A low-friction continuous multi-angle automatic face shooting system

Abstract

The application discloses a continuous multi-angle automatic face photographing system with low friction and shock resistance, which comprises a face calibration module, a continuous photographing module and a core processing module. The continuous photographing module comprises an arc-shaped convex groove hanging type three-wheel sliding system and a multi-angle continuous automatic face photographing algorithm. The arc-shaped convex groove hanging type three-wheel sliding system is composed of an arc-shaped convex groove track and a hanging type three-wheel track running mechanism. The track running mechanism with a depth camera is controlled to move on the convex groove track through the multi-angle continuous automatic face photographing algorithm, and continuous automatic photographing of the face of a patient is realized. The face calibration module controls a calibration track running mechanism according to a depth camera picture and a face calibration algorithm, so that the face of the patient is always kept in the center of the picture. The core processing module provides computing power and memory support for correct implementation of algorithms in the equipment, and displays continuous photographing results.

Description

TECHNICAL FIELD

[0001] The application relates to a low-friction shockproof system with a multi-angle continuous automatic face shooting function. BACKGROUND

[0002] Obtaining intuitive disease information by observing the face of a patient is a common inquiry method for hospitals, especially in the fields of ophthalmology, otolaryngology, dermatology and traditional Chinese medicine. With the development of artificial intelligence and electronic computer technology and the deepening of medical informatization, the demand for patient face image collection is increasing in clinical practice. At present, the main human face image collection equipment is a non-rail camera device, which has the following three forms: 1. Single camera fixed collection: a variety of support structures are used to fix the patient's head position, and the patient needs to change the head posture according to the doctor's instruction to shoot different angle face images; 2. Multi-camera collection: the patient does not move, and different angle face images of the patient are shot by installing multi-angle depth cameras. 3. Face tracking cloud platform. The first method needs the doctor to guide the patient to move the head posture, which is time-consuming and laborious, and the patient may feel uncomfortable in the fixed state. The second method has high equipment cost, large size, and is cumbersome, and the operation of the equipment is complicated, and the same patient image collected by different cameras needs to be spliced to form a complete face image. The third method is generally used in the field of entertainment for long-distance shooting, and such equipment can only shoot the front, cannot automatically shoot the side of the person, and often obtains a single image, which cannot be continuously shot.

[0003] The track sliding camera shooting can realize continuous image collection, but there is non-negligible friction and vibration in the sliding process, which affects the quality of the image, and the image used for disease diagnosis often requires high clarity and accuracy. SUMMARY

[0004] The technical problem to be solved by the present application is that the current human face image collection equipment is mainly collected by a non-rail camera device, the single camera fixed collection system needs the doctor to guide the patient to move the head posture, the shooting efficiency is low, and the patient's head fixed experience is poor; the multi-camera collection system has high equipment cost, large size and is cumbersome, the operation of the equipment is complicated, and a complete face image can be formed only by image splicing technology; the face tracking cloud platform is generally used for long-distance front shooting, cannot automatically shoot the side of the person, and often obtains a single image, which cannot be continuously shot; the track sliding camera shooting can realize continuous image collection, but there is non-negligible friction and vibration, which affects the quality of the image, and the image used for disease diagnosis often requires high clarity and accuracy.

[0005] In order to solve the above technical problems, the technical scheme of the present application is to provide a shockproof low-friction continuous multi-angle automatic face shooting system, characterized in that it comprises an arc-shaped convex groove track and a hanging three-wheel track running mechanism hung on the arc-shaped convex groove track, wherein:

[0006] The hanging three-wheel track running mechanism is hung on the arc-shaped convex groove track through a three-wheel hanging structure, and the hanging three-wheel track running mechanism is in a parallel hanging state with the arc-shaped convex groove track. The hanging three-wheel track running mechanism comprises a driving structure, a bearing structure and a driven structure arranged in sequence from top to bottom.

[0007] The driving structure is embedded with the arc-shaped convex groove track through a driving V-shaped pulley driven by a driving motor.

[0008] The bearing structure is used for fixing a depth camera and serving as a relay structure connecting the driving structure and the driven structure.

[0009] The driven structure comprises two driven V-shaped pulleys arranged at an angle. The two driven V-shaped pulleys are located directly below the driving V-shaped pulley and jointly constitute the three-wheel hanging structure with the driving V-shaped pulley. At the same time, the two driven V-shaped pulleys are connected together at an angle to adapt to the curvature of the arc-shaped convex groove track.

[0010] The depth camera and the driving motor are connected with a core processing module. A multi-angle continuous automatic face shooting algorithm is run on the core processing module. The algorithm comprises a control part, a shooting part and an angle calculation part. The control part is used for sending control signals and driving the driving motor. The shooting part is used for reading pictures taken by the depth camera in real time and continuously shooting according to current angle information. The angle calculation part estimates the shooting angle by reading the face nose tip key point coordinates and depth data in the pictures taken by the depth camera in real time, using a depth distance calculation method based on coordinate system transformation, so as to obtain the angle and position of the depth camera.

[0011] Preferably, the main body cross section of the arc-shaped convex groove track is a double-sided flat hexagon. The arc-shaped convex groove track is formed by rotating the main body cross section at a certain inner diameter and outer diameter.

[0012] Preferably, the data and power lines of the depth camera are connected to the system through a wire inlet on one side of the bearing structure.

[0013] Preferably, the driven structure is assembled with the bearing structure through a connecting screw rod.

[0014] Preferably, the main pulley and the driven pulley have the same structure, and adopt an asymmetric inner-outer embedding structure, including an inner embedding circle and an outer embedding circle with a larger radius than the inner embedding circle, and the small end of the inner embedding circle is connected with the small end of the outer embedding circle.

[0015] Preferably, the depth distance calculation method comprises the following steps:

[0016] Supposing that the line between the depth camera and the nose tip of the face is perpendicular to the tangent of the arc-shaped convex groove track, the nose tip key point coordinate is (x k ,y k ), the depth is D k , and the user is located at the center of the arc-shaped convex groove track with a radius of R;

[0017] A two-dimensional coordinate system is established with the depth camera as the origin, and the line between the initial position of the depth camera and the nose tip of the face is taken as the x-axis, so that the nose tip position is represented as (0, D k ) at this time;

[0018] When the hanging type three-wheel track running mechanism moves, the nose tip key point coordinate changes in the coordinate system, and the calculation method of the new coordinate (x′ k ,y′ k ) is as shown in the following formula:

[0019]

[0020] Wherein, R(θ) is a rotation matrix like , D′ k is the nose tip depth at the new coordinate, and the angle θ estimation formula is as shown in the following formula:

[0021]

[0022] According to the comparison between the real-time calculated angle estimation information and the set angle: when the angles are inconsistent, the control part is triggered to send a control signal to drive the stepping motor to move the hanging type three-wheel track running mechanism on the arc-shaped convex groove slide rail; and when the angles are consistent, the shooting part is triggered to perform a shooting action to save the current picture.

[0023] Preferably, the arc-shaped convex groove track is arranged on the calibration track running mechanism, and the arc-shaped convex groove track is driven by the calibration track running mechanism to move up and down on the calibration track; the calibration track running mechanism is connected with a face calibration module, the face calibration module controls the calibration track running mechanism according to the picture shot by the depth camera combined with a face calibration algorithm, so that the face of the person being shot is always kept in the center of the picture shot by the depth camera, and the face calibration algorithm realizes the calibration function based on the face key point information.

[0024] The application designs an arc convex groove hanging type three-wheel sliding system with continuous automatic face shooting function, which adopts an arc convex groove as a sliding track of a depth camera to reduce friction, simultaneously adopts a hanging type three-wheel running mechanism driven by a single wheel to realize stable movement of the depth camera on the convex groove track, and realizes continuous automatic shooting of the face of a patient from different angles through a multi-view shooting algorithm of a built-in controller.

[0025] Compared with the prior art, the application has the following innovations:

[0026] 1. The arc convex groove hanging type three-wheel sliding system for continuous automatic face shooting. The structure adopts a semi-arc convex groove as a sliding track of a depth camera to reduce friction, simultaneously adopts a hanging type three-wheel running mechanism driven by a single wheel to realize stable movement of the depth camera on the convex groove track. The convex groove side structure is a double-sided flat hexagon to reduce friction between a V-shaped pulley and the convex groove track while ensuring stable movement of the depth camera, and the asymmetric V-shaped pulley structure reduces the contact area between the pulley embedded circle and the track, thereby further reducing the friction when the pulley slides. The three-wheel track running mechanism is in parallel hanging structure with the arc convex groove track, the driving V-shaped pulley is embedded in the upper part of the semi-arc convex groove, and the two driven V-shaped pulleys are embedded in the lower part of the semi-arc convex groove at a certain connection angle, which can ensure the stability of the center of gravity of the track running mechanism and prevent vibration caused by the slight unevenness of the track surface from affecting the quality of the shooting image. The sliding system is the key structure for realizing the continuous automatic face shooting algorithm, and supports flexible adjustment of the radian and radius to adapt to different actual applications and business scenarios.

[0027] 2. A low-latency, high-precision, scalable multi-angle continuous automatic face shooting algorithm. The continuous automatic face shooting algorithm includes three parts: the control part is used to send control signals and drive the step motor installed on the track running mechanism; the shooting part reads the depth camera picture in real time and continuously shoots according to the current angle information; the angle calculation part reads the face nose key point coordinates and depth data in the picture in real time, estimates the shooting angle by using a depth distance calculation method based on coordinate system transformation, and obtains the angle and position estimation of the camera. The algorithm has the characteristics of low-latency control, high-precision shooting, and shooting strategy expansion according to actual needs. BRIEF DESCRIPTION OF DRAWINGS

[0028] Figure 1Illustrates the bilateral flat-top hexagonal cross section;

[0029] Figures 2a to 2c is a three-view drawing of the arc convex groove track, wherein, Figure 2a is a top view, Figure 2b is a front view, Figure 2c is a left view;

[0030] Figure 3 is a structure diagram of the V-shaped asymmetric three-pulley track running mechanism;

[0031] Figure 4 Illustrates the asymmetric V-shaped pulley;

[0032] Figure 5 is a sectional view of the embedded manner;

[0033] Figure 6 is a top view of the track running mechanism;

[0034] Figure 7 is a continuous shooting algorithm framework diagram;

[0035] Figure 8 is a continuous multi-angle automatic face shooting system equipment appearance structure with anti-vibration and low friction

[0036] Figures 9a to 9d Illustrates the performance verification experiment, wherein: Figure 9a is the initial position of the track running mechanism, and the perspective direction is a top view; Figure 9b 、 Figure 9c is the second and third positions of the track running mechanism in the simulation face continuous shooting experiment; Figure 9d is the final reset of the track running mechanism to the initial position;

[0037] Figure 10 Illustrates the multi-angle continuous automatic face shooting process. DETAILED DESCRIPTION

[0038] The application will be further described below in conjunction with specific embodiments. It should be understood that these embodiments are only used to illustrate the application and not used to limit the scope of the application. In addition, it should be understood that after reading the content taught by the application, those skilled in the art can make various modifications or changes to the application, and these equivalent forms also fall within the scope defined by the appended claims of the application.

[0039] The embodiment of the application discloses a continuous multi-angle automatic face shooting system with anti-vibration and low friction, which comprises:

[0040] I. Arc convex groove hanging type three-wheel sliding system

[0041] The arc-shaped groove hook-and-loop three-wheel sliding system consists of an arc-shaped groove track and a hook-and-loop three-wheel track running mechanism.

[0042] 1.1 Arc-shaped groove track

[0043] The main body of the arc-shaped groove track has a cross-section of a double-sided flat-topped hexagon. Figure 1 As shown, an arc-shaped groove track, formed by rotating this cross-section around a certain inner and outer diameter, provides a support track for the sliding of the depth camera. Compared to a track with an elliptical cross-section, the flat-top structure on the upper and lower sides of this arc-shaped groove track reduces friction between the track and the pulley while enhancing the stability of the pulley running on the groove track. The overall three-view diagram of the arc-shaped groove track is shown below. Figures 2a to 2c As shown. Furthermore, the regular polygonal structure reduces the design difficulty of the track running mechanism to some extent.

[0044] 1.2 Attached Three-Wheeled Rail Running Mechanism

[0045] The attached three-wheeled track running mechanism is a vertical three-pulley structure, consisting of three parts from top to bottom: an active structure, a load-bearing structure, and a driven structure. Figure 3 As shown. The upper section is the active structure, which engages with the arc-shaped groove track via pulleys and is controlled by a drive motor to move the track running mechanism on the arc-shaped groove track. The middle section is the support structure, mainly used to fix the depth camera. There is a cable inlet 1 on one side of the support structure, through which the data and power cables of the depth camera are connected. The lower section is the driven structure, mainly composed of two V-shaped driven pulleys arranged at a certain angle. Both driven pulleys are directly below the V-shaped active pulleys, ensuring the correct engagement and stability of the track running mechanism during operation. At the top of the driven structure is a connecting screw 2, the length of which is adjustable. The support structure and the driven structure are assembled together through the connecting screw 2. There are openings on both sides of the connecting screw 2, where auxiliary screws are installed to limit the movement of the driven structure. By adjusting the length of the connecting screw 2, the track running mechanism is correctly engaged on the arc-shaped groove track. The entire three-wheeled track running mechanism has two key mechanical structural designs: the V-shaped asymmetrical pulley structure and the three-wheeled engagement structure.

[0046] 1.2.1 V-shaped asymmetric pulley structure

[0047] The pulleys of the attached three-wheeled track running mechanism adopt a V-shaped asymmetrical pulley structure, such as... Figure 4As shown in the figure, it is suitable for both the driving pulley and the driven pulley. The pulley adopts an asymmetric inner-outer embedded structure, the radius of the inner embedded circle 4 is smaller than that of the outer embedded circle 5, reducing the contact area when the inner embedded circle 4 contacts with the arc convex groove track, thereby reducing the friction when the pulley slides. In addition, after the pulley is fixed to the track running mechanism body through the mounting hole 3, the three-wheeled track running mechanism and the arc convex groove track are in a parallel hanging state. Due to the asymmetric structure of the pulley shaft and the large radius of the outer embedded circle 5, the center of gravity of the pulley can be offset to the center of the arc convex groove track, which can reduce the friction of the inner embedded circle while ensuring the stability of the center of gravity of the track running mechanism, preventing vibration caused by the slight unevenness of the track surface from affecting the quality of the captured image.

[0048] 1.2.2 Three-wheeled hanging structure

[0049] The cross-sectional view of the hanging three-wheeled track running mechanism is shown in Figure 5 The V-shaped driving pulley 6 is embedded in the top 7 of the arc convex groove track, and the two V-shaped driven pulleys 8 are installed at the bottom of the arc convex groove track 10 and are embedded in the arc convex groove track 10 in a hanging manner. The driving pulley is driven by the control motor to drive the two driven pulleys at the bottom to move on the arc convex groove track 10. Since the driven pulley does not generate gravitational friction on the arc convex groove track 10, and the number of sliding pulleys on the upper part of the arc convex groove track 10 is less than that of the driven pulleys on the lower part, this structure design further reduces the sliding friction of the pulley. In addition, according to the size of the arc convex groove track 10, the two driven pulleys installed at the bottom of the arc convex groove track 10 are connected together at a certain angle, as shown in Figure 6 to adapt to the track curvature and avoid shaking and deformation when the pulley moves along the arc convex groove track 10 due to the mismatch between the moving angle and the track curvature, improving the overall motion stability.

[0050] II. Multi-angle continuous automatic face shooting algorithm

[0051] The framework of the multi-angle continuous automatic face shooting algorithm is shown in Figure 7 The algorithm consists of three parts: the control part is used to send control signals and drive the step motor installed on the track running mechanism; the shooting part reads the depth camera picture in real time and performs continuous shooting according to the current angle information; the angle calculation part estimates the shooting angle by reading the face nose key point coordinates and depth data in the picture in real time, using a depth distance calculation method based on coordinate system transformation, to obtain the angle and position estimation of the camera.

[0052] The principle of the depth distance calculation method based on coordinate system transformation to estimate the shooting angle is as follows:

[0053] Assuming that the line connecting the depth camera and the face nose is perpendicular to the tangent of the arc convex groove track, and the face nose key point coordinates are (x, y), the depth data is z, and the camera is located at the origin of the coordinate system, the depth distance calculation method based on coordinate system transformation is as follows: k,y k ), with a depth of D k The user sits at the center of the arc-shaped convex track with radius R. A two-dimensional coordinate system is established with the facial camera as the origin. The line connecting the initial position of the depth camera and the tip of the nose is taken as the x-axis. Then, the position of the nose tip can be represented as (0, D). k When the track-running mechanism moves, the coordinates of the key point of the nose tip on the face will change in this coordinate system, and the new coordinates (x′) will change. k ,y′ k The calculation method for ) is shown in equation (1):

[0054]

[0055] Where R(θ) is of the form... The rotation matrix, D′ k Given the nasal tip depth at the new coordinates, the angle θ estimation formula is shown in equation (2):

[0056]

[0057] The angle estimation information calculated in real time according to formula (2) is compared with the set angle. When the angles are inconsistent, the control part is triggered to send a control signal to drive the stepper motor to make the track running mechanism move on the arc-shaped convex groove slide rail. When the angles are consistent, the shooting part is triggered to perform the shooting action to save the current picture.

[0058] In practical applications, the arc of the scanning track design can vary from 0 to 360 degrees, and the radius can also be changed. The continuous shooting algorithm can set different shooting angles according to different business needs, thereby forming different shooting strategies to adapt to different application scenarios.

[0059] This invention relates to a vibration-resistant, low-friction, continuous multi-angle automatic facial imaging system, the appearance and structure of which are as follows: Figure 8 As shown in the figure: 10: Arc-shaped groove track; 11: Hanging three-wheel track running mechanism with a depth camera embedded in the middle; 12: Calibration track; 13: Calibration track running mechanism; 14: Equipment base with an embedded microprocessor installed in it, and the interactive screen is embedded on the surface of the base facing the doctor.

[0060] To achieve the full functionality of the system, the facial calibration module controls the track running mechanism based on the depth camera image and facial calibration algorithm, thereby keeping the patient's face centered in the image. The facial calibration module consists of a calibration track and its running mechanism in hardware. A stepper motor at the bottom of the calibration track controls its vertical movement on the track, and the calibration function is achieved based on facial key point information using the facial calibration algorithm.

[0061] The core processing module consists of an embedded microprocessor, an interactive screen, and a device base with a power supply. The embedded microprocessor provides computing power for the algorithm implementation in the two modules mentioned above, and provides memory support for runtime caching and data storage. The interactive screen provides a visual interaction method for the doctor-oriented interactive software.

[0062] To verify the accuracy and stability of the proposed dynamic single-drive arc-shaped convex groove sliding system, a simulated continuous facial imaging experiment was designed to verify the system's performance. The experimental procedure is as follows: Figures 9a to 9d As shown.

[0063] At the start of the experiment, the track-running mechanism was positioned at the initial position of the arc-shaped groove track, with a nearly smooth cylinder placed at the center of the track. After the experiment began, the track-running mechanism first moved 30° to the left to reach the second position, then moved 60° to the right to reach the third position, and finally moved 30° to the left to return to the initial position to complete the simulated continuous facial imaging experiment and verify the stability and accuracy of the system during operation.

[0064] In the simulated continuous facial scanning experiment, the depth camera provides real-time feedback of the detection distance d from the cylinder to the image sensor. n , and the actual distance d nr The ratio can be converted into a percentage to obtain the stability index of the shooting system. The calculation method is shown in equation (3):

[0065]

[0066] Furthermore, assuming that the theoretical distance traveled by the track-operated mechanism to complete one full facial image capture is d. c d c The distance calculated based on the radius of the curved track and the specified shooting angle is d; while the actual moving distance of the track running mechanism is d. cr d cr It can be measured using a flexible measuring tape. Therefore, the accuracy ρ of the imaging system is defined as...

[0067]

[0068] The results obtained from 20 simulated continuous facial imaging experiments are shown in Table 1.

[0069] Table 1 Performance verification experiment results

[0070]

[0071]

[0072] From Table 1, the average stability of the dynamic single-drive V-shaped convex groove sliding system reaches 96.25%, and the accuracy reaches 96.92%, indicating that in the process of simulating continuous face shooting, the detection distance of the track running mechanism and the target object is relatively small compared with the actual distance, and the track running mechanism has good stability; the theoretical running distance of the track running mechanism is relatively small compared with the actual distance, indicating that the stepping motor basically does not have the abnormal situation of missing steps during operation, and has good shooting accuracy.

[0073] The implementation process of the multi-angle continuous automatic face shooting is divided into three steps: face calibration, continuous shooting and result presentation, and the detailed process is as shown in Figure 10

[0074] 1. Face calibration

[0075] Step 1: The device process is started, and the processor reads the real-time picture of the current depth camera.

[0076] Step 2: Detect the face key points in the picture and analyze the nose tip position key point coordinates in real time.

[0077] Step 3: Calculate the deviation of the nose tip position coordinates from the preset position, and generate a control signal.

[0078] Step 4: Convert the control signal into a motor driving signal to drive the calibration track running mechanism to move to reduce the deviation until the nose tip position is located at the center of the picture.

[0079] 2. Automatic shooting

[0080] Step 1: Read and analyze the face picture of the depth camera to obtain the face nose tip key point coordinates and depth information;

[0081] Step 2: Estimate the angle according to formula 2 by combining the new key point coordinates and depth information under the current angle.

[0082] Step 3: When the angle is inconsistent with the specified shooting angle, trigger the control part to send a control signal to drive the stepping motor to move the track running mechanism on the arc convex groove sliding rail; when the angle is consistent, trigger the shooting part to execute the shooting action and save the current picture.

[0083] Step 4: Determine whether the shooting is finished, if not, jump to Step 1 and continue to execute.

[0084] 3. Result presentation

[0085] Step 1: Read the continuous automatic face shooting data results saved locally and output and display.

[0086] Step 2: Present on the interactive screen in a visual form via the doctor interactive software, and complete the entire continuous automatic face shooting process.​

Claims

1. A shockproof low-friction continuous multi-angle automatic face shooting system, characterized in that, The application relates to a kind of arc-shaped convex slot tracks and hanging three-wheel track running mechanism, wherein: The main body cross section of the arc-shaped convex slot track is a double-sided flat-top hexagon; The hanging three-wheel track running mechanism is hung on the arc-shaped convex slot track by a three-wheel hanging structure, and the hanging three-wheel track running mechanism is in parallel with the arc-shaped convex slot track; The hanging three-wheel track running mechanism comprises a driving structure, a bearing structure and a driven structure arranged from top to bottom; The driving structure is embedded with the arc-shaped convex slot track by a driving V-shaped pulley driven by a driving motor; The bearing structure is used to fix a depth camera and as a relay structure connecting the driving structure and the driven structure; The driven structure comprises two driven V-shaped pulleys arranged at an angle, and the two driven V-shaped pulleys are located directly below the driving V-shaped pulley and jointly form the three-wheel hanging structure with the driving V-shaped pulley; Meanwhile, the two driven V-shaped pulleys are connected at an angle to adapt to the curvature of the arc-shaped convex slot track; The driving V-shaped pulley and the driven V-shaped pulley have the same structure and adopt an asymmetric inner-outer embedding structure, comprising an inner embedding circle and an outer embedding circle with a larger radius than the inner embedding circle, and the small end of the inner embedding circle is connected with the small end of the outer embedding circle; The depth camera and the driving motor are connected with a core processing module, and a multi-angle continuous automatic face shooting algorithm is run on the core processing module; The multi-angle continuous automatic face shooting algorithm comprises a control part, a shooting part and an angle calculation part: the control part is used to send control signals and drive the driving motor; The shooting part is used to read the picture taken by the depth camera in real time, and continuously shoot according to the current angle information; The angle calculation part estimates the shooting angle by reading the face nose key point coordinates and depth data in the picture taken by the depth camera in real time, and estimates the angle and position of the depth camera by using a depth distance calculation method based on coordinate system transformation.

2. A shockproof, low-friction, continuous multi-angle automatic face photographing system according to claim 1, characterized in that, The arc-shaped convex slot track is formed by rotating the main body cross section with a certain inner diameter and outer diameter.

3. A shockproof, low-friction, continuous multi-angle automatic face photographing system according to claim 1, characterized in that, The data and power line of the depth camera are connected to the system through a wire inlet on one side of the bearing structure.

4. A shock-absorbing, low-friction, continuous multi-angle automatic face photographing system according to claim 1, characterized in that, The driven structure is assembled with the bearing structure through a connecting screw rod.

5. A shock-absorbing, low-friction, continuous multi-angle automatic face photographing system according to claim 1, characterized in that, The application further comprises a calibration track running mechanism, and the arc-shaped convex slot track is arranged on the calibration track running mechanism and moves up and down along the calibration track driven by the calibration track running mechanism; The calibration track running mechanism is connected with a face calibration module, and the face calibration module controls the calibration track running mechanism according to the picture taken by the depth camera combined with a face calibration algorithm, so that the face of the person being shot is always kept in the center of the picture taken by the depth camera, and the face calibration algorithm realizes the calibration function based on face key point information.

Images