Human eye pupil center position positioning method and system based on double cameras

Abstract

The invention discloses a human eye pupil center position positioning method and system based on double cameras, and the method comprises the steps: constructing an eyeball three-dimensional positioning system through combining a first camera and a second camera; a to-be-detected human eye image is obtained through an eyeball three-dimensional positioning system, human eye pupil position positioning is carried out through a dynamic segmentation algorithm of pupil symmetry, and coordinates of a human eye pupil center in the world are determined; according to the coordinates of the human eye pupil center in the world, target position calibration is carried out through the cross target, and the three-dimensional coordinate position of the human eye pupil center in the world space is determined. According to the method, accurate positioning of the pupil center can be achieved, and the pupil diameter can be calculated more accurately. The human eye pupil center positioning method and system based on the double cameras can be widely applied to the technical field of pupil center positioning.

Description

Technical Field

[0001] This invention relates to the field of pupil center positioning technology, and in particular to a method and system for locating the center position of the human eye pupil based on dual cameras. Background Technology

[0002] Existing instruments such as tonometers, fundus cameras, corneal topographers, and corneal topography instruments all require precise localization of the pupil center to accurately measure eye parameters. Among related technologies, methods for locating and tracking the pupil center primarily involve projection, laser triangulation, and binocular stereo ranging. Projection involves projecting a circular or two semi-circular light ring onto the cornea; the reflected image is then captured by a camera. However, when tears or long eyelashes interfere with the cornea, the circular reflection image can be distorted or missing, potentially leading to inaccurate center localization. When the cornea has significant astigmatism, the circular reflection image becomes elliptical, making it difficult to locate the circular region using Hough circle transform. Projection requires an internal camera within the measurement system, and the small size of the projection target ring makes it unsuitable for large-aperture measurement systems. Laser triangulation can only determine the depth of the laser reflection, not the other two dimensions. Neural network models require significant upfront investment of time for manual annotation and training of large datasets. Furthermore, this method has high computational complexity and cannot meet the requirements of real-time eye tracking. Binocular vision requires stereo vision calibration to determine the transformation matrix between world coordinates, which requires a complex 3D reconstruction process. The reconstruction requires combining coordinates from two cameras, and the cameras can only observe the side of the human eye, which can easily lead to errors in determining the size of the pupil. Summary of the Invention

[0003] To address the aforementioned technical problems, the present invention aims to provide a method and system for locating the center of the human eye pupil based on dual cameras, which can achieve accurate positioning of the pupil center and facilitate more precise calculation of the pupil diameter.

[0004] The first technical solution adopted in this invention is: a method for locating the center position of the human eye pupil based on dual cameras, comprising the following steps: A three-dimensional eye positioning system was constructed by combining the first and second cameras. The eye image of the subject is acquired by a three-dimensional eye positioning system and the position of the pupil is located by a dynamic segmentation algorithm based on pupil symmetry, thus determining the coordinates of the center of the pupil in the world. Based on the coordinates of the center of the human pupil in the world, the target position is determined by using a crosshair target to determine the three-dimensional coordinates of the center of the human pupil in world space.

[0005] Furthermore, the three-dimensional eye positioning system specifically includes a first camera and a second camera, wherein the first camera and the second camera are placed at a spatial right angle, and the optical axes of the first camera and the second camera converge at the position of the human eye to be tested. The first camera is located to the side of the three-dimensional eye positioning system, tilted, with its optical axis intersecting the human eye. The second camera is located directly below the three-dimensional eye positioning system, tilted upwards, with its optical axis intersecting the human eye. Both the first camera and the second camera have an illumination source, a short focal length lens, and a CCD sensor.

[0006] Furthermore, the step of acquiring an image of the human eye through a three-dimensional eye positioning system and locating the pupil position using a dynamic segmentation algorithm based on pupil symmetry to determine the coordinates of the pupil center in the world specifically includes: Based on the three-dimensional eye positioning system, the first image of the human eye under test is acquired through the first camera; Adaptive threshold coarse localization processing is performed on the first human eye image to obtain the preliminary pupil position; Based on the ROI pupil center fine localization method, the initial human eye pupil position is processed for secondary localization to determine the pupil center position of the human eye on the side camera. Based on geometric principles and combined with the position of the center of the human eye's pupil on a side camera, the z-axis coordinates of the center of the human eye's pupil in the world are determined. Based on the eye-based three-dimensional positioning system, an image of the second subject's eye is acquired using a second camera; Determine the binarization segmentation threshold of the pupil grayscale, perform binarization processing on the second human eye image to be tested, fit the pupil to an ellipse, and obtain the center position of the pupil of the human eye on the lower camera. Based on geometric principles and combined with the position of the center of the human eye's pupil on the camera below, the x-axis and y-axis coordinates of the center of the human eye's pupil in the world are determined; By combining the z-axis coordinates, x-axis coordinates, and y-axis coordinates of the center of the human pupil in the world, the coordinates of the center of the human pupil in the world can be obtained.

[0007] Furthermore, the step of performing adaptive threshold coarse localization processing on the first test human eye image to obtain the preliminary pupil position specifically includes: The first test eye image is preprocessed by median filtering and normalization to obtain the preprocessed first test eye image. The Otsu dynamic threshold segmentation and connected component filtering method is used to perform coarse localization and bounding selection on the preprocessed first human eye image to obtain the preliminary pupil region. Obtain the initial centroid of the pupil region, and use the centroid as the center point to determine the initial position of the human eye pupil within a preset range.

[0008] Furthermore, the step of performing secondary localization processing on the initial pupil position based on the ROI-based pupil center fine localization method to determine the pupil center position on the side camera specifically includes: The preliminary human eye pupil position is normalized by median filtering to obtain the human eye pupil position image after median filtering and normalization; A grayscale histogram analysis was performed on the human eye pupil position image after median filtering and normalization. The binarization segmentation threshold was determined based on the grayscale of the pupil. Non-pupil regions were then eliminated using a pupil template image to obtain the eliminated human eye pupil position image. Contour extraction is performed on the image of the human eye pupil position after elimination, the pupil is restored, and the center position of the human eye pupil on the side camera is determined.

[0009] Furthermore, the step of determining the three-dimensional coordinates of the center of the human pupil in world space by using a crosshair target to calibrate the target position based on the coordinates of the center of the human pupil in the world specifically includes: Using the z-axis coordinate and x-axis coordinate of the center of the human pupil in the world, the difference between the x-axis and z-axis coordinates of the center of the human pupil in world space can be obtained through the principle of similar triangles. For the y-axis coordinates of the center of the human pupil in the world, combined with the tilt angles of the first and second cameras, the y-axis coordinates of the center of the human pupil in world space are obtained through the principle of right triangles. By combining the y-axis coordinate of the center of the human pupil in world space and the difference between the x-axis and z-axis coordinates of the center of the human pupil in world space, the three-dimensional coordinate position of the center of the human pupil in world space can be determined.

[0010] Furthermore, the specific expression for calculating the difference between the x-axis and z-axis coordinates of the center of the human pupil in world space is as follows: In the above formula, This represents the distance between the pupil of the human eye and its center position on the x-axis of the world coordinate system. This represents the distance between the pupil of the human eye and its center position on the z-axis of the world coordinate system. This represents the pixel value indicating the horizontal offset of the human eye's pupil from the image center in the second camera. This represents the pixel value indicating the vertical offset of the human eye's pupil from the image center in the first camera. This indicates the distance between the center position and the camera. This indicates the focal length of the camera lens.

[0011] Furthermore, the specific expression for calculating the y-axis coordinate position of the center of the human eye pupil in world space is as follows: In the above formula, This represents the y-axis offset of the human eye's pupil from its center position in the world. This represents the offset distance of the human eye's pupil from its center position in the world coordinates along the camera's projection direction. Indicates the camera tilt angle. This represents the angle between the vertical offset distance of the image and the center of the image.

[0012] The second technical solution adopted in this invention is: a human eye pupil center position positioning system based on dual cameras, comprising: The first module is used to combine the first camera and the second camera to construct a three-dimensional eye positioning system; The second module is used to acquire the image of the human eye under test through the three-dimensional positioning system of the eyeball and to locate the position of the human eye pupil through the dynamic segmentation algorithm of pupil symmetry, and to determine the coordinates of the center of the human eye pupil in the world. The third module is used to determine the three-dimensional coordinates of the center of the human pupil in world space by using a crosshair target to calibrate the target position based on the coordinates of the center of the human pupil in the world.

[0013] The beneficial effects of the method and system of this invention are as follows: This invention constructs a three-dimensional eye positioning system by combining a first camera and a second camera. It further acquires images of the human eye under test through this system and uses a dynamic segmentation algorithm based on pupil symmetry to locate the pupil position, determining the coordinates of the pupil center in the world. The system structure simplifies the calculation of three-dimensional coordinates, directly mapping the coordinate axes of a single camera to a single direction in the world coordinate system. Finally, based on the world coordinates of the pupil center, a crosshair target is used to calibrate the target position, determining the three-dimensional coordinate position of the pupil center in world space. Based on the corresponding geometric principles, the positional difference of the target relative to the world coordinate center is calculated. The use of a crosshair target to calibrate the parallelism between the world coordinate center position and the camera facilitates more accurate calculation of the pupil diameter. Attached Figure Description

[0014] Figure 1 This is a flowchart of the steps of a method for locating the center position of the human eye pupil based on dual cameras according to the present invention; Figure 2 This is a structural block diagram of a human eye pupil center position positioning system based on dual cameras according to the present invention; Figure 3 This is a schematic diagram of the three-dimensional eye positioning system provided in a specific embodiment of the present invention; Figure 4 This is a schematic diagram of the three-dimensional positioning logic provided in a specific embodiment of the present invention; Figure 5 This is a schematic diagram of the human eye image to be tested provided in a specific embodiment of the present invention; Figure 6 This is a schematic diagram of the character target calibration process provided in a specific embodiment of the present invention; Figure 7 This is a schematic diagram of the cross-shaped target calibration principle provided in a specific embodiment of the present invention; Figure 8 This is a schematic diagram of the geometric principle of a dual-camera system provided in a specific embodiment of the present invention; Figure 9 This is a schematic diagram of the pupil diameter of multiple volunteers as shown in a specific embodiment of the present invention.

[0015] Reference numerals: 1. Illumination source; 2. Short focal length lens; 3. CCD sensor; 4. Human eye. Detailed Implementation

[0016] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments. The step numbers in the following embodiments are only for ease of explanation and do not limit the order of the steps. The execution order of each step in the embodiments can be adapted according to the understanding of those skilled in the art.

[0017] First, it should be noted that the main methods for locating and tracking the center of the pupil in related technologies are projection methods, laser triangulation methods, and binocular stereo ranging methods. The projection method involves projecting a circular or two semi-circular light ring onto the cornea, where it is reflected and captured by a camera. By fitting an expression to the circular reflection image, the central pixel coordinates of the cornea are calculated, and two-dimensional centering of the eyeball can be achieved based on these coordinates.

[0018] Triangulation laser ranging mainly consists of a laser source, an imaging system, and a position-sensitive photodetector. A converging lens focuses the laser beam emitted from the source, directing it at a specific angle to the target object, which is then imaged on the photodetector. The position of the target object on the photodetector determines whether it is within the working distance.

[0019] Binocular stereo ranging is a popular spatial positioning method. Traditional binocular vision is divided into two types: parallel and convergent. Parallel means that two cameras are placed parallel to the target, while convergent means that the optical axes of two cameras are converged on the target. Both methods can perform three-dimensional spatial positioning of the target.

[0020] However, the relevant technologies have the following problems: 1) Annular reflection image distortion may cause the Hough circle transform to fail to find the annular region in the image, resulting in calculation failure.

[0021] 2) Laser triangulation can only locate depth information and cannot calculate three-dimensional spatial coordinates.

[0022] 3) Traditional binocular stereo vision systems have high complexity in 3D reconstruction. Neural network models require a large amount of data to be manually labeled and calibrated, which cannot meet the real-time tracking requirements of eye movement.

[0023] 4) It requires a high degree of symmetry in camera position, has low accuracy, and cannot directly reflect the image position.

[0024] Reference Figure 1 This invention provides a method for locating the center position of the human eye pupil based on a dual-camera system. The method includes the following steps: S100, combining the first and second cameras, constructs a three-dimensional eye positioning system; Specifically, the eye-tracking three-dimensional positioning system includes a first camera and a second camera, wherein the first camera and the second camera are placed at a spatial right angle, and the optical axes of the first camera and the second camera converge at the position of the human eye to be tested. The first camera is located to the side of the eye-tracking three-dimensional positioning system, tilted, with its optical axis intersecting the human eye. The second camera is located directly below the eye-tracking three-dimensional positioning system, tilted upwards, with its optical axis intersecting the human eye. Both the first camera and the second camera have an illumination source, a short focal length lens, and a CCD sensor.

[0025] In this embodiment, as Figure 3 As shown, the experimental setup mainly consists of two camera modules, including an illumination source 1, a short-focal-length lens 2, a CCD sensor 3, and a human eye 4. Figure 1 (a) in the figure is a top view of the system. Figure 1 (b) is a side view of the system. The two camera modules are placed at a right angle in space. Camera A, the first camera, is located on the side of the system and is tilted, with its optical axis intersecting the human eye. Camera B, the second camera, is located directly below the system and is tilted upwards, with its optical axis intersecting the human eye. The optical axes of cameras A and B converge at the human eye.

[0026] S200: Obtain the image of the human eye under test through the three-dimensional positioning system of the eyeball and locate the position of the human eye pupil through the dynamic segmentation algorithm of pupil symmetry, and determine the coordinates of the center of the human eye pupil in the world. S210. Based on the three-dimensional eye positioning system, the first image of the human eye to be tested is acquired through the first camera; S220. Perform adaptive threshold coarse localization processing on the first human eye image to be tested to obtain the preliminary position of the human eye pupil. Specifically, the first test eye image is preprocessed by median filtering and normalization to obtain the preprocessed first test eye image; the preprocessed first test eye image is coarsely localized and bounded by the Otsu dynamic threshold segmentation and connected component filtering method to obtain the preliminary pupil region; the centroid of the preliminary pupil region is obtained, and the preliminary pupil position within the preset range is determined with the centroid as the center point.

[0027] S230. Based on the ROI pupil center fine localization method, the initial human eye pupil position is processed for secondary localization to determine the pupil center position of the human eye on the side camera. Specifically, the initial pupil position is processed by median filtering and normalization to obtain a median-filtered and normalized pupil position image. Gray-level histogram analysis is then performed on the median-filtered and normalized pupil position image. The binarization segmentation threshold is determined based on the pupil's gray level, and non-pupil regions are eliminated using a pupil template image to obtain a eliminated pupil position image. Contour extraction is then performed on the eliminated pupil position image to restore the pupil and determine the center position of the pupil on the side camera.

[0028] S240. Based on geometric principles and combined with the position of the center of the human eye's pupil on the side camera, determine the z-axis coordinate of the center of the human eye's pupil in the world. In this embodiment, the process of three-dimensional positioning coordinates of the eyeball is as follows: Figure 4 As shown, first, an image of a human eye is captured using cameras A and B. For the image from camera A, as follows... Figure 5 As shown in (a), the issue of eyelid occlusion of the pupil needs to be addressed. To this end, a dynamic segmentation algorithm based on pupil symmetry is employed. This algorithm consists of two parts: adaptive threshold coarse localization and ROI pupil center fine localization. Adaptive threshold coarse localization first preprocesses the image with median filtering and normalization, then uses Otsu dynamic thresholding and connected component filtering to roughly locate the pupil region. Next, the centroid of the pupil region is calculated, and a small rectangular bounding box is drawn with the centroid as the center point to locate the approximate area of ​​the pupil in the image. ROI pupil center fine localization is based on the coarsely localized image, and median filtering and normalization are performed again. Gray-level histogram analysis is performed on the image, and a binarized segmentation threshold is determined based on the gray level of the pupil. Non-pupil regions are eliminated using a pupil template image, and finally, contour extraction is performed. For the case of eyelid occlusion of the pupil, after contour extraction, the lower half of the pupil is used to reconstruct it, allowing for a more accurate determination of the pupil center on the side camera. The z-coordinate of the center of the human pupil in the world is determined based on geometric principles.

[0029] S250, based on an eye-based three-dimensional positioning system, acquires images of the second human eye under test using a second camera; S260. Determine the binarization segmentation threshold of the pupil grayscale, perform binarization processing on the second human eye image to be tested, perform ellipse fitting on the pupil, and obtain the center position of the pupil of the human eye on the lower camera. S270. Based on geometric principles and combined with the position of the center of the human eye's pupil on the camera below, determine the x-axis and y-axis coordinates of the center of the human eye's pupil in the world. S280. By combining the z-axis coordinates of the center of the human pupil in the world, the x-axis coordinates of the center of the human pupil in the world, and the y-axis coordinates of the center of the human pupil in the world, the coordinates of the center of the human pupil in the world are obtained.

[0030] In this embodiment, for images from camera B, such as Figure 5 As shown in (b), it is only necessary to determine the binarization segmentation threshold of the pupil grayscale based on the grayscale image analysis, and then binarize the image to fit an ellipse to the pupil. Since the view is from below, and the eyelids of the side camera A are obstructed, the bottom camera B will not be obstructed, so the localization of the pupil has a low error. Subsequently, geometric principles are used to determine the target's x-axis and y-axis coordinates in the world.

[0031] S300. Based on the coordinates of the center of the human pupil in the world, the target position is calibrated using a crosshair target to determine the three-dimensional coordinates of the center of the human pupil in world space.

[0032] Specifically, for the z-axis coordinate and x-axis coordinate of the center of the human pupil in the world, the difference between the x-axis and z-axis coordinates of the center of the human pupil in world space is obtained using the principle of similar triangles; for the y-axis coordinate of the center of the human pupil in the world, the position of the y-axis coordinate of the center of the human pupil in world space is obtained using the principle of right triangles, combined with the tilt angle of the first and second cameras; and by combining the y-axis coordinate of the center of the human pupil in world space and the difference between the x-axis and z-axis coordinates of the center of the human pupil in world space, the three-dimensional coordinate position of the center of the human pupil in world space is determined.

[0033] In this embodiment, system installation requires target location calibration, and the calibration steps are as follows: Figure 6 As shown. Using a crosshair target placed at the working distance, with the center of the crosshair target corresponding to the center of the pupil, the camera position is calibrated, as follows. Figure 7 As shown in the image, Figure 7 The vertical direction in (a) is the vertical z-axis in the world coordinate system. Figure 7 In (b), the vertical direction is the forward / backward y-axis of the world coordinate system, and the horizontal direction is the horizontal x-axis of the world coordinate system.

[0034] The geometric principles of its three directions are as follows: Figure 8 As shown, for reference Figure 8 (a) and Figure 8 In (b), based on similar triangles, we can obtain: The coordinates can then be obtained: This formula can be used to obtain the difference between the x-axis and z-axis coordinates of point P relative to point O in the world coordinate system.

[0035] For the y-axis, such as Figure 8 As shown in (c), and Since they are not parallel, similar triangles cannot be directly used to calculate... : Given that the camera's tilt angle is , We can use a right triangle to find: but: By the Law of Sines: This allows us to obtain three-dimensional coordinates in world space, enabling geometric three-dimensional spatial positioning. The x and y coordinates are determined by camera B, and the z coordinate is determined by camera A.

[0036] Based on pupil images captured by a camera positioned from below, this method is unaffected by factors such as eyelashes, eyelids, or tears, making pupil extraction easier and more direct. Figure 9 As shown, these are images taken from multiple different volunteers. The pupils are fully visible, with virtually no obstruction. The red line indicates the pupil diameter. The camera is positioned directly below the system, and the projection of the optical axis of camera B is parallel to the optical axis of the measurement system. This minimizes pupil distortion at the working distance, resulting in a higher accuracy of the pupil diameter obtained from this image compared to that obtained from a binocular vision camera.

[0037] In summary, compared to existing 3D positioning methods, such as projection methods, which require an internal camera to observe the projected image on the human eye, and cannot be installed on large-aperture imaging systems, the projected image is also susceptible to errors due to astigmatism, eyelids, tears, and eyelashes. While binocular vision only requires two external cameras without affecting the internal structure of the measurement system, it requires complex algorithms to extract 3D coordinates, and its calculation of pupil diameter is less accurate or prone to errors, and is easily affected by eyelids and eyelashes.

[0038] This invention adds two cameras external to the measurement system without affecting its internal structure and simplifies the calculation of three-dimensional coordinates. The coordinate axes of a single camera directly correspond to a single direction in the world coordinate system. Based on the corresponding geometric principles, the position difference of the target relative to the world coordinate center is calculated, and a crosshair target is used to calibrate the parallelism between the world coordinate center and the camera. Taking advantage of the characteristic that camera distortion decreases towards the center of the image, a coarse-to-fine positioning logic is designed to achieve accurate positioning of the pupil center. With this invention, the camera provides a more complete view of the pupil, and a reasonable algorithm can be used to calculate the pupil diameter with greater precision.

[0039] Therefore, the embodiments of the present invention have the following advantages compared with the prior art: 1) A direct and simple dual-camera positioning system is proposed, which can still achieve eye alignment even when long eyelashes cause obstruction.

[0040] 2) It has high real-time performance and can track eye tremors during the measurement process in real time, thereby improving measurement accuracy.

[0041] 3) The three-dimensional coordinates can be reconstructed from a geometric perspective, simplifying the positioning algorithm. It can also solve the positioning errors in the x and y directions caused by eyelid occlusion, and the coordinate positioning is independent and is not affected by other cameras.

[0042] 4) Lower cost and computing resources enable higher accuracy of assisted positioning, allowing testers to directly observe the accuracy of the 3D positioning position on the camera.

[0043] 5) The lower camera can capture a more comprehensive view of the pupil and accurately measure the pupil diameter.

[0044] 6) It does not affect the structure of the measurement system and performs three-dimensional spatial positioning of the target.

[0045] Reference Figure 2 A dual-camera-based human eye pupil center positioning system, comprising: The first module 201 is used to combine the first camera and the second camera to construct a three-dimensional eye positioning system; The second module 202 is used to acquire the image of the human eye under test through the three-dimensional positioning system of the eyeball and to locate the position of the human eye pupil through the dynamic segmentation algorithm of pupil symmetry, and to determine the coordinates of the center of the human eye pupil in the world. The third module 203 is used to determine the three-dimensional coordinates of the center of the human pupil in world space by using a crosshair target to calibrate the target position based on the coordinates of the center of the human pupil in the world.

[0046] The content of the above method embodiments is applicable to this system embodiment. The specific functions implemented in this system embodiment are the same as those in the above method embodiments, and the beneficial effects achieved are also the same as those achieved in the above method embodiments.

[0047] The above is a detailed description of the preferred embodiments of the present invention. However, the present invention is not limited to the embodiments described. Those skilled in the art can make various equivalent modifications or substitutions without departing from the spirit of the present invention. All such equivalent modifications or substitutions are included within the scope defined by the claims of this application.

Claims

1. A method for locating the center position of the human eye's pupil based on dual cameras, characterized in that, Includes the following steps: A three-dimensional eye positioning system was constructed by combining the first and second cameras. The eye image of the subject is acquired by a three-dimensional eye positioning system and the position of the pupil is located by a dynamic segmentation algorithm based on pupil symmetry, thus determining the coordinates of the center of the pupil in the world. Based on the coordinates of the center of the human pupil in the world, the target position is determined by using a crosshair target to determine the three-dimensional coordinates of the center of the human pupil in world space.

2. The method for locating the center position of the human eye pupil based on a dual-camera system according to claim 1, characterized in that, The three-dimensional eye positioning system specifically includes a first camera and a second camera, wherein the first camera and the second camera are placed at a spatial right angle, and the optical axes of the first camera and the second camera converge at the position of the human eye to be tested. The first camera is located to the side of the three-dimensional eye positioning system and is tilted, with its optical axis intersecting the human eye. The second camera is located directly below the three-dimensional eye positioning system and is tilted upwards, with its optical axis intersecting the human eye. Both the first camera and the second camera have an illumination source, a short focal length lens, and a CCD sensor.

3. The method for locating the center position of the human eye pupil based on a dual-camera system according to claim 2, characterized in that, The step of acquiring an image of the human eye through a three-dimensional eye positioning system and locating the pupil position using a dynamic segmentation algorithm based on pupil symmetry to determine the coordinates of the pupil center in the world specifically includes: Based on the three-dimensional eye positioning system, the first image of the human eye under test is acquired through the first camera; Adaptive threshold coarse localization processing is performed on the first human eye image to obtain the preliminary pupil position; Based on the ROI pupil center fine localization method, the initial human eye pupil position is processed for secondary localization to determine the pupil center position of the human eye on the side camera. Based on geometric principles and combined with the position of the center of the human eye's pupil on a side camera, the z-axis coordinates of the center of the human eye's pupil in the world are determined. Based on the eye-based three-dimensional positioning system, an image of the second subject's eye is acquired using a second camera; Determine the binarization segmentation threshold of the pupil grayscale, perform binarization processing on the second human eye image to be tested, fit the pupil to an ellipse, and obtain the center position of the pupil of the human eye on the lower camera. Based on geometric principles and combined with the position of the center of the human eye's pupil on the camera below, the x-axis and y-axis coordinates of the center of the human eye's pupil in the world are determined; By combining the z-axis coordinates, x-axis coordinates, and y-axis coordinates of the center of the human pupil in the world, the coordinates of the center of the human pupil in the world can be obtained.

4. The method for locating the center position of the human eye pupil based on a dual-camera system according to claim 3, characterized in that, The step of performing adaptive threshold coarse localization processing on the first human eye image to obtain the preliminary pupil position specifically includes: The first test eye image is preprocessed by median filtering and normalization to obtain the preprocessed first test eye image. The Otsu dynamic threshold segmentation and connected component filtering method is used to perform coarse localization and bounding selection on the preprocessed first human eye image to obtain the preliminary pupil region. Obtain the initial centroid of the pupil region, and use the centroid as the center point to determine the initial position of the human eye pupil within a preset range.

5. The method for locating the center position of the human eye pupil based on a dual-camera system according to claim 4, characterized in that, The method for fine localization of the pupil center based on the ROI performs secondary localization processing on the initial pupil position to determine the center position of the pupil on the side camera. This step specifically includes: The preliminary human eye pupil position is normalized by median filtering to obtain the human eye pupil position image after median filtering and normalization; A grayscale histogram analysis was performed on the human eye pupil position image after median filtering and normalization. The binarization segmentation threshold was determined based on the grayscale of the pupil. Non-pupil regions were then eliminated using a pupil template image to obtain the eliminated human eye pupil position image. Contour extraction is performed on the image of the human eye pupil position after elimination, the pupil is restored, and the center position of the human eye pupil on the side camera is determined.

6. The method for locating the center position of the human eye pupil based on a dual-camera system according to claim 5, characterized in that, The step of determining the three-dimensional coordinates of the center of the human pupil in world space by using a crosshair target to calibrate the target position specifically includes: Using the z-axis coordinate and x-axis coordinate of the center of the human pupil in the world, the difference between the x-axis and z-axis coordinates of the center of the human pupil in world space can be obtained through the principle of similar triangles. For the y-axis coordinates of the center of the human pupil in the world, combined with the tilt angles of the first and second cameras, the y-axis coordinates of the center of the human pupil in world space are obtained through the principle of right triangles. By combining the y-axis coordinate of the center of the human pupil in world space and the difference between the x-axis and z-axis coordinates of the center of the human pupil in world space, the three-dimensional coordinate position of the center of the human pupil in world space can be determined.

7. The method for locating the center position of the human eye pupil based on a dual-camera system according to claim 6, characterized in that, The specific expression for calculating the difference between the x-axis and z-axis coordinates of the center of the human pupil in world space is as follows: In the above formula, This represents the distance between the pupil of the human eye and its center position on the x-axis of the world coordinate system. This represents the distance between the pupil of the human eye and its center position on the z-axis of the world coordinate system. This represents the pixel value indicating the horizontal offset of the human eye's pupil from the image center in the second camera. This represents the pixel value indicating the vertical offset of the human eye's pupil from the image center in the first camera. This indicates the distance between the center position and the camera. This indicates the focal length of the camera lens.

8. The method for locating the center position of the human eye pupil based on a dual-camera system according to claim 7, characterized in that, The specific expression for calculating the y-axis coordinate position of the center of the human pupil in world space is as follows: In the above formula, This represents the y-axis offset of the human eye's pupil from its center position in the world. This represents the offset distance of the human eye's pupil from its center position in the world coordinates along the camera's projection direction. Indicates the camera tilt angle. This represents the angle between the vertical offset distance of the image and the center of the image.

9. A human eye pupil center positioning system based on dual cameras, characterized in that, Includes the following modules: The first module is used to combine the first camera and the second camera to construct a three-dimensional eye positioning system; The second module is used to acquire the image of the human eye under test through the three-dimensional positioning system of the eyeball and to locate the position of the human eye pupil through the dynamic segmentation algorithm of pupil symmetry, and to determine the coordinates of the center of the human eye pupil in the world. The third module is used to determine the three-dimensional coordinates of the center of the human pupil in world space by using a crosshair target to calibrate the target position based on the coordinates of the center of the human pupil in the world.

Images