Face recognition device in high dark light environment and recognition method thereof
By combining lidar and infrared tracking cameras in the face recognition device, the height of lidar is measured and adjusted to extract facial features, and the problem of recognition failure in high-dark light environments is solved, achieving a higher recognition success rate.
Patent Information
- Application Number
- CN202411804294.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-10
- Publication Date
- 2025-03-14
AI Technical Summary
In high dark light environments, it is difficult for face recognition devices to effectively recognize facial features of people with higher heights, resulting in failed recognition.
A device combining lidar and infrared tracking camera is used to measure the height of the human body through lidar and adjust the height, so that the lidar approaches the human head, emits laser pulses to the human body at any time, and extracts facial features depth information at different angles.
The success rate of face recognition is significantly improved in high dark light environments, especially when the lidar and the human face are at the same height, the facial features obtained are the most complete and detailed, and the recognition success rate is the highest.
Smart Images
Figure CN119617251A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of face recognition technology, and in particular to a face recognition device and a recognition method thereof in a high-light environment. Background Art
[0002] Face recognition is a biometric technology that identifies people based on their facial features. It uses a camera or camcorder to collect images or video streams containing faces, automatically detects and tracks faces in images, and then performs facial recognition on the detected faces. It is also commonly called portrait recognition or facial recognition.
[0003] The application of depth cameras and deep learning algorithms in facial recognition is the key to improving the accuracy and efficiency of facial recognition.
[0004] Deep learning algorithms, especially convolutional neural networks (CNNs), have made significant progress in the field of facial recognition. By training on large-scale datasets, CNNs can learn features with good discrimination. These feature vectors can effectively represent facial images, thereby improving the accuracy of facial recognition. For example, deep learning models such as VGG, ResNet, and FaceNet have been widely used in face recognition tasks. Depth cameras, as a new type of input device, can capture richer facial feature information, including three-dimensional spatial information, which is of great significance for improving the accuracy and robustness of facial recognition. Depth cameras provide more data dimensions for facial recognition by capturing the three-dimensional structural information of objects, allowing the algorithm to more accurately identify and understand facial features. The application of this technology is not limited to traditional facial recognition, but also includes face liveness detection, expression recognition and many other aspects.
[0005] Combining depth cameras and deep learning algorithms can achieve more advanced facial recognition capabilities. For example, by processing the data obtained from the depth camera with a deep learning algorithm, high-precision facial feature extraction and recognition can be achieved. This technology not only improves the accuracy of facial recognition, but also improves the efficiency and user-friendliness of the system. In addition, the research on optimization and acceleration methods of deep learning algorithms further improves the real-time performance of face recognition systems, allowing this technology to be better applied in practical scenarios.
[0006] Depth cameras are mainly divided into three types: structured light, binocular vision and TOF (time of flight). Structured light technology projects controllable light spots, light strips or light surface structures onto the surface of the object being measured, and the image sensor obtains the image, and calculates the three-dimensional coordinates of the face through the system geometry and triangulation principles. Binocular vision technology simulates the principle of binocular vision imaging by installing two cameras, and uses two cameras to extract information including three-dimensional position for depth perception. TOF (time of flight) technology calculates the distance of an object by measuring the flight time of light. It usually uses a lidar to emit multiple laser pulses to illuminate the object, and then measures the time required for the reflected pulses to calculate the distance, thereby establishing a three-dimensional facial model. Among the related technologies, structured light and TOF (time of flight) are active ranging, and binocular vision is passive ranging. The low-light recognition capability of active ranging is relatively stronger; however, in high-light environments, the active ranging method will still have a certain impact on the accuracy of face recognition due to insufficient light, shadows, and reflections. In addition, if there is a tilt angle when recognizing the face, it will also affect the recognition result. The reason is that the shape and position of the facial features captured at different angles will be different, which in turn affects the extraction and recognition of facial features. Specifically, when the height of the owner of the face is higher than the camera, people only need to lower their eyelids and look down to easily look at the camera, but at this time the face is not facing the camera, and the camera shoots at an upward angle, which is easy to affect the result of facial recognition. If coupled with a high-light environment, facial recognition failure is likely to occur; and when the height of the owner of the face is lower than the camera, because people are used to staring at the camera, in order to easily look at the camera with both eyes, this will prompt people to look up at the camera, otherwise they need to lift their eyelids, and lifting their eyelids is equivalent to actively opening their eyes, which will cause discomfort to the eyes. In this case, the face is almost facing the camera, and the angle difference when the camera shoots the face is relatively small. In order to increase the probability of successful facial recognition, some facial recognition devices have a common countermeasure of setting an auxiliary light source. When a human body approaches, the light source automatically turns on and fills in the light. The disadvantage of this method is that the light of the auxiliary light source is strong, which will cause discomfort to the human eye and affect the actual experience of the facial recognition device. Summary of the invention
[0007] The present invention aims to provide a face recognition device in a high-light environment to improve the problem that face recognition failure is likely to occur when the height of the human body is high in the high-light environment.
[0008] The present invention is achieved through the following technical solutions: A face recognition device for a high-dark-light environment comprises a box, a side wall of the box is provided with a notch in the vertical direction, a vertical rod is vertically arranged inside the box, a lifting platform is slidably arranged on the vertical rod along the length direction of the vertical rod, a laser radar is also arranged on the lifting platform, and the scanning end of the laser radar is directed toward the outside of the box through the notch; a lifting mechanism is arranged inside the box, and the lifting mechanism is used to drive the lifting platform to rise and fall.
[0009] Furthermore, a horizontal gimbal is provided on the lifting platform, the laser radar is located on the horizontal gimbal, and an infrared tracking camera is also provided on the horizontal gimbal next to the laser radar. The infrared tracking camera is electrically connected to the horizontal gimbal, and the lens of the infrared tracking camera is facing the outside of the box through the notch, and the infrared tracking camera and the laser radar are in the same direction, and the infrared tracking camera and the laser radar can rotate in the notch with the horizontal gimbal.
[0010] Furthermore, the outer cover of the laser radar and the infrared tracking camera is provided with a glass cover, the glass cover is located on the horizontal pan-tilt platform, and the glass cover can rotate in the notch along with the horizontal pan-tilt platform; The lifting mechanism includes a lifting motor and a screw rod, the screw rod is vertically and rotatably arranged inside the box, one end of the screw rod passes through the top wall of the box and is coaxially connected to the output shaft of the lifting motor, the screw rod passes through the lifting platform and is threadedly connected to the lifting platform, and the thread of the screw rod is right-handed; A dust removal mechanism is provided on the box body at the lower part of the notch, and the dust removal mechanism includes a first spur gear, an intermediate rod, a second spur gear, a transmission rod, a third spur gear, a first bevel gear, a second bevel gear, a rotating shaft horizontally and rotatably arranged outside the box body, and a plurality of brushes; a cylindrical section is provided on the top of the screw rod, the first spur gear is sleeved on the cylindrical section of the screw rod, the intermediate rod is vertically and rotatably arranged inside the box body, the second spur gear is sleeved on the top of the intermediate rod, and the first spur gear is meshed with the second spur gear; the transmission rod is vertically and rotatably arranged outside the box body, the third spur gear is sleeved on the top of the transmission rod, and the second spur gear is meshed with the third spur gear; the first bevel gear is sleeved on the lower part of the transmission rod, the second bevel gear is sleeved on one end of the rotating shaft, and the first bevel gear is meshed with the second bevel gear, and a plurality of brushes are arranged on the rotating shaft at intervals along the circumferential direction of the rotating shaft, and the plurality of brushes can clean a side wall of the glass cover facing the outside of the box body as the rotating shaft rotates.
[0011] Furthermore, a dust collection mechanism is provided inside the box body, and the dust collection mechanism includes a dust collecting bin, an exhaust fan and a dust filter bag; the dust collecting bin is located inside the notch below the horizontal pan head, one end of the dust collecting bin extends to the outside of the box body, an opening is provided on the upper wall of the dust collecting bin, and a side wall of the glass cover facing the outside of the box body is located directly above the upper opening of the dust collecting bin; an arc-shaped cover is provided on the upper cover of one side of the dust collecting bin located outside the box body, the rotating shaft passes through the arc-shaped cover, and the plurality of brushes are all located inside the arc-shaped cover, and the bending direction of the arc-shaped inner wall of the arc-shaped cover corresponds to the rotation trajectory of the brush; The exhaust fan is arranged inside the box body, an exhaust pipe is arranged between the air inlet of the exhaust fan and the arc-shaped cover, and the dust filter bag is sleeved on the air outlet of the exhaust fan; The lifting platform is provided with an extension plate extending downward, the extension plate is provided with a horizontal plate, the horizontal pan head is located on the horizontal plate, the glass cover can be lowered to the inside of the dust collecting bin, when any of the brushes cleans the glass cover, one end of the arc-shaped outer wall of the arc-shaped cover is connected to the dust collecting bin, and the other end is close to the side wall of the glass cover facing the outside of the box.
[0012] Further, the arc-shaped inner wall of the arc-shaped cover is provided with a plurality of dust removal plates at intervals along the bending direction of the arc-shaped inner wall, one end of each of the plurality of dust removal plates is connected to the arc-shaped inner wall of the arc-shaped cover, and the other end of each of the plurality of dust removal plates points to the rotating shaft, and the plate surfaces of each of the plurality of dust removal plates are parallel to the rotating shaft; a plurality of passage grooves are provided on each of the plurality of dust removal plates at intervals along the length direction of the rotating shaft, and the plurality of brushes can pass through the passage grooves of all dust removal plates when rotating; Two adjacent dust removal plates along the rotation track direction of the brush are combined with the corresponding wall surface of the arc-shaped cover to form a dust collecting chamber, and the side wall of each dust collecting chamber is connected to one of the exhaust pipes.
[0013] The present invention also provides a recognition method of the above-mentioned face recognition device, comprising the following steps: S1. Regularly lower the height of the laser radar through the lifting mechanism, clean the dust on the side wall surface of the glass cover facing the outside of the box through the dust removal mechanism, then collect and process the cleaned dust through the dust suction mechanism, and drive the laser radar to rise through the lifting mechanism to wait for face recognition; S2. When a human body appears within the shooting range of the infrared tracking camera, the infrared tracking camera and the horizontal gimbal are used to track and shoot the human body, so that the human body is always located in the middle of the shooting picture, so that the laser radar can quickly face the direction of the human body; S3, obtaining the body data of the human body through laser radar measurement, and then adjusting the height of the laser radar through the lifting mechanism so that the laser radar is close to the head of the human body. The laser radar scans the human body while moving to obtain the facial features of the human body; S4. When the facial features obtained by scanning match the facial features obtained in advance, the lifting mechanism stops running and the face recognition is successful.
[0014] Furthermore, in step S3, the body data includes the overall three-dimensional modeling data, and the current body data is compared with the pre-acquired body data, combined with the comparison of facial features, and then the face recognition result is determined and output: When the requirements for face recognition are high, if the facial features match but the body data does not match, face recognition fails. If both the facial features and the body data match, face recognition succeeds.
[0015] Furthermore, in steps S3 and S4, when the facial features obtained by the laser radar approaching the human head do not match the pre-acquired facial features, the laser radar is controlled to rise and fall reciprocally through the lifting mechanism, so that the laser radar scans the human face up and down multiple times to obtain depth information of the face at different angles. At the same time, the extracted facial features are compared with the pre-acquired facial features at any time. When face recognition is successful, the lifting mechanism stops running and the laser radar stops scanning.
[0016] Compared with the prior art, the present invention has the following advantages and beneficial effects: In the present invention, in a high-dark-light environment, when a tall person uses a face recognition device, the height of the person is measured by a laser radar, and then the lifting platform and the laser radar thereon are driven up by a lifting mechanism to make the laser radar close to the person's head. The laser radar emits laser pulses to the person at any time during the movement to extract depth information of facial features at different angles in real time. The closer the laser radar is to the person's head, the more complete the facial features obtained are, and the easier it is to successfully identify. In particular, when the laser radar and the person's face are at the same height, the facial features obtained are the most complete and detailed, and have the highest similarity to the pre-acquired facial features, and are easiest to successfully identify. Since the laser radar can perform facial recognition in a dark-light environment and obtain the depth information of the face, adjusting the scanning angle of the laser radar is conducive to improving the problem of facial recognition failure that is prone to occur when the height is higher than the laser radar in a high-dark-light environment. BRIEF DESCRIPTION OF THE DRAWINGS
[0017] The drawings described herein are used to provide a further understanding of the embodiments of the present invention, constitute a part of this application, and do not constitute a limitation of the embodiments of the present invention.
[0018] In the attached picture: Figure 1 It is a structural schematic diagram of the present invention; Figure 2 It is a partial cross-sectional view of the present invention; Figure 3 For the present invention Figure 2 Enlarged view of part A in the middle.
[0019] Marks and corresponding parts names in the attached drawings: 1. Box body; 2. Notch; 3. Vertical rod; 4. Lifting platform; 5. Laser radar; 6. Horizontal pan-tilt head; 7. Infrared tracking camera; 8. Glass cover; 9. Lifting motor; 10. Screw; 11. First spur gear; 12. Middle rod; 13. Second spur gear; 14. Transmission rod; 15. Third spur gear; 16. First bevel gear; 17. Second bevel gear; 18. Rotating shaft; 19. Brush; 20. Bottom box; 21. Top box; 22. Dust collecting bin; 23. Exhaust fan; 24. Dust filter bag; 25. Arc cover; 26. Extension plate; 27. Horizontal plate; 28. Dust removal plate; 29. Passageway; 30. Exhaust duct. DETAILED DESCRIPTION
[0020] In order to make the purpose, technical scheme and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the embodiments and drawings. The schematic implementation modes and descriptions of the present invention are only used to explain the present invention and are not intended to limit the present invention. It should be noted that the present invention is already in the actual development and use stage. Example 1
[0021] A face recognition device for a high-dark light environment comprises a box body 1, a notch 2 is formed on one side wall of the box body 1 in the vertical direction, a vertical rod 3 is vertically arranged inside the box body 1, a lifting platform 4 is slidingly arranged on the vertical rod 3 along the length direction of the vertical rod 3, a laser radar 5 is also arranged on the lifting platform 4, and a scanning end of the laser radar 5 is directed toward the outside of the box body 1 through the notch 2; a lifting mechanism is arranged inside the box body 1, and the lifting mechanism is used to drive the lifting platform 4 to rise and fall.
[0022] In this solution, in a high-light environment, when a tall person uses the face recognition device, the height of the person is measured by the laser radar 5, and then the lifting platform 4 and the laser radar 5 thereon are driven up by the lifting mechanism to make the laser radar 5 close to the head of the person. The laser radar 5 emits laser pulses to the person at any time during the movement to extract the depth information of facial features at different angles in real time. The closer the laser radar 5 is to the head of the person, the more complete the facial features obtained are, and the easier it is to successfully identify. In particular, when the laser radar 5 is at the same height as the face of the person, the facial features obtained are the most complete and detailed, and have the highest similarity to the pre-acquired facial features, and are easiest to successfully identify. Since the laser radar 5 can perform facial recognition in a low-light environment and obtain the depth information of the face, by adjusting the scanning angle of the laser radar 5, it is helpful to improve the problem of facial recognition failure that is prone to occur when the height is higher than the laser radar 5 in a high-light environment.
[0023] Furthermore, a horizontal platform 6 is provided on the lifting platform 4, and the laser radar 5 is located on the horizontal platform 6. An infrared tracking camera 7 is also provided on the horizontal platform 6 beside the laser radar 5. The infrared tracking camera 7 is electrically connected to the horizontal platform 6. The lens of the infrared tracking camera 7 faces the outside of the box 1 through the notch 2, and the infrared tracking camera 7 and the laser radar 5 are in the same direction. The infrared tracking camera 7 and the laser radar 5 can rotate in the notch 2 with the horizontal platform 6. When a human body approaches the box 1 and enters the shooting range of the infrared tracking camera 7, the tracking camera function of the horizontal platform 6 and the infrared tracking camera 7 will be triggered, and the infrared tracking camera 7 can make the image of the human body always located in the middle of the shooting picture during the tracking camera process. Since the scanning direction of the laser radar 5 is consistent with the shooting direction of the infrared tracking camera 7, it is convenient for the scanning end of the laser radar 5 to be quickly adjusted to the direction facing the human body, so that the angle of the laser radar 5 when scanning the facial features of the human body is more suitable, so as to improve the success rate of facial recognition.
[0024] Furthermore, the outer cover of the laser radar 5 and the infrared tracking camera 7 is provided with a glass cover 8, and the glass cover 8 is located on the horizontal pan-tilt platform 6, and the glass cover 8 can rotate with the horizontal pan-tilt platform 6 in the gap 2; the lifting mechanism includes a lifting motor 9 and a screw rod 10, the screw rod 10 is vertically and rotatably arranged inside the box body 1, one end of the screw rod 10 passes through the top wall of the box body 1 and is coaxially connected to the output shaft of the lifting motor 9, and the screw rod 10 passes through the lifting platform 4 and is threadedly connected to the lifting platform 4; a dust removal mechanism is arranged on the box body 1 at the lower part of the gap 2, and the dust removal mechanism includes a first spur gear 11, an intermediate rod 12, a second spur gear 13, a transmission rod 14, a third spur gear 15, a first bevel gear 16, a second bevel gear 17, a rotating shaft 18 horizontally and rotatably arranged outside the box body 1, and a plurality of brushes 19, and the rotating shaft 18 is close to the lower part of the box body 1; A cylindrical section is arranged at the top of the screw rod 10, the first spur gear 11 is sleeved on the cylindrical section of the screw rod 10, the intermediate rod 12 is vertically and rotatably arranged inside the box body 1, the second spur gear 13 is sleeved on the top of the intermediate rod 12, and the first spur gear 11 is meshed with the second spur gear 13; a bottom box 20 and a top box 21 are arranged outside the box body 1, the bottom box 20 is located at the lower part of the side wall of the box body 1 where the notch 2 is located, and the top box 21 is located at the upper part of the side wall of the box body 1 where the notch 2 is located, the top end of the transmission rod 14 passes through the bottom wall of the top box 21 and extends to the inside of the top box 21 and is rotatably connected to the box wall of the top box 21, and the bottom end of the transmission rod 14 passes through the top wall of the bottom box 20 and extends to the inside of the bottom box 20 and is rotatably connected to the box wall of the bottom box 20 The transmission rod 14 is vertically arranged, the third spur gear 15 is sleeved on the top of the transmission rod 14, and the second spur gear 13 is meshed with the third spur gear 15, and the second spur gear 13 and the third spur gear 15 are protected by the box body 1 and the top box 21; the first bevel gear 16 is sleeved on the lower part of the transmission rod 14, the second bevel gear 17 is sleeved on one end of the rotating shaft 18, and the first bevel gear 16 is meshed with the second bevel gear 17, and the first bevel gear 16 and the second bevel gear 17 are both located inside the bottom box 20; a plurality of brushes 19 are arranged on the rotating shaft 18 at intervals along the circumference of the rotating shaft 18, and the plurality of brushes 19 can clean the side wall of the glass cover 8 facing the outside of the box body 1 as the rotating shaft 18 rotates. One end of the rotating shaft 18 passes through the bottom box 20 and extends to the inside of the bottom box 20 and is rotatably connected to the box wall of the bottom box 20.
[0025] Among them, the glass cover 8 can be made of some glass materials with high light transmittance, such as ultra-white glass, which is beneficial to reduce the energy attenuation caused by the laser pulse emitted by the laser radar 5 passing through the glass cover 8.
[0026] The screw rod 10 preferably adopts a screw rod 10 with right-handed threads. In other embodiments, the screw rod 10 may also adopt a left-handed thread. Relatively speaking, the dust removal effect of the screw rod 10 with right-handed threads is better. By processing the dust on the outer wall of the glass cover 8, it is helpful to reduce the influence of dust on the transmission and reception of the laser radar 5, which is beneficial to improve the accuracy of the facial information extracted by the laser radar 5, and thus improve the success rate of facial recognition.
[0027] Here are the reasons: The conventional position of the laser radar 5 is set above the dust removal mechanism, so that the dust removal mechanism will not affect its conventional facial recognition work; if a left-handed screw 10 is used, when performing dust removal work, the screw 10 is driven by the lifting motor 9 to rotate clockwise (from a bird's-eye view). Under the guidance of the vertical rod 3, the screw 10 rotates to drive the lifting platform 4 and the horizontal pan / tilt 6, laser radar 5, infrared tracking camera 7 and glass cover 8 thereon to descend. At this time, the screw 10 drives the second spur gear 13 and the intermediate rod 12 to rotate counterclockwise through the first spur gear 11, and the second spur gear 13 drives the third spur gear 15 and the transmission rod 14 to rotate clockwise. The transmission rod 14 rotates through the first bevel gear 1 6 drives the second bevel gear 17 and the rotating shaft 18 to rotate, and the rotating shaft 18 drives the brush 19 to rotate, and the plurality of brushes 19 clean the corresponding outer wall of the glass cover 8 from top to bottom (from top to bottom here refers to the cleaning direction of the brush 19 when it rotates, rather than the brush 19 cleaning from the top of the glass cover 8); that is, when the glass cover 8 moves downward, the brush 19 also cleans downward, that is, the moving direction of the glass cover 8 is the same as the cleaning direction of the brush 19, the lower part of the glass cover 8 is cleaned first, and when the glass cover 8 continues to move downward, the dust cleaned subsequently is driven downward by the brush 19, which easily causes the dust to adhere to the cleaned part below the glass cover 8 again, thereby causing secondary pollution.
[0028] However, if a right-handed screw rod 10 is used, when the glass cover 8 moves downward, the brush 19 will clean upward, and the dust cleaned subsequently will be carried upward by the brush 19, which is relatively less likely to cause the dust to move downward and adhere to the cleaned part below the glass cover 8, that is, the moving direction of the glass cover 8 is opposite to the cleaning direction of the brush 19; when the glass cover 8 moves upward, the brush 19 will clean downward, and the part of the glass cover 8 that is cleaned for the second time will move upward, and the dust cleaned for the second time will be driven downward by the brush 19, which is less likely to cause secondary contamination of the glass cover 8.
[0029] Furthermore, a dust collection mechanism is provided on the box body 1, and the dust collection mechanism includes a dust collecting bin 22, an exhaust fan 23 and a dust filter bag 24; the dust collecting bin 22 is located inside the notch 2 below the horizontal pan head 6, one end of the dust collecting bin 22 extends to the outside of the box body 1, an opening is provided on the upper wall of the dust collecting bin 22, and a side wall of the glass cover 8 facing the outside of the box body 1 is located directly above the upper opening of the dust collecting bin 22; the dust collecting bin 22 is located on one side of the outside of the box body 1, and an arc-shaped cover 25 is provided on the upper cover, and an opening is provided below the arc-shaped cover 25, the rotating shaft 18 passes through the arc-shaped cover 25 and is rotatably connected to the side wall of the arc-shaped cover 25, and multiple brushes 19 are all located inside the arc-shaped cover 25, and the bending direction of the arc-shaped outer wall of the arc-shaped cover 25 corresponds to the rotation trajectory of the brush 19 ; The exhaust fan 23 is arranged inside the box body 1, and an exhaust pipe 30 is arranged between the air inlet of the exhaust fan 23 and the arc-shaped cover 25, and the dust filter bag 24 is sleeved on the air outlet of the exhaust fan 23; an extension plate 26 extending downward is arranged on the lifting platform 4, and a horizontal plate 27 is arranged on the extension plate 26, and the horizontal pan-tilt platform 6 is located on the horizontal plate 27; when the lifting platform 4 moves downward, the glass cover 8 can be lowered into the dust receiving bin 22, and the horizontal plate 27, the horizontal pan-tilt platform 6 and the glass cover 8 extend downward into the dust receiving bin 22 through the extension plate 26. When any brush 19 cleans the glass cover 8, one end of the arc-shaped outer wall of the arc-shaped cover 25 is connected to the dust receiving bin 22, and the other end is close to the side wall of the glass cover 8 facing the outside of the box body 1.
[0030] Most of the dust generated when the dust removal mechanism cleans the glass cover 8 is confined to the dust receiving bin 22 and the arc cover 25, and the raised dust can be promptly extracted by the exhaust fan 23 and stored in the dust filter bag 24, which can prevent the dust from moving inside the dust receiving bin 22 and the arc cover 25 with the multiple brushes 19 to a certain extent, reducing the probability of dust adhering to the outer wall of the glass cover 8 again, thereby improving the dust removal effect on the glass cover 8.
[0031] Furthermore, the arc-shaped inner wall of the arc-shaped cover 25 is provided with a plurality of dust removal plates 28 at intervals along the bending direction of the arc-shaped inner wall, one end of each of the dust removal plates 28 is connected to the arc-shaped inner wall of the arc-shaped cover 25, and the other end points to the rotating shaft 18, and the plate surfaces of each of the dust removal plates 28 are parallel to the rotating shaft 18; a plurality of passage grooves 29 are provided on each of the dust removal plates 28 at intervals along the length direction of the rotating shaft 18, and a plurality of brushes 19 can pass through the passage grooves 29 of all dust removal plates 28 when rotating; two adjacent dust removal plates 28 along the rotating trajectory direction of the brush 19 are enclosed with the corresponding wall surface of the arc-shaped cover 25 to form a dust collecting chamber, and the side wall of each dust collecting chamber is connected to an exhaust pipe 30. Figure 3 In the figure, there are two dust removal plates 28 that appear to be suspended only to express the positions and postures of the multiple dust removal plates 28 inside the arc cover 25.
[0032] In this solution, when the bristles of each brush 19 pass through multiple passage grooves 29, the bristles contact the inner walls of the passage grooves 29, which makes it easy to remove the dust adhering to the bristles, so that the dust enters the dust collecting chamber corresponding to the arc cover 25, and the dust floating in the air inside each dust collecting chamber can be extracted in time through the exhaust fan 23 and the exhaust pipe 30, and most of the dust can be extracted and stored inside the dust filter bag 24.
[0033] The present invention also provides a recognition method of the above-mentioned face recognition device, comprising the following steps: S1. Regularly lower the height of the laser radar 5 and the glass cover 8 through the lifting mechanism, clean the dust on the side wall surface of the glass cover 8 facing the outside of the box 1 through the dust removal mechanism, then collect and process the cleaned dust through the dust suction mechanism, and drive the laser radar 5 to rise through the lifting mechanism to wait for face recognition; The lifting mechanism and the dust removal mechanism work together, and a lifting motor 9 drives the laser radar 5 to lift and lower, and the lifting motor 9 drives the rotating shaft 18 and the brush 19 to clean the glass cover 8, thereby reducing the number of motors used, which is beneficial to reducing the manufacturing cost of the face recognition device and the power consumption during the use of the face recognition device; in addition, there is another linkage mechanism between the lifting mechanism and the dust removal mechanism, that is, when the glass cover 8 descends, the brush 19 cleans the glass cover 8 upward, and when the glass cover 8 rises, the brush 19 cleans the glass cover 8 downward, which can reduce the secondary pollution of the glass cover 8 caused by dust. By regularly cleaning the outer wall of the glass cover 8 outside the laser radar 5 and reducing the dust adhering to the outer wall of the glass cover 8, the glass cover 8 can be kept in good working condition, which is beneficial to improving its success rate in recognizing faces; S2. When a human body appears within the shooting range of the infrared tracking camera 7, the infrared tracking camera 7 and the horizontal gimbal 6 are used to track and shoot the human body, so that the human body is always located in the middle of the shooting screen, and the laser radar 5 can quickly face the direction of the human body; this is beneficial for the laser radar 5 to scan at an angle facing the human body's face.
[0034] S3, obtaining the body data of the human body through the laser radar 5, and then adjusting the height of the laser radar 5 through the lifting mechanism so that the laser radar 5 is close to the head of the human body. The laser radar 5 scans the human body while moving to obtain the facial features of the human body; S4. When the facial features obtained by scanning match the facial features obtained in advance, the lifting mechanism stops running and the face recognition is successful.
[0035] Furthermore, the body data includes the three-dimensional modeling data of the whole human body. The current body data is compared with the pre-acquired body data, combined with the comparison of facial features, and then the face recognition result is determined and output: In cases where the requirements for face recognition are high, if the facial features match but the body data does not match, face recognition fails; in the above case, face recognition is passed only when both the facial features and the body data match.
[0036] Furthermore, in steps S3 and S4, when the facial features obtained by the laser radar 5 in the process of approaching the human head do not match the facial features obtained in advance, the laser radar 5 is controlled to rise and fall reciprocally through the lifting mechanism, so that the laser radar 5 scans the human face up and down multiple times to obtain depth information of the face at different angles. At the same time, the extracted facial features are compared with the pre-acquired facial features at any time. When face recognition is successful, the lifting mechanism stops running and the laser radar 5 stops scanning.
[0037] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.
Claims
1. A face recognition device in a high-light environment, characterized by: The invention comprises a box (1), a notch (2) is provided on one side wall of the box (1) in the vertical direction, a vertical rod (3) is vertically arranged inside the box (1), a lifting platform (4) is slidably arranged on the vertical rod (3) along the length direction of the vertical rod (3), a laser radar (5) is also arranged on the lifting platform (4), and the scanning end of the laser radar (5) is directed to the outside of the box (1) through the notch (2); a lifting mechanism is arranged inside the box (1), and the lifting mechanism is used to drive the lifting platform (4) to rise and fall.
2. The face recognition device in a high-light and low-light environment according to claim 1, characterized in that: The lifting platform (4) is provided with a horizontal pan-tilt platform (6), the laser radar (5) is located on the horizontal pan-tilt platform (6), and an infrared tracking camera (7) is also provided on the horizontal pan-tilt platform (6) next to the laser radar (5). The infrared tracking camera (7) is electrically connected to the horizontal pan-tilt platform (6), and the lens of the infrared tracking camera (7) is directed toward the outside of the box (1) through the notch (2). The infrared tracking camera (7) and the laser radar (5) are directed in the same direction, and the infrared tracking camera (7) and the laser radar (5) can rotate in the notch (2) along with the horizontal pan-tilt platform (6).
3. The face recognition device in a high-light and low-light environment according to claim 2, characterized in that: The outer cover of the laser radar (5) and the infrared tracking camera (7) is provided with a glass cover (8), and the glass cover (8) is located on the horizontal pan-tilt platform (6), and the glass cover (8) can rotate in the notch (2) along with the horizontal pan-tilt platform (6); The lifting mechanism comprises a lifting motor (9) and a screw rod (10), wherein the screw rod (10) is vertically and rotatably arranged inside the box (1), one end of the screw rod (10) passes through the top wall of the box (1) and is coaxially connected to the output shaft of the lifting motor (9), and the screw rod (10) passes through the lifting platform (4) and is threadedly connected to the lifting platform (4), and the thread of the screw rod (10) is right-handed; A dust removal mechanism is arranged on the box body (1) at the lower part of the notch (2), and the dust removal mechanism comprises a first spur gear (11), an intermediate rod (12), a second spur gear (13), a transmission rod (14), a third spur gear (15), a first bevel gear (16), a second bevel gear (17), a rotating shaft (18) horizontally and rotatably arranged outside the box body (1), and a plurality of brushes (19); a cylindrical section is arranged on the top of the screw rod (10), the first spur gear (11) is sleeved on the cylindrical section of the screw rod (10), the intermediate rod (12) is vertically and rotatably arranged inside the box body (1), the second spur gear (13) is sleeved on the top of the intermediate rod (12), and the first spur gear (11) and the second bevel gear (17) are sleeved on the top of the intermediate rod (12). The transmission rod (14) is vertically and rotatably arranged outside the housing (1), the third spur gear (15) is sleeved on the top of the transmission rod (14), and the second spur gear (13) is meshed with the third spur gear (15); the first bevel gear (16) is sleeved on the lower part of the transmission rod (14), the second bevel gear (17) is sleeved on one end of the rotating shaft (18), and the first bevel gear (16) is meshed with the second bevel gear (17); a plurality of brushes (19) are arranged on the rotating shaft (18) at intervals along the circumference of the rotating shaft (18), and the plurality of brushes (19) can rotate with the rotating shaft (18) to clean a side wall of the glass cover (8) facing the outside of the housing (1).
4. The face recognition device in a high-light and low-light environment according to claim 3, characterized in that: A dust collecting mechanism is arranged inside the box body (1), and the dust collecting mechanism comprises a dust collecting bin (22), an exhaust fan (23) and a dust filter bag (24); the dust collecting bin (22) is located inside the notch (2) below the horizontal platform (6), one end of the dust collecting bin (22) extends to the outside of the box body (1), an upper wall of the dust collecting bin (22) is provided with an opening, and a side wall of the glass cover (8) facing the outside of the box body (1) is located directly above the upper opening of the dust collecting bin (22); an arc-shaped cover (25) is arranged above one side of the dust collecting bin (22) located outside the box body (1), the rotating shaft (18) passes through the arc-shaped cover (25), and the plurality of brushes (19) are all located inside the arc-shaped cover (25), and the bending direction of the arc-shaped inner wall of the arc-shaped cover (25) corresponds to the rotation trajectory of the brushes (19); The exhaust fan (23) is arranged inside the box (1), an exhaust pipe (30) is arranged between the air inlet of the exhaust fan (23) and the arc-shaped cover (25), and the dust filter bag (24) is sleeved on the air outlet of the exhaust fan (23); The lifting platform (4) is provided with an extension plate (26) extending downwards, the extension plate (26) is provided with a horizontal plate (27), the horizontal platform (6) is located on the horizontal plate (27), the glass cover (8) can be lowered into the dust receiving bin (22), when any of the brushes (19) cleans the glass cover (8), one end of the arc-shaped outer wall of the arc-shaped cover (25) is connected to the dust receiving bin (22), and the other end is close to the side wall of the glass cover (8) facing the outside of the box body (1).
5. The face recognition device in a high-light and low-light environment according to claim 4, characterized in that: The arc-shaped inner wall of the arc-shaped cover (25) is provided with a plurality of dust removal plates (28) at intervals along the curvature direction of the arc-shaped inner wall, one end of each of the dust removal plates (28) is connected to the arc-shaped inner wall of the arc-shaped cover (25), and the other end of each of the dust removal plates (28) points to the rotating shaft (18), and the plate surfaces of each of the dust removal plates (28) are parallel to the rotating shaft (18); a plurality of passage grooves (29) are provided on each of the dust removal plates (28) at intervals along the length direction of the rotating shaft (18), and the plurality of brushes (19) can pass through the passage grooves (29) of all the dust removal plates (28) when rotating; Two adjacent dust removal plates (28) along the rotation track direction of the brush (19) are combined with the corresponding wall surface of the arc-shaped cover (25) to form a dust collecting chamber, and the side wall of each dust collecting chamber is connected to an exhaust pipe (30).
6. A recognition method for a face recognition device, characterized in that: The identification method is applied to any one of claims 1 to 5, comprising the following steps: S1. Regularly lower the height of the laser radar through the lifting mechanism, clean the dust on the side wall surface of the glass cover facing the outside of the box through the dust removal mechanism, then collect and process the cleaned dust through the dust suction mechanism, and drive the laser radar to rise through the lifting mechanism to wait for face recognition; S2. When a human body appears within the shooting range of the infrared tracking camera, the infrared tracking camera and the horizontal gimbal are used to track and shoot the human body, so that the human body is always located in the middle of the shooting picture, so that the laser radar can quickly face the direction of the human body; S3, obtaining the body data of the human body through laser radar measurement, and then adjusting the height of the laser radar through the lifting mechanism so that the laser radar is close to the head of the human body. The laser radar scans the human body while moving to obtain the facial features of the human body; S4. When the facial features obtained by scanning match the facial features obtained in advance, the lifting mechanism stops running and the face recognition is successful.
7. The face recognition method according to claim 6, characterized in that: In step S3, the body data includes the overall three-dimensional modeling data, and the current body data is compared with the pre-acquired body data, combined with the comparison of facial features, and then the face recognition result is determined and output: When the requirements for face recognition are high, if the facial features match but the body data does not match, face recognition fails. If both the facial features and the body data match, face recognition succeeds.
8. The face recognition method according to claim 7, characterized in that: In steps S3 and S4, when the facial features obtained by the laser radar approaching the human head do not match the pre-acquired facial features, the laser radar is controlled to rise and fall through the lifting mechanism, so that the laser radar scans the human face up and down multiple times to obtain depth information of the face at different angles. At the same time, the extracted facial features are compared with the pre-acquired facial features at any time. When face recognition is successful, the lifting mechanism stops running and the laser radar stops scanning.