CN109740501A - A kind of Work attendance method and device of recognition of face - Google Patents
A kind of Work attendance method and device of recognition of face Download PDFInfo
- Publication number
- CN109740501A CN109740501A CN201811625248.1A CN201811625248A CN109740501A CN 109740501 A CN109740501 A CN 109740501A CN 201811625248 A CN201811625248 A CN 201811625248A CN 109740501 A CN109740501 A CN 109740501A
- Authority
- CN
- China
- Prior art keywords
- face
- attendance
- angle
- picture
- unit
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 29
- 230000001815 facial effect Effects 0.000 claims abstract description 53
- 238000004458 analytical method Methods 0.000 claims abstract description 17
- 239000000284 extract Substances 0.000 claims abstract description 13
- PXFBZOLANLWPMH-UHFFFAOYSA-N 16-Epiaffinine Natural products C1C(C2=CC=CC=C2N2)=C2C(=O)CC2C(=CC)CN(C)C1C2CO PXFBZOLANLWPMH-UHFFFAOYSA-N 0.000 claims abstract description 9
- 238000011426 transformation method Methods 0.000 claims abstract description 8
- 238000000605 extraction Methods 0.000 claims description 20
- 238000004321 preservation Methods 0.000 claims description 6
- 230000001537 neural effect Effects 0.000 claims 2
- 238000001514 detection method Methods 0.000 abstract description 9
- 238000013527 convolutional neural network Methods 0.000 abstract description 6
- 238000005516 engineering process Methods 0.000 description 5
- 238000013135 deep learning Methods 0.000 description 3
- 239000002699 waste material Substances 0.000 description 3
- 238000013473 artificial intelligence Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 240000007594 Oryza sativa Species 0.000 description 1
- 235000007164 Oryza sativa Nutrition 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 210000003128 head Anatomy 0.000 description 1
- 238000001727 in vivo Methods 0.000 description 1
- 230000006698 induction Effects 0.000 description 1
- 235000009566 rice Nutrition 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Landscapes
- Image Analysis (AREA)
- Image Processing (AREA)
- Collating Specific Patterns (AREA)
Abstract
The present invention relates to a kind of method and apparatus of recognition of face, by acquiring facial image from multi-angle；After the Face datection in facial image is come out using depth convolutional neural networks method respectively, face registration, alignment are carried out using facial image of the affine transformation method to multiple angles, quickly analysis is carried out respectively extracts face characteristic, face characteristic under multiple angles is compared with data in attendance face information database respectively, judge whether it is to personnel in attendance face information library, avoid the resource of system, avoid non-essential analysis detection, the recognition of face for quickly carrying out non-formula, improves recognition efficiency.
Description
Technical field
The present invention relates to field of attendance, the Work attendance method and device of especially a kind of recognition of face.
Background technique
With the arriving in " brush face " epoch, face recognition technology constantly obtains new research achievement in artificial intelligence field,
Based on the recognition methods of depth convolutional neural networks model, the precision of recognition of face is improved.Now, more and more artificial intelligence
Energy research achievement enters in daily life and work, and human face identification work-attendance checking is a kind of application of face recognition technology, phase
It checks card for traditional induction, fingerprint attendance etc., the characteristic that human face identification work-attendance checking can not be substituted based on face, intuitive, friendly
It can also prevent simultaneously for phenomenon of checking card.
Face recognition technology is a concrete application of the computer vision field on Work attendance method, uses deep learning
Recognition of face is carried out, MTCNN is that common Face datection and face alignment model need to extract suitable face characteristic
Optimize convolutional neural networks using suitable loss.Work attendance method based on recognition of face can effectively improve the attendance pipe of enterprise
Reason mode, the operation of specification staff attendance prevent to occur improving the efficiency of attendance for the phenomenon that checking card.Based on depth convolution mind
Face characteristic is extracted through network, a large amount of face characteristic informations are saved in database, carries out recognition of face using the feature of extraction.
The face picture that acquisition equipment captures is carried out feature extraction and detected from database similarly to spend highest face.
Currently, most of the acquisition equipment of recognition of face is formula, and need to each picture captured
Continuous analysis identification is carried out, wastes the resource of system significantly, non-essential analysis detection reduces server process efficiency.
Be primarily present following defect: (1) recognition of face of attendance is formula mostly, needs to carry out In vivo detection, such as open one's mouth or
Left-right rotation head is living person with guarantee identification, this sequence of operations can waste the regular hour, so can not accomplish true
It is efficient in positive meaning；(2) face that server captures acquisition equipment is constantly analyzed, even the same person
Multiple continuous face pictures, will also be tested and analyzed one by one, it is not necessary to identification operation waste the system resource of server,
Greatly reduce the operational efficiency of system；(3) the attendance mode of base recognition of face, ultra-large training data improve face
The precision of identification, but be still a big challenge in the recognition speed of practical application and efficiency.
Therefore, the resource for how avoiding system avoids non-essential analysis detection, and the face for quickly carrying out non-formula is known
Not, improving recognition efficiency is that urgent problem to be solved is needed in human face identification work-attendance checking.
Summary of the invention
Technical problem to be solved by the invention is to provide a kind of method and devices of recognition of face.
In order to solve the above-mentioned technical problem, the technical solution adopted by the present invention are as follows: a kind of Work attendance method of recognition of face, packet
Include following steps:
S1. facial image is acquired from multi-angle: acquiring multiple angle facial images to attendance personnel for one；
S2. it exports face picture placed in the middle: applying depth convolution refreshing respectively collected all multiple angle facial images
After network method comes out the Face datection in facial image, using affine transformation method to the facial images of multiple angles into
Pedestrian's face registration, alignment, then export multiple angle human face pictures placed in the middle；
S3. feature-extraction analysis: carrying out quickly analyzing respectively and extract face characteristic to multiple angle human face pictures placed in the middle,
Export the face characteristic under multiple angles；
S4. face alignment: the face characteristic under multiple angles is compared with data in attendance face information database respectively
Right, the face characteristic and similarity judged whether there is under any one angle reaches threshold value, if going to step S5；Otherwise, defeated
Error false information then terminates；
S5. attendance success: the corresponding staff attendance success of label current face's feature then goes to step S1.
Further, further include the steps that S00 before step S1) establish attendance face information database, the S00)
Specifically include step:
S001. multi-angle acquires face picture: successively acquiring the multi-orientation Face picture each to attendance personnel；
S002. picture quality judges: judging whether reach judgement to face quality in each face picture of attendance personnel
The requirement of standard then enters step S003 if then saving the picture, otherwise deletes the picture；The judgment criteria includes
One or more of: whether it is more than whether setting value, Eulerian angles are greater than setting that whether resolution ratio is more than setting value, clarity
Value；
S003. feature and personal information are extracted: face characteristic being extracted to the picture of preservation, establishes face characteristic and personal letter
The corresponding association of breath；
S004. give up into next: deleting the picture；The face characteristic of extraction and the write-in of associated personal information are examined
Diligent face information database.
Further, the S1) multi-angle acquisition face picture step, specifically:
By being 1.7≤H≤2.5 meter in same plane, height H, deviation is no more than 0.6 meter, to attendance personnel movement
Horizontal direction and the angle of acquisition are 120 °≤θ≤180 ° no more than 60 degree, relative angle θ and are 1 to attendance personnel's distance L
The multi-angle of≤L≤5 meter successively acquires each face picture to attendance personnel.
Further, the S4, specifically comprises the following steps:
S41. feature loads: respectively by face characteristic in the face characteristic and attendance face information database under multiple angles
It is loaded into memory；
S42. aspect ratio pair: the face characteristic under multiple angles is respectively with data in attendance face information database in memory
In be compared；
S43. judge similarity: the face characteristic and similarity judged whether there is under any one angle reaches threshold value, if
Go to step S5；Otherwise, output error message then terminates.
It further, include step before the S1,
S01) the face video stream to attendance personnel is acquired from multi-angle；
The step S1 acquires at least one set of to attendance people specifically, intercepting from the face video stream that multi-angle acquires
Facial image under multiple angles of member.
The present invention also provides a kind of devices of recognition of face, including,
Facial image unit is acquired from multi-angle: acquiring multiple angle facial images to attendance personnel for one；
Output face picture element unit cell placed in the middle: apply depth convolution refreshing respectively collected all multiple angle facial images
After network method comes out the Face datection in facial image, using affine transformation method to the facial images of multiple angles into
Pedestrian's face registration, alignment, then export multiple angle human face pictures placed in the middle；
Feature-extraction analysis unit: multiple angle human face pictures placed in the middle are carried out quickly analyzing respectively and extract face spy
Sign, exports the face characteristic under multiple angles；
Face alignment unit: the face characteristic under multiple angles is compared with data in attendance face information database respectively
Right, the face characteristic and similarity judged whether there is under any one angle reaches threshold value, if going to attendance success unit；It is no
Then, output error message then terminates；
Attendance success unit: the corresponding staff attendance success of label current face's feature then goes to from multi-angle and acquires
Facial image unit.
It further, further include establishing attendance face information database before acquiring facial image unit from multi-angle
Unit is established human face data library unit and is specifically included:
Multi-angle acquires face picture unit: successively acquiring the multi-orientation Face picture each to attendance personnel；
Picture quality judging unit: judge whether reach judgement mark to face quality in each face picture of attendance personnel
Quasi- requirement then enters if then saving the picture and extracts feature and personal information unit, otherwise delete the picture；It is described
Judgment criteria includes one or more of: whether it is more than that setting value, Eulerian angles are that whether resolution ratio is more than setting value, clarity
It is no to be greater than the set value；
It extracts feature and personal information unit: face characteristic being extracted to the picture of preservation, establishes face characteristic and personal letter
The corresponding association of breath；
Give up into next Zhang Danyuan: deleting the picture；The face characteristic of extraction and the write-in of associated personal information are examined
Diligent face information database.
Further, the multi-angle acquires face picture unit, specifically:
By being 1.7≤H≤2.5 meter in same plane, height H, deviation is no more than 0.6 meter, to attendance personnel movement
Horizontal direction and the angle of acquisition are 120 °≤θ≤180 ° no more than 60 degree, relative angle θ and are 1 to attendance personnel's distance L
The multi-angle of≤L≤5 meter successively acquires each face picture to attendance personnel.
Further, the face alignment unit, specifically includes:
Feature loading unit: respectively by face characteristic in the face characteristic and attendance face information database under multiple angles
It is loaded into memory；
Feature comparing unit: the face characteristic under multiple angles is respectively with data in attendance face information database in memory
In be compared；
Judge similarity unit: the face characteristic and similarity judged whether there is under any one angle reaches threshold value, if
Go to attendance success unit；Otherwise, output error message then terminates.
Further, further include before the acquisition facial image unit from multi-angle,
A face video stream unit to attendance personnel is acquired from multi-angle；
The multi-angle acquisition facial image unit acquires specifically, intercepting from the face video stream that multi-angle acquires
At least one set is to the face figure under multiple angles of attendance personnel.
The Work attendance method and device of a kind of recognition of face provided by the invention, by acquiring facial image from multi-angle；Point
Not Ying Yong depth convolutional neural networks method by the Face datection in facial image come out after, using affine transformation method to multiple
The facial image of angle carries out face registration, alignment, carries out quickly analysis respectively and extracts face characteristic, the face under multiple angles
Feature is compared with data in attendance face information database respectively, judge whether be to personnel in attendance face information library,
The resource for avoiding system avoids non-essential analysis detection, quickly carries out the recognition of face of non-formula, improves identification effect
Rate.
Detailed description of the invention
Fig. 1 is the non-cooperation fast face attendance overall flow figure figure of the present invention；
Fig. 2 establishes attendance face information database flowchart for one embodiment of the invention；
Fig. 3 is the acquisition deployed with devices figure that multi-angle of the present invention acquires picture；
Fig. 4 is that one embodiment of the invention multi-angle acquires flow chart of steps before facial image；
Fig. 5 is the face alignment flow chart of one embodiment of the invention；
Fig. 6 is the specific flow chart of the human face identification work-attendance checking method of one embodiment of the invention.
Specific embodiment
In order to describe the technical content, the structural feature, the achieved object and the effect of this invention in detail, below in conjunction with embodiment
And attached drawing is cooperated to be explained in detail.
The most critical design of the present invention is: carrying out quickly analysis by acquiring facial image from multi-angle and extracting face
Feature is sentenced to be compared respectively with face characteristic in attendance face information database with the face characteristic under multiple angles
Whether disconnected is to personnel in attendance face information library.
The present invention proposes a kind of Work attendance method of recognition of face, as shown in Figure 1,
Include the following steps:
S1. facial image is acquired from multi-angle: acquiring multiple angle facial images to attendance personnel for one；
S2. it exports face picture placed in the middle: applying depth convolution refreshing respectively collected all multiple angle facial images
After network method comes out the Face datection in facial image, using affine transformation method to the facial images of multiple angles into
Pedestrian's face registration, alignment, then export multiple angle human face pictures placed in the middle；
Picture containing face will use depth convolutional neural networks that Face datection present in picture is come out and carried out
Face alignment, i.e., " ajust face ".After convolutional neural networks, from the available face frame of output information and 5 features
Point is allowed to be aligned by the affine transformation in OpenCV.
S3. feature-extraction analysis: carrying out quickly analyzing respectively and extract face characteristic to multiple angle human face pictures placed in the middle,
Export the face characteristic under multiple angles；
S4. face alignment: the face characteristic under multiple angles is compared with data in attendance face information database respectively
Right, the face characteristic and similarity judged whether there is under any one angle reaches threshold value, if going to step S5；Otherwise, defeated
Error false information then terminates；
S5. attendance success: the corresponding staff attendance success of label current face's feature then goes to step S1.
Complete as can be seen from the above description, the beneficial effects of the present invention are: multi-orientation Face feature is quickly analyzed
Face characteristic is extracted, the face characteristic under multiple angles is compared with data in attendance face information database respectively, judges
It whether is to avoid the resource of system to personnel in attendance face information library, avoid non-essential analysis detection, quickly carry out non-match
Box-like recognition of face, improves recognition efficiency.
Embodiment 1:
As shown in fig. 6, further including the steps that S00 before step S1) attendance face information database is established, such as Fig. 2 institute
Show, the S00) specifically include step:
S001. multi-angle acquires face picture: successively acquiring the multi-orientation Face picture each to attendance personnel；
S002. picture quality judges: judging whether reach judgement to face quality in each face picture of attendance personnel
The requirement of standard then enters step S003 if then saving the picture, otherwise deletes the picture；The judgment criteria includes
One or more of: whether it is more than whether setting value, Eulerian angles are greater than setting that whether resolution ratio is more than setting value, clarity
Value；The size of image resolution ratio determines face characteristic information amount, and the face information that resolution ratio more the Supreme People's Procuratorate measures is more, but consumes
Shi Yuegao；Human face characteristic point mainly has (eyes, nose, mouth), and the side face occurred in image, the face that is blocked are commented by human face posture
Estimate (Eulerian angles) decision, for Eulerian angles in left-right rotary corner < 15 degree, upper and lower pitch angle < 15 degree are more reasonable, the direct mistake of the face that is blocked
It filters.Therefore selection criteria are as follows: resolution ratio 720P, clarity is high, and (left-right rotary corner is less than 15 degree, upper and lower pitch angle for Eulerian angles
Less than 15 degree) complete face.
S003. feature and personal information are extracted: face characteristic being extracted to the picture of preservation, establishes face characteristic and personal letter
The corresponding association of breath；
S004. give up into next: deleting the picture；The face characteristic of extraction and the write-in of associated personal information are examined
Diligent face information database.
By establishing attendance face information database, by feature and the individual of picture and extraction under the multiple angles of face
Information is stored in library, to facilitate aspect ratio pair when being compared.
Embodiment 2:
The S1) multi-angle acquisition face picture step, specifically:
By being 1.7≤H≤2.5 meter in same plane, height H, deviation is no more than 0.6 meter, to attendance personnel movement
Horizontal direction and the angle of acquisition are 120 °≤θ≤180 ° no more than 60 degree, relative angle θ and are 1 to attendance personnel's distance L
The multi-angle of≤L≤5 meter successively acquires each face picture to attendance personnel.
Specifically, the acquisition deployment of multi-angle acquisition face picture includes: the position for acquiring equipment, the quantity n (n of equipment
>=2), at a distance from mobile target etc., the height, angle, n that the position of equipment is divided into equipment again are set for the direction of equipment, equipment
Standby relative position etc..The deployment of equipment by taking the quantity n=3 of equipment as an example, the position of deployed with devices mainly have equipment height,
The relative position of angle, n equipment, in 1.7≤H≤2.5 meter, altitude range and employee's is averaged the optimal altitude range of equipment
Height is related, the appropriate adjustment up and down with the difference of average height；The angle of deployed with devices and the height of equipment are related, according to
The height of equipment and the average height of employee and appropriate adjustment, in order to guarantee to capture the integrality of face, when face is passed by
Horizontal direction and equipment angle most perfect condition be no more than 60 degree；The relative position of n equipment occupies the face captured
Middle rate has a great impact, as shown in figure 3, equipment is respectively deployed in positive direction, left and the right side of employee's advance as n=3
Side just forms a semicircle, and left equipment and the opposite angle of right equipment are 180 degree, the relative angle of equipment room at this time
Optimum range is 120 ° -180 °.
Multi-angle acquisition, the quantity for acquiring equipment is n, n >=2, and the deployment of n equipment is substantially at same plane, according to
Actual production environment slightly deviation, but deviation is no more than 0.6 meter.The relative angle of n equipment room is up to 180 °, with
The reduction of number of devices n and suitably reduce, but ideal range be not less than 120 °.The best value of the quantity n of equipment is 4,
I.e. employee's direction of advance just before, just left, positive right, dead astern dispose a face acquisition equipment respectively, such as camera can be with
Comprehensive to capture face, either working is still come off duty, and can realize that multi-angle acquires face picture.
Embodiment 3:
The S4, specifically comprise the following steps: as shown in figure 5,
S41. feature loads: respectively by face characteristic in the face characteristic and attendance face information database under multiple angles
It is loaded into memory；
Specifically, a series of face pictures that can be captured according to the frame per second of video to acquisition equipment, identification 15 per second
Frame face, is set to 60ms, and the people in this time break may be considered the same person, chooses that quality is preferable, people
Face a series of pictures placed in the middle carries out face characteristic extraction, and these features are loaded into memory.
S42. aspect ratio pair: the face characteristic under multiple angles is respectively with data in attendance face information database in memory
In be compared；
S43. judge similarity: the face characteristic and similarity judged whether there is under any one angle reaches threshold value, if
Go to step S5；Otherwise, output error message then terminates.
Face picture and face feature extraction are acquired by multi-angle, feature is loaded into memory, then reads attendance face
Face characteristic in attendance face information database is equally loaded into memory, server is facilitated quickly to read by information database
Be compared.
Embodiment 4:
It include step before the S1, as shown in figure 4,
S01) the face video stream to attendance personnel is acquired from multi-angle；
The step S1 acquires at least one set of to attendance people specifically, intercepting from the face video stream that multi-angle acquires
Facial image under multiple angles of member.
From multi-angle acquisition video flowing, corresponding face acquisition and candid photograph detection device, this method application deep learning
Technology, the automatic human body detected in video, acquisition captures face and carries out attendance in real time, greatly improves speed.
The present invention also proposes a kind of Work attendance device of recognition of face, comprising:
Facial image unit is acquired from multi-angle: acquiring multiple angle facial images to attendance personnel for one；
Output face picture element unit cell placed in the middle: apply depth convolution refreshing respectively collected all multiple angle facial images
After network method comes out the Face datection in facial image, using affine transformation method to the facial images of multiple angles into
Pedestrian's face registration, alignment, then export multiple angle human face pictures placed in the middle；
Feature-extraction analysis unit: multiple angle human face pictures placed in the middle are carried out quickly analyzing respectively and extract face spy
Sign, exports the face characteristic under multiple angles；
Face alignment unit: the face characteristic under multiple angles is compared with data in attendance face information database respectively
Right, the face characteristic and similarity judged whether there is under any one angle reaches threshold value, if going to attendance success unit；It is no
Then, output error message then terminates；
Attendance success unit: the corresponding staff attendance success of label current face's feature then goes to from multi-angle and acquires
Facial image unit.
As can be seen from the above description, the beneficial effects of the present invention are: quickly analysis is carried out to multi-orientation Face feature and is extracted
Face characteristic, the face characteristic under multiple angles are compared with data in attendance face information database respectively, judge whether
It is to avoid the resource of system to personnel in attendance face information library, avoid non-essential analysis detection, quickly carry out non-formula
Recognition of face, improve recognition efficiency.
Embodiment 5:
Further include the unit for establishing attendance face information database before acquiring facial image unit from multi-angle, establishes
Human face data library unit specifically includes:
Multi-angle acquires face picture unit: successively acquiring the multi-orientation Face picture each to attendance personnel；
Picture quality judging unit: judge whether reach judgement mark to face quality in each face picture of attendance personnel
Quasi- requirement then enters if then saving the picture and extracts feature and personal information unit, otherwise delete the picture；It is described
Judgment criteria includes one or more of: whether it is more than that setting value, Eulerian angles are that whether resolution ratio is more than setting value, clarity
It is no to be greater than the set value；The size of image resolution ratio determines that face characteristic information amount, resolution ratio get over the face information that the Supreme People's Procuratorate measures
It is more, but it is time-consuming higher；Human face characteristic point mainly has (eyes, nose, mouth), and the side face that occurs in image, be blocked face
Determine that Eulerian angles are in left-right rotary corner < 15 degree, and upper and lower pitch angle < 15 degree are more reasonable, quilt by human face posture assessment (Eulerian angles)
Face is blocked directly to filter.Therefore selection criteria are as follows: resolution ratio 720P, clarity is high, and (left-right rotary corner is less than 15 for Eulerian angles
Degree, upper and lower pitch angle is less than 15 degree) complete face.
It extracts feature and personal information unit: face characteristic being extracted to the picture of preservation, establishes face characteristic and personal letter
The corresponding association of breath；
Give up into next Zhang Danyuan: deleting the picture；The face characteristic of extraction and the write-in of associated personal information are examined
Diligent face information database.
By establishing attendance face information database, by feature and the individual of picture and extraction under the multiple angles of face
Information is stored in library, to facilitate aspect ratio pair when being compared.
Embodiment 6:
The multi-angle acquires face picture unit, specifically:
By being 1.7≤H≤2.5 meter in same plane, height H, deviation is no more than 0.6 meter, to attendance personnel movement
Horizontal direction and the angle of acquisition are 120 °≤θ≤180 ° no more than 60 degree, relative angle θ and are 1 to attendance personnel's distance L
The multi-angle of≤L≤5 meter successively acquires each face picture to attendance personnel.
Specifically, the acquisition deployment of multi-angle acquisition face picture includes: the position for acquiring equipment, the quantity n (n of equipment
>=2), at a distance from mobile target etc., the height, angle, n that the position of equipment is divided into equipment again are set for the direction of equipment, equipment
Standby relative position etc..The deployment of equipment by taking the quantity n=3 of equipment as an example, the position of deployed with devices mainly have equipment height,
The relative position of angle, n equipment, for the optimal altitude range of equipment in rice, altitude range is related with the average height of employee, with
The difference of average height and appropriate adjustment up and down；The angle of deployed with devices and the height of equipment are related, according to the height of equipment
With the average height of employee and appropriate adjustment, the horizontal direction in order to guarantee to capture the integrality of face, when face is passed by
It is no more than 60 degree with the angle most perfect condition of equipment；There is the face captured rate placed in the middle in the relative position of n equipment very big
Influence, as n=3, equipment be respectively deployed in employee advance positive direction, the left and right, just formed a semicircle, this
Shi Zuofang equipment and the opposite angle of right equipment are 180 degree, and the relative angle optimum range of equipment room is 120 ° -180 °.
Multi-angle acquisition, the quantity for acquiring equipment is n, n >=2, and the deployment of n equipment is substantially at same plane, according to
Actual production environment slightly deviation, but deviation is no more than 0.6 meter.The relative angle of n equipment room is up to 180 °, with
The reduction of number of devices n and suitably reduce, but ideal range be not less than 120 °.The best value of the quantity n of equipment is 4,
I.e. employee's direction of advance just before, just left, positive right, dead astern dispose a face acquisition equipment respectively, such as camera can be with
Comprehensive to capture face, either working is still come off duty, and can realize that multi-angle acquires face picture.
Embodiment 7:
The face alignment unit, specifically includes:
Feature loading unit: respectively by face characteristic in the face characteristic and attendance face information database under multiple angles
It is loaded into memory；
Specifically, a series of face pictures that can be captured according to the frame per second of video to acquisition equipment, identification 15 per second
Frame face, is set to 60ms, and the people in this time break may be considered the same person, chooses that quality is preferable, people
Face a series of pictures placed in the middle carries out face characteristic extraction, and these features are loaded into memory.
Feature comparing unit: the face characteristic under multiple angles is respectively with data in attendance face information database in memory
In be compared；
Judge similarity unit: the face characteristic and similarity judged whether there is under any one angle reaches threshold value, if
Go to attendance success unit；Otherwise, output error message then terminates.
Face picture and face feature extraction are acquired by multi-angle, feature is loaded into memory, then reads attendance face
Face characteristic in attendance face information database is equally loaded into memory, server is facilitated quickly to read by information database
Be compared.
Embodiment 8:
Further include before the acquisition facial image unit from multi-angle,
A face video stream unit to attendance personnel is acquired from multi-angle；
The multi-angle acquisition facial image unit acquires specifically, intercepting from the face video stream that multi-angle acquires
At least one set is to the facial image under multiple angles of attendance personnel.
From multi-angle acquisition video flowing, corresponding face acquisition and candid photograph detection device, this method application deep learning
Technology, the automatic human body detected in video, acquisition captures face and carries out attendance in real time, greatly improves speed.
The above description is only an embodiment of the present invention, is not intended to limit the scope of the invention, all to utilize this hair
Equivalent structure or equivalent flow shift made by bright specification and accompanying drawing content is applied directly or indirectly in other relevant skills
Art field, is included within the scope of the present invention.
Claims (10)
1. a kind of Work attendance method of recognition of face, characterized by the following steps:
S1. facial image is acquired from multi-angle: acquiring multiple angle facial images to attendance personnel for one；
S2. it exports face picture placed in the middle: depth convolutional Neural net is applied respectively to collected all multiple angle facial images
After network method comes out the Face datection in facial image, people is carried out using facial image of the affine transformation method to multiple angles
Face registration, alignment, then export multiple angle human face pictures placed in the middle；
S3. feature-extraction analysis: carrying out quickly analyzing respectively and extract face characteristic to multiple angle human face pictures placed in the middle, output
Face characteristic under multiple angles；
S4. face alignment: the face characteristic under multiple angles is compared with data in attendance face information database respectively, sentences
The disconnected face characteristic whether having under any one angle and similarity reach threshold value, if going to step S5；Otherwise, output error
Information then terminates；
S5. attendance success: the corresponding staff attendance success of label current face's feature then goes to step S1.
2. the Work attendance method of recognition of face as described in claim 1, it is characterised in that: before step S1 further include S00) it builds
The step of vertical attendance face information database, the S00) specifically include step:
S001. multi-angle acquires face picture: successively acquiring the multi-orientation Face picture each to attendance personnel；
S002. picture quality judges: judging whether reach judgment criteria to face quality in each face picture of attendance personnel
Requirement then enter step S003 if then saving the picture, otherwise delete the picture；The judgment criteria includes following
One or more: whether it is more than whether setting value, Eulerian angles are greater than the set value that whether resolution ratio is more than setting value, clarity；
S003. feature and personal information are extracted: face characteristic being extracted to the picture of preservation, establishes face characteristic and personal information pair
It should be associated with；
S004. give up into next: deleting the picture；Attendance people is written into the face characteristic of extraction and associated personal information
Face information database.
3. the Work attendance method of recognition of face as claimed in claim 2, it is characterised in that: the S1) multi-angle acquisition face figure
Piece step, specifically:
By being 1.7≤H≤2.5 meter in same plane, height H, deviation is no more than 0.6 meter, to attendance personnel's mobile and horizontal
The angle of direction and acquisition is 120 °≤θ≤180 no more than 60 degree, relative angle θ°, with to attendance personnel's distance L be 1≤L≤
5 meters of multi-angle successively acquires each face picture to attendance personnel.
4. the Work attendance method of recognition of face as claimed in claim 3, it is characterised in that: the S4 specifically comprises the following steps:
S41. feature loads: respectively loading face characteristic in the face characteristic and attendance face information database under multiple angles
Into memory；
S42. aspect ratio pair: face characteristic under multiple angles respectively with data in attendance face information database in memory into
Row compares；
S43. judge similarity: the face characteristic and similarity judged whether there is under any one angle reaches threshold value, turns if having
To step S5；Otherwise, output error message then terminates.
5. the Work attendance method of recognition of face as claimed in claim 4, it is characterised in that: it include step before the S1,
S01) the face video stream to attendance personnel is acquired from multi-angle；
The step S1 acquires at least one set of to attendance personnel's specifically, intercepting from the face video stream that multi-angle acquires
Facial image under multiple angles.
6. a kind of Work attendance device of recognition of face, it is characterised in that: include:
Facial image unit is acquired from multi-angle: acquiring multiple angle facial images to attendance personnel for one；
Output face picture element unit cell placed in the middle: depth convolutional Neural net is applied respectively to collected all multiple angle facial images
After network method comes out the Face datection in facial image, people is carried out using facial image of the affine transformation method to multiple angles
Face registration, alignment, then export multiple angle human face pictures placed in the middle；
Feature-extraction analysis unit: carrying out quickly analyzing respectively and extract face characteristic to multiple angle human face pictures placed in the middle, defeated
Face characteristic under multiple angles out；
Face alignment unit: the face characteristic under multiple angles is compared with data in attendance face information database respectively,
The face characteristic and similarity judged whether there is under any one angle reaches threshold value, if going to attendance success unit；Otherwise,
Output error message then terminates；
Attendance success unit: the corresponding staff attendance success of label current face's feature then goes to from multi-angle and acquires face
Elementary area.
7. the Work attendance device of recognition of face as claimed in claim 6, it is characterised in that: acquiring facial image list from multi-angle
Further include the unit for establishing attendance face information database before member, establish human face data library unit and specifically include:
Multi-angle acquires face picture unit: successively acquiring the multi-orientation Face picture each to attendance personnel；
Picture quality judging unit: judge whether reach judgment criteria to face quality in each face picture of attendance personnel
It is required that then entering if then saving the picture and extracting feature and personal information unit, otherwise delete the picture；The judgement
Standard includes one or more of: whether it is more than whether setting value, Eulerian angles are big that whether resolution ratio is more than setting value, clarity
In setting value；
It extracts feature and personal information unit: face characteristic being extracted to the picture of preservation, establishes face characteristic and personal information pair
It should be associated with；
Give up into next Zhang Danyuan: deleting the picture；Attendance people is written into the face characteristic of extraction and associated personal information
Face information database.
8. the Work attendance device of recognition of face as claimed in claim 7, it is characterised in that: the multi-angle acquires face picture list
Member, specifically:
By being 1.7≤H≤2.5 meter in same plane, height H, deviation is no more than 0.6 meter, to attendance personnel's mobile and horizontal
The angle of direction and acquisition be no more than 60 degree, relative angle θ be 120 °≤θ≤180 °, and to attendance personnel's distance L be 1≤L≤
5 meters of multi-angle successively acquires each face picture to attendance personnel.
9. the Work attendance device of recognition of face as claimed in claim 8, it is characterised in that: the face alignment unit, it is specific to wrap
It includes:
Feature loading unit: face characteristic in the face characteristic and attendance face information database under multiple angles is loaded respectively
Into memory；
Feature comparing unit: face characteristic under multiple angles respectively with data in attendance face information database in memory into
Row compares；
Judge similarity unit: the face characteristic and similarity judged whether there is under any one angle reaches threshold value, turns if having
To attendance success unit；Otherwise, output error message then terminates.
10. the Work attendance device of recognition of face as claimed in claim 9, it is characterised in that: described to acquire face figure from multi-angle
As further including before unit,
A face video stream unit to attendance personnel is acquired from multi-angle；
The multi-angle acquisition facial image unit acquires at least specifically, intercepting from the face video stream that multi-angle acquires
One group of facial image waited under multiple angles of attendance personnel.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811625248.1A CN109740501A (en) | 2018-12-28 | 2018-12-28 | A kind of Work attendance method and device of recognition of face |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811625248.1A CN109740501A (en) | 2018-12-28 | 2018-12-28 | A kind of Work attendance method and device of recognition of face |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109740501A true CN109740501A (en) | 2019-05-10 |
Family
ID=66361941
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811625248.1A Pending CN109740501A (en) | 2018-12-28 | 2018-12-28 | A kind of Work attendance method and device of recognition of face |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109740501A (en) |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110569809A (en) * | 2019-09-11 | 2019-12-13 | 淄博矿业集团有限责任公司 | coal mine dynamic face recognition attendance checking method and system based on deep learning |
| CN110852703A (en) * | 2019-10-22 | 2020-02-28 | 佛山科学技术学院 | Attendance checking method, system, equipment and medium based on side face multi-feature fusion face recognition |
| CN110889392A (en) * | 2019-12-06 | 2020-03-17 | 北京百度网讯科技有限公司 | Method and device for processing face image |
| CN111612930A (en) * | 2020-04-14 | 2020-09-01 | 安徽中迅徽软科技有限公司 | Attendance system and method |
| CN111696222A (en) * | 2020-06-19 | 2020-09-22 | 打工快线网络科技(苏州)有限责任公司 | Intelligent attendance system based on comprehensive identification |
| CN111783581A (en) * | 2020-06-19 | 2020-10-16 | 何金荣 | Iris identification method and related device |
| CN112001219A (en) * | 2020-06-19 | 2020-11-27 | 国家电网有限公司技术学院分公司 | Multi-angle multi-face recognition attendance checking method and system |
| CN112149517A (en) * | 2020-08-31 | 2020-12-29 | 三盟科技股份有限公司 | Face attendance checking method and system, computer equipment and storage medium |
| CN112597854A (en) * | 2020-12-15 | 2021-04-02 | 重庆电子工程职业学院 | Non-matching type face recognition system and method |
| CN113095289A (en) * | 2020-10-28 | 2021-07-09 | 重庆电政信息科技有限公司 | Massive image preprocessing network method based on urban complex scene |
| CN113469024A (en) * | 2021-06-29 | 2021-10-01 | 深圳市商汤科技有限公司 | Face attendance checking method and device, electronic equipment and computer readable storage medium |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103984941A (en) * | 2014-06-10 | 2014-08-13 | 深圳市赛为智能股份有限公司 | Face recognition checking-in method and device thereof |
| CN106407966A (en) * | 2016-11-28 | 2017-02-15 | 南京理工大学 | Face identification method applied to checking attendance |
| CN107480658A (en) * | 2017-09-19 | 2017-12-15 | 苏州大学 | Face identification device and method based on multi-angle video |
| CN107563280A (en) * | 2017-07-24 | 2018-01-09 | 南京道熵信息技术有限公司 | Face identification method and device based on multi-model |
| CN108416865A (en) * | 2018-04-03 | 2018-08-17 | 郑州云海信息技术有限公司 | A kind of facial recognition data automatic correcting method and system based on cloud computing |
| CN108765611A (en) * | 2018-05-21 | 2018-11-06 | 中兴智能视觉大数据技术（湖北）有限公司 | A kind of dynamic human face identification Work attendance management system and its management method |
-
2018
- 2018-12-28 CN CN201811625248.1A patent/CN109740501A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103984941A (en) * | 2014-06-10 | 2014-08-13 | 深圳市赛为智能股份有限公司 | Face recognition checking-in method and device thereof |
| CN106407966A (en) * | 2016-11-28 | 2017-02-15 | 南京理工大学 | Face identification method applied to checking attendance |
| CN107563280A (en) * | 2017-07-24 | 2018-01-09 | 南京道熵信息技术有限公司 | Face identification method and device based on multi-model |
| CN107480658A (en) * | 2017-09-19 | 2017-12-15 | 苏州大学 | Face identification device and method based on multi-angle video |
| CN108416865A (en) * | 2018-04-03 | 2018-08-17 | 郑州云海信息技术有限公司 | A kind of facial recognition data automatic correcting method and system based on cloud computing |
| CN108765611A (en) * | 2018-05-21 | 2018-11-06 | 中兴智能视觉大数据技术（湖北）有限公司 | A kind of dynamic human face identification Work attendance management system and its management method |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110569809A (en) * | 2019-09-11 | 2019-12-13 | 淄博矿业集团有限责任公司 | coal mine dynamic face recognition attendance checking method and system based on deep learning |
| CN110852703A (en) * | 2019-10-22 | 2020-02-28 | 佛山科学技术学院 | Attendance checking method, system, equipment and medium based on side face multi-feature fusion face recognition |
| CN110852703B (en) * | 2019-10-22 | 2023-05-23 | 佛山科学技术学院 | Attendance checking method, system, equipment and medium based on face multi-feature fusion face recognition |
| CN110889392A (en) * | 2019-12-06 | 2020-03-17 | 北京百度网讯科技有限公司 | Method and device for processing face image |
| CN110889392B (en) * | 2019-12-06 | 2022-06-10 | 北京百度网讯科技有限公司 | Method and device for processing face image |
| CN111612930A (en) * | 2020-04-14 | 2020-09-01 | 安徽中迅徽软科技有限公司 | Attendance system and method |
| CN112001219A (en) * | 2020-06-19 | 2020-11-27 | 国家电网有限公司技术学院分公司 | Multi-angle multi-face recognition attendance checking method and system |
| CN111783581A (en) * | 2020-06-19 | 2020-10-16 | 何金荣 | Iris identification method and related device |
| CN111696222A (en) * | 2020-06-19 | 2020-09-22 | 打工快线网络科技(苏州)有限责任公司 | Intelligent attendance system based on comprehensive identification |
| CN112001219B (en) * | 2020-06-19 | 2024-02-09 | 国家电网有限公司技术学院分公司 | Multi-angle multi-face recognition attendance checking method and system |
| CN112149517A (en) * | 2020-08-31 | 2020-12-29 | 三盟科技股份有限公司 | Face attendance checking method and system, computer equipment and storage medium |
| CN113095289A (en) * | 2020-10-28 | 2021-07-09 | 重庆电政信息科技有限公司 | Massive image preprocessing network method based on urban complex scene |
| CN112597854A (en) * | 2020-12-15 | 2021-04-02 | 重庆电子工程职业学院 | Non-matching type face recognition system and method |
| CN113469024A (en) * | 2021-06-29 | 2021-10-01 | 深圳市商汤科技有限公司 | Face attendance checking method and device, electronic equipment and computer readable storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109740501A (en) | A kind of Work attendance method and device of recognition of face | |
| CN105551104B (en) | A kind of automatic attendance checking system in middle and primary schools classroom differentiated based on monitoring image seat | |
| Lim et al. | Automated classroom monitoring with connected visioning system | |
| CN106169071A (en) | A kind of Work attendance method based on dynamic human face and chest card recognition and system | |
| WO2019127273A1 (en) | Multi-person face detection method, apparatus, server, system, and storage medium | |
| CN109858371A (en) | The method and device of recognition of face | |
| CN105844206A (en) | Identity authentication method and identity authentication device | |
| CN108960067A (en) | Real-time train driver motion recognition system and method based on deep learning | |
| CN108875461A (en) | A kind of human face in-vivo detection method and device | |
| CN108596041A (en) | A kind of human face in-vivo detection method based on video | |
| CN110414381A (en) | Tracing type face identification system | |
| CN107480586B (en) | Face characteristic point displacement-based biometric photo counterfeit attack detection method | |
| CN103605971A (en) | Method and device for capturing face images | |
| CN106709438A (en) | Method for collecting statistics of number of people based on video conference | |
| CN101976356A (en) | Face identification method for surfing identification policy, and identification system thereof | |
| WO2021217764A1 (en) | Human face liveness detection method based on polarization imaging | |
| Huang et al. | Attendance system based on dynamic face recognition | |
| Chowdhury et al. | Development of an automatic class attendance system using cnn-based face recognition | |
| CN107516076A (en) | Portrait identification method and device | |
| Agarwal et al. | Face recognition based smart and robust attendance monitoring using deep CNN | |
| Gill et al. | Attendance Management System Using Facial Recognition and Image Augmentation Technique | |
| Kainz et al. | Visual system for student attendance monitoring with non-standard situation detection | |
| CN109886084A (en) | Face authentication method, electronic equipment and storage medium based on gyroscope | |
| Kakran et al. | Identification and Recognition of face and number Plate for Autonomous and Secure Car Parking | |
| CN111191563A (en) | Face recognition method and system based on data sample and test data set training |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190510 |
|
| RJ01 | Rejection of invention patent application after publication |