighting calculating:
Wherein, h
ibe the sex the value of the confidence of i head portrait after quality is corrected, h
i>0.5 represents the male sex, h
i<0.5 represents women, and H represents exponential weighting result of calculation;
When H<0, final sex result is judged to be the male sex, and when H>0, final sex result is judged to be women.
Weights in described weighted calculation are set as respectively: people's little weights of being bold are 0.35, and people's face angle weights are 0.45, and people's face fog-level weights are 0.20.
Because having adopted a plurality of people's faces in video sequence, the present invention judges someone's sex, solved to get and when individual static images carries out gender classification, be subject to site environment to disturb larger problem, and solved by quality of human face image analysis and the computing of index weights the problem that on-the-spot complex environment disturbs recognition accuracy.
Accompanying drawing explanation
Fig. 1 is workflow schematic diagram in the present invention.
Below in conjunction with the drawings and specific embodiments, the present invention is further described.
Embodiment
As Fig. 1, a kind of face gender identification method based on multiparameter exponential weighting of the present invention, specifically comprises the steps:
Step 1, by the people's face to same person in video sequence, follow the tracks of, gather a plurality of facial images that obtain same person in video monitoring scene:
By RTSP stream media protocol, connecting camera head obtains stream medium data and decodes, complete after decoding, end user's face detection algorithm detects the people's face in video sequence, when present frame detects new people's face, using the interval time of the coordinate position of people's face and interframe as input, use kalman wave filter predict this people's face in next frame by the coordinate range there will be, then next frame image is carried out to the detection of people's face, if the position that people's face occurs is positioned at the position range of kalman filtering threshold, think that people's face of working as forefathers' face and previous frame belongs to same person, according to above-mentioned judgment mode, from a plurality of images of Real-time Collection, obtain a plurality of facial images of same person,
Step 2, quality of human face image analysis, calculate the mass parameter that belongs to a plurality of facial images of same person in above-mentioned video sequence:
Image quality parameter has determined this people's face weight in gender classification computing in the back, for a plurality of facial images that belong to same person in above-mentioned video sequence, first calculate three mass parameters of the fog-level of the size of people's face, the angle of people's face and people's face, then these three mass parameters are weighted to summation, obtain final quality of human face image value c, weights in the present embodiment weighted calculation are set as respectively: people's little weights of being bold are 0.35, people's face angle weights are 0.45, and people's face fog-level weights are 0.20;
Step 3, employing LBP feature are described people's face, use support vector machine to carry out face gender classification as face characteristic sorter:
First, ready multiple people's face training samples that completed Sex-linked marker are carried out to LBP feature extraction, recycling face characteristic sorter is trained these features, obtains sex disaggregated model, comprises male sex faceform and women faceform in this Gender Classification model; When carrying out face gender classification, first load this Gender Classification model, people's face to new collection carries out modeling, again the LBP characteristic information of a plurality of facial images of same person in the video monitoring scene gathering in step 1 is input to respectively in face characteristic sorter and is calculated, obtain the sex the value of the confidence x of every facial image, when sex the value of the confidence is greater than 0.5, be the male sex, it is women that sex the value of the confidence is less than at 0.5 o'clock, completes the face gender classification to every facial image;
Step 4, according to index weights, people's face is carried out to gender analysis:
First the N that is ready to belong to same person opens facial image, calculates the mass calibration sex the value of the confidence h of each facial image, and computing method are as follows:
Wherein c represents quality of human face image value, and x is illustrated in the face gender the value of the confidence of calculating acquisition in step 3;
Obtain after mass calibration sex the value of the confidence h, then carry out exponential weighting calculating:
Wherein, h
ibe the sex the value of the confidence of i head portrait after quality is corrected, h
i>0.5 represents the male sex, h
i<0.5 represents women, and H represents exponential weighting result of calculation;
When H<0, final sex result is judged to be the male sex, and when H>0, final sex result is judged to be women.
The above, it is only preferred embodiment of the present invention, not technical scope of the present invention is imposed any restrictions, therefore any trickle modification, equivalent variations and modification that every foundation technical spirit of the present invention is done above embodiment all still belong in the scope of technical solution of the present invention.
Claims (3)
1. the face gender identification method based on multiparameter exponential weighting, it is characterized in that: by the people's face to same person in video sequence, follow the tracks of, gather a plurality of facial images of same person, a plurality of facial images that collect are carried out to quality of human face image analysis, using quality of human face image analysis result as gender classification accuracy weights, calculate the face gender the value of the confidence after graphical analysis is corrected, and the face gender the value of the confidence of a plurality of facial images of same person is carried out to exponential weighting computing, according to result of calculation, finally judge face gender.
2. a kind of face gender identification method based on multiparameter exponential weighting according to claim 1, is characterized in that specifically comprising the steps:
Step 1, by the people's face to same person in video sequence, follow the tracks of, gather a plurality of facial images that obtain same person in video monitoring scene:
By RTSP stream media protocol, connecting camera head obtains stream medium data and decodes, complete after decoding, end user's face detection algorithm detects the people's face in video sequence, when present frame detects new people's face, using the interval time of the coordinate position of people's face and interframe as input, use kalman wave filter predict this people's face in next frame by the coordinate range there will be, then next frame image is carried out to the detection of people's face, if the position that people's face occurs is positioned at the position range of kalman filtering threshold, think that people's face of working as forefathers' face and previous frame belongs to same person, according to above-mentioned judgment mode, from a plurality of images of Real-time Collection, obtain a plurality of facial images of same person,
Step 2, quality of human face image analysis, calculate the mass parameter that belongs to a plurality of facial images of same person in above-mentioned video sequence:
For a plurality of facial images that belong to same person in above-mentioned video sequence, first calculate three mass parameters of the fog-level of the size of people's face, the angle of people's face and people's face, then these three mass parameters are weighted to summation, obtain final quality of human face image value c;
Step 3, employing LBP feature are described people's face, use support vector machine to carry out face gender classification as face characteristic sorter:
First, ready multiple people's face training samples that completed Sex-linked marker are carried out to LBP feature extraction, recycling face characteristic sorter is trained these features, obtains sex disaggregated model, comprises male sex faceform and women faceform in this Gender Classification model; When carrying out face gender classification, first load this Gender Classification model, people's face to new collection carries out modeling, again the LBP characteristic information of a plurality of facial images of same person in the video monitoring scene gathering in step 1 is input to respectively in face characteristic sorter and is calculated, obtain the sex the value of the confidence x of every facial image, when sex the value of the confidence is greater than 0.5, be the male sex, it is women that sex the value of the confidence is less than at 0.5 o'clock, completes the face gender classification to every facial image;
Step 4, according to index weights, people's face is carried out to gender analysis:
First the N that is ready to belong to same person opens facial image, calculates the mass calibration sex the value of the confidence h of each facial image, and computing method are as follows:
Wherein c represents quality of human face image value, and x is illustrated in the face gender the value of the confidence of calculating acquisition in step 3;
Obtain after mass calibration sex the value of the confidence h, then carry out exponential weighting calculating:
Wherein, h
ibe the sex the value of the confidence of i head portrait after quality is corrected, h
i>0.5 represents the male sex, h
i<0.5 represents women, and H represents exponential weighting result of calculation;
When H<0, final sex result is judged to be the male sex, and when H>0, final sex result is judged to be women.
3. a kind of face gender identification method based on multiparameter exponential weighting according to claim 1, it is characterized in that the weights in described weighted calculation are set as respectively: people's little weights of being bold are 0.35, people's face angle weights are 0.45, and people's face fog-level weights are 0.20.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410230388.4A CN104036236B (en) | 2014-05-27 | 2014-05-27 | A kind of face gender identification method based on multiparameter exponential weighting |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410230388.4A CN104036236B (en) | 2014-05-27 | 2014-05-27 | A kind of face gender identification method based on multiparameter exponential weighting |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104036236A true CN104036236A (en) | 2014-09-10 |
| CN104036236B CN104036236B (en) | 2017-03-29 |
Family
ID=51467002
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410230388.4A Active CN104036236B (en) | 2014-05-27 | 2014-05-27 | A kind of face gender identification method based on multiparameter exponential weighting |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104036236B (en) |
Cited By (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104701684A (en) * | 2015-04-01 | 2015-06-10 | 中国矿业大学(北京) | Intelligent socket system for preventing infant from getting electric shock in dark environment |
| CN106203306A (en) * | 2016-06-30 | 2016-12-07 | 北京小米移动软件有限公司 | The Forecasting Methodology at age, device and terminal |
| CN106446821A (en) * | 2016-09-20 | 2017-02-22 | 北京金山安全软件有限公司 | Method and device for identifying gender of user and electronic equipment |
| CN107169433A (en) * | 2017-05-10 | 2017-09-15 | 成都优孚达信息技术有限公司 | A kind of face identification method based on Streaming Media |
| CN107945168A (en) * | 2017-11-30 | 2018-04-20 | 上海联影医疗科技有限公司 | The processing method and magic magiscan of a kind of medical image |
| CN108090428A (en) * | 2017-12-08 | 2018-05-29 | 广西师范大学 | A kind of face identification method and its system |
| CN108229322A (en) * | 2017-11-30 | 2018-06-29 | 北京市商汤科技开发有限公司 | Face identification method, device, electronic equipment and storage medium based on video |
| CN108875731A (en) * | 2017-12-28 | 2018-11-23 | 北京旷视科技有限公司 | Target identification method, device, system and storage medium |
| CN108960205A (en) * | 2018-08-06 | 2018-12-07 | 广州开瑞信息科技有限公司 | A kind of Intelligent human-face recognition methods and system |
| CN109190449A (en) * | 2018-07-09 | 2019-01-11 | 北京达佳互联信息技术有限公司 | Age recognition methods, device, electronic equipment and storage medium |
| CN111223225A (en) * | 2020-02-11 | 2020-06-02 | 厦门瑞为信息技术有限公司 | Detection passing system integrating temperature measurement and face recognition gate machine partner |
| CN111325173A (en) * | 2020-02-28 | 2020-06-23 | 腾讯科技（深圳）有限公司 | Hair type identification method and device, electronic equipment and storage medium |
| WO2020238321A1 (en) * | 2019-05-27 | 2020-12-03 | 北京字节跳动网络技术有限公司 | Method and device for age identification |
| CN112926542A (en) * | 2021-04-09 | 2021-06-08 | 博众精工科技股份有限公司 | Performance detection method and device, electronic equipment and storage medium |
| CN112926478A (en) * | 2021-03-08 | 2021-06-08 | 新疆爱华盈通信息技术有限公司 | Gender identification method, system, electronic device and storage medium |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101510254A (en) * | 2009-03-25 | 2009-08-19 | 北京中星微电子有限公司 | Method for updating gender classifier in image analysis and the gender classifier |
| CN103310187A (en) * | 2012-03-13 | 2013-09-18 | 霍尼韦尔国际公司 | Face image prioritization based on face quality analysis |
-
2014
- 2014-05-27 CN CN201410230388.4A patent/CN104036236B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101510254A (en) * | 2009-03-25 | 2009-08-19 | 北京中星微电子有限公司 | Method for updating gender classifier in image analysis and the gender classifier |
| CN103310187A (en) * | 2012-03-13 | 2013-09-18 | 霍尼韦尔国际公司 | Face image prioritization based on face quality analysis |
| US20130243268A1 (en) * | 2012-03-13 | 2013-09-19 | Honeywell International Inc. | Face image prioritization based on face quality analysis |
Non-Patent Citations (1)
| Title |
|---|
| 朱长仁: ""复杂背景下的多姿态人脸识别技术研究"", 《中国博士学位论文全文数据库 信息科技辑》 * |
Cited By (21)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104701684A (en) * | 2015-04-01 | 2015-06-10 | 中国矿业大学(北京) | Intelligent socket system for preventing infant from getting electric shock in dark environment |
| CN104701684B (en) * | 2015-04-01 | 2017-07-18 | 中国矿业大学(北京) | A kind of child's electric shock intelligent socket system anti-in dark situation |
| CN106203306A (en) * | 2016-06-30 | 2016-12-07 | 北京小米移动软件有限公司 | The Forecasting Methodology at age, device and terminal |
| CN106446821A (en) * | 2016-09-20 | 2017-02-22 | 北京金山安全软件有限公司 | Method and device for identifying gender of user and electronic equipment |
| CN107169433A (en) * | 2017-05-10 | 2017-09-15 | 成都优孚达信息技术有限公司 | A kind of face identification method based on Streaming Media |
| CN108229322B (en) * | 2017-11-30 | 2021-02-12 | 北京市商汤科技开发有限公司 | Video-based face recognition method and device, electronic equipment and storage medium |
| CN107945168A (en) * | 2017-11-30 | 2018-04-20 | 上海联影医疗科技有限公司 | The processing method and magic magiscan of a kind of medical image |
| CN108229322A (en) * | 2017-11-30 | 2018-06-29 | 北京市商汤科技开发有限公司 | Face identification method, device, electronic equipment and storage medium based on video |
| WO2019105337A1 (en) * | 2017-11-30 | 2019-06-06 | 北京市商汤科技开发有限公司 | Video-based face recognition method, apparatus, device, medium and program |
| CN107945168B (en) * | 2017-11-30 | 2021-12-10 | 上海联影医疗科技股份有限公司 | Medical image processing method and medical image processing system |
| US11068697B2 (en) | 2017-11-30 | 2021-07-20 | Beijing Sensetime Technology Development Co., Ltd | Methods and apparatus for video-based facial recognition, electronic devices, and storage media |
| CN108090428A (en) * | 2017-12-08 | 2018-05-29 | 广西师范大学 | A kind of face identification method and its system |
| CN108875731A (en) * | 2017-12-28 | 2018-11-23 | 北京旷视科技有限公司 | Target identification method, device, system and storage medium |
| CN109190449A (en) * | 2018-07-09 | 2019-01-11 | 北京达佳互联信息技术有限公司 | Age recognition methods, device, electronic equipment and storage medium |
| CN108960205A (en) * | 2018-08-06 | 2018-12-07 | 广州开瑞信息科技有限公司 | A kind of Intelligent human-face recognition methods and system |
| WO2020238321A1 (en) * | 2019-05-27 | 2020-12-03 | 北京字节跳动网络技术有限公司 | Method and device for age identification |
| CN111223225A (en) * | 2020-02-11 | 2020-06-02 | 厦门瑞为信息技术有限公司 | Detection passing system integrating temperature measurement and face recognition gate machine partner |
| CN111325173A (en) * | 2020-02-28 | 2020-06-23 | 腾讯科技（深圳）有限公司 | Hair type identification method and device, electronic equipment and storage medium |
| CN112926478A (en) * | 2021-03-08 | 2021-06-08 | 新疆爱华盈通信息技术有限公司 | Gender identification method, system, electronic device and storage medium |
| CN112926542A (en) * | 2021-04-09 | 2021-06-08 | 博众精工科技股份有限公司 | Performance detection method and device, electronic equipment and storage medium |
| CN112926542B (en) * | 2021-04-09 | 2024-04-30 | 博众精工科技股份有限公司 | Sex detection method and device, electronic equipment and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104036236B (en) | 2017-03-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104036236A (en) | Human face gender recognition method based on multi-parameter index weighting | |
| CN104091176B (en) | Portrait comparison application technology in video | |
| WO2020215961A1 (en) | Personnel information detection method and system for indoor climate control | |
| CN111460962B (en) | Face recognition method and face recognition system for mask | |
| CN104143079B (en) | The method and system of face character identification | |
| CN104881637B (en) | Multimodal information system and its fusion method based on heat transfer agent and target tracking | |
| WO2018119668A1 (en) | Method and system for recognizing head of pedestrian | |
| CN102214291B (en) | Method for quickly and accurately detecting and tracking human face based on video sequence | |
| CN109819208A (en) | A kind of dense population security monitoring management method based on artificial intelligence dynamic monitoring | |
| CN105160318A (en) | Facial expression based lie detection method and system | |
| CN107590452A (en) | A kind of personal identification method and device based on gait and face fusion | |
| CN109697430A (en) | The detection method that working region safety cap based on image recognition is worn | |
| CN108647625A (en) | A kind of expression recognition method and device | |
| CN109711370A (en) | A kind of data anastomosing algorithm based on WIFI detection and face cluster | |
| CN104166841A (en) | Rapid detection identification method for specified pedestrian or vehicle in video monitoring network | |
| CN109522853A (en) | Face datection and searching method towards monitor video | |
| CN105740780A (en) | Method and device for human face in-vivo detection | |
| CN109214373A (en) | A kind of face identification system and method for attendance | |
| CN106548148A (en) | The recognition methodss of unknown face and system in video | |
| CN102567716B (en) | Face synthetic system and implementation method | |
| TW201201115A (en) | Facial expression recognition systems and methods and computer program products thereof | |
| CN106886216A (en) | Robot automatic tracking method and system based on RGBD Face datections | |
| CN102043953A (en) | Real-time-robust pedestrian detection method aiming at specific scene | |
| CN105447432A (en) | Face anti-fake method based on local motion pattern | |
| CN103996033B (en) | Human identification method based on tracking of human face five-sense-organ coordinates |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |