A human eye positioning and forehead support lifting linkage control system and method

Abstract

The application discloses a human eye positioning and forehead support lifting linkage control system and method. The system comprises: an infrared sensing module for acquiring an induction signal transmitted by an infrared distance sensor; a binocular vision positioning module for periodically acquiring human face information and human eye position information after acquiring an object signal transmitted by the infrared distance sensor; a double condition and gate confirmation module for periodically generating a lifting control start signal according to the object signal transmitted by the infrared distance sensor and the human face information; and a lifting control module for periodically generating a lifting signal with a preset threshold as a fixed control period until a positioning in position condition is met. The system of the application can solve the technical problems of low positioning accuracy, poor linkage and no feedback in the prior art, realize rapid capture of human eye position and accurate and real-time following of the forehead support, and thus improve the efficiency and accuracy of medical detection.

Description

Technical Field

[0001] This application relates to the field of human eye positioning technology, specifically to a human eye positioning and forehead support lifting linkage control system and method. Background Technology

[0002] In the fields of ophthalmology and optometry, such as fundus photography, optical coherence tomography (OCT), and refraction, accurately and stably positioning the subject's head, especially the eyes, within the instrument's working area is the primary prerequisite for ensuring the accuracy of test results and operational safety.

[0003] Currently, traditional head immobilization devices typically employ manually adjustable forehead and chin supports. During use, the subject rests their chin and forehead against the supports, and the operator manually rotates the adjustment knob based on experience, while simultaneously observing the subject's eye position until the eyes are roughly aligned with the instrument's baseline. However, this traditional approach has the following significant technical drawbacks:

[0004] Low positioning accuracy and inefficiency: Relying on the operator's visual experience and manual fine-tuning is not only time-consuming, but also makes it difficult for the human eye to accurately and repeatedly position the device at the same height, which easily introduces human error.

[0005] Poor linkage: There is no correlation between the raising and lowering of the forehead support and the position of the human eye. The operator needs to go through a back-and-forth process of adjustment, observation and readjustment, and it is impossible to achieve real-time and automatic following.

[0006] Lack of feedback mechanism: The system is an open-loop control system, which cannot know the actual position of the forehead support or whether the human eye has reached the ideal focus area. It cannot make real-time corrections and confirmations to the adjustment process, making it difficult to meet the precision requirements of modern high-end medical equipment. Summary of the Invention

[0007] The purpose of this invention is to provide a human eye positioning and forehead support lifting linkage control system to at least solve one of the above-mentioned technical problems.

[0008] One aspect of the present invention provides a human eye positioning and forehead support lifting linkage control system, the human eye positioning and forehead support lifting linkage control system comprising:

[0009] An infrared sensing module is used to acquire the sensing signal transmitted by an infrared ranging sensor.

[0010] A binocular vision positioning module is used to periodically acquire face information and eye position information after acquiring the object signal transmitted by the infrared ranging sensor.

[0011] A dual-condition AND gate confirmation module is used to periodically generate a lifting control start signal based on the object signal and face information transmitted by the infrared ranging sensor.

[0012] The lifting control module is used to generate lifting signals cyclically according to the lifting control start signal with a preset threshold as the fixed control cycle until the positioning condition is met.

[0013] Optionally, the binocular vision positioning module includes:

[0014] The initial phase difference calculation module is used to acquire the original RGB facial image synchronously acquired by the binocular camera, and generate left and right grayscale images and continuous phase difference distribution based on the original RGB facial image synchronously acquired by the binocular camera.

[0015] A coarse face region localization module is used to generate coarse face region localization information based on left and right grayscale images and continuous phase difference distribution. The coarse face region localization information includes the left sub-image of the face, the right sub-image of the face, and the coordinates of the bounding rectangle of the face region.

[0016] An eye feature region precise positioning module is used to generate eye feature region positioning information based on the coarse positioning information of the face region. The eye feature region positioning information includes a left eye region sub-image, a right eye region sub-image, and the pixel coordinates of the left and right eyes in the left image.

[0017] A key point extraction module is used to generate key point pixel coordinate information based on eye feature region positioning information.

[0018] The three-dimensional coordinate calculation module is used to generate the three-dimensional coordinates of the eye center and the vertical pupil height deviation based on the key point pixel coordinate information and the continuous phase difference distribution.

[0019] Optionally, the dual-condition AND gate confirmation module includes:

[0020] The synchronization module is used to synchronize the sensing signal transmitted by the infrared ranging sensor and the key point pixel coordinate information, so as to obtain a set of effective input signals after synchronization.

[0021] The first condition judgment module is used to generate first condition judgment information based on the set of valid input signals after synchronization.

[0022] The second condition judgment module is used to generate second condition judgment information based on the key point pixel coordinate information and the physiological geometric ratio constant of the human eye.

[0023] The final determination module is used to generate a lifting control start signal based on the first condition judgment information and the second condition judgment information.

[0024] Optionally, the lifting control module includes:

[0025] An initialization module is used to load the initial information of the lifting mechanism;

[0026] The periodic cycle control module is used to control the lifting mechanism according to the lifting control start signal and the human eye position information, so that the lifting mechanism can reach the preset conditions.

[0027] Optionally, the periodic cycle control module includes:

[0028] The pupil height deviation segmentation and target step length mapping module is used to generate the target step length and the motor running direction indicator for the current cycle based on the human eye position information for the current cycle.

[0029] A single-cycle motor drive pulse count calculation module is used to generate the number of drive pulses in the current cycle based on the target step size of the current cycle.

[0030] A pulse control signal sending module is used to generate a square wave pulse signal according to the number of drive pulses in the current cycle, and send the square wave pulse signal and the motor running direction indicator in the current cycle to the stepper motor.

[0031] A real-time location tracking module is used to obtain the actual distance traveled in the current period.

[0032] Optionally, the periodic cycle control module further includes:

[0033] The positioning condition judgment module is used to determine whether the position is in place based on the position information of the human eye.

[0034] Optionally, the full cost prediction model is the LightGBM full cost prediction model;

[0035] The initial phase difference calculation module includes:

[0036] The original image alignment module is used to align the timestamps of the original images acquired by the binocular camera to obtain an aligned image group.

[0037] A grayscale conversion module is used to perform grayscale conversion on the aligned image group to obtain a grayscale image group;

[0038] A frequency domain matrix acquisition module is used to perform Fourier transform on the grayscale image group to obtain the frequency domain matrix of the left image and the frequency domain matrix of the right image.

[0039] A phase difference matrix calculation module is used to generate a phase difference matrix between the left and right images based on the frequency domain matrix of the left image and the frequency domain matrix of the right image.

[0040] A phase difference distribution acquisition module is used to perform phase unwrapping on the phase difference matrix of the left and right images to obtain a continuous phase difference distribution.

[0041] Optionally, the face region coarse localization module includes:

[0042] The gradient value calculation module is used to calculate the gradient values ​​of continuous phase differences distributed in the horizontal and vertical directions, thereby obtaining the phase difference gradient magnitude of each pixel. The phase difference gradient magnitudes of each pixel form a gradient magnitude image.

[0043] The binarization module is used to perform binarization processing on the gradient magnitude image to obtain a binarized image.

[0044] A foreground connected component acquisition module is used to perform connected component analysis on a binarized image to obtain various foreground connected components.

[0045] A face candidate region filtering module is used to obtain face candidate regions based on each connected component.

[0046] Optionally, the precise localization module for eye feature regions includes:

[0047] An eye candidate region acquisition module is used to acquire preset eye candidate regions;

[0048] The left sub-image acquisition module for the eye is used to extract the corresponding sub-image of the candidate eye region in the left sub-image of the face as the left sub-image of the eye;

[0049] A horizontal projection curve acquisition module is used to perform horizontal integral projection calculation on the left sub-image of the eye to obtain a horizontal projection curve.

[0050] A smoothing module is used to smooth the horizontal projection curve to obtain a smoothed horizontal projection curve.

[0051] A horizontal center position acquisition module is used to acquire the horizontal center positions of the left and right eyes based on a smoothed horizontal projection curve.

[0052] The region segmentation module is used to extract the horizontal sub-regions of the left and right eyes respectively in the left sub-image of the eyes, with the horizontal center position of the left and right eyes as the center. The horizontal sub-regions of the left and right eyes include the horizontal sub-region of the left eye and the horizontal sub-region of the right eye.

[0053] A region sub-image acquisition module is used to generate a left-eye region sub-image and a right-eye region sub-image based on the left-eye horizontal sub-region and the right-eye horizontal sub-region.

[0054] This application also provides a method for linking eye positioning with forehead support lifting, the method comprising:

[0055] Acquire the sensing signal transmitted by the infrared ranging sensor;

[0056] After acquiring the object signal transmitted by the infrared ranging sensor, facial information and eye position information are periodically acquired.

[0057] The system periodically generates a lifting control start signal based on the object signal and facial information transmitted by the infrared ranging sensor.

[0058] Based on the lifting control start signal, lifting signals are generated cyclically with a preset threshold as the fixed control cycle until the positioning condition is met.

[0059] This application provides a human eye positioning and forehead support lifting linkage control system that can solve the technical problems of low positioning accuracy, poor linkage and no feedback in the prior art, realize the rapid capture of human eye position and accurate, real-time tracking of forehead support, thereby improving the efficiency and accuracy of medical testing. Attached Figure Description

[0060] Figure 1 This is a schematic diagram of a human eye positioning and forehead support lifting linkage control system according to an embodiment of this application.

[0061] Figure 2 This is a front view structural diagram of a positioning device according to an embodiment of this application.

[0062] Figure 3 This is a side view of a positioning device according to an embodiment of this application. Detailed Implementation

[0063] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions in the embodiments of this application will be described in more detail below with reference to the accompanying drawings. In the drawings, the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The described embodiments are some, but not all, embodiments of this application. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application. The embodiments of this application will be described in detail below with reference to the accompanying drawings.

[0064] See Figure 2 as well as Figure 3 , Figure 2 as well as Figure 3 The image shown is a positioning device applicable to this application.

[0065] In this embodiment, Figure 2 as well as Figure 3 The positioning device structure shown includes:

[0066] Base: Made of cast iron, with four adjustable leveling feet at the bottom to adjust the overall level of the system and ensure the verticality of the lifting movement. The base has internal space reserved for the installation of an electrical control box to accommodate the main control unit and power supply module.

[0067] Upright column: Made of aluminum alloy profile, vertically fixed to the center of the upper surface of the base. The upright column has a rectangular cross-section to ensure bending rigidity. Two parallel T-shaped guide rail mounting slots are opened along the axial direction on the front surface of the upright column for fixing the vertical guide rails.

[0068] Vertical guide rails: Two high-precision linear ball bearing guide rails are used, which are fixed to the two T-shaped mounting slots of the column by bolts. The guide rails are 300mm long and cover the entire lifting stroke of the headrest.

[0069] Slider assembly: Contains two ball bearing sliders that match the guide rail, allowing the sliders to slide along the guide rail without backlash. The front surface of the sliders is bolted to the forehead support bracket, converting the rotational motion of the motor into the linear lifting motion of the forehead support.

[0070] Forehead support bracket: Made of L-shaped aluminum alloy sheet, the vertical section is fixed to the slider, and the horizontal section extends forward. The upper part of the vertical section of the forehead support bracket has three mounting holes for mounting the infrared range sensor and the binocular vision camera group, respectively; the front end of the horizontal section is connected to the forehead support pad by a buckle.

[0071] Forehead support: Made of medical-grade silicone, the curved design conforms to the contour of the human forehead. It can be detachably connected to the front of the horizontal section of the forehead support bracket via a buckle, making it easy to disinfect and replace.

[0072] Chin Support: Made of the same medical-grade silicone as the forehead support, it is fixed to the center of the front surface of the base via an adjustable support rod, located 150mm directly below the forehead support, forming an upper and lower aligned support structure. The support rod can be manually adjusted within ±50mm in the vertical direction to accommodate test subjects of different heights.

[0073] In this embodiment, the dual-sensor positioning module is integrated into the upper part of the vertical section of the forehead support, with both detection surfaces facing forward and opposite to the subject's face, realizing a two-level sensing mechanism of infrared triggering and binocular precise positioning. The specific structure and connection relationship are as follows:

[0074] Infrared ranging sensor: An infrared ranging module based on the TOF principle, measuring 12mm × 12mm × 5mm, is bolted to the upper center of the vertical section of the forehead support, with the detection surface flush with the front surface of the forehead support. The sensor's signal output is connected to the ADC interface of the main control unit via a shielded cable for real-time detection of the distance between the subject and the forehead support, enabling low-power wake-up and target presence detection of the system.

[0075] Binocular vision camera group: Employs two identical global shutter CMOS cameras with a resolution of 640×480 and a frame rate of 30fps. The two cameras are bolted to the left and right sides of the infrared rangefinder sensor, respectively, with a spacing of 60mm (baseline length), and their optical axes are parallel and perpendicular to the front surface of the forehead rest. The synchronization trigger terminals of the two cameras are connected to the GPIO interface of the main control unit to achieve hardware-level synchronous acquisition; the image data output terminals are connected to the image processing unit of the main control unit via a MIPI interface for acquiring facial images and extracting key eye points.

[0076] Ring-shaped fill light: Eight 850nm infrared LED beads are evenly distributed around the binocular camera array, forming a 40mm diameter ring-shaped fill light structure. The driver of the fill light is connected to the PWM interface of the main control unit and is only lit when the binocular vision module is working. It is used to improve facial image contrast in low-light environments, and the infrared light is invisible and will not irritate the eyes of the subject.

[0077] In this embodiment, the lifting execution module is installed on the top and inside of the column, providing high-precision lifting power and position feedback for the headrest, realizing full closed-loop position control. The specific structure and connection relationship are as follows:

[0078] Stepper motor: A type 42 two-phase hybrid stepper motor with a step angle of 1.8° and a holding torque of 0.4 N·m is used. It is fixed to the top of the column via a motor bracket. The motor's drive input is connected to the output of the motor drive chip in the main control unit to provide lifting power.

[0079] Harmonic reducer: A harmonic reducer with a reduction ratio of 50:1 is used. The input end is coaxially fixed to the motor output shaft through a coupling, and the output end is connected to the upper end of the ball screw through a key. It is used to reduce the motor speed and increase the output torque to ensure the smoothness of lifting and lowering movements.

[0080] Ball screw: A precision ball screw with a lead of 4mm is used, vertically installed at the center of the column, and supported at both ends by angular contact ball bearings. The screw nut is fixed to the slider assembly through a connecting plate, converting the rotational motion of the motor into the linear lifting motion of the slider and the bracket.

[0081] Absolute Encoder: A 16-bit single-turn absolute encoder is used, which is coaxially fixed to the lower end of the ball screw via a coupling. The encoder's signal output is connected to the main control unit via an SPI interface. The encoder can detect the rotation angle of the lead screw in real time and then convert it into the absolute position of the bearing, realizing full closed-loop position feedback without the need for zeroing upon each power-on.

[0082] Upper and lower limit switches: Two mechanical microswitches are used, which are fixed to the upper and lower ends of the vertical guide rail by bolts respectively. The normally closed contacts of the limit switches are connected in series in the motor drive power supply circuit. When the slider triggers the limit switch, the motor power is immediately cut off, realizing hardware-level limit protection. At the same time, the signal output terminal of the limit switch is connected to the GPIO interface of the main control unit to report the limit status to the main control unit.

[0083] The main control and power supply modules are installed in the electrical control box inside the base, providing the system with computing control capabilities and stable power supply. The specific structure and connection relationship are as follows:

[0084] Main control unit: An embedded microcontroller with an ARM Cortex-M7 core and a main frequency of 400MHz, integrating a rich set of interfaces including an image processing unit, motor drive controller, ADC, GPIO, SPI, and MIPI. The main control unit is electrically connected to the infrared ranging sensor, binocular vision camera group, ring fill light, stepper motor driver chip, absolute encoder, and upper and lower limit switches, and is responsible for executing all algorithm logic and control instructions.

[0085] Motor driver chip: A dual-channel stepper motor driver chip is adopted, with a maximum output current of 2A. It communicates with the main control unit through the SPI interface, receives pulse and direction control signals from the main control unit, and drives the stepper motor to run.

[0086] The voltage regulator module adopts an architecture combining switching power supplies and linear voltage regulators. It receives 220V AC input and outputs three DC power supplies: 12V / 5A, 5V / 3A, and 3.3V / 2A. The 12V power supply powers the stepper motor and supplementary lighting, the 5V power supply powers the main control unit and encoder, and the 3.3V power supply powers the infrared sensor and binocular camera.

[0087] Rechargeable battery: A 12V / 2000mAh lithium battery pack is used, which is connected in parallel to the 12V output of the voltage regulator module to provide emergency power when the external power supply is disconnected, so as to ensure that the system can complete the current detection process and reset safely.

[0088] Human-computer interaction interface: Includes an OLED display screen and three physical buttons, mounted on the side of the base. The display screen shows the system status, pupil height deviation value, and positioning results; the three buttons are the power button, manual up button, and manual down button, used for system power on / off and manual adjustment of the forehead rest position.

[0089] like Figure 1 The human eye positioning and forehead support lifting linkage control system shown includes:

[0090] An infrared sensing module is used to acquire the sensing signal transmitted by an infrared ranging sensor.

[0091] A binocular vision positioning module is used to periodically acquire face information and eye position information after acquiring the object signal transmitted by the infrared ranging sensor.

[0092] A dual-condition AND gate confirmation module is used to periodically generate a lifting control start signal based on the object signal and face information transmitted by the infrared ranging sensor.

[0093] The lifting control module is used to generate lifting signals cyclically according to the lifting control start signal with a preset threshold as the fixed control cycle until the positioning condition is met.

[0094] In this embodiment, in the system standby state, only the infrared sensing module samples at a low frequency of 2Hz, and all non-essential modules are powered off and put into hibernation, minimizing the system standby power consumption.

[0095] The raw distance data obtained by the infrared ranging sensor is filtered by a sliding window mean filter algorithm with a length of 5 to remove noise interference caused by sudden changes in ambient light and rapid passing of objects, and outputs a smoothed distance value.

[0096] When the distance values ​​after filtering for 5 consecutive frames all fall within the effective detection range of 80mm-300mm, and the distance fluctuation between adjacent frames is ≤10mm, it is determined that a valid human target is approaching, and a sensing signal is immediately generated; if any condition is not met, it will continue to maintain a low-power standby state.

[0097] In this embodiment, after acquiring the object signal transmitted by the infrared ranging sensor, the binocular vision positioning module simultaneously acquires RGB facial images through the binocular left and right cameras. It adopts a binocular phase difference-guided layered extraction of eye features and a three-dimensional coordinate direct mapping algorithm to achieve high-precision detection of key eye points and three-dimensional coordinate calculation. Finally, it outputs the three-dimensional coordinates of the eye center and the vertical pupil height deviation ΔH, providing core input for subsequent forehead support lifting control.

[0098] In this embodiment, the binocular vision positioning module includes:

[0099] The initial phase difference calculation module is used to acquire the original RGB facial image synchronously acquired by the binocular camera, and generate left and right grayscale images and continuous phase difference distribution based on the original RGB facial image synchronously acquired by the binocular camera.

[0100] In this embodiment, the initial phase difference calculation module includes:

[0101] The original image alignment module is used to align the timestamps of the original images acquired by the binocular camera to obtain an aligned image group.

[0102] Specifically, hardware-level timestamp alignment is performed on the original images captured by the left and right cameras in the binocular camera to ensure that the acquisition time difference between the left and right images is less than 1ms, thereby eliminating phase shift caused by motion blur.

[0103] A grayscale conversion module is used to perform grayscale conversion on the aligned image group to obtain a grayscale image group;

[0104] In this embodiment, the grayscale conversion formula is:

[0105] ;

[0106] R, G, and B represent the pixel values ​​of the red, green, and blue channels, respectively, preserving the image's brightness information while reducing computational load.

[0107] In this embodiment, the grayscale image group includes grayscale images from the left camera and grayscale images from the right camera.

[0108] A frequency domain matrix acquisition module is used to perform Fourier transform on the grayscale image group to obtain the frequency domain matrix of the left image and the frequency domain matrix of the right image.

[0109] Specifically, two-dimensional Fourier transforms are performed on the grayscale images from the left and right cameras respectively to obtain the frequency domain matrix of the left image. and the frequency domain matrix of the right image , where u and v are the frequency components in the horizontal and vertical directions, respectively.

[0110] A phase difference matrix calculation module is used to generate a phase difference matrix between the left and right images based on the frequency domain matrix of the left image and the frequency domain matrix of the right image.

[0111] Specifically, the phase difference matrix is ​​calculated using the following formula:

[0112] ;

[0113] in This represents taking the principal argument of a complex number, with a range of... .

[0114] A phase difference distribution acquisition module is used to perform phase unwrapping on the phase difference matrix of the left and right images to obtain a continuous phase difference distribution.

[0115] Specifically, the phase difference matrix is ​​unwrapped using a region-growing-based unwrapping algorithm. Starting from the region with the smallest phase gradient, the algorithm gradually expands to the entire image plane, eliminating the discontinuities caused by phase jumps and obtaining a continuous phase difference distribution. .

[0116] A coarse face region localization module is used to generate coarse face region localization information based on left and right grayscale images and continuous phase difference distribution. The coarse face region localization information includes the left sub-image of the face, the right sub-image of the face, and the coordinates of the bounding rectangle of the face region.

[0117] In this embodiment, the face region coarse localization module includes:

[0118] The gradient value calculation module is used to calculate the gradient values ​​of continuous phase differences distributed in the horizontal and vertical directions, thereby obtaining the phase difference gradient magnitude of each pixel. The phase difference gradient magnitudes of each pixel form a gradient magnitude image.

[0119] Specifically, the phase difference distribution is calculated using the following formula. The gradient values ​​in the horizontal and vertical directions are denoted as follows: and The gradient value is calculated using the following formula:

[0120] ;

[0121] ;

[0122] Calculate the phase difference gradient magnitude for each pixel:

[0123] ;

[0124] The gradient magnitude reflects the degree of change of the spatial point corresponding to the pixel in the depth direction.

[0125] The binarization module is used to perform binarization processing on the gradient magnitude image to obtain a binarized image.

[0126] Specifically, the gradient magnitude image is binarized, and a threshold is set. ,in To find the maximum value in the gradient magnitude image, select gradient magnitudes greater than [value missing]. Pixels that are less than or equal to the foreground pixel are marked as foreground pixels. The pixels marked are the background pixels.

[0127] A foreground connected component acquisition module is used to perform connected component analysis on a binarized image to obtain various foreground connected components.

[0128] Specifically, connected component analysis is performed on the binarized image, and the 8-neighborhood connectivity criterion is used to extract all foreground connected components.

[0129] A face candidate region filtering module is used to obtain face candidate regions based on each connected component.

[0130] Specifically, the area, aspect ratio of the bounding rectangle, and center coordinates of each connected region are calculated. Connected regions with an area between 10,000 and 50,000 pixels and an aspect ratio between 0.8 and 1.2 are selected and marked as candidate face regions.

[0131] If multiple face candidate regions exist, the connected component with the center coordinates closest to the image center is selected as the final face region. The corresponding sub-images of this region in the left and right grayscale images are extracted and denoted as the left face sub-image. And the right image of a human face It also stores the coordinates of the bounding rectangle of the face region.

[0132] An eye feature region precise positioning module is used to generate eye feature region positioning information based on the coarse positioning information of the face region. The eye feature region positioning information includes a left eye region sub-image, a right eye region sub-image, and the pixel coordinates of the left and right eyes in the left image.

[0133] In this embodiment, the precise positioning module for eye feature regions includes:

[0134] An eye candidate region acquisition module is used to acquire preset eye candidate regions;

[0135] Specifically, based on the geometric proportions of the face (the geometric proportions used in this application adopt the standard facial proportion rules of third and fifth division commonly used in the field of computer vision, which are inherent laws of human facial anatomy), in the left sub-image of the face... The candidate eye region is determined using the following method:

[0136] Establish a local coordinate system with the top-left corner of the bounding rectangle of the face region as the origin. The coordinates of the top-left corner of the eye candidate region are: The coordinates of the lower right corner are , where W and H are the width and height of the bounding rectangle of the face region, respectively.

[0137] The left sub-image acquisition module for the eye is used to extract the corresponding sub-image of the candidate eye region in the left sub-image of the face as the left sub-image of the eye;

[0138] Specifically, the candidate eye region is extracted from the left sub-image of the face. The corresponding sub-image in the image is denoted as the left sub-image of the eye. ;

[0139] A horizontal projection curve acquisition module is used to perform horizontal integral projection calculation on the left sub-image of the eye to obtain a horizontal projection curve.

[0140] Specifically, for the left sub-image of the eye Perform horizontal integral projection calculations to obtain the horizontal projection curve. , where x is the pixel coordinate in the horizontal direction. This is the sum of the grayscale values ​​of all pixels in this column.

[0141] A smoothing module is used to smooth the horizontal projection curve to obtain a smoothed horizontal projection curve.

[0142] Specifically, for the horizontal projection curve Smoothing is performed using a sliding window mean filter with a length of 7 to eliminate curve fluctuations caused by noise.

[0143] A horizontal center position acquisition module is used to acquire the horizontal center positions of the left and right eyes based on a smoothed horizontal projection curve.

[0144] Specifically, two local minima are obtained in the smoothed horizontal projection curve. These two points correspond to the horizontal center positions of the left and right eyes, respectively, and are denoted as... and ,in, .

[0145] The region segmentation module is used to extract the horizontal sub-regions of the left and right eyes respectively in the left sub-image of the eyes, with the horizontal center position of the left and right eyes as the center. The horizontal sub-regions of the left and right eyes include the horizontal sub-region of the left eye and the horizontal sub-region of the right eye.

[0146] Specifically, with Centered on the left sub-image of the eye Extract the width as The rectangular region whose height is equal to the full height of the left sub-image of the eye is denoted as the left eye horizontal sub-region. .

[0147] by Centered on the left sub-image of the eye Extract the width as The rectangular region whose height is the full height of the left sub-image of the eye is denoted as the horizontal sub-region of the right eye. .

[0148] A region sub-image acquisition module is used to generate a left-eye region sub-image and a right-eye region sub-image based on the left-eye horizontal sub-region and the right-eye horizontal sub-region.

[0149] Specifically, for the horizontal sub-region of the left eye Perform vertical integral projection calculations to obtain the vertical projection curve. , where y is the pixel coordinate in the vertical direction. This is the sum of the grayscale values ​​of all pixels in that row.

[0150] For vertical projection curves Perform the same sliding window mean filtering smoothing process with a length of 7 to eliminate curve fluctuations caused by noise.

[0151] Find the smoothed vertical projection curve The local minimum point in the image, which corresponds to the vertical center of the left eye, is denoted as . .

[0152] For the right eye horizontal sub-region Perform vertical integral projection calculations to obtain the vertical projection curve. , where y is the pixel coordinate in the vertical direction. This is the sum of the grayscale values ​​of all pixels in that row.

[0153] For vertical projection curves Perform the same sliding window mean filtering smoothing process with a length of 7 to eliminate curve fluctuations caused by noise.

[0154] Find the smoothed vertical projection curve The local minimum point in the image, which corresponds to the vertical center of the right eye, is denoted as . .

[0155] At this point, the complete center coordinates of the left eye in the left sub-image of the face have been obtained. And the complete center coordinates of the right eye in the left sub-image of the face. .

[0156] by Centered on the left sub-image of the face Extract a rectangular region of size 60×60 pixels, and denote it as the left eye region sub-image. .

[0157] by Centered on the left sub-image of the face Extract a rectangular region of size 60×60 pixels, and denote it as the right eye region sub-image. .

[0158] Based on the epipolar constraint relationship of the binocular camera, in the right sub-image of the face... In the middle, the sub-image of the left eye region Extract a rectangular region of size 60×60 pixels from the same horizontal row (same Y coordinate), denoted as . .

[0159] Based on the epipolar constraint relationship of the binocular camera, in the right sub-image of the face... In the middle, the sub-image of the right eye region Extract a rectangular region of size 60×60 pixels from the same horizontal row (same Y coordinate), denoted as . .

[0160] Finally, the left eye region sub-image, the right eye region sub-image, and the pixel coordinates of the left and right eyes in the left image are obtained.

[0161] In this embodiment, the 60×60 pixel area is determined based on the average size of a normal adult eye in a 640×480 resolution image. This area can fully encompass key features such as the pupil and inner and outer corners of the eyes, while not introducing excessive background noise. For other resolutions or for teenagers, adjustments can be made as needed.

[0162] A key point extraction module is used to generate key point pixel coordinate information based on eye feature region positioning information.

[0163] Specifically, the grayscale of the left eye region sub-image is inverted to obtain an inverted image, which transforms the pupil region (originally a low grayscale region) into a high grayscale region, facilitating subsequent detection.

[0164] • Perform threshold segmentation on the inverted image, setting the threshold. ,in To invert the maximum gray value in the image, select the gray value greater than... The pixels are marked as pupil candidate pixels, and a pupil binarized image is obtained.

[0165] Morphological opening operations are performed on the binarized pupil image, using a 3×3 rectangular structuring element to eliminate small noise points and isolated pixels while preserving the overall shape of the pupil.

[0166] Connectivity analysis was performed on the processed binarized pupil image. The largest connected component was extracted as the pupil region, and the centroid coordinates of this connected component were calculated. These centroid coordinates represent the pixel coordinates of the left pupil center in the left image. .

[0167] Sub-image in the left eye region In the middle, with the center of the pupil Establish a polar coordinate system with the origin as the origin, and calculate the polar angle of each pixel. And the polar radius r.

[0168] Along polar angle From 0 to In terms of direction, find the two local minima of the polar radius r, which correspond to the inner and outer corners of the left eye, respectively. The specific method is: for each polar angle... Calculate the rate of change of gray values ​​in that direction, and the point with the largest rate of change of gray values ​​is the corner of the eye.

[0169] Record the pixel coordinates of the inner corner of the left eye in the left image. outer corner pixel coordinates .

[0170] Using the same method, in the right eye region sub-image Extract the center pixel coordinates of the right pupil Inner corner pixel coordinates outer corner pixel coordinates .

[0171] Using the same processing flow, in and The pixel coordinates of the corresponding key points of the left and right eyes in the right image are extracted respectively, including the pixel coordinates of 6 key points, including the center of the pupils of both eyes and the inner and outer corners of the eyes, in the left and right images.

[0172] The three-dimensional coordinate calculation module is used to generate the three-dimensional coordinates of the eye center and the vertical pupil height deviation based on the key point pixel coordinate information and the continuous phase difference distribution.

[0173] Specifically, for each key eye point, its pixel coordinates (u, v) in the left image are obtained from the continuous phase difference distribution. Read the phase difference value corresponding to this point. .

[0174] Based on the calibration parameters of the stereo camera, a direct mapping relationship between phase difference and depth is established. The mapping formula is as follows:

[0175] ;

[0176] Where Z is the depth coordinate of the keypoint in the camera coordinate system, K is the phase difference-depth conversion coefficient, and C is a constant offset. The values ​​of K and C are predetermined by a standard calibration board during the system calibration phase. During the calibration process, phase difference data at different distances are collected, and the optimal values ​​of K and C are obtained through linear fitting.

[0177] The formula for calculating the X and Y coordinates of the key point in the camera coordinate system is as follows:

[0178] ;

[0179] ;

[0180] in and Let these be the coordinates of the camera's principal point. and The focal lengths of the camera in the x and y directions are derived from the camera intrinsic parameter matrix.

[0181] Repeat the above steps to calculate the 3D coordinates (X, Y, Z) of all 6 eye key points in the camera coordinate system.

[0182] The average of the three-dimensional coordinates of the centers of both pupils is taken as the reference coordinates of the eye position center. .

[0183] The Y-axis (vertical direction) coordinate of the eye position center With respect to the preset instrument reference eye level The difference is calculated to obtain the pupil height deviation:

[0184] .

[0185] The binocular vision localization module of this application employs a binocular phase difference-guided layered extraction algorithm for eye features and a direct 3D coordinate mapping algorithm. Through a three-level localization mechanism—coarse localization of the face region using phase difference gradient, precise localization of the eye region using integral projection, and key point extraction using polar coordinate analysis—it achieves a pixel accuracy of ≤0.8px for eye key point detection and a vertical localization accuracy of ±0.15mm, surpassing existing technologies. The algorithm exhibits strong adaptability to changes in lighting, facial expressions, and slight head posture variations, maintaining high detection accuracy even in low-light environments and under partial occlusion.

[0186] In this embodiment, the dual-condition AND gate confirmation module includes:

[0187] In this embodiment, the dual-condition AND gate confirmation module further includes a verification module, which is specifically as follows:

[0188] All input signals are timestamped at the hardware level. The timestamp of each signal is strictly aligned with the system global clock T, with a time resolution of 1ms, thus eliminating clock drift errors between different modules.

[0189] A ring-shaped signal synchronization buffer with a length of 3 binocular vision sampling periods (approximately 100ms) was established to achieve time matching between the low-frequency infrared ranging signal (2Hz, period 500ms) and the high-frequency binocular vision signal (30Hz, period 33ms).

[0190] The synchronization module is used to synchronize the sensing signal transmitted by the infrared ranging sensor and the key point pixel coordinate information, so as to obtain a set of effective input signals after synchronization.

[0191] Specifically, for each binocular vision sampling time Find the distance in the infrared ranging signal buffer. Recent effective infrared ranging data The time difference condition is met. This is used as the infrared ranging value corresponding to that moment.

[0192] Mark the validity of each input signal: if the infrared ranging data Within the physical measurement range of 0-1000mm, it is marked as valid; if the coordinates of 6 key eye points... All key points within the 640×480 image range are marked as valid; if the coordinates of any key point are outside the image range, they are marked as invalid.

[0193] Finally, obtain the set of valid input signals after synchronization. ,in This refers to the sampling time for binocular vision.

[0194] The first condition judgment module is used to generate first condition judgment information based on the set of valid input signals after synchronization.

[0195] Specifically, the first conditional judgment is as follows:

[0196] Initialize the first condition cumulative counter Confidence level of condition 1 .

[0197] For each sampling time Determine whether the infrared ranging data satisfies the valid interval constraint: ;

[0198] If the above inequality holds, the cumulative counter... Increment by 1; if not true, increment the counter. Reset immediately to zero, confidence level Reset to 0.

[0199] When the cumulative counter When calculating the confidence level for condition 1:

[0200] ;

[0201] in It is a fixed constant used to control the growth rate of the confidence level. When the value is 3, a stable confidence level of 0.73 can be achieved after 8 consecutive effective periods.

[0202] when If the value is ≥0.8, the first condition is satisfied; otherwise, the first condition is not satisfied.

[0203] The second condition judgment module is used to generate second condition judgment information based on the key point pixel coordinate information and the physiological geometric ratio constant of the human eye.

[0204] The second condition is as follows:

[0205] Perform basic effectiveness screening: if any key eye point Invalid; the second condition is directly determined to be unmet.

[0206] If the basic screening is passed, extract the pixel coordinates of 6 key points, assuming they are: the center of the left pupil. The inner corner of the left eye The outer corner of the left eye Right pupil center The inner corner of the right eye outer corner of the right eye .

[0207] Calculate the pixel spacing between the centers of the pupils of both eyes :

[0208] ;

[0209] Calculate the pixel spacing between the inner and outer corners of the left eye:

[0210] ;

[0211] Calculate the pixel spacing between the inner and outer corners of the right eye. :

[0212] ;

[0213] Calculate the vertical distance from the center of the left pupil to the line connecting the inner and outer corners of the eye. :

[0214] First, derive the equation of the straight line connecting the inner and outer corners of the eyes:

[0215] ,in: ; ; ;

[0216] Then calculate the perpendicular distance using the formula for the distance from a point to a line:

[0217] ;

[0218] The same method was used to calculate the vertical distance from the center of the right pupil to the line connecting the inner and outer corners of the eye. .

[0219] Indirect head posture assessment based on eye geometry: Calculate the angle between the line connecting the centers of both pupils and the horizontal direction. Used to determine the pitch and yaw attitude of the head:

[0220] ;

[0221] in The unit is radians, and the range is... .

[0222] Perform physiological-geometric consistency and posture constraint verification for the human eye, and determine whether all of the following inequalities hold simultaneously: ;

[0223] If all inequalities are true simultaneously, the second condition is satisfied; otherwise, the second condition is not satisfied.

[0224] The final determination module is used to generate a lifting control start signal based on the first condition judgment information and the second condition judgment information.

[0225] In this embodiment, the final determination method is as follows:

[0226] Initialize the two-condition cumulative counter Lift control enable signal .

[0227] For each sampling time ,like =1 and =1, dual-condition cumulative counter Increment by 1; otherwise, increment the biconditional cumulative counter. Immediately reset to zero; the lift control enable signal En is forcibly set to 0.

[0228] When the dual-condition accumulator counter and At that time, output the lifting control enable signal. Otherwise, output 0.

[0229] Once the enable signal En is set to 1, as long as and Keep it as 1 and The enable signal remains valid; if any condition is not met, the enable signal is immediately set to 0 within 1ms, cutting off the motor drive circuit.

[0230] In this embodiment, This represents the state where the first condition is met. This represents the condition for satisfying the second condition; 1 indicates satisfaction, and 0 indicates non-satisfaction.

[0231] The dual-condition AND gate confirmation module of this application indirectly determines the head's pitch and yaw attitude by calculating the angle between the line connecting the centers of the pupils and the horizontal direction. This replaces the traditional face pose angle calculation, avoids complex three-dimensional pose estimation calculations, and maintains a pose constraint accuracy of 15°. Furthermore, it abandons the traditional instantaneous logic AND gate judgment and, through the calculation of confidence level adjusted by an accumulator counter and a fixed constant, can effectively filter instantaneous infrared interference signals with a duration of less than 250ms, reducing the false trigger probability to less than one in a million.

[0232] Through basic validity screening, physiological and geometric consistency verification, and dual-condition timing accumulation three-level protection, comprehensive protection against false triggering is achieved, ensuring that the system will not produce dangerous lifting or lowering movements in any complex environment.

[0233] In this embodiment, the binocular vision positioning module, infrared sensing module, and dual-condition AND gate confirmation module all operate according to the same cycle as the lifting control module.

[0234] In this embodiment, the lifting control module is the core execution unit of the system. After the dual-condition AND gate confirmation module outputs a valid enable signal, it performs a cyclic iteration in units of a preset fixed control period T. In each cycle, it completes a complete input sampling, deviation calculation, pulse generation and position feedback verification, driving the forehead rest to complete the precise linkage adjustment with the eye position deviation.

[0235] In this embodiment, the lifting control module includes:

[0236] An initialization module is used to load the initial information of the lifting mechanism.

[0237] In this embodiment, the initialization is as follows:

[0238] After the system is powered on, all pre-calibrated fixed control parameters are loaded. All parameters remain constant during operation without any dynamic modification or adjustment.

[0239] Initialize the motor drive port, absolute encoder read port, and limit switch input port, and set the motor to self-locking mode.

[0240] Initialize global control state variables: Control cycle counter C = 0, positioning completion flag. Abnormal alarm signs .

[0241] Generate a global synchronization clock signal with a clock frequency of Each rising edge of the clock triggers a control cycle execution.

[0242] In this embodiment, the lifting control module further includes:

[0243] The periodic cycle control module is used to control the lifting mechanism according to the lifting control start signal and the human eye position information, so that the lifting mechanism can reach the preset conditions.

[0244] In this embodiment, the periodic loop control module includes:

[0245] A control cycle triggering and input signal validity verification module, comprising:

[0246] When the rising edge of the global synchronization clock arrives, the control cycle counter C is incremented by 1, and the control flow of the current cycle begins to be executed.

[0247] Read the status of the lifting control enable signal En: If En=0 (invalid), immediately stop all motor drive signal output, the motor remains in a self-locking state, the control cycle counter C is cleared, and waits for the next rising edge of the clock to re-detect the enable signal; if En=1 (valid), continue to execute the subsequent verification process.

[0248] Pupil height deviation Perform a physical validity check to determine if it falls within the preset reasonable range [-50mm, 50mm]. If it exceeds the range, mark it as invalid and set an abnormal alarm flag. Immediately stop control and output an abnormal alarm signal; if within the range, mark it as a valid control input.

[0249] Real-time position values ​​of absolute encoder Perform limit switch validity verification to determine whether it falls within the position range corresponding to the upper and lower limit switches on the guide rail. If the position exceeds the range, hardware-level protection will be triggered immediately, cutting off the motor drive power; if the position is within the range, it will be marked as valid position feedback.

[0250] Prioritize hardware limit protection judgment: If the upper limit switch signal is... (Trigger) and motor running direction (Up), or lower limit switch signal (Trigger) and motor running direction -1 (decreasing) immediately stops all pulse signal output, triggers motor hardware self-locking, and sets an abnormal alarm flag. .

[0251] The pupil height deviation segmentation and target step length mapping module is used to generate the target step length and the motor running direction indicator for the current cycle based on the human eye position information for the current cycle.

[0252] Specifically, the pupil height deviation segmentation and target step size mapping module includes:

[0253] Calculate the absolute value of the pupil height deviation in this cycle. This serves as the sole basis for segmented mapping.

[0254] according to The value of the threshold is mapped to four predefined fixed deviation segments, each corresponding to a unique fixed motor movement step size L. All segment thresholds and step size values ​​are calibrated and determined by a standard calibration board before the system leaves the factory and remain constant during operation without any dynamic adjustment.

[0255] The specific segmentation mapping rules are as follows:

[0256] when When the deviation is greater than 2mm, the segment is adjusted quickly to accommodate large deviations, with a target step size of L=L1 (e.g., L1=0.5mm).

[0257] When 1mm < When the deviation is ≤2mm, the corresponding segment is adjusted to stabilize the deviation, with a target step size L=L2 (e.g., L2=0.2mm).

[0258] When 0.2mm < When the deviation is ≤1mm, the corresponding small deviation fine adjustment segment is used, with a target step size L=L3 (e.g., L3=0.05mm).

[0259] when When the value is ≤0.2mm, the corresponding positioning segment is in place, and the target step size L=0mm.

[0260] At the same time, according to The sign determines the direction of motor operation in this cycle: when When >0, the motor running direction indicator Dir=1 (the radiator rises); when When <0, Dir=-1 (Errone decreases); when When =0, Dir=0 (motor stops).

[0261] A single-cycle motor drive pulse count calculation module is used to generate the number of drive pulses in the current cycle based on the target step size of the current cycle.

[0262] In this embodiment, the single-cycle motor drive pulse count calculation module includes:

[0263] Motor inherent step angle 1. The subdivision factor M and the lead screw lead P are both fixed mechanical parameters of the lifting mechanism, which remain unchanged during operation. The lead screw lead P represents the distance that the lead screw drives the overhead support to move vertically for each revolution of the motor.

[0264] Based on the target step size L for this cycle, calculate the number of motor drive pulses N required for this cycle using the following formula:

[0265] ;

[0266] Where N is the number of drive pulses per cycle, and the calculation result is rounded to the nearest integer; L is the target step size, in millimeters (mm); M is the motor microstepping factor, which has no unit. 1 represents the inherent step angle of the motor, in degrees (°); P represents the lead of the lead screw, in millimeters (mm).

[0267] When the target step size L=0mm, the number of drive pulses N in a single cycle is 0, and no pulse signal is output. The final output is the number of drive pulses N in this cycle and the motor running direction indicator Dir for this cycle.

[0268] A pulse control signal sending module is used to generate a square wave pulse signal according to the number of drive pulses in the current cycle, and send the square wave pulse signal and the motor running direction indicator in the current cycle to the stepper motor.

[0269] Specifically, during the remaining time of this control cycle, N square wave pulse signals and corresponding direction control signals Dir are output according to a preset fixed pulse frequency (e.g., f=1000Hz) to drive the stepper motor.

[0270] A real-time location tracking module is used to obtain the actual distance traveled in the current period.

[0271] Specifically, during the pulse output process, the position value S(t) of the absolute encoder is read in real time at a frequency of 1kHz, and the initial position at the beginning of the cycle is recorded. and the actual position at the end of this period .

[0272] Calculate the actual movement distance of the forehead support during this period. :

[0273] ;

[0274] Thanks to the fully closed-loop position feedback using an absolute encoder, the actual position S(t) can be accurately measured even if the motor experiences slight step loss. The next control cycle will be based on the latest pupil height deviation output from the binocular vision positioning module. The target step size is recalculated, and the position error of the previous cycle is automatically compensated.

[0275] In this embodiment, the periodic loop control module further includes:

[0276] The positioning condition judgment module is used to determine whether the position is in place based on the position information of the human eye.

[0277] Specifically, the positioning condition judgment module includes:

[0278] Execution of positioning accuracy determination: When the absolute value of the pupil height deviation in this cycle meets the following conditions... When the location is confirmed to be in place, a location completion flag is set. Immediately stop outputting pulse signals, trigger the motor to self-lock, and maintain the current position of the forehead support.

[0279] If the positioning conditions are not met Then set a location completion flag. =0, wait for the next rising edge of the global synchronization clock, automatically enter the next control cycle, and repeat the execution of each module (such as the binocular vision positioning module and the dual-condition AND gate confirmation module).

[0280] When the positioning is complete. When =1, the system exits the main control loop and enters the position holding state.

[0281] In existing technologies, if a vision module is used to work continuously, it not only consumes a lot of power, but is also easily affected by ambient light, moving objects and other interference, resulting in false triggering. If a simple infrared trigger is used, it can only detect the presence of an object, but cannot distinguish between a human body and other objects, resulting in a very high risk of false triggering and failing to meet the safety requirements of medical devices.

[0282] This application solves this problem through an inventive architecture that combines infrared sensing pre-triggered detection with dual-condition AND gate joint verification. Specifically, when the system is in standby mode, only the infrared ranging module samples at a low frequency, while all other high-power modules are completely powered off and put into hibernation, achieving extremely low-power standby.

[0283] By abandoning the traditional instantaneous logic AND gate judgment, a timing cumulative counter and confidence calculation mechanism are introduced to filter instantaneous interference. At the same time, a second condition check is performed by combining the inherent physiological geometric proportions of the human eye. In principle, the possibility of non-human targets triggering the system is eliminated, and medical-grade trigger reliability is achieved while ensuring low power consumption.

[0284] Existing binocular vision positioning technology relies on global stereo matching, which involves a large amount of computation and is difficult to implement in real time on embedded devices; it is prone to mismatches, leading to errors in depth calculation; and it is sensitive to changes in lighting and partial occlusion, resulting in poor robustness.

[0285] This application addresses the aforementioned issues through a three-tiered architecture: phase difference gradient face coarse screening, integral projection eye precise localization, and polar coordinate key point extraction.

[0286] By using the phase difference gradient magnitude to quickly filter face regions, the subsequent processing range is reduced to less than 1 / 10 of the original image, significantly reducing the amount of computation.

[0287] Establishing a direct mapping relationship between phase difference and depth eliminates the need for traditional point-by-point stereo matching, thus completely avoiding mismatching problems;

[0288] The algorithm is naturally robust to changes in lighting, facial expressions, and slight head pose changes, without relying on complex image preprocessing and deep learning models.

[0289] In existing technologies, head posture determination requires the calculation of three-dimensional posture angles, which involves a large amount of computation and low accuracy. Simple eye positioning completely ignores changes in head posture, and when the subject's head is tilted excessively, it will lead to serious positioning errors and affect the accuracy of the detection results.

[0290] This application utilizes the inherent physiological and geometric features of the human eye to achieve head posture determination:

[0291] The pitch and yaw attitude of the head can be indirectly determined by calculating the angle between the line connecting the centers of the pupils of both eyes and the horizontal direction.

[0292] Based on the inherent physiological proportions of the human eye, such as interpupillary distance and palpebral fissure length, and vertical distance between the pupil and the corner of the eye, the extracted key points of the eye are verified for consistency.

[0293] The entire process does not require complex 3D pose calculations. Without increasing any additional computational load, it effectively eliminates invalid positioning results caused by excessive head tilt, thus improving the robustness of the system.

[0294] This application also provides a method for linking eye positioning with forehead support lifting, the method comprising:

[0295] Acquire the sensing signal transmitted by the infrared ranging sensor;

[0296] After acquiring the object signal transmitted by the infrared ranging sensor, facial information and eye position information are periodically acquired.

[0297] The system periodically generates a lifting control start signal based on the object signal and facial information transmitted by the infrared ranging sensor.

[0298] Based on the lifting control start signal, lifting signals are generated cyclically with a preset threshold as the fixed control cycle until the positioning condition is met.

[0299] Although the present invention has been described in detail above with general descriptions and specific embodiments, modifications or improvements can be made to it, which will be obvious to those skilled in the art. Therefore, all such modifications or improvements made without departing from the spirit of the present invention fall within the scope of protection claimed by the present invention.

Claims

1. A human eye positioning and forehead support lifting linkage control system, characterized in that, The human eye positioning and forehead support lifting linkage control system includes: An infrared sensing module is used to acquire the sensing signal transmitted by an infrared ranging sensor. A binocular vision positioning module is used to periodically acquire face information and eye position information after acquiring the object signal transmitted by the infrared ranging sensor. A dual-condition AND gate confirmation module is used to periodically generate a lifting control start signal based on the object signal and face information transmitted by the infrared ranging sensor. The lifting control module is used to generate lifting signals cyclically according to the lifting control start signal with a preset threshold as the fixed control cycle until the positioning condition is met.

2. The human eye positioning and forehead support lifting linkage control system as described in claim 1, characterized in that, The binocular vision positioning module includes: The initial phase difference calculation module is used to acquire the original RGB facial image synchronously acquired by the binocular camera, and generate left and right grayscale images and continuous phase difference distribution based on the original RGB facial image synchronously acquired by the binocular camera. A coarse face region localization module is used to generate coarse face region localization information based on left and right grayscale images and continuous phase difference distribution. The coarse face region localization information includes the left sub-image of the face, the right sub-image of the face, and the coordinates of the bounding rectangle of the face region. An eye feature region precise positioning module is used to generate eye feature region positioning information based on the coarse positioning information of the face region. The eye feature region positioning information includes a left eye region sub-image, a right eye region sub-image, and the pixel coordinates of the left and right eyes in the left image. A key point extraction module is used to generate key point pixel coordinate information based on eye feature region positioning information. The three-dimensional coordinate calculation module is used to generate the three-dimensional coordinates of the eye center and the vertical pupil height deviation based on the key point pixel coordinate information and the continuous phase difference distribution.

3. The human eye positioning and forehead support lifting linkage control system as described in claim 2, characterized in that, The dual-condition AND gate confirmation module includes: The synchronization module is used to synchronize the sensing signal transmitted by the infrared ranging sensor and the key point pixel coordinate information, so as to obtain a set of effective input signals after synchronization. The first condition judgment module is used to generate first condition judgment information based on the set of valid input signals after synchronization. The second condition judgment module is used to generate second condition judgment information based on the key point pixel coordinate information and the physiological geometric ratio constant of the human eye. The final determination module is used to generate a lifting control start signal based on the first condition judgment information and the second condition judgment information.

4. The human eye positioning and forehead support lifting linkage control system as described in claim 3, characterized in that, The lifting control module includes: An initialization module is used to load the initial information of the lifting mechanism; The periodic cycle control module is used to control the lifting mechanism according to the lifting control start signal and the human eye position information, so that the lifting mechanism can reach the preset conditions.

5. The human eye positioning and forehead support lifting linkage control system as described in claim 4, characterized in that, The periodic loop control module includes: The pupil height deviation segmentation and target step length mapping module is used to generate the target step length and the motor running direction indicator for the current cycle based on the human eye position information for the current cycle. A single-cycle motor drive pulse count calculation module is used to generate the number of drive pulses in the current cycle based on the target step size of the current cycle. A pulse control signal sending module is used to generate a square wave pulse signal according to the number of drive pulses in the current cycle, and send the square wave pulse signal and the motor running direction indicator in the current cycle to the stepper motor. A real-time location tracking module is used to obtain the actual distance traveled in the current period.

6. The human eye positioning and forehead support lifting linkage control system as described in claim 5, characterized in that, The periodic loop control module further includes: The positioning condition judgment module is used to determine whether the position is in place based on the position information of the human eye.

7. The human eye positioning and forehead support lifting linkage control system as described in claim 2, characterized in that, The initial phase difference calculation module includes: The original image alignment module is used to align the timestamps of the original images acquired by the binocular camera to obtain an aligned image group. A grayscale conversion module is used to perform grayscale conversion on the aligned image group to obtain a grayscale image group; A frequency domain matrix acquisition module is used to perform Fourier transform on the grayscale image group to obtain the frequency domain matrix of the left image and the frequency domain matrix of the right image. A phase difference matrix calculation module is used to generate a phase difference matrix between the left and right images based on the frequency domain matrix of the left image and the frequency domain matrix of the right image. A phase difference distribution acquisition module is used to perform phase unwrapping on the phase difference matrix of the left and right images to obtain a continuous phase difference distribution.

8. The human eye positioning and forehead support lifting linkage control system as described in claim 7, characterized in that, The coarse facial region localization module includes: The gradient value calculation module is used to calculate the gradient values ​​of continuous phase differences distributed in the horizontal and vertical directions, thereby obtaining the phase difference gradient magnitude of each pixel. The phase difference gradient magnitudes of each pixel form a gradient magnitude image. The binarization module is used to perform binarization processing on the gradient magnitude image to obtain a binarized image. A foreground connected component acquisition module is used to perform connected component analysis on a binarized image to obtain various foreground connected components. A face candidate region filtering module is used to obtain face candidate regions based on each connected component.

9. The human eye positioning and forehead support lifting linkage control system as described in claim 8, characterized in that, The precise localization module for eye feature regions includes: An eye candidate region acquisition module is used to acquire preset eye candidate regions; The left sub-image acquisition module for the eye is used to extract the corresponding sub-image of the candidate eye region in the left sub-image of the face as the left sub-image of the eye; A horizontal projection curve acquisition module is used to perform horizontal integral projection calculation on the left sub-image of the eye to obtain a horizontal projection curve. A smoothing module is used to smooth the horizontal projection curve to obtain a smoothed horizontal projection curve. A horizontal center position acquisition module is used to acquire the horizontal center positions of the left and right eyes based on a smoothed horizontal projection curve. The region segmentation module is used to extract the horizontal sub-regions of the left and right eyes respectively in the left sub-image of the eyes, with the horizontal center position of the left and right eyes as the center. The horizontal sub-regions of the left and right eyes include the horizontal sub-region of the left eye and the horizontal sub-region of the right eye. A region sub-image acquisition module is used to generate a left-eye region sub-image and a right-eye region sub-image based on the left-eye horizontal sub-region and the right-eye horizontal sub-region.

10. A method for linking human eye positioning with forehead support lifting, characterized in that, The method for linking human eye positioning with forehead support lifting includes: Acquire the sensing signal transmitted by the infrared ranging sensor; After acquiring the object signal transmitted by the infrared ranging sensor, facial information and eye position information are periodically acquired. The system periodically generates a lifting control start signal based on the object signal and facial information transmitted by the infrared ranging sensor. Based on the lifting control start signal, lifting signals are generated cyclically with a preset threshold as the fixed control cycle until the positioning condition is met.

Images