CN108364374A - Face access control device based on deep learning and method - Google Patents
Face access control device based on deep learning and method Download PDFInfo
- Publication number
- CN108364374A CN108364374A CN201711455571.4A CN201711455571A CN108364374A CN 108364374 A CN108364374 A CN 108364374A CN 201711455571 A CN201711455571 A CN 201711455571A CN 108364374 A CN108364374 A CN 108364374A
- Authority
- CN
- China
- Prior art keywords
- face
- module
- access control
- image
- video
- 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
Classifications
-
- G—PHYSICS
- G07—CHECKING-DEVICES
- G07C—TIME OR ATTENDANCE REGISTERS; REGISTERING OR INDICATING THE WORKING OF MACHINES; GENERATING RANDOM NUMBERS; VOTING OR LOTTERY APPARATUS; ARRANGEMENTS, SYSTEMS OR APPARATUS FOR CHECKING NOT PROVIDED FOR ELSEWHERE
- G07C9/00—Individual registration on entry or exit
- G07C9/30—Individual registration on entry or exit not involving the use of a pass
- G07C9/32—Individual registration on entry or exit not involving the use of a pass in combination with an identity check
- G07C9/37—Individual registration on entry or exit not involving the use of a pass in combination with an identity check using biometric data, e.g. fingerprints, iris scans or voice recognition
-
- G—PHYSICS
- G07—CHECKING-DEVICES
- G07C—TIME OR ATTENDANCE REGISTERS; REGISTERING OR INDICATING THE WORKING OF MACHINES; GENERATING RANDOM NUMBERS; VOTING OR LOTTERY APPARATUS; ARRANGEMENTS, SYSTEMS OR APPARATUS FOR CHECKING NOT PROVIDED FOR ELSEWHERE
- G07C9/00—Individual registration on entry or exit
- G07C9/30—Individual registration on entry or exit not involving the use of a pass
- G07C9/38—Individual registration on entry or exit not involving the use of a pass with central registration
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Human Computer Interaction (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The present invention relates to access control fields, provide a kind of face access control device based on deep learning, including face database, image acquisition part, Face datection portion, recognition of face portion and access control portion.A kind of face access control method based on deep learning is also provided, is included the following steps：S1 by the face information typing of internal staff and is stored using face database；S2, using image acquisition part come collection site image；Whether S3 there is facial image in the image scene using Face datection portion to detect acquisition；The facial image detected is compared with the face information in the face database using recognition of face portion by S4；S5 is received the comparison information that the recognition of face portion is transmitted using access control portion and controls whether gate is opened.The present invention realizes gate function, can be suitable for the numerous occasions of personnel such as entire cell, school and workshop by being compared with the face information in face database after acquisition, detection and identification.
Description
Technical field
The present invention relates to access control fields, specially face access control device and method based on deep learning.
Background technology
With the fast development of science and technology, access control system has been deployed to the places such as many company and enterprises, school, workshop,
Current access control system is required for greatly the personnel to come in and go out to wear corresponding badge, and swiping the card could enter, or need to input fingerprint letter
Breath, as long as also a kind of infrared induction is to the close meeting automatic door opening of personnel, these scenes can meet basic gate inhibition's demand, but have
Other problems, if personnel forget that band badge, fingerprint input need to take, infrared detection can not rule out Migrant women, at this moment as can certainly
The face information of dynamic testing staff simultaneously judges whether to belong to internal staff, and belonging to internal staff, then access control carries out enabling behaviour
Make so that it is convenient to personnel enter, it can better access control system.
Application No. is the Chinese invention patents of CN201410809509.0《A kind of three-dimensional face gate identification system》, master
It wants feature to be to include man face image acquiring system, image procossing client, wireless communication module, comparison server, data storage
The image of acquisition is reached image procossing client, image by server, alarm and door lock microcontroller, man face image acquiring system
Processing client handles image by the tool built in itself and is merged as a secondary three-dimensional face images, after synthesis
The three-dimensional face database to prestore in three-dimensional face images and comparison server is compared, and if there is what is be consistent, is then sent out out
Otherwise door instruction sends warning instruction, decides whether to open the door by owner to door lock microcontroller.The present invention is than Quick Response Code identification technology
Safety higher acquires facial image using dual camera, then synthesizes 3-D view, compared to the three of utilization laser scanning
Dimension image cost is lower, and the scope of application is wider in the family.But the processing of its face information increases visitor all at client
Family end handles the pressure of data, it is impossible to reach the processing of big data quantity, so home scenarios can only be mainly used in, in addition adopt
The data volume that need to be preserved with 3-D view is also larger.
Invention content
The purpose of the present invention is to provide a kind of face access control device and method based on deep learning, at least can be with
Solve segmental defect in the prior art.
To achieve the above object, the embodiment of the present invention provides a kind of technical solution：A kind of face door based on deep learning
Prohibit control device, including face database, image acquisition part, Face datection portion, recognition of face portion and access control portion；
The face database for typing and stores internal staff's face information；
Described image acquisition portion is used for collection site image；
The Face datection portion, for detecting in the image scene acquired whether facial image occur；
The recognition of face portion, for carrying out the face information in the facial image detected and the face database
It compares；
The access control portion, for receiving comparison information that the recognition of face portion is transmitted and controlling whether gate is opened
It opens.
Further, described image acquisition portion includes at least video acquisition module, video pre-filtering module, video encoding module
Module, picture capture module, network transmission module；
The video acquisition module, for acquiring real-time video；
The video pre-filtering module, for carrying out white balance and exposure-processed to the real-time video；
The video encoding module carries out compressed encoding for the video data stream to the real-time video；
The picture captures module, and the frame picture for the facial image to detecting carries out compressed encoding and picture is made
The face picture of format；
The network transmission module, for leading to the video data after the face picture and compressed encoding that detect
Network transmission is crossed to other each portions, other described each sections Wei not image acquisition part, Face datection portion, recognition of face portion and door
Prohibit control unit.
Further, the Face datection portion includes at least realtime graphic receiving module, image analysis module and face inspection
Survey module；
The realtime graphic receiving module, for receiving the collected real-time image scene of image acquisition part；
Described image parsing module whether there is face, the face of appearance for parsing in the image scene received
Whether it is to occur for the first time, prevents a face from repeatedly triggering identification；
The face detection module for detecting the face information in the image scene received, and positions the face
Coordinate of the information in the face database.
Further, the recognition of face portion includes at least facial image receiving module, facial image comparing module and people
Face alignment algorithm module；
The facial image receiving module, for receiving the facial image；
The facial image comparing module, for comparing the data in the data and face database of the facial image of reception
It is right, judge the face for whether having satisfaction in face database；
Face alignment algoritic module, for being compared by the algorithm of deep learning.
Further, described image acquisition portion further includes video camera, and the video camera is installed on the gate.
The embodiment of the present invention also provides another technical solution：A kind of face access control method based on deep learning,
Include the following steps：
S1 by the face information typing of internal staff and is stored using face database in advance；
S2, using image acquisition part come collection site image；
S3, the image scene of acquisition is detected using Face datection portion, and judges whether facial image occur；
S4 obtains the facial image that the Face datection does not detect using recognition of face portion, and by the facial image with
Face information in the face database is compared, and obtains comparison result；
The comparison result is sent to access control portion by S5, and the access control portion receives the comparing result and controls
Whether open at gate processed.
Further, in the S2 steps, acquisition the specific steps are：
S21 acquires real-time video using video acquisition module；
S22 carries out white balance and exposure-processed using video pre-filtering module to the real-time video；
S23 carries out compressed encoding using video encoding module to the video data stream of the real-time video；
S24 captures module to carry out compressed encoding to the frame picture of the facial image detected and figure is made using picture
The face picture of piece format；
S25, using network transmission module by the video data after the face picture and compressed encoding that detect
By network transmission give other each portions, other described each sections not Wei image acquisition part, Face datection portion, recognition of face portion and
Access control portion.
Further, in the S3 steps, detection the specific steps are：
S31 receives the collected real-time image scene of image acquisition part using realtime graphic receiving module；
S32 whether there is face in the image scene using image analysis module to parse reception, and if it exists, then
Continue following step, if being not present, stops the step；
S33, the face information in the image scene using face detection module to detect reception, and position the face
Coordinate of the information in the face database, then the result is sent to the recognition of face portion.
Further, in the S4 steps, identification the specific steps are：
S41 receives the facial image using facial image receiving module；
S42 is passed through the data in the data and face database of the facial image of reception using facial image comparing module
The algorithm of deep learning is compared；When percentage of the comparison result more than or equal to setting, then it is identified that the face belongs to institute
The personnel in face database are stated, are denoted as correct result, and the correct result is passed in the access control portion；When comparison result is small
In the percentage of setting, then the personnel that the face is not belonging in the face database are identified, are denoted as error result, and by the mistake
As a result it passes in the access control portion.
Further, in the S5 steps, the access control portion receives the information that the recognition of face portion is transmitted, if the ratio
The condition opened the door is met to information, then controls gate unlatching, if the comparison information is unsatisfactory for door-opening condition, is not turned on gate.
Compared with prior art, the beneficial effects of the invention are as follows：It need not be handled using client, directly acquisition, inspection
It surveys and can be compared with the face information in face database after identifying, then control whether to open the door by access control portion, locate
Manage rate it is fast and can batch processing, be not only suitable for the less occasion of the personnel such as family, additionally it is possible to be suitable for entire cell,
The numerous occasion of the personnel such as school and workshop can effectively improve the security performance of existing access control system and increase existing gate inhibition system
The applicable place of system.
Description of the drawings
Fig. 1 is a kind of First Principle frame of the face access control device based on deep learning provided in an embodiment of the present invention
Figure；
Fig. 2 is a kind of image acquisition part of the face access control device based on deep learning provided in an embodiment of the present invention
Flow chart；
Fig. 3 is a kind of Face datection portion of the face access control device based on deep learning provided in an embodiment of the present invention
Flow chart；
Fig. 4 is a kind of face access control device recognition of face portion based on deep learning provided in an embodiment of the present invention
Flow chart；
Fig. 5 is a kind of second principle frame of the face access control device based on deep learning provided in an embodiment of the present invention
Figure；
Fig. 6 is that a kind of face access control device based on deep learning provided in an embodiment of the present invention is schemed using several sets
As the flow chart in acquisition portion, several set Face datection portions and several set access control portions；
Fig. 7 is a kind of step flow of the face access control method based on deep learning provided in an embodiment of the present invention
Figure.
Specific implementation mode
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation describes, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is based on
Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts all other
Embodiment shall fall within the protection scope of the present invention.
The embodiment of the present invention provides a kind of face access control device based on deep learning, this control device includes face
Library, image acquisition part, Face datection portion, recognition of face portion and access control portion；Wherein, face database is in typing and storage
Portion's personnel's face information；Image acquisition part is used for collection site image；The Face datection portion is used to detect the described existing of acquisition
Whether there is facial image in field picture；The facial image that the recognition of face portion is used to detect and the face database
In face information be compared；The access control portion is for receiving the comparison information and control that the recognition of face portion is transmitted
Whether gate is opened, it includes access control module.The present apparatus need not be handled using client, directly acquisition, detect with
And can be compared with the face information in face database after identification, then control whether to open the door by access control portion, processing speed
Rate it is fast and can batch processing, be not only suitable for the less occasion of the personnel such as family, additionally it is possible to be suitable for entire cell, school with
And the occasion that the personnel such as workshop are numerous, the security performance of existing access control system can be effectively improved and increase existing access control system
It is applicable in place.
Specifically, it referring to Fig. 1, image scene passes to image acquisition part 101, is integrated in video camera, is mainly responsible for
Acquire real-time video picture, used for follow-up each portion, 101 collected image scene of image acquisition part by Face datection portion come into
Row detection, wherein Face datection portion have video data analysis module 102, it collected to image acquisition part can in real time be regarded
Frequency carries out the algorithm of deep learning using GPU hardware according to analyzing processing is carried out, and whether there is face in video pictures to detect
Information, following Face datection portion judge whether to detect face 103, are then to carry out down, otherwise come back to Image Acquisition
Face information is preserved by transmission module 104 and is transmitted to by network down the recognition of face on backstage into guild by portion 101
Portion, the picture receiving module 105 in recognition of face portion receive the face picture information of transmission module, and by picture face partition mould
Block 106 evaluates the scoring of the face information in the face picture received, judges whether face partition is more than the M of setting
107, which sets according to actual conditions, carries out recognition of face again when M value of the scoring more than setting, is otherwise not processed,
To save flow, speed is changed up, next by face recognition module 108 according to above-mentioned appraisal result, when face partition is more than M
When, then can face picture be subjected to recognition of face, judged whether than the face 109 in middle database, according to face recognition module
108 as a result, judge the face whether in face database, if not in face database, check frequency of occurrence 110,
When not in database, but in real-time database ratio, counting the number of appearance, and judge that face occurrence number is more than the threshold of setting
Value N 111, if it exceeds N (N values are set according to actual conditions, can be set as three times, four inferior different numbers) announcement will be triggered
Alert module 113 is alerted, and is sent out face alarm and frequency warning information by network by first network transmission module 114
It sees off, if in face database, gate is opened by 112 access control module 115 of access control module, further includes client
End module 116, it can be according to first network transmission module 114 as a result, showing that different information alerts is used on the client
Family.
Optimize above-mentioned image acquisition part, please refer to Fig.1 and Fig. 2, image acquisition part include at least video acquisition module 202,
Video pre-filtering module 203, video encoding module module 204, picture capture module 207, the second network transmission module 208；Institute
Video acquisition module 202 is stated, for acquiring real-time video；The video pre-filtering module 203, for the real-time video into
Row white balance and exposure-processed；The video encoding module 204 is pressed for the video data stream to the real-time video
Reduce the staff code；The picture captures module 207, and the frame picture for the facial image to detecting carries out compressed encoding and figure is made
The face picture of piece format；Second network transmission module 208, for compiling the face picture detected and compression
Code after video data by network transmission give other each portions, other described each sections not Wei image acquisition part, Face datection portion,
Recognition of face portion and access control portion.Other than these modules, some existing moulds can also be increased according to actual conditions
Block.
Specifically, image acquisition part further includes video camera, which installs on the gate, the camera lens of video camera
201 by the picture catching to the sensor target surface of video camera for passing by the video camera, and video acquisition module 202 acquires camera lens transmission
The picture to come over, i.e. real-time video, video pre-filtering module 203 handle the picture of acquisition, mainly complete white balance, expose
The processing such as light, video encoding module 204 carry out compressed encoding, Face datection to treated the image of video pre-filtering module 203
Module to treated, analyze by image, and whether detection wherein carries out comprising face video preprocessing module treated image
Analysis, whether detection includes wherein face, and judges whether to detect face 206, is carried out for the result of face detection module
Further detection, judges whether to detect face 206, and when judging result is yes, picture captures module 207 to detecting face figure
The frame picture of picture carries out compressed encoding and the face picture of picture format is made, the picture format can be JPG, PNG, GIF with
And the public picture format such as TIFF, the video data after video encoding module 204 is encoded using the second network transmission module 208
Capture the face picture data after module 207 encodes with picture and other each portions be sent to by network, when transmission use gigabit with
Too network interface card 209 is sent, and can save sending time, improves efficiency.
Optimize above-mentioned Face datection portion, please refers to Fig.1 and Fig. 3, above-mentioned video data analysis module are refined as scheming in real time
As receiving module 302, image analysis module 303 and face detection module 304；The realtime graphic receiving module, for connecing
Receive the collected real-time image scene of image acquisition part；Described image parsing module, for parsing the scene photo received
It whether there is face as in, whether the face of appearance is to occur for the first time, prevents a face from repeatedly triggering identification；The face
Detection module for detecting the face information in the image scene received, and positions the face information in the face database
In coordinate.Other than these modules, some existing modules can also be increased according to actual conditions.
Specifically, the real-time video packet of the acquisition of above-mentioned image acquisition part 101 is received using realtime graphic receiving module 302,
302 real time video data of realtime graphic receiving module is parsed by image analysis module 303, it is main judge previous frame whether someone
Whether face, the face in picture occurred, and the multiple detection trigger of face were prevented, then by face detection module 304, to image analysis
The data of module 303 are detected, and the algorithm of deep learning are mainly carried out by GPU to realize Face datection, and judge whether
Detect face 305, it is no, then above-mentioned image acquisition part 101 is come back to, is, then enters recognition of face portion and access control portion
The step of in.
Optimize above-mentioned recognition of face portion, please refers to Fig.1 and Fig. 4, recognition of face portion include at least facial image and receive mould
Block, facial image comparing module and face alignment algoritic module；The facial image receiving module, for receiving the face
Image；The facial image comparing module, for the data of the facial image of reception to be compared with the data in face database,
Judge the face for whether having satisfaction in face database；Face alignment algoritic module, for being compared by the algorithm of deep learning.
Other than these modules, some existing modules can also be increased according to actual conditions.
Specifically, the facial image that above-mentioned Face datection portion detects, the figure are received using facial image receiving module 403
As can capture 207 processed picture of module by picture, facial image comparing module 404 be then used, according to face figure
As the face picture data that receiving module 403 obtains, picture is compared, picture is sent to face alignment arithmetic server
405 are compared, then arithmetic server is waited for return to the result compared, wherein face alignment arithmetic server 405 uses GPU
Deep learning algorithm carry out recognition of face, next judge whether the face score for comparing out has 406 more than P, be, then
Result is transmitted to access control portion.There is electric controller, it can control gate and automatically open in access control portion.
Another embodiment of the present apparatus, as shown in figure 5, the picture that transmission can be passed by video camera by camera lens passes to video camera
On sensor target surface, through the real-time video of camera lens, image analysis module detection image acquisition portion regards for image acquisition part acquisition
Whether frequency evidence, detection include wherein face information, judge whether to include face, according to the testing result of image analysis module,
Judge in video whether face information, such as detect face, then enter recognition of face server, which includes
The picture of face recognition module, the face that will determine that carries out recognition of face, is compared with face database, whether judges the face
In face database, judge whether than middle face, according to face recognition module as a result, judging real-time face whether in face database
In, alarm is if it is sent, and access control is wanted to open the door, using alarm module by face in the ratio of face recognition module
Warning information sent by network, then, access control module carries out enabling behaviour according to the warning information of alarm module
Make, gate is controlled using Electronic control, the instruction for receiving access control module carries out opening door operation.
Another embodiment of the present apparatus, Fig. 6 are that access control device is examined using several set image acquisition parts, several set faces
The flow chart in survey portion and several set access control portions is examined using more set image acquisition parts, several set faces in the present embodiment
Survey portion and more set access control portions, but only with a set of recognition of face portion as identification.Their specific implementation
Process is same as above, and is repeated no more.The present embodiment can be used for the numerous occasion of the personnel such as entire cell, school and workshop, can be effective
Ground improves the security performance of existing access control system and increases the applicable place of existing access control system, realizes the effect of batch processing
Fruit.
Referring to Fig. 7, the embodiment of the present invention also provides a kind of face access control method based on deep learning, including such as
Lower step：S1 by the face information typing of internal staff and is stored using face database in advance；S2 is adopted using image acquisition part
Collect image scene；S3, the image scene of acquisition is detected using Face datection portion, and judges whether facial image occur；S4 is adopted
The facial image that the Face datection does not detect is obtained with face identification part, and will be in the facial image and the face database
Face information is compared, and obtains comparison result；The comparison result is sent to access control portion, the access control by S5
Portion receives the comparing result and controls whether gate is opened.This method need not be handled using client, directly acquisition, inspection
It surveys and can be compared with the face information in face database after identifying, then control whether to open the door by access control portion, locate
Manage rate it is fast and can batch processing, be not only suitable for the less occasion of the personnel such as family, additionally it is possible to be suitable for entire cell,
The numerous occasion of the personnel such as school and workshop can effectively improve the security performance of existing access control system and increase existing gate inhibition system
The applicable place of system.
As the prioritization scheme of the embodiment of the present invention, in S2 steps, acquisition the specific steps are：S21, using video acquisition
Module acquires real-time video；S22 carries out at white balance and exposure the real-time video using video pre-filtering module
Reason；S23 carries out compressed encoding using video encoding module to the video data stream of the real-time video；S24, using picture
Module is captured to carry out compressed encoding to the frame picture of the facial image detected and the face picture of picture format is made；S25,
Video data after the face picture and compressed encoding that detect is passed through by network transmission using network transmission module
Give other each portions, other described each sections Wei not image acquisition part, Face datection portion, recognition of face portion and access control portion.
As the prioritization scheme of the embodiment of the present invention, in S3 steps, detection the specific steps are：S31, using realtime graphic
Receiving module receives the collected real-time image scene of image acquisition part；S32 is received using image analysis module to parse
The image scene in whether there is face, and if it exists, then continue following step, if being not present, stop the step；
S33, the face information in the image scene using face detection module to detect reception, and the face information is positioned in institute
The coordinate in face database is stated, then the result is sent to the recognition of face portion.
As the prioritization scheme of the embodiment of the present invention, in S4 steps, identification the specific steps are：S41, using facial image
Receiving module receives the facial image；S42, the data for the facial image that will be received using facial image comparing module with
Data in face database are compared by the algorithm of deep learning；When percentage of the comparison result more than or equal to setting, then
It is identified the personnel that the face belongs in the face database, is denoted as correct result, and the correct result is passed into the gate inhibition and is controlled
In portion processed；When percentage of the comparison result less than setting, then it is identified the personnel that the face is not belonging in the face database, is denoted as
Error result, and the error result is passed in the access control portion.
As the prioritization scheme of the embodiment of the present invention, in S5 steps, the access control portion receives the recognition of face portion
The information transmitted controls gate unlatching if the comparison information meets the condition opened the door to get to the information of correct result, if
The comparison information is unsatisfactory for door-opening condition to get to the information of error result, then is not turned on gate.In the present embodiment, also may be used
Open gate to adopt manually, can avoid it is injured on the face or wear masks etc. special circumstances trouble in human face recognition the case where
Occur.
It although an embodiment of the present invention has been shown and described, for the ordinary skill in the art, can be with
Understanding without departing from the principles and spirit of the present invention can carry out these embodiments a variety of variations, modification, replace
And modification, the scope of the present invention is defined by the appended.
Claims (10)
1. a kind of face access control device based on deep learning, it is characterised in that：Including face database, image acquisition part, people
Face test section, recognition of face portion and access control portion；
The face database for typing and stores internal staff's face information；
Described image acquisition portion is used for collection site image；
The Face datection portion, for detecting in the image scene acquired whether facial image occur；
The recognition of face portion, for comparing the facial image detected with the face information in the face database
It is right；
The access control portion, for receiving comparison information that the recognition of face portion is transmitted and controlling whether gate is opened.
2. a kind of face access control device based on deep learning as described in claim 1, it is characterised in that：Described image
Acquisition portion includes at least video acquisition module, video pre-filtering module, video encoding module module, picture and captures module, network
Transmission module；
The video acquisition module, for acquiring real-time video；
The video pre-filtering module, for carrying out white balance and exposure-processed to the real-time video；
The video encoding module carries out compressed encoding for the video data stream to the real-time video；
The picture captures module, and the frame picture for the facial image to detecting carries out compressed encoding and picture format is made
Face picture；
The network transmission module, for the video data after the face picture and compressed encoding that detect to be passed through net
Network is transferred to other each portions, other described each sections Wei not image acquisition part, Face datection portion, recognition of face portion and gate inhibition's control
Portion processed.
3. a kind of face access control device based on deep learning as described in claim 1, it is characterised in that：The face
Test section includes at least realtime graphic receiving module, image analysis module and face detection module；
The realtime graphic receiving module, for receiving the collected real-time image scene of image acquisition part；
Described image parsing module whether there is face for parsing in the image scene received, whether the face of appearance
To occur for the first time, prevent a face from repeatedly triggering identification；
The face detection module for detecting the face information in the image scene received, and positions the face information
Coordinate in the face database.
4. a kind of face access control device based on deep learning as described in claim 1, it is characterised in that：The face
Identification part includes at least facial image receiving module, facial image comparing module and face alignment algoritic module；
The facial image receiving module, for receiving the facial image；
The facial image comparing module, for the data of the facial image of reception to be compared with the data in face database,
Judge the face for whether having satisfaction in face database；
Face alignment algoritic module, for being compared by the algorithm of deep learning.
5. a kind of face access control device based on deep learning as described in claim 1, it is characterised in that：Described image
Acquisition portion further includes video camera, and the video camera is installed on the gate.
6. a kind of face access control method based on deep learning, which is characterized in that include the following steps：
S1 by the face information typing of internal staff and is stored using face database in advance；
S2, using image acquisition part come collection site image；
S3, the image scene of acquisition is detected using Face datection portion, and judges whether facial image occur；
S4 obtains the facial image that the Face datection does not detect using recognition of face portion, and by the facial image with it is described
Face information in face database is compared, and obtains comparison result；
The comparison result is sent to access control portion by S5, and the access control portion receives the comparing result and controls big
Whether door is opened.
7. a kind of face access control method based on deep learning as claimed in claim 6, which is characterized in that the S2 steps
In rapid, acquisition the specific steps are：
S21 acquires real-time video using video acquisition module；
S22 carries out white balance and exposure-processed using video pre-filtering module to the real-time video；
S23 carries out compressed encoding using video encoding module to the video data stream of the real-time video；
S24 captures module to carry out compressed encoding to the frame picture of the facial image detected and picture lattice are made using picture
The face picture of formula；
S25 is passed through the video data after the face picture and compressed encoding that detect using network transmission module
Network transmission gives other each portions, other described each sections Wei not image acquisition part, Face datection portion, recognition of face portion and gate inhibition
Control unit.
8. a kind of face access control method based on deep learning as claimed in claim 6, which is characterized in that the S3 steps
In rapid, detection the specific steps are：
S31 receives the collected real-time image scene of image acquisition part using realtime graphic receiving module；
S32 whether there is face in the image scene using image analysis module to parse reception, and if it exists, then continue
Following step stops the step if being not present；
S33, the face information in the image scene using face detection module to detect reception, and position the face information
Coordinate in the face database, then the result is sent to the recognition of face portion.
9. a kind of face access control method based on deep learning as claimed in claim 6, which is characterized in that the S4 steps
In rapid, identification the specific steps are：
S41 receives the facial image using facial image receiving module；
Data in the data and face database of the facial image of reception are passed through depth by S42 using facial image comparing module
The algorithm of study is compared；When percentage of the comparison result more than or equal to setting, then it is identified that the face belongs to the people
Personnel in face library are denoted as correct result, and the correct result are passed in the access control portion；It is set when comparison result is less than
Fixed percentage is then identified the personnel that the face is not belonging in the face database, is denoted as error result, and by the error result
It passes in the access control portion.
10. a kind of face access control method based on deep learning as claimed in claim 6, which is characterized in that the S5
In step, the access control portion receives the information that the recognition of face portion is transmitted, if the comparison information meets the condition opened the door,
Gate unlatching is then controlled, if the comparison information is unsatisfactory for door-opening condition, is not turned on gate.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711455571.4A CN108364374A (en) | 2017-12-28 | 2017-12-28 | Face access control device based on deep learning and method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711455571.4A CN108364374A (en) | 2017-12-28 | 2017-12-28 | Face access control device based on deep learning and method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108364374A true CN108364374A (en) | 2018-08-03 |
Family
ID=63010310
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711455571.4A Pending CN108364374A (en) | 2017-12-28 | 2017-12-28 | Face access control device based on deep learning and method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108364374A (en) |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109243021A (en) * | 2018-08-28 | 2019-01-18 | 余利 | Deeply learning type intelligent door lock system and device based on user experience analysis |
| CN109377614A (en) * | 2018-10-29 | 2019-02-22 | 重庆中科云丛科技有限公司 | A kind of face gate inhibition recognition methods, system, computer storage medium and equipment |
| CN109409258A (en) * | 2018-10-11 | 2019-03-01 | 深圳市恩钛控股有限公司 | A kind of Intelligent human-face detection identifying system based on deep learning face recognition algorithm |
| CN109615750A (en) * | 2018-12-29 | 2019-04-12 | 深圳市多度科技有限公司 | The recognition of face control method and device of door access machine, access control equipment, storage medium |
| CN109636990A (en) * | 2019-01-08 | 2019-04-16 | 武汉烽火众智数字技术有限责任公司 | A kind of cell electric vehicle management system and method based on RFID detection and recognition of face |
| CN110321835A (en) * | 2019-07-01 | 2019-10-11 | 杭州创匠信息科技有限公司 | Face guard method, system and equipment |
| CN110930624A (en) * | 2019-12-06 | 2020-03-27 | 深圳北斗国芯科技有限公司 | Safety in production monitored control system based on big dipper |
| CN111243142A (en) * | 2020-02-04 | 2020-06-05 | 浙江大华技术股份有限公司 | Failure processing method and device for intelligent door lock and storage medium |
| CN111710085A (en) * | 2020-05-27 | 2020-09-25 | 南京金陵塑胶化工有限公司 | Production workshop safety management method and system based on facial recognition |
| CN112598836A (en) * | 2020-12-14 | 2021-04-02 | 天时地理(深圳)智能科技有限公司 | Face recognition comparison system and method based on deep learning |
| CN112634520A (en) * | 2020-12-12 | 2021-04-09 | 上海应用技术大学 | Protection material distribution system based on face recognition technology |
| CN113780088A (en) * | 2021-08-11 | 2021-12-10 | 北京超维世纪科技有限公司 | Industrial patrol robot real-time detection algorithm for human invasion based on deep learning |
| CN114360183A (en) * | 2021-12-29 | 2022-04-15 | 慧之安信息技术股份有限公司 | Automatic alarm method and device based on deep learning |
Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN202003369U (en) * | 2011-03-21 | 2011-10-05 | 周曦 | Dynamic human face recognition system |
| CN103606216A (en) * | 2013-12-11 | 2014-02-26 | 周卫荣 | School entrance guard system |
| CN104102409A (en) * | 2013-04-12 | 2014-10-15 | 三星电子（中国）研发中心 | Scenario adaptation device and method for user interface |
| CN204131643U (en) * | 2014-10-15 | 2015-01-28 | 富盛科技股份有限公司 | A kind of image collecting device with face characteristic abstraction function |
| CN104376679A (en) * | 2014-11-24 | 2015-02-25 | 苏州立瓷电子技术有限公司 | Intelligent household pre-warning method |
| CN104392578A (en) * | 2014-11-24 | 2015-03-04 | 苏州立瓷电子技术有限公司 | Household fireproofing and guard against theft system with early warning function |
| CN104917963A (en) * | 2015-05-25 | 2015-09-16 | 深圳市金立通信设备有限公司 | Image processing method and terminal |
| CN104933426A (en) * | 2015-07-10 | 2015-09-23 | 李泰� | Self-service photo taking system and realization method thereof |
| CN105117692A (en) * | 2015-08-05 | 2015-12-02 | 福州瑞芯微电子股份有限公司 | Real-time face identification method and system based on deep learning |
| CN105427421A (en) * | 2015-11-16 | 2016-03-23 | 苏州市公安局虎丘分局 | Entrance guard control method based on face recognition |
| CN106203561A (en) * | 2016-08-31 | 2016-12-07 | 贵州远方生态环保科技股份有限公司 | A kind of intelligence environmental sanitation system public lavatory monitoring management system |
| CN106803301A (en) * | 2017-03-28 | 2017-06-06 | 广东工业大学 | A kind of recognition of face guard method and system based on deep learning |
| CN107067504A (en) * | 2016-12-25 | 2017-08-18 | 北京中海投资管理有限公司 | A kind of recognition of face safety-protection system and a suspect's detection and method for early warning |
-
2017
- 2017-12-28 CN CN201711455571.4A patent/CN108364374A/en active Pending
Patent Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN202003369U (en) * | 2011-03-21 | 2011-10-05 | 周曦 | Dynamic human face recognition system |
| CN104102409A (en) * | 2013-04-12 | 2014-10-15 | 三星电子（中国）研发中心 | Scenario adaptation device and method for user interface |
| CN103606216A (en) * | 2013-12-11 | 2014-02-26 | 周卫荣 | School entrance guard system |
| CN204131643U (en) * | 2014-10-15 | 2015-01-28 | 富盛科技股份有限公司 | A kind of image collecting device with face characteristic abstraction function |
| CN104376679A (en) * | 2014-11-24 | 2015-02-25 | 苏州立瓷电子技术有限公司 | Intelligent household pre-warning method |
| CN104392578A (en) * | 2014-11-24 | 2015-03-04 | 苏州立瓷电子技术有限公司 | Household fireproofing and guard against theft system with early warning function |
| CN104917963A (en) * | 2015-05-25 | 2015-09-16 | 深圳市金立通信设备有限公司 | Image processing method and terminal |
| CN104933426A (en) * | 2015-07-10 | 2015-09-23 | 李泰� | Self-service photo taking system and realization method thereof |
| CN105117692A (en) * | 2015-08-05 | 2015-12-02 | 福州瑞芯微电子股份有限公司 | Real-time face identification method and system based on deep learning |
| CN105427421A (en) * | 2015-11-16 | 2016-03-23 | 苏州市公安局虎丘分局 | Entrance guard control method based on face recognition |
| CN106203561A (en) * | 2016-08-31 | 2016-12-07 | 贵州远方生态环保科技股份有限公司 | A kind of intelligence environmental sanitation system public lavatory monitoring management system |
| CN107067504A (en) * | 2016-12-25 | 2017-08-18 | 北京中海投资管理有限公司 | A kind of recognition of face safety-protection system and a suspect's detection and method for early warning |
| CN106803301A (en) * | 2017-03-28 | 2017-06-06 | 广东工业大学 | A kind of recognition of face guard method and system based on deep learning |
Cited By (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109243021A (en) * | 2018-08-28 | 2019-01-18 | 余利 | Deeply learning type intelligent door lock system and device based on user experience analysis |
| CN109409258A (en) * | 2018-10-11 | 2019-03-01 | 深圳市恩钛控股有限公司 | A kind of Intelligent human-face detection identifying system based on deep learning face recognition algorithm |
| CN109409258B (en) * | 2018-10-11 | 2022-05-24 | 深圳市恩钛控股有限公司 | Intelligent face detection and recognition system based on deep learning face recognition algorithm |
| CN109377614A (en) * | 2018-10-29 | 2019-02-22 | 重庆中科云丛科技有限公司 | A kind of face gate inhibition recognition methods, system, computer storage medium and equipment |
| CN109615750B (en) * | 2018-12-29 | 2021-12-28 | 深圳市多度科技有限公司 | Face recognition control method and device for access control machine, access control equipment and storage medium |
| CN109615750A (en) * | 2018-12-29 | 2019-04-12 | 深圳市多度科技有限公司 | The recognition of face control method and device of door access machine, access control equipment, storage medium |
| CN109636990A (en) * | 2019-01-08 | 2019-04-16 | 武汉烽火众智数字技术有限责任公司 | A kind of cell electric vehicle management system and method based on RFID detection and recognition of face |
| CN110321835A (en) * | 2019-07-01 | 2019-10-11 | 杭州创匠信息科技有限公司 | Face guard method, system and equipment |
| CN110930624A (en) * | 2019-12-06 | 2020-03-27 | 深圳北斗国芯科技有限公司 | Safety in production monitored control system based on big dipper |
| CN111243142A (en) * | 2020-02-04 | 2020-06-05 | 浙江大华技术股份有限公司 | Failure processing method and device for intelligent door lock and storage medium |
| CN111710085A (en) * | 2020-05-27 | 2020-09-25 | 南京金陵塑胶化工有限公司 | Production workshop safety management method and system based on facial recognition |
| CN112634520A (en) * | 2020-12-12 | 2021-04-09 | 上海应用技术大学 | Protection material distribution system based on face recognition technology |
| CN112598836A (en) * | 2020-12-14 | 2021-04-02 | 天时地理(深圳)智能科技有限公司 | Face recognition comparison system and method based on deep learning |
| CN113780088A (en) * | 2021-08-11 | 2021-12-10 | 北京超维世纪科技有限公司 | Industrial patrol robot real-time detection algorithm for human invasion based on deep learning |
| CN114360183A (en) * | 2021-12-29 | 2022-04-15 | 慧之安信息技术股份有限公司 | Automatic alarm method and device based on deep learning |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108364374A (en) | Face access control device based on deep learning and method | |
| CN110942545B (en) | Dense person entrance guard control system and method based on face recognition and video fence | |
| KR101215948B1 (en) | Image information masking method of monitoring system based on face recognition and body information | |
| WO2020078229A1 (en) | Target object identification method and apparatus, storage medium and electronic apparatus | |
| CN206515931U (en) | A kind of face identification system | |
| CN110569720B (en) | Audio and video intelligent identification processing method based on audio and video processing system | |
| CN112396658B (en) | Indoor personnel positioning method and system based on video | |
| CN106204815A (en) | A kind of gate control system based on human face detection and recognition | |
| CN110110575A (en) | A kind of personnel leave post detection method and device | |
| CN109299683A (en) | A kind of security protection assessment system based on recognition of face and behavior big data | |
| CN101556717A (en) | ATM intelligent security system and monitoring method | |
| CN102521578A (en) | Method for detecting and identifying intrusion | |
| CN102917207A (en) | Motion sequence based abnormal motion vision monitoring system | |
| CN106951846A (en) | A kind of face 3D models typing and recognition methods and device | |
| CN101976356A (en) | Face identification method for surfing identification policy, and identification system thereof | |
| CN106780875A (en) | A kind of recognition of face gate inhibition integrated apparatus and its recognition methods | |
| CN110852306A (en) | Safety monitoring system based on artificial intelligence | |
| CN111539338A (en) | Pedestrian mask wearing control method, device, equipment and computer storage medium | |
| KR101092472B1 (en) | Video indexing system using surveillance camera and the method thereof | |
| CN112232211A (en) | Intelligent video monitoring system based on deep learning | |
| CN109243029A (en) | A kind of intelligent access control system | |
| CN107978136A (en) | A kind of method that false alarm automatic identification is realized in internet alarm system | |
| CN106296949A (en) | A kind of gate control system based on H264 video decoding techniques and method | |
| CN114125806B (en) | Wireless camera detection method based on cloud storage mode of wireless network flow | |
| CN114898443A (en) | Face data acquisition method and device |
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: 20180803 |
|
| RJ01 | Rejection of invention patent application after publication |