CN102419819A - Method and system for recognizing human face image - Google Patents
Method and system for recognizing human face image Download PDFInfo
- Publication number
- CN102419819A CN102419819A CN2010105172326A CN201010517232A CN102419819A CN 102419819 A CN102419819 A CN 102419819A CN 2010105172326 A CN2010105172326 A CN 2010105172326A CN 201010517232 A CN201010517232 A CN 201010517232A CN 102419819 A CN102419819 A CN 102419819A
- Authority
- CN
- China
- Prior art keywords
- template
- user
- face
- image
- recognition
- 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
- 238000000034 method Methods 0.000 title claims abstract description 67
- 238000012545 processing Methods 0.000 claims abstract description 12
- 230000001815 facial effect Effects 0.000 claims description 41
- 230000008569 process Effects 0.000 claims description 18
- 239000000284 extract Substances 0.000 claims description 16
- 238000000605 extraction Methods 0.000 claims description 16
- 238000010606 normalization Methods 0.000 claims description 12
- 230000006870 function Effects 0.000 claims description 9
- 230000004044 response Effects 0.000 claims description 8
- 238000005286 illumination Methods 0.000 claims description 7
- 238000006243 chemical reaction Methods 0.000 claims description 5
- 239000012141 concentrate Substances 0.000 claims description 4
- 238000001914 filtration Methods 0.000 claims description 4
- 230000008676 import Effects 0.000 claims description 3
- 230000008901 benefit Effects 0.000 abstract description 3
- 230000000295 complement effect Effects 0.000 abstract description 2
- 239000004065 semiconductor Substances 0.000 abstract 1
- 238000012163 sequencing technique Methods 0.000 abstract 1
- 238000005457 optimization Methods 0.000 description 9
- 230000008878 coupling Effects 0.000 description 8
- 238000010168 coupling process Methods 0.000 description 8
- 238000005859 coupling reaction Methods 0.000 description 8
- 238000005516 engineering process Methods 0.000 description 6
- 238000013461 design Methods 0.000 description 5
- 230000008859 change Effects 0.000 description 2
- 230000005764 inhibitory process Effects 0.000 description 2
- 230000013011 mating Effects 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000002950 deficient Effects 0.000 description 1
- 238000000151 deposition Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 210000000887 face Anatomy 0.000 description 1
- 239000012467 final product Substances 0.000 description 1
- 230000008707 rearrangement Effects 0.000 description 1
Images
Landscapes
- Collating Specific Patterns (AREA)
- Image Analysis (AREA)
Abstract
The invention relates to a method and a system for recognizing a human face image to solve problems of low recognition speed and low recognition rate in the prior art. In the method, a user template library is set up firstly and then recognition is performed; and for the user template library, during user registration, a user changes N kinds of attitudes, and three frames of images are collected and characteristic templates are extracted at each attitude, wherein a template sequencing principle is as follows: A, the selected templates represent the unselected templates as far as possible, that is to say, the selected templates have the maximum similarity with other candidate templates; and B, the selected templates are far from the templates which are selected before as much as possible, that is to say, the selected templates keep the minimum similarity with the templates which are selected before. In the system, an embedded microprocessor is used as a central processing unit of the system, a CMOS (Complementary Metal-Oxide-Semiconductor Transistor) sensor is used for collecting human face images, a human face infrared lighting device is used as an auxiliary device, and a narrow-band infrared filter is used to filter visible light. According to the method and the system for recognizing the human face image, disclosed by the invention, the problems of low recognition speed and low recognition rate can be solved efficiently, and both the speed and the recognition rate can reach a commercial application level. The system has the advantages of usability under any light condition, capability of removing interference from the ambient light, good reliability, and recognition rate as well as recognition speed both of which reach a satisfactory level.
Description
Technical field
The present invention designs a kind of living things feature recognition method, particularly relates to a kind of facial image recognition method and system.
Background technology
At present, the facial image recognition method that is used for person identification is different, but these methods all low, Flame Image Process of ubiquity discrimination and the slow problem of recognition speed.This makes that face recognition technology is difficult to be applied in the embedded system.
Summary of the invention
The objective of the invention is to overcome the defective of prior art, propose and realized a kind ofly can when significantly improving processing speed, keeping the facial image recognition method of better discrimination; The object of the invention also is to be provided for implementing the Face Image Recognition System of this method.
Be to realize above-mentioned purpose, inventor's face image identifying method is to set up the user template storehouse earlier, discerns as follows again: 1) gather facial image to be identified, 2) extract template to be identified, 3) user's recognition of face; In order to realize complete people's face identification system, the registration of user people's face and two major functions of recognition of face need be provided.
Said user's recognition of face is: template to be identified first template contrast with each user in the user template storehouse, obtain the user list A1 of all similarity marks greater than first lower threshold, arrange from big to small by the similarity mark; If A1 is sky then recognition failures; If first user's similarity mark is greater than upper limit threshold among the A1; Then discern and successfully return respective user,, then carry out next step: 2-5 the template comparison of template to be identified with each user among the A1 if not; Obtain the user list A2 of all marks, arrange from big to small by the similarity mark greater than second lower threshold; If A2 is sky then recognition failures; If first user's similarity mark is then discerned and is successfully returned respective user, if not greater than upper limit threshold among the A2; Then carry out next step: the 6-15 template contrast of template to be identified with each user among the A2; Obtain the user list A3 of all similarity marks, if A3 is sky, then recognition failures greater than recognition threshold; Arrange from big to small by the similarity mark, recognition function is returned respective user;
Said user template storehouse is when the registered user; 5 kinds of attitudes of user's conversion; Under each attitude, gather three two field pictures and extract feature templates, obtain 15 feature templates altogether, and through concentrate select progressively to go out the method ordering of template from candidate template; The template that ordering is selected based on following two principle: A. is represented the non-selected template that goes out as far as possible, and promptly the candidate template of it and other has maximum similarity; B. the template of selecting is as far as possible away from the template that has chosen, and promptly it maintains minimum similarity with modeling plate;
Said upper limit threshold is greater than recognition threshold, and recognition threshold is greater than second lower threshold, and second lower threshold is greater than first lower threshold.It has and can when significantly improving processing speed, keep the advantage of better discrimination.
As optimization, discern successfully after, if template to be identified and user's similarity mark are when setting the study threshold value than bigger one of upper limit threshold; Then waiting to learn template to the template of current collection in worksite as one learns as follows: import template to be learnt into to the user template storehouse; Form new template to be sorted to the template of waiting to learn template and user registration, ordering: if last template is a template to be learnt, then respective user is returned in the study failure; If no; Then remove last template, deposit database in, learning success returns respective user.
As optimization, said setting study threshold value is 100.
As optimization, said first lower threshold is 43, and upper limit threshold is that 90, the second lower thresholds are 60, and recognition threshold is 80.
As optimization, the sort method of said 15 feature templates is: select first template, make that the similarity mark average of itself and other template is maximum; Move on to it in the modeling plate then, select second template again, make that the similarity mark average of itself and other template is maximum; Move on to it in the modeling plate then, select the 3rd template again, make that the similarity mark average of itself and other template is maximum; Move on to it in the modeling plate then, and the like up to there not being candidate template.
As optimization, the similarity mark of selected template and other templates calculates and gets according to following formula:
C＝a1×c1-a2×c2；
In the formula: c1 representes the similarity mark average of this template and other candidate template, and c2 representes this template and the similarity mark average of modeling plate, and a1, a2 are two parameters.
As optimization, parameter a1 value is 5/9, and parameter a2 value is 4/9.
As optimization, said user template storehouse is built up through the following step: 1) gather user's facial image, 2) extraction user characteristics template, 3) registered user;
Said 1) collection facial image to be identified and 1) gathering user's facial image is: use the infrared LED lighting source to be shone gathering people's face, in gatherer process, also visible light is filtered;
Said 2) extraction template to be identified and 2) steps in sequence of extraction user characteristics template is: 1) detect people's face; 2) location eye position; 3) regular facial image; 4) assessment quality of human face image; 5) Gabor feature extraction.
As optimization, said 1) detecting people's face is: on integrogram, detect the harr characteristic, use the AdaBoost algorithm to carry out people's face then and detect; Said 2) the location eye position is: through the symmetry of people's face and the darker characteristic of gray scale of position of human eye, orient in the picture people position of right and left eyes on the face; Said 3) regular facial image is: at first according to the position of right and left eyes, image is rotated, makes right and left eyes be in same horizontal line, convergent-divergent entire image then is the place normalization to of a right and left eyes fixing width; Further, according to the position of eyes, go out the image of whole people's face according to fixing width and height intercepting; At last image is carried out the part and strengthens, in the removal of images because the luminance difference that uneven illumination etc. produce; Obtain the image of people's face key position at last; Said 4) the assessment quality of human face image is: in the above image is carried out in the normalized processing procedure; Calculate acutance, contrast and the gray level of image simultaneously; And these three indexs of having portrayed picture quality simultaneously have only the threshold value separately that surpasses setting in advance just to carry out next step processing; Said 5) the Gabor feature extraction is: the image of one group of two-dimensional Gabor wavelet basis that we use N frequency range, a M direction after to normalization carries out filtering; Thereby obtain the response energy of image under this group wavelet basis; After the response energy carried out normalization and handle, face characteristic template to the end.
As optimization, the image of the said people's of obtaining face key position is: getting right and left eyes distance is 75 pixels, from move 35 pixels, 38 locations of pixels that move to left begin, getting width is 151, highly is 151 image;
Said N >=8, M >=16.
Be used to realize that the Face Image Recognition System of the method for the invention is to use embedded microprocessor to connect the cmos sensor of gathering facial image as CPU, the CPU of system; And the infrared illuminator that is aided with people's face realizes the functions of use under any light condition, adopts the arrowband infrared fileter to filter the interference that visible light is got rid of surround lighting.It is complete, a reliable face identification system, and its discrimination and recognition speed all reach gratifying level, can be applied in the systems such as work attendance, gate inhibition.Gather the face images of users aspect, a good system, the image of its collection should be got rid of interference and noise clear as far as possible, as far as possible, because this accuracy to identification is extremely important.Consider the service condition of a true application system, promptly this system may be installed under the various photoenvironments, even in night, uses, so we can not depend on environment suitable, illumination uniformly can be provided.Based on this reason, specialized designs infrared LED lighting source solves this problem among the present invention, simultaneously because infrared ray promptly can not constitute harm to the user, because its invisibility can not cause interference to the user yet.
After adopting technique scheme; Inventor's face image identifying method carries out user's registration and identification; Can effectively solve the low and slow problem of recognition speed of discrimination, and can be when significantly improving processing speed, its speed and discrimination all can reach the commercial applications level.
Face Image Recognition System of the present invention has the use under any light condition of ability, can get rid of the interference that can fall surround lighting, good reliability, and discrimination and recognition speed all reach the advantage of satisfactory level.
Description of drawings
Fig. 1 is the principle schematic of Face Image Recognition System of the present invention;
Fig. 2 is the main schematic flow sheet of Face Image Recognition System of the present invention;
Fig. 3 is the template extraction schematic flow sheet of inventor's face image identifying method;
Fig. 4 is user's registering flow path synoptic diagram of inventor's face image identifying method;
Fig. 5 is the template ordering schematic flow sheet of inventor's face image identifying method;
Fig. 6 is the User Recognition schematic flow sheet of inventor's face image identifying method;
Fig. 7 is the face template self study schematic flow sheet of inventor's face image identifying method.
Embodiment
Shown in Fig. 2-7, inventor's face image identifying method is to set up the user template storehouse earlier, discerns as follows again: 1) gather facial image to be identified, 2) extract template to be identified, 3) user's recognition of face;
Said user's recognition of face is: first template contrast of template to be identified with each user in the user template storehouse, obtain all similarity marks greater than 43 user list A1, arrange from big to small by the similarity mark; If A1 is sky then recognition failures; If first user's similarity mark is greater than upper limit threshold 90 among the A1; Then discern and successfully return respective user,, then carry out next step: 2-5 the template comparison of template to be identified with each user among the A1 if not; Obtain all marks greater than 60 user list A2, arrange from big to small by the similarity mark; If A2 is sky then recognition failures, if first user's similarity mark is then discerned and successfully returned respective user greater than upper limit threshold 90 among the A2; If no; Then carry out next step: with the contrast of each user's among the A2 6-15 template, obtain the user list A3 of all similarity marks, if A3 is sky then recognition failures to template to be identified greater than threshold value 80; Arrange from big to small by the similarity mark, recognition function is returned respective user;
Said user template storehouse is when the registered user; 5 kinds of attitudes of user's conversion; Under each attitude, gather three two field pictures and extract feature templates, obtain 15 feature templates altogether, and through concentrate select progressively to go out the method ordering of template from candidate template; The template that ordering is selected based on following two principle: A. is represented the non-selected template that goes out as far as possible, and promptly the candidate template of it and other has maximum similarity; B. the template of selecting is as far as possible away from the template that has chosen, and promptly it maintains minimum similarity with modeling plate.
After discerning successfully; If template to be identified and user's similarity mark greater than than the bigger setting threshold 100 of said upper limit threshold 90 time, are then waited to learn template to the template of current collection in worksite as one and are learnt as follows: import template to be learnt into to the user template storehouse, form new template to be sorted to the template of waiting to learn template and user registration; Ordering: if last template is a template to be learnt; Then respective user is returned in the study failure, if not, then removes last template; Deposit database in, learning success returns respective user.
The sort method of said 15 feature templates is: select first template, make that the similarity mark average of itself and other template is maximum, move on to it in the modeling plate then; Select second template again, make that the similarity mark average of itself and other template is maximum, move on to it in the modeling plate then; Select the 3rd template again; Make that the similarity mark average of itself and other template is maximum, move on to it in the modeling plate then, and the like up to there not being candidate template.
The similarity mark of selected template and other templates calculates and gets according to following formula:
C＝a1×c1-a2×c2；
In the formula: c1 representes the similarity mark average of this template and other candidate template, and c2 representes this template and the similarity mark average of modeling plate, and a1, a2 are two parameters; Wherein parameter a1 value is 5/9, and parameter a2 value is 4/9.
Said user template storehouse is built up through the following step: 1) gather user's facial image, 2) extraction user characteristics template, 3) registered user;
Said 1) collection facial image to be identified and 1) gathering user's facial image is: use the infrared LED lighting source to be shone gathering people's face, in gatherer process, also visible light is filtered;
Said 2) extraction template to be identified and 2) steps in sequence of extraction user characteristics template is: 1) detect people's face; 2) location eye position; 3) regular facial image; 4) assessment quality of human face image; 5) Gabor feature extraction.
Said 1) detecting people's face is: on integrogram, detect the harr characteristic, use the AdaBoost algorithm to carry out people's face then and detect; Said 2) the location eye position is: through the symmetry of people's face and the darker characteristic of gray scale of position of human eye, orient in the picture people position of right and left eyes on the face; Said 3) regular facial image is: at first according to the position of right and left eyes, image is rotated, makes right and left eyes be in same horizontal line, convergent-divergent entire image then is the place normalization to of a right and left eyes fixing width; Further, according to the position of eyes, go out the image of whole people's face according to fixing width and height intercepting; At last image is carried out the part and strengthens, in the removal of images because the luminance difference that uneven illumination etc. produce; Obtain the image of people's face key position at last; Said 4) the assessment quality of human face image is: in the above image is carried out in the normalized processing procedure; Calculate acutance, contrast and the gray level of image simultaneously; And these three indexs of having portrayed picture quality simultaneously have only the threshold value separately that surpasses setting in advance just to carry out next step processing; Said 5) the Gabor feature extraction is: the image of one group of two-dimensional Gabor wavelet basis that we use N frequency range, a M direction after to normalization carries out filtering; Thereby obtain the response energy of image under this group wavelet basis; After the response energy carried out normalization and handle, face characteristic template to the end.The image of the said people's of obtaining face key position is: getting right and left eyes distance is 75 pixels, from move 35 pixels, 38 locations of pixels that move to left begin, getting width is 151, highly is 151 image; Said N >=8, M >=16.
More specifically be:, the registration of user people's face and two major functions of recognition of face need be provided in order to realize complete people's face identification system.
Enrollment process, be a facial image of gathering extract a feature templates (among the biostatistics biometrics, term " template " and the biological characteristic that extracts that template just is meant) after, be saved in the database; And comparison process then is after extracting feature templates to the facial image of gathering, and the template of depositing in the database during with registration is compared one by one, up to the template that finds coupling.
1, gathers facial image
The key of these two functions is to gather face images of users, and extracts the characteristic of people's face from this image.Gather the face images of users aspect, a good system, the image of its collection should be got rid of interference and noise clear as far as possible, as far as possible, because this accuracy to identification is extremely important.Consider the service condition of a true application system, promptly this system may be installed under the various photoenvironments, even in night, uses, so we can not depend on environment suitable, illumination uniformly can be provided.Based on this reason, specialized designs infrared LED lighting source solves this problem among the present invention, simultaneously because infrared ray promptly can not constitute harm to the user, because its invisibility can not cause interference to the user yet.
2, extract feature templates
Extract the process of feature templates from original image, the biometrics identification technology most critical part normally, face recognition technology also is like this.Among the present invention, we realize as the next characteristic flow process of extracting, and can extract the feature templates of people's face quickly and efficiently.
1) detects people's face: on integrogram, detect the harr characteristic, use the AdaBoost algorithm to carry out people's face then and detect;
2) location eye position:, orient in the picture people position of right and left eyes on the face through the symmetry of people's face and the darker characteristic of gray scale of position of human eye;
3) regular facial image: we at first according to the position of right and left eyes, are rotated image, make right and left eyes be in same horizontal line, and convergent-divergent entire image then is the place normalization to of a right and left eyes fixing width; Further, we go out the image of whole people's face according to the position of eyes according to fixing width and height intercepting; We carry out the part to image and strengthen at last, in the removal of images because the luminance difference that uneven illumination etc. produce.We get right and left eyes distance is 75 pixels, from move 35 pixels, 38 locations of pixels that move to left begin, getting width is 151, highly is 151 image, just in time is people's face key component image.
4) assessment quality of human face image: we find, the quality of quality of human face image has very big influence for the identification accuracy.The characteristic that bad picture quality extracts is just unreliable.Therefore in order to improve the performance of system, we propose suitable requirement to picture quality.In the above image is carried out in the normalized processing procedure, calculate acutance, contrast and the gray level of image simultaneously, these three quality that index has been portrayed image simultaneously.We set corresponding threshold, have only the processing of just carrying out next step above these threshold values.
5) Gabor feature extraction: the image of one group of two-dimensional Gabor wavelet basis that we use 8 frequency ranges, 16 directions after to normalization carries out filtering, thereby obtains the response energy of image under this group wavelet basis.After the response energy carried out normalization and handle, we got face characteristic template to the end.
3, registered user
Because people's face is three-dimensional, the image difference that people's face is gathered under different attitudes can be very big, because the image of an attitude only can be portrayed the two dimensional character of people's face under this attitude, the very difficult three-dimensional feature that gives expression to people's face fully comes.If, will inevitably obtain very low discrimination so we only register a user according to the image of a frame people face.Therefore we need gather a plurality of images of people's face in a plurality of attitudes, just can obtain people's face three-dimensional feature as much as possible.Only in this way, in the application of reality, just can obtain acceptable discrimination.
Further handle, common face recognition technology is the image that merges a plurality of attitudes, therefrom analyzes the geometric expression of messenger's face three-dimensional feature, comes the construction feature template according to this, Here it is so-called 3D face recognition technology.Yet this method can be brought very big computing cost, and neither be very accurate, in embedded system is impossible realize.We find that through people's face picture of analyzing these different attitudes the similarity of the image that the attitude of same people's face is close is very high, and the similarity of the image of the different people face of different attitudes is then very low.The present invention designs a kind of method in view of the above, and this method is carried out user's registration and identification, can effectively solve the low and slow problem of recognition speed of discrimination.
We require 5 kinds of attitudes of user's conversion when the registered user, system can gather three two field pictures and extract feature templates under each attitude.We can obtain 15 feature templates altogether like this.The process of gathering is through voice suggestion user conversion attitude, and system gathers automatically and calculates simultaneously, and this makes that this process can't be difficult to very much use.Because in actual use; Often the user can strict not follow the variation of the time generation attitude of voice suggestion; User's attitude also not necessarily meets our requirement fully, so we can not guarantee that 15 templates that collect at last are every kind of 3 templates of 5 kinds of attitudes accurately.This template that makes us can not extract every kind of attitude is simply carried out subsequent treatment as representative.The present invention deposits these 15 templates in database simply, but to they advanced line orderings.
Our method ordering through concentrate select progressively to go out template from candidate template, based on two principles:
A. the template of selecting is represented the non-selected template that goes out as far as possible, and promptly the candidate template of it and other has maximum similarity;
B. the template of selecting is as far as possible away from the template that has chosen, and promptly it maintains minimum similarity with modeling plate; These two principles have guaranteed that preceding 5 templates can represent the face characteristic of 5 attitudes basically.10 templates then are the repetition to preceding face die plate basically, also have some to replenish.Through experiment, we obtain the parameter a1 in the top flow process, and the best value of a2 is a1=5/9, a2=4/9.
5, recognition of face
In identifying, gathering facial image is the same with the process of extracting feature templates during with registrant's face, and certain we just discern behind collection one two field picture immediately at this moment.
The result of two people's face feature templates coupling is a similarity scope at 0～120 mark.If this mark is 120, just represent that two people's faces are complementary fully; Be that zero expression does not match fully.In order to improve discrimination, we can be 120 as the foundations of judging same people's face, but selected threshold value, when similarity mark during greater than this threshold value, just conclude that two face templates come from people's face.According to the result who in large-scale face database, tests, threshold value is 80 o'clock, and our system can obtain the false recognition rate about 1/100000.We adopt this threshold value in common application.
The coupling flow process that we design is divided into three phases according to our storage order to the user fingerprints template: first is discerned in first template of each user; But get a lower threshold value (first lower threshold); For example 43; We can filter out a part of user and come out like this, carry out the comparison of subordinate phase.Therefore after the coupling through first stage, probably there is 70% user to be filtered, participates in the user only remaining about 30% of the comparison of subordinate phase.In the subordinate phase comparison, we choose a higher threshold value (second lower threshold), and for example 60, to have 95% user like this can be filtered, and therefore participates in the user only remaining about 5% of the comparison of phase III.Three phases, the carrying out of all the other 10 templates of our user of in the end select 5% compared one by one, result to the end.Can find out that from this step the complexity that the template matches of this flow process is calculated is:
Template matches calculation times=C+C*30%*4+C*30%*4*5%*10=2.8*C
Wherein C is the number of database.With 1000 people is example, and we need carry out 2800 template couplings at most and calculate.In fact, our test shows, the overwhelming majority finally can not unmatched template can not get any possible matching result usually in the coupling of subordinate phase, the coupling that therefore can not get into the phase III.In other words, the computation complexity in the actual process can than top formula calculate much lower.
We also set a upper limit threshold; For example 90, in the comparison in each stage, in case this upper limit threshold that surpasses of a template and our on-the-spot template comparison is arranged; Just can think immediately and find the template of mating, and need not continue the comparison of next procedure.Like this in the result that can obtain apace mating.
Therefore; Though the present invention has preserved 15 templates for each user in database; But through top flow process, promptly let 15 templates play a role, greatly improved discrimination; Because there is not the number of times of large increase template comparison, total system finally can also be discerned apace simultaneously.
6, the study of face template
As time goes on the characteristic of people's face also can produce some and change for example fat and thin change.Therefore our system preferably can in use be reflected to these variations on user's the registration template.So as time goes by, user's registration template also can that is to say to have autolearn feature along with keeping renewal, is unlikely to after people's face variation of user runs up to a certain degree, to cause system finally can't discern.
We set a strategy; In the process that the user uses; Identification case of successful under, if the coupling mark greater than a given upper limit threshold, for example 100 minutes; Then we be sure of that current recognition result is absolutely correct, and we learn the template of current collection in worksite as a template to be learnt like this.
In this flow process, we adopt with registration during template the same method to registered 15 templates of user with wait to learn template and sort together, remove then and come a last template.
Behind the learning success, 15 templates that the storehouse that Updates Information writes this user's rearrangement get final product.
Algorithm flow and technical scheme that the present invention describes are applied on our many moneys equipment.These equipment adopt the Embedded Application processor P XA310 of Marvel, and its speed and discrimination have all reached the commercial applications level.
Shown in Fig. 1-2; The present invention is used to realize that the Face Image Recognition System of said method of the present invention is to use embedded microprocessor to connect the cmos sensor 2 of gathering facial image as CPU or microprocessor core core 1, CPU or the microprocessor core core 1 of system; And the infrared illuminator infrared LED light source 3 that is aided with people's face realizes the functions of use under any light condition, adopts arrowband infrared fileter 4 to filter the interference that visible lights are got rid of surround lighting.Among Fig. 1: the user is 5, and the infrared light of light source irradiation is 61, and the infrared light of people's face reflection is 62.Wherein: CPU or microprocessor core core adopt the Embedded Application processor P XA310 of Marvel.
More specifically be: the present invention uses embedded microprocessor to gather facial image as CPU, the connection cmos sensor of system, and is aided with infrared illumination realization use at night, filters the interference that visible light is got rid of surround lighting.Thereby the present invention realizes complete, a reliable face identification system, and its discrimination and recognition speed all reach gratifying level, can be applied in the systems such as work attendance, gate inhibition.
Claims (11)
1. a facial image recognition method is characterized in that setting up the user template storehouse earlier, discerns as follows: 1) gather facial image to be identified, 2 again) extract template to be identified, 3) user's recognition of face;
Said user's recognition of face is: template to be identified first template contrast with each user in the user template storehouse, obtain the user list A1 of all similarity marks greater than first lower threshold, arrange from big to small by the similarity mark; If A1 is sky then recognition failures; If first user's similarity mark is greater than upper limit threshold among the A1; Then discern and successfully return respective user,, then carry out next step: 2-5 the template comparison of template to be identified with each user among the A1 if not; Obtain the user list A2 of all marks, arrange from big to small by the similarity mark greater than second lower threshold; If A2 is sky then recognition failures; If first user's similarity mark is then discerned and is successfully returned respective user, if not greater than upper limit threshold among the A2; Then carry out next step: the 6-15 template contrast of template to be identified with each user among the A2; Obtain the user list A3 of all similarity marks, if A3 is sky, then recognition failures greater than recognition threshold; Arrange from big to small by the similarity mark, recognition function is returned respective user;
Said user template storehouse is when the registered user; 5 kinds of attitudes of user's conversion; Under each attitude, gather three two field pictures and extract feature templates, obtain 15 feature templates altogether, and through concentrate select progressively to go out the method ordering of template from candidate template; The template that ordering is selected based on following two principle: A. is represented the non-selected template that goes out as far as possible, and promptly the candidate template of it and other has maximum similarity; B. the template of selecting is as far as possible away from the template that has chosen, and promptly it maintains minimum similarity with modeling plate;
Said upper limit threshold is greater than recognition threshold, and recognition threshold is greater than second lower threshold, and second lower threshold is greater than first lower threshold.
2. according to the said facial image recognition method of claim 1, it is characterized in that discerning successfully after, if template to be identified and user's similarity mark are when setting the study threshold value than bigger one of upper limit threshold; Then waiting to learn template to the template of current collection in worksite as one learns as follows: import template to be learnt into to the user template storehouse; Form new template to be sorted to the template of waiting to learn template and user registration, ordering: if last template is a template to be learnt, then respective user is returned in the study failure; If no; Then remove last template, deposit database in, learning success returns respective user.
3. according to the said facial image recognition method of claim 2, it is characterized in that said setting study threshold value is 100.
4. according to the said facial image recognition method of claim 1, it is characterized in that said first lower threshold is 43, upper limit threshold is that 90, the second lower thresholds are 60, and recognition threshold is 80.
5. according to claim 1 or 2 or 3 or 4 said facial image recognition methods, it is characterized in that the sort method of said 15 feature templates is: select first template, make that the similarity mark average of itself and other template is maximum; Move on to it in the modeling plate then; Select second template again, make that the similarity mark average of itself and other template is maximum, move on to it in the modeling plate then; Select the 3rd template again; Make that the similarity mark average of itself and other template is maximum, move on to it in the modeling plate then, and the like up to there not being candidate template.
6. according to the said facial image recognition method of claim 5, it is characterized in that the similarity mark of selected template and other templates calculates and gets according to following formula:
C＝a1×c1-a2×c2；
In the formula: c1 representes the similarity mark average of this template and other candidate template, and c2 representes this template and the similarity mark average of modeling plate, and a1, a2 are two parameters.
7. according to the said facial image recognition method of claim 6, it is characterized in that parameter a1 value is 5/9, parameter a2 value is 4/9.
8. according to claim 1 or 2 or 3 or 4 said facial image recognition methods, it is characterized in that said user template storehouse builds up through the following step: 1) gather user's facial image, 2) extract user characteristics template, 3) registered user;
Said 1) collection facial image to be identified and 1) gathering user's facial image is: use the infrared LED lighting source to be shone gathering people's face, in gatherer process, also visible light is filtered;
Said 2) extraction template to be identified and 2) steps in sequence of extraction user characteristics template is: 1) detect people's face; 2) location eye position; 3) regular facial image; 4) assessment quality of human face image; 5) Gabor feature extraction.
9. said according to Claim 8 facial image recognition method is characterized in that said 1) detect people's face and be: on integrogram, detect the harr characteristic, use the AdaBoost algorithm to carry out people's face then and detect; Said 2) the location eye position is: through the symmetry of people's face and the darker characteristic of gray scale of position of human eye, orient in the picture people position of right and left eyes on the face; Said 3) regular facial image is: at first according to the position of right and left eyes, image is rotated, makes right and left eyes be in same horizontal line, convergent-divergent entire image then is the place normalization to of a right and left eyes fixing width; Further, according to the position of eyes, go out the image of whole people's face according to fixing width and height intercepting; At last image is carried out the part and strengthens, in the removal of images because the luminance difference that uneven illumination etc. produce; Obtain the image of people's face key position at last; Said 4) the assessment quality of human face image is: in the above image is carried out in the normalized processing procedure; Calculate acutance, contrast and the gray level of image simultaneously; And these three indexs of having portrayed picture quality simultaneously have only the threshold value separately that surpasses setting in advance just to carry out next step processing; Said 5) the Gabor feature extraction is: the image of one group of two-dimensional Gabor wavelet basis that we use N frequency range, a M direction after to normalization carries out filtering; Thereby obtain the response energy of image under this group wavelet basis; After the response energy carried out normalization and handle, face characteristic template to the end.
10. according to the said facial image recognition method of claim 9; The image that it is characterized in that the said people's of obtaining face key position is: getting the right and left eyes distance is 75 pixels; From move 35 pixels, 38 locations of pixels that move to left begin, getting width is 151, highly is 151 image;
Said N >=8, M >=16.
11. be used to realize the Face Image Recognition System of the said method of claim 1; It is characterized in that using embedded microprocessor to connect the cmos sensor of gathering facial image as CPU, the CPU of system; And the infrared illuminator that is aided with people's face realizes the functions of use under any light condition, adopts the arrowband infrared fileter to filter the interference that visible light is got rid of surround lighting.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201010517232.6A CN102419819B (en) | 2010-10-25 | 2010-10-25 | Method and system for recognizing human face image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201010517232.6A CN102419819B (en) | 2010-10-25 | 2010-10-25 | Method and system for recognizing human face image |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102419819A true CN102419819A (en) | 2012-04-18 |
| CN102419819B CN102419819B (en) | 2014-10-08 |
Family
ID=45944222
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201010517232.6A Active CN102419819B (en) | 2010-10-25 | 2010-10-25 | Method and system for recognizing human face image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102419819B (en) |
Cited By (22)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102945366A (en) * | 2012-11-23 | 2013-02-27 | 海信集团有限公司 | Method and device for face recognition |
| CN102968645A (en) * | 2012-10-24 | 2013-03-13 | 蔡翔 | Method for improving face recognition accuracy rate and adaptability through updating of images |
| CN103280005A (en) * | 2013-04-16 | 2013-09-04 | 无锡市崇安区科技创业服务中心 | Entrance guard system |
| CN103315754A (en) * | 2013-07-01 | 2013-09-25 | 深圳市飞瑞斯科技有限公司 | Method and device for detecting fatigue |
| CN104021387A (en) * | 2014-04-04 | 2014-09-03 | 南京工程学院 | Face image illumination processing method based on visual modeling |
| CN105260706A (en) * | 2015-09-15 | 2016-01-20 | 山东大学 | Head gesture detection method based on image comparison and heading gesture system |
| CN106489153A (en) * | 2016-10-13 | 2017-03-08 | 厦门中控生物识别信息技术有限公司 | A kind of face register method and face registering apparatus |
| WO2017080291A1 (en) * | 2015-11-13 | 2017-05-18 | 广东欧珀移动通信有限公司 | Fingerprint recognition method, method and device for updating fingerprint template, and mobile terminal |
| WO2017080278A1 (en) * | 2015-11-13 | 2017-05-18 | 广东欧珀移动通信有限公司 | Method and device for fingerprint recognition and terminal device |
| WO2017148291A1 (en) * | 2016-03-04 | 2017-09-08 | 腾讯科技（深圳）有限公司 | Human face recognition-based sign-in system, method and device, and server |
| CN107578398A (en) * | 2017-07-25 | 2018-01-12 | 浙江力太科技有限公司 | A kind of method for improving rotational symmetry figure discrimination |
| CN108446387A (en) * | 2018-03-22 | 2018-08-24 | 百度在线网络技术（北京）有限公司 | Method and apparatus for updating face registration library |
| CN108536293A (en) * | 2018-03-29 | 2018-09-14 | 北京字节跳动网络技术有限公司 | Man-machine interactive system, method, computer readable storage medium and interactive device |
| CN109003377A (en) * | 2018-09-13 | 2018-12-14 | 深圳阜时科技有限公司 | Access control system and control method with 3D face identification functions |
| CN109871859A (en) * | 2018-09-28 | 2019-06-11 | 北京矩视智能科技有限公司 | One kind automatically generating training set of images system |
| CN109993024A (en) * | 2017-12-29 | 2019-07-09 | 技嘉科技股份有限公司 | Authentication means, auth method and computer-readable storage medium |
| CN109993028A (en) * | 2017-12-29 | 2019-07-09 | 技嘉科技股份有限公司 | Human face recognition device and method, the method for promoting image identification rate |
| CN110263744A (en) * | 2019-06-26 | 2019-09-20 | 苏州万店掌网络科技有限公司 | The method for improving noninductive face identification rate |
| CN111178162A (en) * | 2019-12-12 | 2020-05-19 | 北京迈格威科技有限公司 | Image recognition method and device, computer equipment and storage medium |
| CN111325059A (en) * | 2018-12-14 | 2020-06-23 | 技嘉科技股份有限公司 | Face recognition method, device and computer readable medium |
| CN113192358A (en) * | 2021-04-26 | 2021-07-30 | 贵州车秘科技有限公司 | Parking management system based on thermal imaging technology in intelligent parking field and use method thereof |
| CN113313078A (en) * | 2021-07-02 | 2021-08-27 | 昆明理工大学 | Lightweight night infrared image pedestrian detection method and system based on model optimization |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101216884A (en) * | 2007-12-29 | 2008-07-09 | 北京中星微电子有限公司 | A method and system for face authentication |
| US7555148B1 (en) * | 2004-01-22 | 2009-06-30 | Fotonation Vision Limited | Classification system for consumer digital images using workflow, face detection, normalization, and face recognition |
| CN101587485A (en) * | 2009-06-15 | 2009-11-25 | 无锡骏聿科技有限公司 | Face information automatic login method based on face recognition technology |
-
2010
- 2010-10-25 CN CN201010517232.6A patent/CN102419819B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7555148B1 (en) * | 2004-01-22 | 2009-06-30 | Fotonation Vision Limited | Classification system for consumer digital images using workflow, face detection, normalization, and face recognition |
| CN101216884A (en) * | 2007-12-29 | 2008-07-09 | 北京中星微电子有限公司 | A method and system for face authentication |
| CN101587485A (en) * | 2009-06-15 | 2009-11-25 | 无锡骏聿科技有限公司 | Face information automatic login method based on face recognition technology |
Cited By (32)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102968645A (en) * | 2012-10-24 | 2013-03-13 | 蔡翔 | Method for improving face recognition accuracy rate and adaptability through updating of images |
| CN102945366B (en) * | 2012-11-23 | 2016-12-21 | 海信集团有限公司 | A kind of method and device of recognition of face |
| CN102945366A (en) * | 2012-11-23 | 2013-02-27 | 海信集团有限公司 | Method and device for face recognition |
| CN103280005A (en) * | 2013-04-16 | 2013-09-04 | 无锡市崇安区科技创业服务中心 | Entrance guard system |
| CN103315754A (en) * | 2013-07-01 | 2013-09-25 | 深圳市飞瑞斯科技有限公司 | Method and device for detecting fatigue |
| CN103315754B (en) * | 2013-07-01 | 2015-11-25 | 深圳市飞瑞斯科技有限公司 | A kind of fatigue detection method and device |
| CN104021387B (en) * | 2014-04-04 | 2017-03-08 | 南京工程学院 | The facial image illumination processing method of view-based access control model modeling |
| CN104021387A (en) * | 2014-04-04 | 2014-09-03 | 南京工程学院 | Face image illumination processing method based on visual modeling |
| CN105260706A (en) * | 2015-09-15 | 2016-01-20 | 山东大学 | Head gesture detection method based on image comparison and heading gesture system |
| US10460149B2 (en) | 2015-11-13 | 2019-10-29 | Guangdong Oppo Mobile Telecommunications Corp., Ltd. | Method and apparatus for updating fingerprint templates, and mobile terminal |
| WO2017080291A1 (en) * | 2015-11-13 | 2017-05-18 | 广东欧珀移动通信有限公司 | Fingerprint recognition method, method and device for updating fingerprint template, and mobile terminal |
| WO2017080278A1 (en) * | 2015-11-13 | 2017-05-18 | 广东欧珀移动通信有限公司 | Method and device for fingerprint recognition and terminal device |
| WO2017148291A1 (en) * | 2016-03-04 | 2017-09-08 | 腾讯科技（深圳）有限公司 | Human face recognition-based sign-in system, method and device, and server |
| US10664580B2 (en) | 2016-03-04 | 2020-05-26 | Tencent Technology (Shenzhen) Company Limited | Sign-in system, method, apparatus and server based on facial recognition |
| CN106489153A (en) * | 2016-10-13 | 2017-03-08 | 厦门中控生物识别信息技术有限公司 | A kind of face register method and face registering apparatus |
| WO2018068261A1 (en) * | 2016-10-13 | 2018-04-19 | 厦门中控智慧信息技术有限公司 | Face registration method and face registration device |
| CN107578398A (en) * | 2017-07-25 | 2018-01-12 | 浙江力太科技有限公司 | A kind of method for improving rotational symmetry figure discrimination |
| CN109993024A (en) * | 2017-12-29 | 2019-07-09 | 技嘉科技股份有限公司 | Authentication means, auth method and computer-readable storage medium |
| CN109993028A (en) * | 2017-12-29 | 2019-07-09 | 技嘉科技股份有限公司 | Human face recognition device and method, the method for promoting image identification rate |
| CN108446387A (en) * | 2018-03-22 | 2018-08-24 | 百度在线网络技术（北京）有限公司 | Method and apparatus for updating face registration library |
| CN108536293A (en) * | 2018-03-29 | 2018-09-14 | 北京字节跳动网络技术有限公司 | Man-machine interactive system, method, computer readable storage medium and interactive device |
| CN108536293B (en) * | 2018-03-29 | 2020-06-30 | 北京字节跳动网络技术有限公司 | Man-machine interaction system, man-machine interaction method, computer-readable storage medium and interaction device |
| CN109003377A (en) * | 2018-09-13 | 2018-12-14 | 深圳阜时科技有限公司 | Access control system and control method with 3D face identification functions |
| CN109871859A (en) * | 2018-09-28 | 2019-06-11 | 北京矩视智能科技有限公司 | One kind automatically generating training set of images system |
| CN111325059A (en) * | 2018-12-14 | 2020-06-23 | 技嘉科技股份有限公司 | Face recognition method, device and computer readable medium |
| CN110263744B (en) * | 2019-06-26 | 2021-05-11 | 苏州万店掌网络科技有限公司 | Method for improving non-sense face recognition rate |
| CN110263744A (en) * | 2019-06-26 | 2019-09-20 | 苏州万店掌网络科技有限公司 | The method for improving noninductive face identification rate |
| CN111178162A (en) * | 2019-12-12 | 2020-05-19 | 北京迈格威科技有限公司 | Image recognition method and device, computer equipment and storage medium |
| CN111178162B (en) * | 2019-12-12 | 2023-11-07 | 北京迈格威科技有限公司 | Image recognition method, device, computer equipment and storage medium |
| CN113192358A (en) * | 2021-04-26 | 2021-07-30 | 贵州车秘科技有限公司 | Parking management system based on thermal imaging technology in intelligent parking field and use method thereof |
| CN113313078A (en) * | 2021-07-02 | 2021-08-27 | 昆明理工大学 | Lightweight night infrared image pedestrian detection method and system based on model optimization |
| CN113313078B (en) * | 2021-07-02 | 2022-07-08 | 昆明理工大学 | Lightweight night infrared image pedestrian detection method and system based on model optimization |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102419819B (en) | 2014-10-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102419819B (en) | Method and system for recognizing human face image | |
| CN108520216B (en) | Gait image-based identity recognition method | |
| US10204262B2 (en) | Infrared imaging recognition enhanced by 3D verification | |
| Sun et al. | Improving iris recognition accuracy via cascaded classifiers | |
| CN109558810B (en) | Target person identification method based on part segmentation and fusion | |
| CN102822865A (en) | Face recognition device and face recognition method | |
| CN107066942A (en) | A kind of living body faces recognition methods and system | |
| CN103295016A (en) | Behavior recognition method based on depth and RGB information and multi-scale and multidirectional rank and level characteristics | |
| CN103473530A (en) | Adaptive action recognition method based on multi-view and multi-mode characteristics | |
| CN110119726A (en) | A kind of vehicle brand multi-angle recognition methods based on YOLOv3 model | |
| CN104504395A (en) | Method and system for achieving classification of pedestrians and vehicles based on neural network | |
| CN101794374A (en) | Method and system for identifying a person using their finger-joint print | |
| CN110363087B (en) | Long-baseline binocular face in-vivo detection method and system | |
| CN105894655A (en) | Method for detecting and recognizing bill under complex environments based on RGB-D camera | |
| CN110796101A (en) | Face recognition method and system of embedded platform | |
| WO2013075295A1 (en) | Clothing identification method and system for low-resolution video | |
| CN112132157B (en) | Gait face fusion recognition method based on raspberry pie | |
| CN106529441B (en) | Depth motion figure Human bodys' response method based on smeared out boundary fragment | |
| CN112434545A (en) | Intelligent place management method and system | |
| CN113920498B (en) | Point cloud 3D object detection method based on multilayer feature pyramid | |
| Xiao et al. | Fusion of iris and periocular biometrics for cross-sensor identification | |
| CN113420582B (en) | Anti-fake detection method and system for palm vein recognition | |
| CN116434029B (en) | Drinking detection method | |
| Yang et al. | Detection and segmentation of latent fingerprints | |
| CN109711232A (en) | Deep learning pedestrian recognition methods again based on multiple objective function |
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 | ||
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20181015 Address after: 523710 26 Pingshan 188 Industrial Avenue, Tangxia Town, Dongguan, Guangdong Patentee after: ZKTECO Co.,Ltd. Address before: 518129 Bantian, HUAWEI base five and north central control building, Longgang, Shenzhen, Guangdong. Patentee before: SHENZHEN ZK SOFTWARE BIOMETRIC IDENTIFICATION TECHNOLOGY CO.,LTD. |
|
| PE01 | Entry into force of the registration of the contract for pledge of patent right | ||
| PE01 | Entry into force of the registration of the contract for pledge of patent right |
Denomination of invention: Method and system for recognizing human face image Effective date of registration: 20181211 Granted publication date: 20141008 Pledgee: Dongguan Tangxia Branch of Agricultural Bank of China Ltd. Pledgor: ZKTECO Co.,Ltd. Registration number: 2018440000362 |
|
| CP01 | Change in the name or title of a patent holder | ||
| CP01 | Change in the name or title of a patent holder |
Address after: 523710, 26, 188 Industrial Road, Pingshan Town, Guangdong, Dongguan, Tangxia Patentee after: Entropy Technology Co.,Ltd. Address before: 523710, 26, 188 Industrial Road, Pingshan Town, Guangdong, Dongguan, Tangxia Patentee before: ZKTECO Co.,Ltd. |
|
| CP02 | Change in the address of a patent holder | ||
| CP02 | Change in the address of a patent holder |
Address after: No.32, Pingshan Industrial Road, Tangxia Town, Dongguan City, Guangdong Province, 523710 Patentee after: Entropy Technology Co.,Ltd. Address before: 523710 26 Pingshan 188 Industrial Avenue, Tangxia Town, Dongguan, Guangdong Patentee before: Entropy Technology Co.,Ltd. |
|
| PC01 | Cancellation of the registration of the contract for pledge of patent right | ||
| PC01 | Cancellation of the registration of the contract for pledge of patent right |
Date of cancellation: 20210916 Granted publication date: 20141008 Pledgee: Dongguan Tangxia Branch of Agricultural Bank of China Ltd. Pledgor: ZKTECO Co.,Ltd. Registration number: 2018440000362 |
|
| PE01 | Entry into force of the registration of the contract for pledge of patent right | ||
| PE01 | Entry into force of the registration of the contract for pledge of patent right |
Denomination of invention: Face image recognition method and system Effective date of registration: 20210924 Granted publication date: 20141008 Pledgee: Dongguan Tangxia Branch of Agricultural Bank of China Ltd. Pledgor: Entropy Technology Co.,Ltd. Registration number: Y2021980009839 |
|
| PC01 | Cancellation of the registration of the contract for pledge of patent right | ||
| PC01 | Cancellation of the registration of the contract for pledge of patent right |
Date of cancellation: 20220930 Granted publication date: 20141008 Pledgee: Dongguan Tangxia Branch of Agricultural Bank of China Ltd. Pledgor: Entropy Technology Co.,Ltd. Registration number: Y2021980009839 |