CN1686051A - Canthus and pupil location method based on VPP and improved SUSAN - Google Patents
Canthus and pupil location method based on VPP and improved SUSAN Download PDFInfo
- Publication number
- CN1686051A CN1686051A CNA2005100256663A CN200510025666A CN1686051A CN 1686051 A CN1686051 A CN 1686051A CN A2005100256663 A CNA2005100256663 A CN A2005100256663A CN 200510025666 A CN200510025666 A CN 200510025666A CN 1686051 A CN1686051 A CN 1686051A
- Authority
- CN
- China
- Prior art keywords
- pupil
- canthus
- vpf
- circle
- zone
- 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
Images
Landscapes
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
A method based on VPF and improved SVSAN for locating the canthi and pupil in order to recognize human face, sex, expression, age, etc includes such steps as determining the horizontal coordinate of pupil center according to the maximum of VPF, determining the left and right boundaries of pupil (the diameter and radius of pupil) according to the positions of the maximal and minimal first-order derivative of VPF, matching the circle with template to obtain the circle center of pupil, and determining the positions of canthi according to improved SUSCAN operator.
Description
Technical field
The present invention relates to a kind of recognition of face that is applied to, the eyeball attitude estimates, Expression Recognition based on VPF (variance projection function) and the pupil of improved SUSAN (minimum monodrome fragment attracts to examine) and the localization method at canthus, relate to image processing field.
Background technology
The human face characteristic point detection is the key technology during recognition of face, Expression Recognition, sex identification etc. are used, and the localized accuracy of its characteristic point position directly has influence on the precision of identification, and eye feature is particularly important in face characteristic.Therefore, the position of locating eyes exactly can be improved the precision of identification widely.The location of current eye position is main according to the analysis to gradation of image, image border.Because the influence of complex background and illumination, only relying on gray scale and edge, to locate eyes be the very work of difficulty but in most of the cases.
Find by prior art documents, (Rein-Lien Hsu such as Rein-Lien Hsu, MohamedAbdel-Mottaleb, Anil K.Jain Face Detection In Color images IEEETRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, VOL 24, NO.5.MAY 2002) once the statistics by great amount of samples provided the distribution of eye areas at each component of YCbCr color space, and locate the position of human eye roughly according to the distribution of these components, but for different illumination, the attitude of ethnic group and people's face, this method still can not provide correct positioning.
Summary of the invention
The objective of the invention is at the deficiencies in the prior art, propose a kind of based on the pupil of VPF and improved SUSAN and the localization method at canthus, make the pupil position location of its foundation and the method for canthus location positioning, can be used for recognition of face, sex identification, estimation of Age, fields such as eyes attitude estimation.
The present invention is achieved by the following technical solutions, the present invention orients the abscissa of the center of pupil according to the maximum of VPF, the border, the left and right sides of determining pupil according to the maximum and the minima position of VPF first derivative then, also just obtained the abscissa of pupil center location accurately, border, the pupil left and right sides is the diameter of pupil in the distance of abscissa direction, thereby obtain the radius of pupil, the circle that is the pupil radius with a radius at pupil center location abscissa place carries out template matching from top to bottom then, because the pupil corresponding gray is smaller, therefore making in the template average gray reach minimum position in the process that template moves up and down is exactly the vertical coordinate of pupil center location, and then obtains the center (the pupil center of circle) of pupil; According to the position at improved SUSAN operator detection canthus, pupil center, pupil radius, the location of also just having finished eyes after all locate the position at canthus.
The present invention includes following steps:
1. the smoothing processing of eye areas is carried out smoothing denoising to eye areas utilization median filter, improves the precision of subsequent treatment;
2. each row VPF of the image after calculating smoothly, the abscissa of location pupil center location;
This string in the pupil center location correspondence, this zone is the colour of skin above the pupil, its corresponding gray strengthens, the pupil corresponding gray of this section of mediating is less, and also be that the area of skin color corresponding gray is bigger below the pupil, maximizing is the abscissa of the center of pupil on the whole VPF that is calculated.
3. determine the abscissa in the pupil center of circle, the vertical coordinate in the pupil center of circle, the diameter of pupil, the pupil that draws circle;
The abscissa in the described pupil center of circle, by calculating the derivative of VPF in second step, this gray value that lists pixel is very approaching in the left side of pupil left margin, and its variance ratio is less, and this string of right side is on the pupil position, variance ratio is bigger, and variance has a sudden change in pupil left margin position, changes from small to big, just its derivative value is bigger, and for just, search for left from pupil center location, the maximum present position of VPF derivative is the position of pupil left margin; In like manner, search for to the right from pupil center location, the minima present position of VPF derivative is pupil right margin present position.
The ordinate in the described pupil center of circle; Seeking the ordinate in the pupil center of circle realizes by the following method: the circular shuttering that is PD with a diameter at the abscissa place of pupil center location moves from top to bottom; Calculate simultaneously the average gray under this template; Because the gray value that the pixel on the pupil is corresponding is very little; When template just covers on the pupil; Its average gray value reaches minimum of a value; Therefore the ordinate position that makes the interior pixel average gray of template reach minimum of a value is the ordinate in the pupil center of circle
The diameter of described pupil, because pupil is a circle, the left and right sides boundary position of pupil all finds, the distance between its left and right sides boundary position is the diameter of pupil.
4. on eye areas, use improved SUSAN operator and find some alternative canthus;
Operator is detected at described improved SUSAN angle, is meant: the corresponding rule of utilization is removed pseudo-canthus point, thereby obtains real canthus point, and concrete operation rule is as follows:
If on the A. alternative zone within border, the pupil left and right sides, this canthus point is a pseudo-canthus point so, should remove;
B. for the left eye angle, the average gray value in its fixed size zone, left side (area of skin color) should be less than the average gray value in fixed size zone, the right (white of the eye zone), therefore, for eye image, do not satisfy the angle point of above-mentioned condition on the zone that the pupil left margin is turned left and all should remove; In like manner, for the right eye angle, the average gray value in its fixed size zone, the right (area of skin color) should be less than the average gray value in its fixed size zone, left side (white of the eye zone), and for eye image, the pupil right margin angle point that does not satisfy above-mentioned condition on the zone of turning right all should be removed;
If C. a plurality of angle points are arranged in the zone of a fixed size, the angle point of Grad minimum should be removed so because real canthus point is in the eyes intersection of two contour edges up and down, and this position Grad should be very big.
Method of the present invention can obtain higher accuracy rate.Owing to used a gray value that lists each pixel, simultaneously they have been carried out statistical analysis, the position of finding the variable quantity maximum is a pupil center location, pupil left and right sides boundary position, simultaneously traditional SUSAN operator has been carried out corresponding improvement and detected eyes, experiment shows that the localized method of pupil and canthus that the present invention proposes is better than the existing localized method of traditional pupil and canthus of only considering pixel level information.In the background complexity, under the not so good situation of illumination condition method of the present invention get advantage can be more obvious.
Description of drawings
Fig. 1 is the image behind eye areas image and the process smothing filtering.
The VPF of each row of image behind Fig. 2 smothing filtering.
The position on the border, the left and right sides of the pupil that Fig. 3 finds according to the derivative of VPF and VPF.
Pupil circle and its center of circle that Fig. 4 draws.
The alternative canthus point that Fig. 5 is found by improved SUSAN operator.
The 5th step operation rules of telling about in Fig. 6 application invention content is removed pseudo-canthus point, obtains real canthus point.
The specific embodiment
Below in conjunction with specific embodiment technical scheme of the present invention is described in further detail.
The facial image database of the eye imaging that embodiment adopts.Whole invention implementation procedure is as follows:
1. the smoothing processing of eye areas image uses median filter to carry out smoothing denoising among the present invention, can further remove noise in the image by this step, can improve canthus and positioning precision of pupils greatly.The image of original image and process smoothing processing as shown in Figure 1.
2. calculate the VPF of each row of the image after level and smooth, just calculate the variance of each row, as shown in Figure 2.
3. calculate the derivative of the rapid middle VPF of previous step.As shown in Figure 3.The pupil that draws simultaneously circle, as shown in Figure 4.
4. the SUSAN operator of application enhancements finds some alternative canthus on eye areas.Experimental result as shown in Figure 5.Thereby the corresponding rule of utilization is removed pseudo-canthus point and is obtained real canthus point, as shown in Figure 6.
Claims (7)
1, a kind of based on the canthus of VPF and improved SUSAN and the localization method of pupil, it is characterized in that, orient the abscissa of the center of pupil according to the maximum of VPF, the border, the left and right sides of determining pupil according to the maximum and the minima position of VPF first derivative then, obtain the abscissa of pupil center location, border, the pupil left and right sides is the diameter of pupil in the distance of abscissa direction, obtain the radius of pupil, the circle that is the pupil radius with a radius at pupil center location abscissa place carries out template matching from top to bottom then, because the pupil corresponding gray is smaller, making average gray in the template reach minimum position in the process that template moves up and down is exactly the vertical coordinate of pupil center location, detect the position at canthus according to improved SUSAN operator, pupil center, the pupil radius, the location of also just having finished eyes after all locate the position at canthus.
2, according to claim 1ly it is characterized in that, comprise the steps: based on the canthus of VPF and improved SUSAN and the localization method of pupil
1. the smoothing processing of eye areas is carried out smoothing denoising to eye areas utilization median filter, improves the precision of subsequent treatment;
2. each row VPF of the image after calculating smoothly, the abscissa of location pupil center location;
3. determine the abscissa in the pupil center of circle, the vertical coordinate in the pupil center of circle, the diameter of pupil, the pupil that draws circle;
4. on eye areas, use improved SUSAN operator and find some alternative canthus.
3, according to claim 1 based on the canthus of VPF and improved SUSAN and the localization method of pupil, it is characterized in that, described step 2., be meant: at this string of pupil center location correspondence, this zone is the colour of skin above the pupil, and its corresponding gray strengthens, and the pupil corresponding gray of this section that mediate is less, and also be that the area of skin color corresponding gray is bigger below the pupil, maximizing is the abscissa of the center of pupil on the whole VPF that is calculated.
4, according to claim 1 based on the canthus of VPF and improved SUSAN and the localization method of pupil, it is characterized in that, the abscissa in the described pupil center of circle, by calculating the derivative of VPF in second step, this gray value that lists pixel is very approaching in the left side of pupil left margin, its variance ratio is less, and this string of right side is on the pupil position, variance ratio is bigger, and variance has a sudden change in pupil left margin position, changes from small to big, just its derivative value is bigger, and for just, search for left from pupil center location, the maximum present position of VPF derivative is the position of pupil left margin; In like manner, search for to the right from pupil center location, the minima present position of VPF derivative is pupil right margin present position.
5, according to claim 1 based on the canthus of VPF and improved SUSAN and the localization method of pupil, it is characterized in that, the vertical coordinate in the described pupil center of circle, seeking the vertical coordinate in the pupil center of circle realizes by the following method: the circular shuttering that is pupil diameter with a diameter at the abscissa place of pupil center location moves from top to bottom, calculate the average gray under this template simultaneously, because the pixel corresponding gray on the pupil is very little, when template just covers on the pupil, its average gray value reaches minima, and the vertical coordinate position that therefore makes the interior pixel average gray of template reach minima is the vertical coordinate in the pupil center of circle.
6, according to claim 1 based on the canthus of VPF and improved SUSAN and the localization method of pupil, it is characterized in that the diameter of described pupil is because pupil is a circle, the left and right sides boundary position of pupil all finds, and the distance between its left and right sides boundary position is the diameter of pupil.
7, according to claim 1 based on the canthus of VPF and improved SUSAN and the localization method of pupil, it is characterized in that operator is detected at described improved SUSAN angle, is meant: the corresponding rule of utilization is removed pseudo-canthus point, thereby obtain real canthus point, concrete operation rule is as follows:
If on the A. alternative zone within border, the pupil left and right sides, this canthus point is a pseudo-canthus point so, should remove;
B. for the left eye angle, the average gray value in its fixed size zone, left side (area of skin color) should be less than the average gray value in fixed size zone, the right (white of the eye zone), therefore, for eye image, do not satisfy the angle point of above-mentioned condition on the zone that the pupil left margin is turned left and all should remove; In like manner, for the right eye angle, the average gray value in its fixed size zone, the right (area of skin color) should be less than the average gray value in its fixed size zone, left side (white of the eye zone), and for eye image, the pupil right margin angle point that does not satisfy above-mentioned condition on the zone of turning right all should be removed;
If C. a plurality of angle points are arranged in the zone of a fixed size, the angle point of Grad minimum should be removed so because real canthus point is in the eyes intersection of two contour edges up and down, and this position Grad should be very big.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2005100256663A CN100382751C (en) | 2005-05-08 | 2005-05-08 | Canthus and pupil location method based on VPP and improved SUSAN |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2005100256663A CN100382751C (en) | 2005-05-08 | 2005-05-08 | Canthus and pupil location method based on VPP and improved SUSAN |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN1686051A true CN1686051A (en) | 2005-10-26 |
| CN100382751C CN100382751C (en) | 2008-04-23 |
Family
ID=35304092
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB2005100256663A Expired - Fee Related CN100382751C (en) | 2005-05-08 | 2005-05-08 | Canthus and pupil location method based on VPP and improved SUSAN |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN100382751C (en) |
Cited By (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101893858A (en) * | 2010-07-15 | 2010-11-24 | 华中科技大学 | Method for controlling distance between eyes of user and screen of electronic equipment |
| CN102362291A (en) * | 2009-04-02 | 2012-02-22 | 爱信精机株式会社 | Face feature point detection device and program |
| CN101639939B (en) * | 2008-07-29 | 2012-06-20 | 株式会社日立制作所 | Image information processing method and apparatus |
| CN102622740A (en) * | 2011-01-28 | 2012-08-01 | 鸿富锦精密工业（深圳）有限公司 | Anti-eye-closure portrait shooting system and method |
| CN101573714B (en) * | 2006-10-02 | 2012-11-14 | 强生消费者公司 | Method and apparatus for identifying facial regions |
| CN103440476A (en) * | 2013-08-26 | 2013-12-11 | 大连理工大学 | Locating method for pupil in face video |
| WO2014121573A1 (en) * | 2013-02-06 | 2014-08-14 | 华中科技大学 | Susan edge detection method and system based on non-local mean |
| CN104573660A (en) * | 2015-01-13 | 2015-04-29 | 青岛大学 | Method for precisely positioning human eyes by SIFT point descriptor |
| CN104809458A (en) * | 2014-12-29 | 2015-07-29 | 华为技术有限公司 | Pupil center positioning method and pupil center positioning device |
| CN105760848A (en) * | 2016-03-04 | 2016-07-13 | 蒋志平 | Pupil positioning method based on annular template convolution |
| WO2017092679A1 (en) * | 2015-12-02 | 2017-06-08 | 中国银联股份有限公司 | Eyeball tracking method and apparatus, and device |
| CN106919933A (en) * | 2017-03-13 | 2017-07-04 | 重庆贝奥新视野医疗设备有限公司 | The method and device of Pupil diameter |
| CN107516067A (en) * | 2017-07-21 | 2017-12-26 | 深圳市梦网百科信息技术有限公司 | A kind of human-eye positioning method and system based on Face Detection |
| CN107808165A (en) * | 2017-10-19 | 2018-03-16 | 南京理工大学 | A kind of infrared image matching process based on SUSAN Corner Detections |
| CN108446658A (en) * | 2018-03-28 | 2018-08-24 | 百度在线网络技术（北京）有限公司 | The method and apparatus of facial image for identification |
| CN109086713A (en) * | 2018-07-27 | 2018-12-25 | 腾讯科技（深圳）有限公司 | Eye recognition method, apparatus, terminal and storage medium |
| CN110051319A (en) * | 2019-04-23 | 2019-07-26 | 七鑫易维(深圳)科技有限公司 | Adjusting method, device, equipment and the storage medium of eyeball tracking sensor |
| WO2021093744A1 (en) * | 2019-11-11 | 2021-05-20 | 中国科学院深圳先进技术研究院 | Method and apparatus for measuring diameter of pupil, and computer-readable storage medium |
| CN116421135A (en) * | 2023-04-27 | 2023-07-14 | 北京京东拓先科技有限公司 | Vision testing method, device, electronic equipment and storage medium |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5231674A (en) * | 1989-06-09 | 1993-07-27 | Lc Technologies, Inc. | Eye tracking method and apparatus |
| JPH08287216A (en) * | 1995-04-18 | 1996-11-01 | Sanyo Electric Co Ltd | In-face position recognizing method |
| CN1204531C (en) * | 2003-07-14 | 2005-06-01 | 中国科学院计算技术研究所 | Human eye location method based on GaborEge model |
-
2005
- 2005-05-08 CN CNB2005100256663A patent/CN100382751C/en not_active Expired - Fee Related
Cited By (27)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101573714B (en) * | 2006-10-02 | 2012-11-14 | 强生消费者公司 | Method and apparatus for identifying facial regions |
| CN101639939B (en) * | 2008-07-29 | 2012-06-20 | 株式会社日立制作所 | Image information processing method and apparatus |
| CN102362291A (en) * | 2009-04-02 | 2012-02-22 | 爱信精机株式会社 | Face feature point detection device and program |
| CN102362291B (en) * | 2009-04-02 | 2013-09-04 | 爱信精机株式会社 | Face feature point detection device and method |
| CN101893858A (en) * | 2010-07-15 | 2010-11-24 | 华中科技大学 | Method for controlling distance between eyes of user and screen of electronic equipment |
| CN102622740A (en) * | 2011-01-28 | 2012-08-01 | 鸿富锦精密工业（深圳）有限公司 | Anti-eye-closure portrait shooting system and method |
| CN102622740B (en) * | 2011-01-28 | 2016-07-20 | 鸿富锦精密工业（深圳）有限公司 | Anti-eye closing portrait system and method |
| WO2014121573A1 (en) * | 2013-02-06 | 2014-08-14 | 华中科技大学 | Susan edge detection method and system based on non-local mean |
| CN103440476A (en) * | 2013-08-26 | 2013-12-11 | 大连理工大学 | Locating method for pupil in face video |
| CN104809458A (en) * | 2014-12-29 | 2015-07-29 | 华为技术有限公司 | Pupil center positioning method and pupil center positioning device |
| CN104809458B (en) * | 2014-12-29 | 2018-09-28 | 华为技术有限公司 | A kind of pupil center's localization method and device |
| CN104573660A (en) * | 2015-01-13 | 2015-04-29 | 青岛大学 | Method for precisely positioning human eyes by SIFT point descriptor |
| WO2017092679A1 (en) * | 2015-12-02 | 2017-06-08 | 中国银联股份有限公司 | Eyeball tracking method and apparatus, and device |
| US10699420B2 (en) | 2015-12-02 | 2020-06-30 | China Unionpay Co., Ltd. | Eyeball tracking method and apparatus, and device |
| CN105760848B (en) * | 2016-03-04 | 2019-05-14 | 蒋志平 | A kind of pupil positioning method based on annular mask convolution |
| CN105760848A (en) * | 2016-03-04 | 2016-07-13 | 蒋志平 | Pupil positioning method based on annular template convolution |
| CN106919933A (en) * | 2017-03-13 | 2017-07-04 | 重庆贝奥新视野医疗设备有限公司 | The method and device of Pupil diameter |
| CN107516067A (en) * | 2017-07-21 | 2017-12-26 | 深圳市梦网百科信息技术有限公司 | A kind of human-eye positioning method and system based on Face Detection |
| CN107516067B (en) * | 2017-07-21 | 2020-08-04 | 深圳市梦网视讯有限公司 | Human eye positioning method and system based on skin color detection |
| CN107808165A (en) * | 2017-10-19 | 2018-03-16 | 南京理工大学 | A kind of infrared image matching process based on SUSAN Corner Detections |
| CN107808165B (en) * | 2017-10-19 | 2021-04-16 | 南京理工大学 | Infrared image matching method based on SUSAN corner detection |
| CN108446658A (en) * | 2018-03-28 | 2018-08-24 | 百度在线网络技术（北京）有限公司 | The method and apparatus of facial image for identification |
| CN109086713A (en) * | 2018-07-27 | 2018-12-25 | 腾讯科技（深圳）有限公司 | Eye recognition method, apparatus, terminal and storage medium |
| CN109086713B (en) * | 2018-07-27 | 2019-11-15 | 腾讯科技（深圳）有限公司 | Eye recognition method, apparatus, terminal and storage medium |
| CN110051319A (en) * | 2019-04-23 | 2019-07-26 | 七鑫易维(深圳)科技有限公司 | Adjusting method, device, equipment and the storage medium of eyeball tracking sensor |
| WO2021093744A1 (en) * | 2019-11-11 | 2021-05-20 | 中国科学院深圳先进技术研究院 | Method and apparatus for measuring diameter of pupil, and computer-readable storage medium |
| CN116421135A (en) * | 2023-04-27 | 2023-07-14 | 北京京东拓先科技有限公司 | Vision testing method, device, electronic equipment and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN100382751C (en) | 2008-04-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN1686051A (en) | Canthus and pupil location method based on VPP and improved SUSAN | |
| Peng et al. | A robust agorithm for eye detection on gray intensity face without spectacles | |
| Gu et al. | Feature points extraction from faces | |
| CN1731416A (en) | Method of quick and accurate human face feature point positioning | |
| CN107330371A (en) | Acquisition methods, device and the storage device of the countenance of 3D facial models | |
| CN107368778A (en) | Method for catching, device and the storage device of human face expression | |
| CN1687957A (en) | Man face characteristic point positioning method of combining local searching and movable appearance model | |
| US8325996B2 (en) | Method and device for locating a human iris in an eye image | |
| CN106127145B (en) | Pupil diameter and tracking | |
| Melli et al. | Comparison of color clustering algorithms for segmentation of dermatological images | |
| CN110163119A (en) | A kind of finger vein identification method and system | |
| CN106778499B (en) | Method for rapidly positioning human iris in iris acquisition process | |
| CN106203375A (en) | A kind of based on face in facial image with the pupil positioning method of human eye detection | |
| CN109840484B (en) | Pupil detection method based on edge filtering, ellipse evaluation and pupil verification | |
| CN110021029B (en) | Real-time dynamic registration method and storage medium suitable for RGBD-SLAM | |
| CN106960199B (en) | Complete extraction method of white eye region of true color eye picture | |
| CN1204531C (en) | Human eye location method based on GaborEge model | |
| CN116051631A (en) | Light spot labeling method and system | |
| CN110929570B (en) | Iris rapid positioning device and positioning method thereof | |
| CN1166313A (en) | Iris recoganizing method | |
| CN114202795A (en) | Method for quickly positioning pupils of old people | |
| Xu et al. | A robust and accurate method for pupil features extra | |
| CN113947805A (en) | Eye shake type classification method based on video image | |
| JP7044504B2 (en) | Image processing device, image processing method and image processing program | |
| WO2016004706A1 (en) | Method for improving iris recognition performance in non-ideal environment |
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 | ||
| C17 | Cessation of patent right | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20080423 Termination date: 20110508 |