A wearable visual distance measurement method

Abstract

The present invention provides a wearable visual distance measurement method, which belongs to the field of visual depth extraction, including a pupil tracking algorithm and a visual distance measurement algorithm. The pupil tracking algorithm tracks the pupil position state of the subject in real time, pre-processes the eye image obtained by the wearable eye movement video acquisition hardware and fits the Hough circle, so as to calibrate the pupil contour and pupil center of the subject in real time; the visual distance measurement algorithm is based on the data of the pupil tracking algorithm, fits the position of the center point of the eyeball through the center calibration algorithm, and imports the established data of the image sensor and eye parameters into the visual depth model, so as to obtain the key information of visual distance measurement, and then calculate the real-time visual depth of the subject through the ranging equation. The present invention adopts the above-mentioned wearable visual distance measurement method to construct a wearable binocular visual distance detection algorithm system for multiple scenarios, and provides accurate real-time data for application scenarios requiring binocular visual distance.

Description

Technical Field

[0001] The present invention relates to the field of visual depth extraction, and in particular to a wearable visual range measurement method. Background Art

[0002] Wearable technology has shown great research value and market potential in many fields, especially in health monitoring and human-computer interaction. Among them, eye tracking devices, as part of smart glasses, will expand rapidly in virtual reality (VR), augmented reality (AR) and medical health fields.

[0003] In terms of health monitoring, wearable eye tracking technology can be used to detect and diagnose neurological diseases such as Alzheimer's disease, autism and multiple sclerosis. It can provide non-invasive diagnostic information in the early stages of the disease by analyzing eye movements, which makes it have broad application prospects in long-term health monitoring. The current research focus of academia and industry is on how to accurately measure eye rotation angles and binocular gaze directions in a low-cost, safe and efficient manner. This type of technology can not only provide new data support for ophthalmic research and treatment, but can also be widely used in consumer electronics to achieve a more personalized interactive experience and promote the popularization and development of eye tracking technology.

[0004] Wearable visual distance measurement cannot be separated from the support of eye tracking technology. Human-computer interaction through eye movement is more efficient than traditional methods. Compared with other technologies, eye tracking technology has the characteristics of directness and naturalness, and has less impact on people themselves, so using vision as an input method will be more efficient and natural. However, the existing eye tracking technology is limited to estimating the sight direction through two-dimensional or three-dimensional mapping through the process of automatically detecting the center position of the pupil or identifying the visual fixation point of a fixed scene. It is often used to determine the two-dimensional position of the visual fixation point on the scene camera screen. Therefore, the existing technology lacks the ability to extract the visual depth information of the fixation point position after the sight estimation (that is, the distance between the human eye and the fixation target), and cannot fit and extract the spatial three-dimensional position information of the fixation target. In addition, although traditional distance measurement sensors such as TOF can accurately measure the distance information of the target object, they are limited to the sensor's direct view of the target object, and cannot follow the eye movement information to perform follow-up measurement of any fixation point in the spatial environment. Additional auxiliary electric mechanisms will also affect the wearability of the device. Summary of the invention

[0005] The purpose of the present invention is to provide a wearable visual distance measurement method, which builds a set of wearable binocular visual distance detection algorithm system for multiple scenarios based on the hardware of eye movement video acquisition, tracks the gaze direction of human eyes and extracts visual depth information, and provides accurate real-time data for application scenarios that require binocular visual distance. Solve the above technical problems and fill the gap in the field of visual depth extraction of eye movement tracking technology.

[0006] To achieve the above object, the present invention provides a wearable visual range measurement method, comprising the following steps:

[0007] S1. Pupil tracking algorithm: Preprocess the eye images obtained by wearable eye movement video acquisition hardware and perform Hough circle fitting to calibrate the pupil contour and pupil center of the subject in real time;

[0008] S11. Data preprocessing: Obtain the eye image of the subject through wearable eye movement video acquisition hardware, preprocess the eye image data through color space conversion, and then convert the preprocessed eye image into a binary image through threshold segmentation to highlight the pupil area;

[0009] S12, pupil edge detection: filter and smooth the binary image obtained in S11, then calculate its horizontal and vertical gradients, and then calculate the gradient amplitude and direction of each image pixel; perform non-maximum suppression on the gradient amplitude of each image pixel, retain the local maximum value to refine the edge features, and then distinguish strong edges from weak edges to ensure that strong edges are retained, and weak edges are retained only when connected to strong edges, and finally form a complete pupil edge line through edge connection;

[0010] S13, Hough circle fitting: Apply the Hough circle fitting algorithm to identify the pupil edge line in S12;

[0011] S14, Real-time pupil tracking and positioning: By constructing a likelihood probability equation, introducing a voting mechanism, accumulating counts in the parameter space (a, b, r), and finding the (a, b, r) combination with the highest number of votes, the pupil center point in the eye image is fitted;

[0012] S2. Using the pupil contour and pupil center data of the subject, the center point position of the eyeball is fitted through a center calibration algorithm, and the established data of the image sensor and eye parameters are imported into a visual depth model to obtain the visual distance measurement information of the eyeball, and then the real-time visual depth of the subject is calculated through a range measurement equation;

[0013] S21, gaze center calibration: according to the pupil contour and pupil center point of the subject obtained in S1, the weighted average position of multiple marking points is calculated to determine the optimal center point coordinates;

[0014] S22, visual depth model: modeling based on the established data of image sensors and eye parameters, calculating the eye sight vector and its corresponding yaw and pitch angle information;

[0015] S23, constructing perspective projection: according to the eye sight vector and its corresponding yaw angle and pitch angle, projecting the spherical point onto the 2D plane through the rotation matrix;

[0016] S24. Calculate visual depth using the distance measurement equation: Calculate the subject's visual distance in real time using the set pupil distance and the calculated eye sight vector parameter.

[0017] Preferably, the specific steps of S1 are:

[0018] S11. Data preprocessing: Obtain the subject's eyeball image through an image acquisition device, and perform color space conversion to preprocess the image data to enhance contrast and simplify subsequent processing. In addition, the image is converted into a binary image through threshold segmentation to highlight the pupil area;

[0019] S12, pupil edge detection: first, filter and smooth the binary image in S1 to reduce noise interference, then calculate the horizontal and vertical gradients of the image, and then calculate the gradient amplitude and direction of each pixel, and then perform non-maximum suppression to retain the local maximum value to refine the edge features. Next, apply double threshold processing to distinguish strong edges from weak edges to ensure that strong edges are retained, and weak edges are retained only when they are connected to strong edges. Finally, through the edge connection process, a complete pupil edge line is formed;

[0020] S13, Hough circle fitting: Apply the Hough circle fitting algorithm to identify the pupil lines in S12;

[0021] S14. Real-time pupil tracking and positioning: By constructing a likelihood probability equation and introducing a voting mechanism, the counts are accumulated in the parameter space (a, b, r) and the (a, b, r) combination with the highest number of votes is found to fit the pupil in the image.

[0022] Preferably, the specific steps of S2 are:

[0023] S21, gaze center calibration: according to the n pupil center points and contours of the subject obtained in S14, the optimal center point coordinates are determined by calculating the weighted average position of multiple marking points;

[0024] S22, visual depth model: Modeling is performed based on the focal length and distance of the acquisition sensor, image size, eyeball radius and pupil distance, etc., to calculate the eyeball sight vector and pitch angle and other information;

[0025] S23, constructing a perspective projection: according to the sight vector and its corresponding yaw angle and pitch angle, projecting the spherical point onto the 2D plane through a rotation matrix;

[0026] S24. Calculate visual depth using the distance measurement equation: Calculate the subject's visual distance in real time using the set pupil distance and the calculated sight line vector and other parameters.

[0027] Preferably, in S11, the data preprocessing formula for color space conversion is:

[0028] Icolor=i×R+j×G+k×B;

[0029] Among them, R is the pixel value of the red channel, G is the pixel value of the green channel, B is the pixel value of the blue channel, and i, j, and k are the color thresholds corresponding to the eye image respectively;

[0030] The formula for threshold segmentation is:

[0031]

[0032] Among them, Ibinary(x,y) represents the binary image pixel value at the coordinate (x,y), Icolor(x,y) is the pixel value at this point after color space conversion, and T is the threshold.

[0033] Preferably, in S12, the filtering and smoothing formula is:

[0034]

[0035] Where Iblur(x,y) refers to the pixel value at the image coordinate (x,y) after filtering and smoothing; σ refers to the standard deviation of the distribution;

[0036] Calculate the horizontal and vertical gradients of the image after filtering and smoothing. The formula is:

[0037]

[0038] Among them, Gx and Gy represent the horizontal and vertical gradient amplitudes, that is, the rate of change of the image in the horizontal and vertical directions; I represents the pixel value of the original image;

[0039] The gradient magnitude and direction of each image pixel is given by:

[0040]

[0041] Among them, G represents the gradient amplitude, that is, the edge strength at a certain point, combining the changes in the horizontal and vertical directions; θ represents the gradient direction, that is, the direction of the edge.

[0042] Preferably, in S13, the Hough circle fitting is specifically as follows:

[0043] Define the parameter space of the circle, the equation of the circle is:

[0044] (xa)2+(yb)2=r2;

[0045] Among them, (a, b) is the coordinate of the center of the circle, and r is the radius;

[0046] For each edge point (xi, yi), after traversing the radius r, the corresponding center (a, b) is calculated as:

[0047] a=xi-rcos(θ), b=yi-rsin(θ).

[0048] Preferably, in S14, the likelihood probability equation is:

[0049]

[0050] Among them, P(E|a,b,r) is the likelihood function, which represents the probability of observing an edge image under a given parameter combination, and σ is the standard deviation of the error.

[0051] Preferably, in S21, each marking point is set to (xi, yi, axis_ratioi), and the weight wi is calculated in such a way that the closer the axis_ratioi is to 1, the higher the weight is. The specific formula is:

[0052]

[0053] Among them, xi and yi are the horizontal and vertical coordinates of the point respectively, and axis_ratioi is the axis ratio of the point;

[0054] The calculation formula of the weighted average coordinate is:

[0055]

[0056] Among them, (best_x, best_y) is the final result, that is, the coordinates of the eye's center point.

[0057] Preferably, in S22, the step of establishing a visual depth model is:

[0058] S221. Set physical parameters, including the focal length and distance of the acquisition sensor, image size, and eyeball radius parameters. The specific settings are as follows:

[0059] Collect sensor parameters: distance d from the device to the origin, focal length f;

[0060] Image parameters: image size img_size, scaling factor scale;

[0061] Eyeball parameters: Eyeball radius: R eye, pupil distance P;

[0062] S222, coordinate conversion, converting the pixel coordinates into the standardized coordinates NDC according to the set image size img_size, and the conversion formula is as follows:

[0063]

[0064] Among them, ndc_x and ndc_y represent the standardized coordinates in the horizontal and vertical directions respectively, and their values ​​range from -1 at the left edge to 1 at the right edge;

[0065] Then, according to the set focal length f and the distance d from the device to the origin, NDC is converted into the ray direction in the world coordinate system. The ray direction ray_dir calculation formula is as follows:

[0066]

[0067] In order to ensure that the ray direction is a unit vector, it is normalized. The normalization formula is as follows:

[0068]

[0069] Among them, ray_dirnorm represents the direction of the ray emitted from the device;

[0070] S223. Solving the equation by setting the coordinates of the sphere center, the eyeball radius Reye and the ray direction ray_dir can obtain the intersection of the ray and the sphere. Calculate the yaw angle Yaw and the pitch angle pitch according to the ray direction ray_dir. The formula is as follows:

[0071] yaw=atan2(x,z);

[0072]

[0073] Where x, y, and z represent the components of the sight vector on the x, y, and z axes, respectively.

[0074] Preferably, in S23, the perspective projection formula is:

[0075]

[0076] Among them, xproj and yproj represent the x and y coordinates after projection respectively.

[0077] Preferably, in S24, the steps of calculating the visual depth using the ranging equation are as follows:

[0078] S241. Calculate the angle between each line of sight and the pupil. The formula is as follows:

[0079]

[0080] Among them, θ represents the angle between the line of sight and the line connecting the pupil (i.e., the x-axis), Reye represents the radius of the eyeball, x represents the coordinate of the center of the pupil on the x-axis at this time, and best_x represents the coordinate of the center of the eyeball on the x-axis; According to this formula, the angle θL between the left eye line of sight and the front and the angle θR between the right eye line of sight and the front can be obtained respectively;

[0081] S242, calculating the viewing distance. The viewing distance can be calculated according to the set pupil distance and the sight angle of each eyeball. The formula is as follows:

[0082]

[0083] Among them, L represents the viewing distance and P represents the pupil distance.

[0084] Therefore, the present invention adopts the above-mentioned wearable visual distance measurement method, and the technical effects are as follows: by constructing a set of wearable visual distance measurement algorithms for multiple scenarios, it is possible to track the gaze direction of the human eye in real time and extract visual depth information, providing data support for application scenarios that require binocular visual distance. This method can achieve eye movement tracking visual distance measurement without the subject changing the head position, overcomes the limitation that traditional TOF sensors can only measure direct-viewing targets, improves the wearability of the device and the flexibility of measurement, and combines mathematical algorithms and visual depth models to achieve real-time ranging of the subject's visual depth, providing important technical support and application value for ophthalmic research, medical health monitoring, virtual reality, and augmented reality. BRIEF DESCRIPTION OF THE DRAWINGS

[0085] Figure 1 A flow chart of a wearable visual range measurement method of the present invention;

[0086] Figure 2 is a flow chart of the visual depth model algorithm of the present invention;

[0087] Figure 3 This is a flow chart of the first embodiment of the present invention;

[0088] Figure 4 This is a flow chart reference of the second embodiment of the present invention. DETAILED DESCRIPTION

[0089] The technical solution of the present invention is further described below through the accompanying drawings and embodiments.

[0090] Unless otherwise defined, technical or scientific terms used in the present invention shall have the common meanings understood by one having ordinary skills in the field to which the present invention belongs.

[0091] Embodiment 1

[0092] like Figure 3 As shown, the present invention provides a wearable visual distance measurement method, wherein the pupil tracking algorithm tracks and locates the pupil position of the subject in real time according to the eye image data, and the specific steps are as follows:

[0093] S11. Data preprocessing: Perform color space conversion on the subject’s eye image data to enhance contrast and simplify subsequent processing. The formula is as follows:

[0094] Icolor=0.2989×R+0.5870×G+0.1140×B;

[0095] In addition, the eye image is converted into a binary image through threshold segmentation to highlight the pupil area. The formula is as follows:

[0096]

[0097] S12, pupil edge detection, firstly filter and smooth the binary image data obtained after preprocessing to reduce noise interference, and calculate its horizontal and vertical gradients, the formula is as follows:

[0098]

[0099] Among them, Gx and Gy represent the horizontal and vertical gradients, that is, the rate of change of the image in the horizontal and vertical directions; I represents the pixel value of the original image.

[0100] Calculate the gradient magnitude and direction of each pixel using the formula:

[0101]

[0102] The gradient amplitude of each image pixel is non-maximum suppressed, and the local maximum is retained to refine the edge features. Then the strong edges and weak edges are distinguished to ensure that the strong edges are retained, and the weak edges are retained only when they are connected to the strong edges. Finally, the complete pupil edge line is formed through edge connection.

[0103] S13, Hough circle fitting, after completing edge detection, the Hough circle fitting algorithm is applied to identify the pupil edge line. It involves defining the parameter space of the circle, where the center coordinates and radius are the key parameters. For each detected edge point, the Hough circle fitting algorithm traverses all possible radius values ​​and calculates the corresponding center position. The Hough circle fitting algorithm is an ideal choice for identifying circular features such as pupils due to its robustness to noise and its ability to handle incomplete edges, which improves the accuracy and reliability of recognition.

[0104] S14, real-time pupil tracking and positioning, by constructing a likelihood probability equation, introducing a voting mechanism, accumulating counts in the parameter space (a, b, r), and finding the (a, b, r) combination with the highest number of votes, so as to fit the pupil center point in the eye image. The likelihood probability equation is:

[0105]

[0106] Embodiment 2

[0107] like Figure 4 As shown, the present invention provides a wearable visual distance measurement method, wherein the visual distance measurement algorithm measures the visual depth of the subject in real time based on the real-time pupil position information and in combination with the established parameters, and the specific steps are as follows:

[0108] S21, fixation center calibration, obtain 300 pupil center points and contours of the subject according to the pupil tracking algorithm, and determine the best center point coordinates by calculating the weighted average position of multiple marking points. Calculate according to the following formula:

[0109]

[0110]

[0111] The coordinates of the subject's eye gaze center point (best_x, best_y) can be obtained.

[0112] S22, visual depth model, first set the acquisition sensor focal length and distance, image size and eyeball radius parameters respectively. The specific settings are as follows:

[0113] Collect sensor parameters: distance from device to origin d = 55, focal length f = 50;

[0114] Image parameters: image size img_size = 640, scaling factor scale = 1;

[0115] Eyeball parameters: eyeball radius Reye = 0.7, pupil distance P = 63.

[0116] According to the set image size img_size, the pixel coordinates are converted to the standardized coordinates NDC, and then according to the set focal length f and the distance d from the image acquisition sensor to the origin, the NDC is converted to rays in the world coordinate system and normalized. The formula is as follows:

[0117]

[0118] Calculate the yaw angle Yaw and pitch angle pitch according to the ray direction ray_dir. The formula is as follows:

[0119] yaw=atan2(x,z);

[0120]

[0121] S23, construct a perspective projection, and project the spherical point onto the 2D plane through a rotation matrix according to the eye sight vector and its corresponding yaw angle and pitch angle. The perspective projection formula is as follows:

[0122]

[0123] S24, the distance measurement equation calculates the visual depth, and the eye distance of the subject is calculated in real time through the set pupil distance and the calculated eye sight vector parameter. The formula is as follows:

[0124]

[0125] Therefore, the present invention adopts the above-mentioned wearable visual distance measurement method, and combines the wearable visual distance measurement method through a pupil tracking algorithm and a visual distance measurement algorithm, so that the pupil position of the subject can be accurately tracked in a real-time eye use environment, and the key eye movement parameters of visual distance measurement can be obtained at the same time, and the visual depth of the subject's eye movement fixation point can be calculated in real time.

[0126] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that they can still modify or replace the technical solution of the present invention with equivalents, and these modifications or equivalent replacements cannot cause the modified technical solution to deviate from the spirit and scope of the technical solution of the present invention.

Claims

1. A wearable visual range measurement method, characterized in that: The following steps are involved: S1. Pupil tracking algorithm: Preprocess the eye images obtained by wearable eye movement video acquisition hardware and perform Hough circle fitting to calibrate the pupil contour and pupil center of the subject in real time; S11. Data preprocessing: Obtain the eye image of the subject through wearable eye movement video acquisition hardware, preprocess the eye image data through color space conversion, and then convert the preprocessed eye image into a binary image through threshold segmentation to highlight the pupil area; S12, pupil edge detection: filter and smooth the binary image obtained in S11, then calculate its horizontal and vertical gradients, and then calculate the gradient amplitude and direction of each image pixel; perform non-maximum suppression on the gradient amplitude of each image pixel, retain the local maximum value to refine the edge features, and then distinguish strong edges from weak edges to ensure that strong edges are retained, and weak edges are retained only when connected to strong edges, and finally form a complete pupil edge line through edge connection; S13, Hough circle fitting: Apply the Hough circle fitting algorithm to identify the pupil edge line in S12; S14, Real-time pupil tracking and positioning: By constructing a likelihood probability equation, introducing a voting mechanism, accumulating counts in the parameter space (a, b, r), and finding the (a, b, r) combination with the highest number of votes, the pupil center point in the eye image is fitted; S2. Using the pupil contour and pupil center data of the subject, the center point position of the eyeball is fitted through a center calibration algorithm, and the established data of the image sensor and eye parameters are imported into a visual depth model to obtain the visual distance measurement information of the eyeball, and then the real-time visual depth of the subject is calculated through a range measurement equation; S21, gaze center calibration: according to the pupil contour and pupil center point of the subject obtained in S1, the weighted average position of multiple marking points is calculated to determine the optimal center point coordinates; S22, visual depth model: modeling based on the established data of image sensors and eye parameters, calculating the eye sight vector and its corresponding yaw and pitch angle information; S23, constructing perspective projection: according to the eye sight vector and its corresponding yaw angle and pitch angle, projecting the spherical point onto the 2D plane through the rotation matrix; S24. Calculate visual depth using the distance measurement equation: Calculate the subject's visual distance in real time using the set pupil distance and the calculated eye sight vector parameter.

2. A wearable visual range measurement method according to claim 1, characterized in that: In S11, the data preprocessing formula for color space conversion is: Icolor=i×R+j×G+k×B; Where R is the pixel value of the red channel, G is the pixel value of the green channel, B is the pixel value of the blue channel, and i, j, and k are the color thresholds corresponding to the eye image respectively; The formula for threshold segmentation is: Among them, Ibinary(x,y) represents the binary image pixel value at the coordinate (x,y), Icolor(x,y) is the pixel value at this point after color space conversion, and T is the threshold.

3. A wearable visual range measurement method according to claim 1, characterized in that: In S12, the filtering and smoothing formula is: Where Iblur(x,y) refers to the pixel value at the image coordinate (x,y) after filtering and smoothing; σ refers to the standard deviation of the distribution; Calculate the horizontal and vertical gradients of the image after filtering and smoothing. The formula is: Among them, Gx and Gy represent the horizontal and vertical gradient amplitudes, that is, the rate of change of the image in the horizontal and vertical directions; I represents the pixel value of the original image; The gradient magnitude and direction of each image pixel is given by: Among them, G represents the gradient amplitude, that is, the edge strength at a certain point, combining the changes in the horizontal and vertical directions; θ represents the gradient direction, that is, the direction of the edge.

4. A wearable visual range measurement method according to claim 1, characterized in that: In S13, the Hough circle fitting is specifically as follows: Define the parameter space of the circle, the equation of the circle is: (xa)2+(yb)2=r2; Among them, (a, b) is the coordinate of the center of the circle, and r is the radius; For each edge point (xi, yi), after traversing the radius r, the corresponding center (a, b) is calculated as: a=xi-rcos(θ), b=yi-rsin(θ).

5. A wearable visual range measurement method according to claim 1, characterized in that: In S14, the likelihood probability equation is: Among them, P(E|a,b,r) is the likelihood function, which represents the probability of observing an edge image under a given parameter combination, and σ is the standard deviation of the error.

6. A wearable visual range measurement method according to claim 1, characterized in that: In S21, each marking point is set to (xi, yi, axis_ratioi), and the weight wi is calculated in such a way that the closer the axis_ratioi is to 1, the higher the weight is. The specific formula is: Among them, xi and yi are the horizontal and vertical coordinates of the point respectively, and axis_ratioi is the axis ratio of the point; The calculation formula of the weighted average coordinate is obtained as follows: Among them, (best_x, best_y) is the final result, that is, the coordinates of the eye's center point.

7. A wearable visual range measurement method according to claim 1, characterized in that: In S22, the steps of establishing a visual depth model are: S221. Set physical parameters, including the focal length and distance of the acquisition sensor, image size, and eyeball radius parameters. The specific settings are as follows: Collect sensor parameters: distance d from the device to the origin, focal length f; Image parameters: image size img_size, scaling factor scale; Eyeball parameters: Eyeball radius: R eye, pupil distance P; S222, coordinate conversion, converting the pixel coordinates into the standardized coordinates NDC according to the set image size img_size, and the conversion formula is as follows: Among them, ndc_x and ndc_y represent the standardized coordinates in the horizontal and vertical directions respectively, and their values ​​range from -1 at the left edge to 1 at the right edge; Then, according to the set focal length f and the distance d from the device to the origin, NDC is converted into the ray direction in the world coordinate system. The ray direction ray_dir calculation formula is as follows: In order to ensure that the ray direction is a unit vector, it is normalized. The normalization formula is as follows: Among them, ray_dirnorm represents the direction of the ray emitted from the device; S223. Solving the equation by setting the coordinates of the sphere center, the eyeball radius Reye and the ray direction ray_dir can obtain the intersection of the ray and the sphere. Calculate the yaw angle Yaw and the pitch angle pitch according to the ray direction ray_dir. The formula is as follows: yaw=atan2(x,z); Where x, y, and z represent the components of the sight vector on the x, y, and z axes, respectively.

8. A wearable visual range measurement method according to claim 1, characterized in that: In S23, the perspective projection formula is: Among them, xproj and yproj represent the x and y coordinates after projection respectively.

9. A wearable visual range measurement method according to claim 1, characterized in that: In S24, the steps of calculating the visual depth using the ranging equation are as follows: S241. Calculate the angle between each line of sight and the pupil. The formula is as follows: Among them, θ represents the angle between the line of sight and the line connecting the pupil (i.e., the x-axis), Reye represents the radius of the eyeball, x represents the coordinate of the center of the pupil on the x-axis at this time, and best_x represents the coordinate of the center of the eyeball on the x-axis; According to this formula, the angle θL between the left eye line of sight and the front and the angle θR between the right eye line of sight and the front can be obtained respectively; S242, calculating the viewing distance. The viewing distance can be calculated according to the set pupil distance and the sight angle of each eyeball. The formula is as follows: Among them, L represents the viewing distance and P represents the pupil distance.

Images