CN114710620A - Multi-angle face recognition equipment based on deep learning - Google Patents

Multi-angle face recognition equipment based on deep learning Download PDF

Info

Publication number
CN114710620A
CN114710620A CN202210306806.8A CN202210306806A CN114710620A CN 114710620 A CN114710620 A CN 114710620A CN 202210306806 A CN202210306806 A CN 202210306806A CN 114710620 A CN114710620 A CN 114710620A
Authority
CN
China
Prior art keywords
face
face recognition
module
angle
recognition device
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.)
Withdrawn
Application number
CN202210306806.8A
Other languages
Chinese (zh)
Inventor
解瑞云
海本斋
牛伟明
张颖
庞笑笑
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Henan Institute of Technology
Original Assignee
Henan Institute of Technology
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Henan Institute of Technology filed Critical Henan Institute of Technology
Priority to CN202210306806.8A priority Critical patent/CN114710620A/en
Publication of CN114710620A publication Critical patent/CN114710620A/en
Withdrawn legal-status Critical Current

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N23/00—Cameras or camera modules comprising electronic image sensors; Control thereof
    • H04N23/60—Control of cameras or camera modules
    • H04N23/61—Control of cameras or camera modules based on recognised objects
    • H04N23/611—Control of cameras or camera modules based on recognised objects where the recognised objects include parts of the human body
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F16—ENGINEERING ELEMENTS AND UNITS; GENERAL MEASURES FOR PRODUCING AND MAINTAINING EFFECTIVE FUNCTIONING OF MACHINES OR INSTALLATIONS; THERMAL INSULATION IN GENERAL
    • F16M—FRAMES, CASINGS OR BEDS OF ENGINES, MACHINES OR APPARATUS, NOT SPECIFIC TO ENGINES, MACHINES OR APPARATUS PROVIDED FOR ELSEWHERE; STANDS; SUPPORTS
    • F16M11/00—Stands or trestles as supports for apparatus or articles placed thereon ; Stands for scientific apparatus such as gravitational force meters
    • F16M11/02—Heads
    • F16M11/04—Means for attachment of apparatus; Means allowing adjustment of the apparatus relatively to the stand
    • F16M11/06—Means for attachment of apparatus; Means allowing adjustment of the apparatus relatively to the stand allowing pivoting
    • F16M11/12—Means for attachment of apparatus; Means allowing adjustment of the apparatus relatively to the stand allowing pivoting in more than one direction
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F16—ENGINEERING ELEMENTS AND UNITS; GENERAL MEASURES FOR PRODUCING AND MAINTAINING EFFECTIVE FUNCTIONING OF MACHINES OR INSTALLATIONS; THERMAL INSULATION IN GENERAL
    • F16M—FRAMES, CASINGS OR BEDS OF ENGINES, MACHINES OR APPARATUS, NOT SPECIFIC TO ENGINES, MACHINES OR APPARATUS PROVIDED FOR ELSEWHERE; STANDS; SUPPORTS
    • F16M11/00—Stands or trestles as supports for apparatus or articles placed thereon ; Stands for scientific apparatus such as gravitational force meters
    • F16M11/02—Heads
    • F16M11/18—Heads with mechanism for moving the apparatus relatively to the stand
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N23/00—Cameras or camera modules comprising electronic image sensors; Control thereof
    • H04N23/60—Control of cameras or camera modules
    • H04N23/66—Remote control of cameras or camera parts, e.g. by remote control devices
    • H04N23/661—Transmitting camera control signals through networks, e.g. control via the Internet
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N23/00—Cameras or camera modules comprising electronic image sensors; Control thereof
    • H04N23/60—Control of cameras or camera modules
    • H04N23/695—Control of camera direction for changing a field of view, e.g. pan, tilt or based on tracking of objects
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N23/00—Cameras or camera modules comprising electronic image sensors; Control thereof
    • H04N23/70—Circuitry for compensating brightness variation in the scene
    • H04N23/74—Circuitry for compensating brightness variation in the scene by influencing the scene brightness using illuminating means

Landscapes

  • Engineering & Computer Science (AREA)
  • Multimedia (AREA)
  • Signal Processing (AREA)
  • General Engineering & Computer Science (AREA)
  • Mechanical Engineering (AREA)
  • Image Analysis (AREA)

Abstract

The invention discloses a multi-angle face recognition device based on deep learning, which comprises: the base is of an arc structure, a first sliding groove is formed in the upper plane of the base, pillars are arranged at positions above the base, which are close to two ends of the base, the other ends of the two pillars are connected with guide rails, and the radian of the guide rails is the same as that of the base; the guide rail is provided with a plurality of sleeves, the sleeves are in clearance sliding connection with the guide rail, the outer side wall above the sleeves is connected with a fixed rod, the other end of the fixed rod is connected with a box body, a face recognition device is arranged above the box body, the outer side wall below the sleeves is connected with a sliding rod, the other end of the sliding rod is connected with a first sliding block, the first sliding block is arranged in a first sliding groove, and the first sliding block is in matching sliding connection with the first sliding groove; the face recognition device is arranged to shoot the face from multiple angles for recognition, the limitation of single-angle recognition is overcome, and recognition errors are reduced.

Description

Multi-angle face recognition equipment based on deep learning
Technical Field
The invention belongs to the field of face recognition equipment, and particularly relates to multi-angle face recognition equipment based on deep learning.
Background
In recent years, with the development of scientific technology, based on the traditional identification mode, the requirements of people on more intelligent and more convenient identity authentication products cannot be met by identity authentication products such as fingerprints and passwords, and the biological characteristic identification technology is a new identity authentication technology, is not lost, is not easy to counterfeit, is convenient to carry, and is gradually favored by people. Compared with other biological characteristics, the face recognition technology is convenient to use and high in accuracy. And the human face recognition method and the human face recognition system have the advantages of meeting the living habits of people identified through eyes, not needing specific acquisition equipment, being low in use cost and the like, and the human face recognition technology is widely applied.
Most face recognition systems at present mainly perform recognition based on a single angle, that is, recognition is performed from the front, and a user is required to face a lens in a close range in a front posture to perform face recognition, so that on one hand, the recognition angle is limited, and on the other hand, if the user does not perform recognition in the front posture, recognition errors and recognition errors may occur. Moreover, the current recognition system is in a dark environment, the camera can possibly capture the face, the face needs to be collected by means of external light, automatic light supplement of the system can not be achieved, and normal use of the system is affected.
Chinese patent application No. 202111160624.6 discloses a training method of a face recognition model, a face recognition method, and an electronic device, the training method including obtaining face sample images and target labels of the face sample images; inputting the face sample image into a feature extraction module in a face recognition model for feature extraction to obtain a plurality of feature maps; inputting the feature maps into a weight determination module in the face recognition model for feature processing, and respectively determining weights corresponding to the feature maps; determining a prediction label of the face recognition model based on a fusion result of the plurality of feature maps and the corresponding weights; and updating parameters of the face recognition model according to the difference between the target label and the predicted label, and determining the target face recognition model. The accuracy of the face recognition model can be improved by giving different weights to different feature maps. Among the above-mentioned prior art, carry out face identification through single angle, have great use limitation, and arouse the discernment mistake easily, influence the discernment accuracy.
Disclosure of Invention
Aiming at the defects of the prior art, the invention aims to provide multi-angle face recognition equipment based on deep learning, multiple face recognition devices are arranged to shoot faces in multiple angles for recognition, the limitation of single-angle recognition is overcome, recognition errors are reduced, three-dimensional space positioning is carried out on the faces of people by arranging a distance sensor and a measuring module, the executing module is used for controlling a first motor and a second motor to automatically rotate the angles of the face recognition devices, the faces in different positions are accurately aligned, the accuracy of multi-angle recognition is further improved, an automatic light mode is arranged, automatic light supplement is carried out in a dark environment, and the use convenience is improved.
The invention provides the following technical scheme:
a multi-angle face recognition equipment based on deep learning: the base is of an arc structure, a first sliding groove is formed in the upper plane of the base, pillars are arranged at positions above the base, which are close to two ends of the base, the other ends of the two pillars are connected with guide rails, the guide rails are of the arc structure, and the radian of the guide rails is the same as that of the base; the guide rail is provided with a plurality of sleeves, the sleeves are in clearance sliding connection with the guide rail, the outer side wall above the sleeves is connected with a fixed rod, the other end of the fixed rod is connected with a box body, a face recognition device is arranged above the box body, the outer side wall below the sleeves is connected with a sliding rod, the other end of the sliding rod is connected with a first sliding block, the first sliding block is arranged in a first sliding groove, and the first sliding block is in matching sliding connection with the first sliding groove;
the box body is connected to the upper end of dead lever, the inside first motor that is equipped with of box body, the output shaft drive of first motor is connected with the pinion, pinion engagement is connected with reduction gear, one side of reduction gear is connected with the drive shaft, the drive shaft extends to the outside of box body, just the drive shaft through the bearing that sets up with the box body rotates and connects, the other end of drive shaft is connected with the fixed block, it is equipped with face identification device to rotate on the fixed block.
Preferably, the position that the fixed block is close to the bottom is equipped with the dwang, and the mounting is connected the position that face recognition device is close to the bottom through the dwang that sets up, the top of fixed block is connected with the second motor, the output shaft of second motor has the lead screw, be equipped with solid board on the lead screw, gu the board is connected with the fixed block, the central point of solid board puts and has seted up the internal thread hole, and the lead screw matches with the internal thread hole and rotates and be connected.
Preferably, two connecting rods are symmetrically arranged on one side of the fixed plate, which is far away from the second motor, the two connecting rods are rotationally connected with the fixed plate through a pin shaft, the other end of the screw rod is connected with a limiting rod, and two ends of the limiting rod are connected with sliding blocks; the other ends of the two connecting rods are connected with rotating pieces, and the connecting rods are rotatably connected with the face recognition device through the rotating pieces; two logical groove has all been seted up along the axial to the connecting rod intermediate position, the sliding block sets up at logical inslot, its sliding block with lead to groove sliding connection.
Preferably, one end of the rotating part, which is close to the face recognition device, is connected with a second sliding block, a second sliding groove is formed in the position, which is close to the top, of the side wall of the face recognition device, and the second sliding groove is connected with the second sliding block in a sliding mode in a matching mode.
Preferably, the number of the face recognition devices is three, each face recognition device comprises a distance sensor and an illumination sensor, the distance sensors and the illumination sensors are connected with a data acquisition card, the data acquisition cards are connected with an FPGA module, and the FPGA module is electrically connected with an illuminating lamp and a camera; the FPGA module is equipped with the wifi module, the wifi module is connected with the wifi converter through wireless communication's mode, the wifi converter is connected with the host computer through serial port communication's mode, turns into digital signal transmission to the host computer through the wifi converter with the image video, the infrared data of gathering, illumination data.
Preferably, the upper computer is connected with a digital resistor module, the digital resistor module is electrically connected with a switch module, and the switch module controls the on and off of the illuminating lamp; the host computer is connected with measuring module, measuring module is through the infrared signal of receiving, and the position of perception people's face in three-dimensional space is fixed a position the people face, and measuring module is connected with control module, and measuring module transmits the digital signal of the three-dimensional space data at people place to control module, and control module transmits digital signal to execution module, the first motor of execution module electric connection and second motor, through the turned angle of the first motor of execution module send instruction control and second motor, change face identification module's angle of raising and control angle.
Preferably, the upper computer is connected with a server, the server is used for identifying the face and is provided with a registration module and an identification module, and the registration module comprises face detection, face alignment, feature extraction and classifier training; the recognition module comprises face detection, face alignment, feature extraction and face recognition; before face recognition, face registration is required, and multi-angle information of the face is uploaded to a server to be stored.
Preferably, the recognition method of the face recognition device includes:
step one, a recognized face enters a region of a face recognition device, three-dimensional space positioning is carried out on the face through a distance sensor, and positioning information is transmitted to an upper computer for processing;
the upper computer analyzes the three-dimensional spatial information of the face through the measuring module, and transmits digital instructions of the three-dimensional spatial information of the face to the control module and the execution module, and the execution module controls the first motor and the second motor to rotate so as to adjust the angles of the three face recognition devices and acquire images or videos; and step three, transmitting the acquired image information of different angles to an upper computer and a server, carrying out key point detection, face alignment and feature extraction on the face through an identification module, then carrying out classifier training comparison identification, judging whether the face leaves an identification area, carrying out image acquisition again if the face leaves the identification area, carrying out feature fusion through the images of three angles if the face does not leave the identification area, and outputting an identification result.
Preferably, in order to improve the accuracy of the shooting angles of the face recognition devices in the face recognition process, three face recognition devices are arranged on the guide rail in total, the three face recognition devices can be adjusted in a sliding mode on the guide rail through the sleeve, the face recognition device in the middle position is taken as the origin of the coordinate system, the horizontal adjustment angle of the face recognition device in the middle position meets (-pi/6), the horizontal adjustment angle of the face recognition device in the right side position meets (pi/3-pi/2), and the horizontal adjustment angle of the face recognition device in the middle position meets (-pi/3-pi/2); the face recognition device adjusts the angle in the vertical direction to meet (-pi/3). Through the method, the recognized face can be shot in a horizontal plane in a 0-180-degree blind angle-free mode, and the shooting accuracy is improved. The first motor and the second motor are servo motors, the first motor drives the pinion to rotate when rotating, the pinion is meshed with the pinion to drive the reduction gear to rotate, the reduction gear drives the fixing block to slowly rotate, and the face recognition device is driven to carry out angle adjustment in the horizontal direction through rotation of the fixing block.
When the face recognition device adjusts the angle in the vertical direction, the second motor drives the screw rod to rotate, and in the rotating process of the screw rod, the fixed plate is connected with the screw rod through the threaded hole,when the lead screw clockwise turning, then the relative slip takes place for second slider and the second spout that the connecting rod tip set up, the gag lever post of lead screw tip takes place to slide through the sliding block that sets up in logical inslot simultaneously, because solid board and fixed block fixed connection, the contained angle that two connecting rods are close to solid board one end diminishes, then the straight line distance between face identification device and the solid board becomes long, through the dwang that the bottom set up during through, face identification device's upper portion leans out, accomplishes the decurrent regulation of angle. If the second motor drives the screw rod to rotate anticlockwise, the directions of all the parts are opposite to the directions, the included angle between the two connecting rods and one end of the fixed plate is enlarged, the linear distance between the face recognition device and the fixed plate is shortened, the upper portion of the face recognition device is inclined inwards, and the upward adjustment of the angle is completed. In order to increase the controllability and accuracy of the angle in the adjustment process, the distance from the fixed plate to the face recognition device in the initial state is l1, the distance after adjustment is l2, the included angle between the connecting rod and the fixed plate is theta 1, the length S of the connecting rod meets the requirement, and S is eta, cos theta 1(l1/n-l2/n)2(ii) a Eta is a relation coefficient, and the value range is 3.22-15.6; the units l1, l2 are cm. The vertical inclination angle θ 2 of the face recognition device satisfies θ 2 ═ arc ((l1-l2)/C), and C is the linear distance between the rotating member and the fixed plate, and is in cm.
When fixing a position people's face through distance sensor, carry out the in-process of measuring through measuring module, in order to increase measuring degree of accuracy, use the face identification device of intermediate position as initial point O, use the direction of face identification device and arc guide rail tangent line to be the X axle, vertical direction is the Y axle, is the Z axle with face vertically direction, then when carrying out facial three-dimensional location, facial coordinate satisfies the relation:
x＝((a1sinθ3)2-(a2cosθ4)2)1/2+a3sinθ5；
y＝((a1cosθ3)2+(a2sinθ4)2)1/2+a3sinθ5；
z＝((a1sinθ3)2-(a2cosθ4)2)1/2+a3sinθ5；
in the above formula, a1For the orthographic projection of the sliding distance of the face recognition device on the guide rail on the X axis, theta3Is the included angle between the face recognition device and the X axis; d2Vertical height, θ, of face and face recognition means4Is the included angle between the face recognition device and the Y axis; d3Is the vertical distance of the face from the face recognition device, theta5Is the angle between the face recognition device and the Z axis. The distance between the face and the three face recognition devices is monitored through the displacement sensor, the measured data are transmitted to the control module, the control module converts the measured data into digital signals, the digital signals are transmitted to the execution module, and the execution module transmits the digital signals to the first motor and the second motor to adjust the horizontal rotation angle and the inclination angle theta 2 of the face recognition devices.
The illumination of facial discernment environment is gathered through the light sensor who sets up, when the illumination is less than the threshold value of settlement, switch module opens the light automatically and carries out the light filling, when ambient illumination is higher than the threshold value of settlement, switch module self-closing light, switch module calculates the power change law of light at the illumination value of host computer settlement and the illumination that the light awakens up, calculates the resistance value change in the digital resistance according to the power change law of light, transmits the change law of resistance value to the host computer, and the host computer controls according to the digital resistance change law of uploading, and the resistance change in the control digital resistance piece, the sub-resistor of digital resistance chip is the linear minimum resistance value that reduces from the maximum resistance along with the illumination is long. Along with the regional luminance environment of facial collection, carry out the light filling by oneself through switch module, host computer storage switch module's lighting data awakens the data of illumination etc. according to switch module and carries out classification training, reachs the regional luminance change law of facial collection, controls the light filling automatically through switch module. Through the above arrangement, the facial information can be conveniently collected in the dark environment, the accuracy of information collection is improved, more intelligent control is realized, and meanwhile, the electric energy can be saved.
Compared with the prior art, the invention has the following beneficial effects:
(1) according to the multi-angle face recognition equipment based on deep learning, the plurality of face recognition devices are arranged to shoot the face at multiple angles for recognition, the limitation of single-angle recognition is overcome, and recognition errors are reduced.
(2) According to the multi-angle face recognition equipment based on deep learning, the distance from the fixed plate to the face recognition device in the initial state is limited, the distance after adjustment is l2, and the angle controllability and accuracy in the adjustment process are further improved through the relation between the included angles of the connecting rod and the fixed plate.
(3) According to the multi-angle face recognition equipment based on deep learning, the face of a person is positioned in a three-dimensional space by arranging the distance sensor and the measuring module, and the executing module controls the first motor and the second motor to automatically rotate the angle of the face recognition device, so that the faces in different positions are accurately aligned, and the accuracy of multi-angle recognition is further improved.
(4) According to the multi-angle face recognition equipment based on deep learning, automatic light supplement is performed through the arranged switch module, so that facial information can be conveniently acquired in a dark environment, the accuracy of information acquisition is improved, more intelligent control is realized, and meanwhile, electric energy can be saved.
(5) According to the multi-angle face recognition equipment based on deep learning, disclosed by the invention, the horizontal rotation angle and the vertical inclination angle of the face recognition device are limited, so that the recognized face can be shot in a horizontal plane at 0-180 degrees without dead angles, and the shooting accuracy is favorably increased.
(6) According to the multi-angle face recognition equipment based on deep learning, due to the fact that the plurality of face recognition devices are arranged and are communicated through wifi, installation and debugging of the equipment are facilitated, the multi-angle face recognition equipment can be a face recognition system which can accommodate a large range and multiple angles, more facial image features can be collected, a person to be recognized does not need to watch a camera, the recognition range is enlarged, and accuracy of face recognition is improved.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings that are required to be used in the embodiments will be briefly described below, it should be understood that the following drawings only illustrate some embodiments of the present invention and therefore should not be considered as limiting the scope, and for those skilled in the art, other related drawings can be obtained according to the drawings without inventive efforts.
Fig. 1 is an overall schematic view of the present invention.
Fig. 2 is a schematic structural diagram of the face recognition device of the present invention.
Fig. 3 is a top view of the face recognition apparatus of the present invention.
Fig. 4 is a system block diagram of an identification device of the present invention.
Fig. 5 is a block diagram of a system of a face recognition device according to the present invention.
FIG. 6 is a block diagram of the control system of the present invention.
FIG. 7 is a block diagram of an identification module of the present invention.
Fig. 8 is a flow chart of face recognition of the present invention.
In the figure: 1. a base; 2. a pillar; 3. a guide rail; 4. a sleeve; 5. a fixing rod; 6. a slide bar; 7. a first chute; 8. a first slider; 9. a face recognition device; 10. a box body; 11. a first motor; 12. a pinion gear; 13. a reduction gear; 14. a drive wheel; 15. a fixed block; 16. rotating the rod; 17. a second motor; 18. a screw rod; 19. fixing the plate; 20. a connecting rod; 21. a through groove; 22. a limiting rod; 23. a slider; 24. a rotating member; 25. a second chute; 26. and a second slider.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the technical solutions of the embodiments of the present invention will be described in detail and completely with reference to the accompanying drawings. It is to be understood that the described embodiments are only a few embodiments of the present invention, and not all embodiments. All other embodiments, which can be obtained by a person skilled in the art without any inventive step based on the embodiments of the present invention, are within the scope of the present invention.
Thus, the following detailed description of the embodiments of the present invention, as presented in the figures, is not intended to limit the scope of the invention, as claimed, but is merely representative of selected embodiments of the invention. All other embodiments, which can be obtained by a person skilled in the art without any inventive step based on the embodiments of the present invention, are within the scope of the present invention.
The first embodiment is as follows:
as shown in fig. 1-3, a multi-angle face recognition apparatus based on deep learning: the base comprises a base 1, wherein the base 1 is of an arc structure, a first sliding groove 7 is formed in the upper plane of the base 1, supporting columns 2 are arranged at positions above the base 1, which are close to two ends of the base, the other ends of the two supporting columns 2 are connected with guide rails 3, the guide rails 3 are of arc structures, and the radian of the guide rails 3 is the same as that of the base 1; the guide rail 3 is provided with a plurality of sleeves 3, the sleeves 3 are in clearance sliding connection with the guide rail 3, the outer side wall of the upper part of each sleeve 3 is connected with a fixed rod 5, the other end of each fixed rod 5 is connected with a box body 10, a face recognition device 9 is arranged above the box body 10, the outer side wall of the lower part of each sleeve 3 is connected with a sliding rod 6, the other end of each sliding rod 6 is connected with a first sliding block 8, each first sliding block 8 is arranged in a first sliding groove 7, and each first sliding block 8 is in matching sliding connection with the first sliding groove 7;
box body 10 is connected to the upper end of dead lever 5, box body 10 inside is equipped with first motor 11, the output shaft drive of first motor 11 is connected with pinion 12, pinion 12 meshing is connected with reduction gear 13, one side of reduction gear 13 is connected with drive shaft 14, drive shaft 14 extends to the outside of box body 10, just drive shaft 14 through the bearing that sets up with box body 10 rotates and connects, the other end of drive shaft 14 is connected with fixed block 15, it is equipped with face recognition device 9 to rotate on the fixed block 15.
The position that fixed block 15 is close to the bottom is equipped with dwang 16, and the mounting is connected the position that face recognition device 9 is close to the bottom through the dwang 16 that sets up, the top of fixed block 15 is connected with second motor 17, second motor 17's output shaft has lead screw 18, be equipped with solid board 19 on the lead screw 18, gu board 19 is connected with fixed block 15, the internal thread hole has been seted up to the central point of solid board 19, and lead screw 18 matches with the internal thread hole and rotates and be connected.
Two connecting rods 20 are symmetrically arranged on one side, away from the second motor 17, of the fixed plate 19, the two connecting rods 20 are rotatably connected with the fixed plate 19 through arranged pin shafts, the other end of the screw rod 18 is connected with a limiting rod 22, and two ends of the limiting rod 22 are connected with sliding blocks 23; the other ends of the two connecting rods 20 are connected with a rotating part 24, and the connecting rods 20 are rotatably connected with the face recognition device 9 through the rotating part 24; two through groove 21 has all been seted up along the axial to connecting rod 20 intermediate position, sliding block 23 sets up in through groove 21, its sliding block 23 with through groove 21 sliding connection.
One end of the rotating piece 24 close to the face recognition device 9 is connected with a second sliding block 26, a second sliding groove 25 is formed in the position, close to the top, of the side wall of the face recognition device 9, and the second sliding groove 25 is connected with the second sliding block 26 in a sliding mode in a matching mode.
Example two:
as shown in fig. 4 to 8, on the basis of the first embodiment, three face recognition devices 9 are provided, each face recognition device 9 includes a distance sensor and an illumination sensor, the distance sensor and the illumination sensor are connected to a data acquisition card, the data acquisition card is connected to an FPGA module, and the FPGA module is electrically connected to an illuminating lamp and a camera; the FPGA module is equipped with the wifi module, the wifi module is connected with the wifi converter through wireless communication's mode, the wifi converter is connected with the host computer through serial port communication's mode, turns into digital signal transmission to the host computer through the wifi converter with the image video, the infrared data of gathering, illumination data.
The upper computer is connected with a digital resistor module, the digital resistor module is electrically connected with a switch module, and the switch module controls the on and off of the illuminating lamp; the host computer is connected with measuring module, measuring module is through the infrared signal of receiving, and the position of perception people's face in three-dimensional space is fixed a position the people's face, and measuring module is connected with control module, and measuring module transmits the digital signal of the three-dimensional space data at people place to control module, and control module transmits digital signal to execution module, the first motor of execution module electric connection 11 and second motor 17, through execution module send instruction control first motor 11 and second motor 17's turned angle, change face identification module's angle of rising and control the angle.
The upper computer is connected with a server, a face is identified in the server, the server is provided with a registration module and an identification module, and the registration module comprises face detection, face alignment, feature extraction and classifier training; the recognition module comprises face detection, face alignment, feature extraction and face recognition; before face recognition, face registration is required, and multi-angle information of the face is uploaded to a server to be stored.
Example three:
on the basis of the first embodiment, in order to improve the accuracy of the shooting angles of the face recognition devices 9 in the face recognition process, three face recognition devices 9 are arranged on the guide rail 3, the three face recognition devices 9 can be adjusted in a sliding manner on the guide rail 3 through the sleeve 3, the face recognition device 9 at the middle position is taken as the origin of the coordinate system, the horizontal adjustment angle of the face recognition device 9 at the middle position meets (-pi/6), the horizontal adjustment angle of the face recognition device 9 at the right side position meets (pi/3-pi/2), and the horizontal adjustment angle of the face recognition device 9 at the middle position meets (-pi/3-pi/2); the face recognition device 9 adjusts the angle in the vertical direction to satisfy (-pi/3). Through the method, the recognized face can be shot in a horizontal plane in a 0-180-degree blind angle-free mode, and the shooting accuracy is improved. The first motor 11 and the second motor 17 are both servo motors, the first motor 11 drives the pinion 12 to rotate when rotating, the pinion 12 is meshed with the reduction gear 13 to drive the reduction gear 13 to rotate, the reduction gear 13 drives the fixing block 15 to slowly rotate, and the face recognition device 9 is driven to carry out angle adjustment in the horizontal direction through rotation of the fixing block 15.
When the face recognition device 9 adjusts the angle in the vertical direction, the second motor 17 drives the screw rod 18 to rotate, and the fixed plate 19 and the screw rod 18 are screwed in the rotating process of the screw rod 18The line jogged joint, when lead screw 18 clockwise turning, then relative slip takes place for second slider 26 and the second spout 25 that the connecting rod 20 tip set up, the gag lever post 22 of lead screw 18 tip takes place to slide in leading to groove 21 through the slider 23 that sets up simultaneously, because solid board 19 and fixed block 15 fixed connection, the contained angle that two connecting rods 20 are close to solid board 19 one end diminishes, then the linear distance between face identification device 9 and the solid board 19 lengthens, through the dwang 16 that the bottom set up during through, the upper portion of face identification device 9 leans out, accomplish the downward regulation of angle. If the second motor 17 drives the screw rod 18 to rotate counterclockwise, the directions of the components are opposite to the above, the included angle between the two connecting rods 20 and the end close to the fixed plate 19 is increased, the linear distance between the face recognition device 9 and the fixed plate 19 is shortened, the upper part of the face recognition device 9 is inclined inwards, and the upward angle adjustment is completed. In order to increase the controllability and accuracy of the angle during the adjustment process, the distance between the fixed plate 19 and the face recognition device 9 in the initial state is l1, the distance after the adjustment is l2, and the included angle between the connecting rod 20 and the fixed plate 19 is theta 1, so that the length S of the connecting rod 20 is satisfied, and S is eta · cos theta 1(l1/n-l2/n)2(ii) a Eta is a relation coefficient, and the value range is 3.22-15.6; the units l1, l2 are cm. The vertical tilt angle θ 2 of the face recognition device 9 is satisfied, where θ 2 is arc ((l1-l2)/C), and C is the linear distance between the rotating member 24 and the fixed plate 19, and is in cm.
Example four
On the basis of the first embodiment, when locating the face of the person through the distance sensor, in the process of measuring through the measuring module, in order to increase the accuracy of measurement, the face recognition device at the intermediate position is taken as the original point O, the direction of the face recognition device and the tangent line of the arc-shaped guide rail is taken as the X axis, the vertical direction is the Y axis, and the direction perpendicular to the face is the Z axis, so that when the face is located three-dimensionally, the coordinates of the face satisfy the following relationship:
x＝((a1sinθ3)2-(a2cosθ4)2)1/2+a3sinθ5；
y＝((a1cosθ3)2+(a2sinθ4)2)1/2+a3sinθ5；
z＝((a1sinθ3)2-(a2cosθ4)2)1/2+a3sinθ5；
in the above formula, a1For the orthographic projection of the sliding distance of the face recognition device on the guide rail on the X axis, theta3Is the included angle between the face recognition device and the X axis; d2Vertical height, θ, of face and face recognition means4Is the included angle between the face recognition device and the Y axis; d3Is the vertical distance of the face from the face recognition device, θ5Is the angle between the face recognition device and the Z axis. The distance between the face and the three face recognition devices is monitored through the displacement sensor, the measured data are transmitted to the control module, the control module converts the measured data into digital signals, the digital signals are transmitted to the execution module, and the execution module transmits the digital signals to the first motor and the second motor to adjust the horizontal rotation angle and the inclination angle theta 2 of the face recognition devices.
The illumination of facial discernment environment is gathered through the illumination sensor who sets up, and when illuminance was less than the threshold value of settlement, switch module opened the light automatically and carried out the light filling, and when ambient illuminance was higher than the threshold value of settlement, switch module self-closing light, switch module calculates the power change law of light at the illuminance value of host computer settlement and the illumination that the light awakened up, calculates the resistance value change in the digital resistance according to the power change law of light, transmits the change law of resistance value to the host computer, and the host computer controls according to the digital resistance change law of uploading, controls the resistance change in the digital resistance piece, and the sub-resistor of digital resistance chip is the linear reduction of being minimum resistance value from the maximum resistance along with the illumination duration. Along with the regional luminance environment of facial collection, carry out the light filling by oneself through switch module, the data of host computer storage switch module are according to switch module awaken the data of illumination etc. and are carried out the classification training, reach the regional luminance change law of facial collection, control the light filling automatically through switch module. Through the setting, the face information collection under the dark environment is facilitated, the information collection accuracy is improved, more intelligent control is realized, and meanwhile, the electric energy can be saved.
EXAMPLE five
On the basis of the first embodiment, the recognition method of the face recognition device comprises the following steps:
step one, a recognized face enters a region of a face recognition device 9, the face is firstly positioned in a three-dimensional space through a distance sensor, and positioning information is transmitted to an upper computer for processing;
step two, the upper computer analyzes the three-dimensional spatial information of the human face through the measurement module, and then transmits a digital instruction of the three-dimensional spatial information of the human face to the control module and the execution module, and the execution module controls the first motor 11 and the second motor 17 to rotate to adjust the angles of the three human face recognition devices 9 so as to collect images or videos;
and step three, transmitting the acquired image information of different angles to an upper computer and a server, carrying out key point detection, face alignment and feature extraction on the face through an identification module, then carrying out classifier training comparison identification, judging whether the face leaves an identification area, carrying out image acquisition again if the face leaves the identification area, carrying out feature fusion through the images of three angles if the face does not leave the identification area, and outputting an identification result.
The device that obtains through above-mentioned technical scheme is a multi-angle face identification equipment based on degree of depth learning, carries out the multi-angle through setting up a plurality of face identification devices and shoots the people's face and discern, overcomes single angle recognition's limitation, reduces the identification error. The distance from the fixed plate to the face recognition device in the initial state is limited, the distance after adjustment is l2, and the controllability and the accuracy of the angle in the adjustment process are further improved through the relation between the included angle of the connecting rod and the fixed plate. And carry out three-dimensional space orientation to people's face through setting up distance sensor and measuring module to through the angle that first motor of execution module control and second motor rotate face identification device automatically, the face in different position is aimed at to the accuracy, has further promoted multi-angle identification's accuracy. Carry out automatic light filling through the switch module who sets up, be convenient for carry out facial information's collection under dark surrounds, promote information acquisition's accuracy, intelligent control more can practice thrift the electric energy simultaneously again. The horizontal rotation angle and the vertical inclination angle of the face recognition device are limited, so that the recognized face can be shot in a horizontal plane at 0-180 degrees without dead angles, and the shooting accuracy is improved. Through setting up between a plurality of face identification device through wifi communication, the installation and the debugging of the equipment of being convenient for can be the face identification system who accepts on a large scale, multi-angle, can gather more facial image characteristics, make by the discernment person need not to watch the camera, increased the recognition scope, promote facial recognition's accuracy.
Other technical solutions not described in detail in the present invention are prior art in the field, and are not described herein again.
The above description is only a preferred embodiment of the present invention, and is not intended to limit the present invention, and it will be apparent to those skilled in the art that various modifications and variations can be made in the present invention; any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims (8)

1. A multi-angle face recognition equipment based on deep learning: the device is characterized by comprising a base (1), wherein the base (1) is of an arc-shaped structure, a first sliding groove (7) is formed in the upper plane of the base (1), supporting columns (2) are arranged at positions, close to two ends, above the base (1), the other ends of the two supporting columns (2) are connected with guide rails (3), the guide rails (3) are of arc-shaped structures, and the radian of the guide rails (3) is identical to that of the base (1); the guide rail (3) is provided with a plurality of sleeves (3), the sleeves (3) are in clearance sliding connection with the guide rail (3), the outer side wall of the upper part of each sleeve (3) is connected with a fixed rod (5), the other end of each fixed rod (5) is connected with a box body (10), a face recognition device (9) is arranged above each box body (10), the outer side wall of the lower part of each sleeve (3) is connected with a sliding rod (6), the other end of each sliding rod (6) is connected with a first sliding block (8), each first sliding block (8) is arranged in a first sliding groove (7), and each first sliding block (8) is in matching sliding connection with the corresponding first sliding groove (7);
box body (10) is connected to the upper end of dead lever (5), box body (10) inside is equipped with first motor (11), the output shaft drive of first motor (11) is connected with pinion (12), pinion (12) meshing is connected with reduction gear (13), one side of reduction gear (13) is connected with drive shaft (14), drive shaft (14) extend to the outside of box body (10), just drive shaft (14) through the bearing that sets up with box body (10) rotate and connect, the other end of drive shaft (14) is connected with fixed block (15), it is equipped with face identification device (9) to rotate on fixed block (15).
2. The multi-angle face recognition device based on deep learning of claim 1, characterized in that, the position that fixed block (15) is close to the bottom is equipped with dwang (16), and the fixed part is connected face recognition device (9) through dwang (16) that sets up and is close to the position of bottom, the top of fixed block (15) is connected with second motor (17), the output shaft of second motor (17) has lead screw (18), be equipped with solid board (19) on lead screw (18), gu board (19) is connected with fixed block (15), the central point of solid board (19) puts and has seted up the internal thread hole, lead screw (18) and internal thread hole matching rotation are connected.
3. The multi-angle face recognition device based on deep learning of claim 2, wherein two connecting rods (20) are symmetrically arranged on one side of the fixed plate (19) far away from the second motor (17), the two connecting rods (20) are rotatably connected with the fixed plate (19) through arranged pin shafts, the other end of the screw rod (18) is connected with a limiting rod (22), and two ends of the limiting rod (22) are connected with sliding blocks (23); the other ends of the two connecting rods (20) are connected with rotating pieces (24), and the connecting rods (20) are rotatably connected with the face recognition device (9) through the rotating pieces (24); two link (20) intermediate position all has seted up logical groove (21) along the axial, sliding block (23) set up in leading to groove (21), its sliding block (23) with lead to groove (21) sliding connection.
4. The deep learning-based multi-angle face recognition device according to claim 3, wherein one end of the rotating member (24) close to the face recognition device (9) is connected with a second sliding block (26), a second sliding groove (25) is formed in the side wall of the face recognition device (9) close to the top, and the second sliding groove (25) is in sliding connection with the second sliding block (26) in a matching manner.
5. The deep learning-based multi-angle face recognition equipment according to any one of claims 1 to 4, wherein three face recognition devices (9) are provided, each face recognition device (9) comprises a distance sensor and an illumination sensor, the distance sensors and the illumination sensors are connected with a data acquisition card, the data acquisition card is connected with an FPGA module, and the FPGA module is electrically connected with an illuminating lamp and a camera; the FPGA module is equipped with the wifi module, the wifi module is connected with the wifi converter through wireless communication's mode, the wifi converter is connected with the host computer through serial port communication's mode, turns into digital signal transmission to the host computer through the wifi converter with the image video, the infrared data of gathering, illumination data.
6. The deep learning-based multi-angle face recognition device according to claim 5, wherein the upper computer is connected with a digital resistor module, the digital resistor module is electrically connected with a switch module, and the switch module controls an illuminating lamp to be turned on and off; the host computer is connected with measuring module, measuring module is through the infrared signal of receiving, and the position of perception people's face in three-dimensional space is fixed a position the people face, and measuring module is connected with control module, and measuring module transmits the digital signal of the three-dimensional space data at people place to control module, and control module transmits digital signal to execution module, the first motor of execution module electric connection (11) and second motor (17), through the turned angle of execution module send instruction control first motor (11) and second motor (17), change face identification module's angle of raising upward and control the angle.
7. The deep learning-based multi-angle face recognition device according to claim 5, wherein the upper computer is connected with a server, a face is recognized in the server, the server is provided with a registration module and a recognition module, and the registration module comprises face detection, face alignment, feature extraction and classifier training; the recognition module comprises face detection, face alignment, feature extraction and face recognition; before face recognition, face registration is required, and multi-angle information of the face is uploaded to a server to be stored.
8. The device for multi-angle face recognition based on deep learning of claim 7, wherein the recognition method of the face recognition device comprises:
step one, a recognized face enters a region of a face recognition device (9), the face is firstly positioned in a three-dimensional space through a distance sensor, and positioning information is transmitted to an upper computer for processing;
after the upper computer analyzes the three-dimensional spatial information of the face through the measurement module, the digital instruction of the three-dimensional spatial information of the face is transmitted to the control module and the execution module, and the execution module controls the first motor (11) and the second motor (17) to rotate to adjust the angles of the three face recognition devices (9) so as to collect images or videos;
and step three, transmitting the acquired image information of different angles to an upper computer and a server, carrying out key point detection, face alignment and feature extraction on the face through an identification module, then carrying out classifier training comparison identification, judging whether the face leaves an identification area, carrying out image acquisition again if the face leaves the identification area, carrying out feature fusion through the images of three angles if the face does not leave the identification area, and outputting an identification result.
CN202210306806.8A 2022-03-25 2022-03-25 Multi-angle face recognition equipment based on deep learning Withdrawn CN114710620A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210306806.8A CN114710620A (en) 2022-03-25 2022-03-25 Multi-angle face recognition equipment based on deep learning

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210306806.8A CN114710620A (en) 2022-03-25 2022-03-25 Multi-angle face recognition equipment based on deep learning

Publications (1)

Publication Number Publication Date
CN114710620A true CN114710620A (en) 2022-07-05

Family

ID=82171623

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210306806.8A Withdrawn CN114710620A (en) 2022-03-25 2022-03-25 Multi-angle face recognition equipment based on deep learning

Country Status (1)

Country Link
CN (1) CN114710620A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115191955A (en) * 2022-09-15 2022-10-18 中国医学科学院皮肤病医院(中国医学科学院皮肤病研究所) Artificial intelligence-based full-angle detection device and detection method for face skin state

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115191955A (en) * 2022-09-15 2022-10-18 中国医学科学院皮肤病医院(中国医学科学院皮肤病研究所) Artificial intelligence-based full-angle detection device and detection method for face skin state

Similar Documents

Publication Publication Date Title
CN101499129B (en) Distant range iris recognition system and method
CN107660039A (en) A kind of lamp control system for identifying dynamic gesture
WO2007064726A2 (en) Position determining apparatus and related method
CN114710620A (en) Multi-angle face recognition equipment based on deep learning
US20030049027A1 (en) Automatic framing camera
CN110811564A (en) Intelligent face image and tongue image acquisition terminal system and face image and tongue image acquisition method
CN212679100U (en) Posture detection system, posture detection device, and table kit
CN102541085A (en) Spherical-bottom device and method for tracking video target and controlling posture of video target
CN208077195U (en) A kind of face recognition device with managing caller function
CN103892790A (en) Manual and automatic intelligent digital synoptophore
CN112417193B (en) Method and system for searching and identifying field cordyceps sinensis
CN110781871A (en) Iris recognition device and recognition method
CN216697347U (en) Face vein collection device and equipment
CN111210192A (en) Intelligent attendance system based on multi-mode feature analysis
CN211124069U (en) Iris recognition device
CN214618613U (en) Intelligent face recognition system
CN2508306Y (en) Vital iris picture collecting device
CN114383077A (en) Intelligent remote monitoring and physiotherapy function table lamp
CN210344844U (en) Face recognition device and electric adjusting device thereof
CN209669746U (en) A kind of automatic lifting stake with Car license recognition function for parking lot
CN206805564U (en) Iris identification device
CN207814815U (en) Display for computer face recognition system
CN216645556U (en) Intelligent body type tester
CN210742961U (en) Anti-cheating iris recognition system for driving test room
CN118067619B (en) Bridge detecting system based on intelligent detecting equipment

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
WW01 Invention patent application withdrawn after publication

Application publication date: 20220705

WW01 Invention patent application withdrawn after publication