A method for calculating a gaze imaging range of a forward-looking camera based on eye tracking

Abstract

This specification discloses a method for calculating the gaze range of a forward-looking camera based on eye tracking, belonging to the field of eye tracking and machine vision technology. It aims to solve the technical problems in existing technologies, such as insufficient accuracy and susceptibility to interference with a single forward-looking camera, weak visualization capabilities of eye tracking, and the inability to establish a precise correlation between the two, thus failing to accurately calculate the actual gaze range of the user. This provides core technical support for scenarios such as sports training, smart wearables, and human-computer interaction. This method is based on monocular calculation, with a clear process and strong engineering feasibility. Specifically, by pre-calibrating the rigid positional relationship between the forward-looking camera, eye-tracking camera, and goggle structure, a visual field cone is constructed with the center of the monocular eyeball as its vertex. Then, through coordinate transformation, the visual field cone is projected onto the imaging plane of the forward-looking camera. The pixel coordinates of the intersection point of the visual field cone edge and the imaging plane are solved, ultimately determining the actual gaze range of the user.

Description

Technical Field

[0001] This invention relates to the field of eye tracking and machine vision technology, specifically to a method for calculating the gaze imaging range of a forward-looking camera based on eye tracking. Background Technology

[0002] In the field of sports training assistive device technology, the combination of forward-looking cameras and eye-tracking technology has become a research hotspot. However, relying solely on a forward-looking camera to capture images of the user's environment suffers from significant accuracy issues. It is easily affected by external lighting and environmental clutter, making it difficult to accurately pinpoint the area of ​​the user's actual focus and thus failing to meet the high-precision requirements for visual focus analysis in sports training. On the other hand, simple eye-tracking devices can only capture the user's eye movements, offering extremely weak visualization capabilities. They cannot correlate abstract eye-tracking data with the actual environmental scene, making it difficult to intuitively present the user's visual focus, significantly diminishing the application value of eye-tracking data. While combining the two can enable the analysis of the user's visual focus, providing data support for sports training guidance and human-computer interaction optimization, existing technologies still have many shortcomings.

[0003] In existing technologies, the forward-looking camera and eye-tracking module often operate independently. This not only fails to address the core issues mentioned above—the accuracy deficiencies and interference problems of the forward-looking camera alone remain unresolved, but also the weak visualization problem of eye tracking is not solved. Furthermore, the two cannot establish a precise correlation, making it impossible to accurately calculate the actual viewing area of ​​the forward-looking camera. For example, in sports training scenarios, when athletes wear goggles integrating a forward-looking camera and eye-tracking module, the forward-looking camera alone is unable to accurately capture the area of ​​the athlete's attention due to environmental interference, while the abstract data from eye tracking is difficult to visualize intuitively. As a result, existing technologies cannot accurately determine the area of ​​the athlete's actual gaze in the forward field of vision (the image captured by the forward-looking camera), making it difficult to analyze key training data such as the athlete's visual attention allocation and movement prediction logic based on gaze information. This severely limits the application value of eye tracking and forward-looking camera technology in sports training.

[0004] Therefore, there is an urgent need for a technical solution that can establish a precise correlation between eye-tracking cameras and forward-looking cameras, solve the problems of insufficient accuracy, susceptibility to interference, and weak eye-tracking visualization capabilities of forward-looking cameras alone, and achieve accurate calculation and intuitive presentation of the imaging range of the forward-looking camera in actual user gaze. Summary of the Invention

[0005] To address the aforementioned shortcomings in existing technologies—such as insufficient accuracy of simple forward-looking cameras, susceptibility to external interference, inability to accurately pinpoint the user's actual area of ​​focus, extremely weak visualization capabilities of simple eye tracking, inability to correlate abstract eye-tracking data with the actual environmental scene, and the inability to establish a precise correlation between the two data points and accurately calculate the imaging range of the forward-looking camera that reflects the user's actual gaze—this invention aims to provide a fundamental patent-level calculation method that achieves precise spatial matching between the eye-tracking field of view and the imaging range of the forward-looking camera, resolving the pain points of existing technologies and providing core support for upper-layer applications.

[0006] It is important to clarify that the core protection point of this invention lies in the collaborative technical solution of "eye-tracking visual field cone construction - precise coordinate mapping - imaging plane projection". This differs from the shortcomings of existing technologies where eye-tracking and forward-looking cameras work independently without a precise correlation mechanism. The invention clearly distinguishes between necessary and unnecessary technical features: rigid position calibration, human eye cone construction, coordinate transformation, projection to obtain pixel coordinates, and determination of the gaze range are necessary technical features (all are indispensable; the core objective cannot be achieved without any of these features); while binocular superposition and parameter fine-tuning are unnecessary technical features (which can be selected according to the actual scenario and do not affect the realization of the core function). This clearly defines the scope of protection of this invention and avoids problems such as omissions, overly narrow or overly broad protection in subsequent claims.

[0007] Furthermore, this invention addresses the core technical problems of existing technologies by designing corresponding technical solutions for each problem and clarifying the corresponding solutions for each technical step. The specific correspondences are as follows: 1. To address the problem of "the inability to establish a precise correlation between eye movement and the forward-looking camera," a step of "fixing and calibrating a rigid positional relationship" is designed. By presetting the fixed positions of the forward-looking camera, eye movement camera, and eyeball, a foundation is provided for the data correlation between the two. 2. To address the problem of "abstract eye movement data that cannot correspond to the actual field of vision," a step of "constructing the visual cone of the human eye's field of vision" is designed. This transforms abstract eye movement data into a concrete visual field boundary, achieving the visualization of eye movement data. 3. To address the problem of "the inconsistency between the eye movement coordinate system and the forward-looking camera coordinate system, making direct mapping impossible," a step of "coordinate transformation" is designed. This achieves a precise correlation between the two coordinate systems through extrinsic parameter transformation. 4. To address the problem of "the inability to accurately lock the imaging area of ​​the forward-looking camera that the user is actually looking at," steps of "projecting to obtain pixel coordinates" and "determining the gaze imaging range" are designed. This projects the boundary of the human eye's field of vision onto the camera's imaging plane, obtaining a precise gaze area and completely solving the core pain points of existing technologies.

[0008] To achieve the above objectives, the present invention adopts the following technical solution: a method for calculating the gaze imaging range of a forward-looking camera based on eye tracking, the core of which lies in the coordinated cooperation of coordinate mapping and gaze range calculation, linking eye tracking data and forward-looking camera imaging data, and accurately calculating the actual gaze imaging range of the user's forward-looking camera.

[0009] The first step is to fix and calibrate the rigid positional relationships: three fixed relationships are preset to ensure calculation accuracy, namely: 1. The extrinsic parameters of the forward-looking camera are fixed relative to the goggle structure, and its extrinsic parameters include the rotation matrix. Translation vector 1. The eye-tracking camera remains unchanged after installation; 2. The external parameters of the eye-tracking camera are fixed relative to the goggle structure and maintain a fixed relative position with the forward-looking camera to ensure the stability of their collaborative work; 3. The center position of the eyeball is determined by a simple point-gazing calibration method after the user wears the device. After calibration, it is considered a known parameter and remains fixed in subsequent calculations.

[0010] Constructing the visual field cone of the human eye: with the center of the eyeball Using [a specific element] as the vertex, construct a visual field pyramid (viewing cone) that precisely corresponds to the actual visible field of view of the human eye. Define the total horizontal viewing angle. Vertical overall perspective Both can be preset according to human visual characteristics (adapting to the normal field of vision of most users), or finely adjusted according to individual user differences; at the same time, the unit vector of the direction of the gaze center is obtained. This vector, obtained by parsing eye-tracking data collected by an eye-tracking camera, represents the user's current gaze orientation; the four edges of the human eye's visual cone (corresponding to the four corners of the visual field) are calculated based on the following formula, fully covering the boundary of the human eye's visual field: .

[0011] Wherein, is the rightward unit vector horizontally from the eyeball. , is the unit vector pointing vertically downwards from the eyeball. Both and the unit vector in the direction of the line of sight center They are perpendicular to each other and together form a three-dimensional orthogonal basis for the eye coordinate system, ensuring the accuracy of visual cone construction.

[0012] It is important to emphasize that this invention employs a "human eye visual field cone" construction method, rather than the "single gaze point mapping" method commonly used in existing technologies. Its innovation lies in the fact that existing single gaze point mapping can only determine the approximate location of the user's gaze, failing to cover the complete visual field and being susceptible to noise in eye-tracking data, resulting in low accuracy. In contrast, the human eye visual field cone (visual cone) constructed in this invention can completely cover the actual visible visual field boundary. Combined with the unit vector of the gaze center direction and orthogonal basis construction, it can achieve precise definition of the visual field boundary, solving the defects of "incomplete gaze range and insufficient accuracy" in existing technologies. Furthermore, existing technologies lack a complete collaborative process of "rigid calibration - visual cone construction - coordinate transformation - projection calculation," often employing fragmented applications of single steps. This invention organically combines these steps, achieving a precise conversion from eye-tracking data to the imaging range throughout the entire process. This is the core innovation that distinguishes it from existing technologies.

[0013] The third step is coordinate transformation: Since eye-tracking data is acquired based on the eye coordinate system, and the forward-facing camera image is based on its own coordinate system, a coordinate transformation is needed to establish the relationship between the two. This involves transforming the four visual cone directions in the eye coordinate system. By presetting extrinsic parameters (rotation matrix) Translation vector Transform to the forward-looking camera coordinate system using the following formula: After transformation, the direction of the visual cone edge in the forward-looking camera coordinate system is obtained. This step realizes the accurate mapping of the human eye's field of view boundary from the eye coordinate system to the forward-looking camera coordinate system, laying the foundation for subsequent projection calculations.

[0014] Step 4, Projection to Calculate Pixel Coordinates: Set the imaging plane of the forward-looking camera to be located in front of the camera. , The focal length is [value missing]. This imaging plane is perpendicular to the optical axis of the forward-looking camera and is the core plane for the camera to acquire environmental images. The coordinates of the eyeball center in the camera coordinate system are [value missing]. As the starting point of the ray, with Let be the direction of the ray, and each ray corresponds to the extension direction of the human eye's field of vision boundary. The intersection point (i.e., pixel coordinates) of each ray with the imaging plane is calculated using the following formula, thus realizing the projection of the human eye's field of vision boundary onto the camera's image: .

[0015] in, The focal length of the forward-looking camera (corresponding to the horizontal and vertical directions respectively). Principal point coordinates (center pixel coordinates of the camera's imaging plane). for The three components (corresponding to the x, y, and z axes in the forward-looking camera coordinate system, respectively); the coordinates of the four view frustum edges are calculated separately to obtain the four pixel coordinates. These four pixel coordinates correspond to the four boundary corners of the human eye's field of view in the camera's image.

[0016] The fifth step is to determine the gaze imaging range: The area enclosed by the above four pixel coordinates is taken as the precise imaging range of the human eye's field of vision in the forward-facing camera image. This range can be represented by a minimum bounding rectangle, a quadrilateral, or a convex hull, depending on the specific application scenario. Among these, the quadrilateral form best matches the actual shape of the human eye's field of vision, the minimum bounding rectangle facilitates subsequent data processing and application, and the convex hull form can adapt to subtle changes in the field of vision boundary, improving the accuracy of the range representation.

[0017] Key takeaways: By pre-calibrating the rigid positional relationship between the forward-looking camera, eye-tracking camera, and eyeball, a visual field cone is constructed with the eyeball as the vertex. The cone's edge is then projected onto the forward-looking camera's imaging plane via coordinate transformation, allowing for precise calculation of the complete area of ​​the human eye's visual field within the camera's imaging range. This method eliminates the need for complex redundant calculations, achieving accurate mapping solely through core formulas. It boasts strong engineering feasibility and is suitable for practical applications such as wearable devices like goggles.

[0018] It is important to emphasize that this method does not limit the specific type of forward-looking camera or the number of cameras in the eye-tracking acquisition device. It is compatible with monocular / binocular eye-tracking acquisition and various types of forward-looking cameras (including fisheye cameras). Furthermore, it supports recalibration after adjustments to the installation position, significantly expanding the scope of protection and preventing circumvention through simple device replacement or installation position changes, thus aligning with the protection requirements of basic patents. The core calculation process of this method allows for flexible omission of unnecessary steps. For example, in low-precision scenarios, the convex hull fitting step can be omitted, and the gaze range can be directly determined using the minimum bounding rectangle. This simplifies the calculation process while still achieving the core functionality, further expanding the applicable scenarios and strengthening the integrity and flexibility of the protection scope, addressing the challenges posed by existing technologies and potential circumvention schemes. Attached image description:

[0019] Figure 1 A flowchart of an instance's method

[0020] Figure 2 This is a spatial diagram illustrating the relationship between the eyeball, the forward-looking camera, and the eye-tracking camera in one example. Labeling: 1-Forward-looking camera; 2-Eye-tracking camera; 3-Center of the eyeball.

[0021] Figure 3 This is an illustration illustrating the effects of the present invention. Reference numerals: 1 - Imaging range of the forward-looking camera; 2 - Imaging range of the human eye. Detailed Implementation

[0022] The technical solutions of the present invention will be clearly and completely described below with reference to the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0023] Reference Figure 3 This embodiment provides a method for calculating the gaze imaging range of a forward-looking camera based on eye tracking, which is used to specifically illustrate the feasibility of the technical solution of the present invention. Its core process is consistent with the technical solution.

[0024] S1 Calibration Operation (corresponding technical solution support mechanism): Acquire the parameters of the forward-looking camera and relevant data from the eye-tracking acquisition device, and complete the correlation calibration between the eye-tracking acquisition device and the forward-looking camera, as well as the eye calibration. Specific parameters and example details used in this embodiment are as follows: A conventional high-definition camera is selected for the forward-looking camera, and the specific parameters are set as follows: focal length... Imaging resolution Pixels, mounted at a horizontal angle of 0° (at the same height as the optical center of the eye-tracking acquisition device), camera intrinsic parameters include lens distortion coefficient. Principal point coordinates Pixels; the eye-tracking acquisition device uses binocular cameras, with a frame rate of [missing information]. The resolution is Pixel.

[0025] The association parameters are preset association parameters between the eye-tracking acquisition device and the forward-looking camera, used to establish the coordinate association between the two to ensure the accuracy of subsequent calculations; combined with Figure 2 (Diagram showing the positional relationship between the eyeball, the forward-looking camera, and the eye-tracking camera). Figure 2 The relative installation positions of the three devices are clearly shown. The eye-tracking camera is symmetrically positioned in front of the eyeball, focusing on the eyeball at close range. The forward-looking camera is mounted above the eye-tracking camera, facing forward at a distance. Their fields of view do not overlap. The associated parameters must match this positional layout, and the optical distortion correction parameters must be consistent with the intrinsic distortion coefficients of the forward-looking camera. The associated parameters also include the rotation matrices of the forward-looking camera and the eye-tracking camera. Translation vector This is used for subsequent coordinate transformations and is consistent with the preset external parameters in the technical solution.

[0026] Eye calibration example: Following the calibration prompts, the user sequentially gazes at several preset calibration points displayed on the screen (evenly distributed across the screen area). The gazing duration at each calibration point is t. The eye-tracking device simultaneously collects binocular eye-tracking data corresponding to each calibration point. The correlation parameters are corrected using least-squares fitting to eliminate individual pupil size differences (in this example, the user's pupil diameter is...). And minor deviations in equipment installation, while simultaneously calibrating the center of the eyeball. The position, once calibrated, is considered a known parameter to ensure the accuracy of subsequent calculations; this step is applicable to monocular / binocular motion cameras and various forward-looking cameras (including fisheye cameras; if a fisheye camera is used, the optical distortion correction parameter can be adjusted to...). (and installation location adjustment scenarios).

[0027] S2 Data Acquisition (corresponding technical solution and supporting mechanism): The eye-tracking acquisition device acquires the user's eye movement data in real time, and the forward-facing camera acquires environmental imaging data in front of the user in real time. Specific examples and preprocessing procedures are as follows. Data Acquisition Example: The eye-tracking acquisition device acquires the coordinates of the center of the user's pupils and the coordinates of the corneal reflection point in real time. An example of the raw eye movement data acquired is: Pupil center coordinates (left eye: Pixels, right eye: (pixels), corneal reflector coordinates (left eye: Pixels, right eye: (pixels). The preprocessing process includes filtering out abnormal noise in the eye-tracking data and correcting imaging distortions from the forward-looking camera to ensure data reliability and provide accurate data support for subsequent fixation point calculation and cone construction.

[0028] S3 gaze point calculation (corresponding technical solution supporting steps): Based on preprocessed eye movement data and calibration-obtained correlation parameters, the pupil center coordinates and corneal reflector coordinates are extracted. The eye rotation angle is calculated using their relative positional relationship, thereby determining the user's gaze direction and approximate gaze point location. This provides a precise foundation for subsequent coordinate mapping and cone construction. Specific examples and formulas are as follows. The pupil-corneal reflector vector method is used to calculate the eye rotation angle. First, the vector difference between the pupil center and the corneal reflector is calculated using the following formula: ,in The coordinates of the pupil center are Here are the coordinates of the corneal reflection point; then, based on the intrinsic parameters of the eye-tracking camera, the pixel vector is converted into the actual angle, and the angle calculation formula is: ,in For eye-tracking camera pixel size, This refers to the focal length of the eye-tracking camera.

[0029] S4 Human Eye Visual Field Cone Construction (Corresponding to the core step of the technical solution): Based on the eyeball center marked in S1 Using the vertex as the point, construct a human eye field pyramid (visual cone), and use the core parameter settings in the technical solution for the total horizontal field of view. Set to 120°, overall vertical viewing angle Set to 80°, the unit vector of the line of sight center direction is obtained from S3. Simultaneously determine the rightward unit vector of the eyeball. Vertical downward unit vector of the eyeball (the two and) (These four edges are perpendicular to each other, forming a three-dimensional orthogonal basis for the eye coordinate system). Following the core formula for the direction of the visual cone edges in the technical solution, the directions of the four edges are calculated by substituting the above parameters. , corresponding to the four corners of the human eye's field of vision.

[0030] S5 projection to determine pixel coordinates (corresponding to the core step of the technical solution): Setting the imaging plane of the front-view camera to be in front of the camera. place ( (The focal length of the forward-looking camera is consistent with the preset parameters of S1), and the coordinates of the eyeball center in the camera coordinate system. As the starting point of the ray, with For the ray direction, following the projection calculation logic in the technical solution, first use the formula... Calculate the distance coefficient from the ray to the imaging plane. (Verify the validity of the intersection point to ensure it lies within the imaging plane), then substitute it into the pixel coordinate formula. Calculate for each of the four visual cone edges.

[0031] S6 determines the gaze imaging range (corresponding to the core step of the technical solution): the four pixel coordinates calculated in S5 are... The enclosed area serves as the precise imaging range of the human eye's field of vision within the forward-facing camera's view. In this embodiment, it is represented by the minimum bounding rectangle, with its boundary determined by the extreme values ​​of the x-axis and y-axis of four pixel coordinates (the x-axis takes the maximum and minimum values ​​of the four x-coordinates, and the y-axis takes the maximum and minimum values ​​of the four y-coordinates). This facilitates subsequent data processing and use by upper-layer applications. For high-precision scenarios, it can be switched to a quadrilateral or convex hull form to better fit the actual shape of the human eye's field of vision. Figure 3 The illustration clearly shows the effect of the present invention, which uses eye-tracking data to accurately calculate the area of ​​a person's field of vision within the imaging range of the forward-looking camera.

[0032] [Optimization Solution Example: Binocular Overlay]

[0033] This optimization scheme is based on the above basic scheme. On the basis of the basic scheme's monocular calculation, it adds a step of superimposing the imaging range of the left and right eyes. The specific supplementary details are as follows.

[0034] Step 1: Following the basic scheme S1-S6 process, using the left and right eyes as the computational monoculars respectively, complete the calibration, data acquisition, fixation point calculation, cone construction, coordinate transformation, projection calculation, and monocular imaging range determination for each eye, thus obtaining the corresponding 4 pixel coordinates for the left eye. The coordinates of the right eye (4 pixels) Among them, binocular calibration and data acquisition can be performed simultaneously. The eye-tracking acquisition device uses binocular cameras to simultaneously acquire data from both eyes, improving efficiency.

[0035] Left and right eye imaging range superposition: The imaging range of the left eye formed by the four pixel coordinates of the left eye and the imaging range of the right eye formed by the four pixel coordinates of the right eye are superimposed. The intersection area of ​​the two is taken by the image algorithm as the imaging range of the forward-looking camera that the user is actually looking at. The intersection area is the area that both eyes are looking at together, which is closer to the actual visual logic of the human eye and effectively improves the calculation accuracy.

[0036] Additional explanation: The binocular superposition step does not require the addition of complex calculation logic and can directly reuse the calculation results of the basic solution. Only the range superposition and intersection extraction steps are added, which has strong engineering feasibility and does not increase the cost of equipment hardware or the computational burden.

[0037] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for calculating the gaze imaging range of a forward-looking camera based on eye tracking, characterized in that, Includes the following steps: Step 1: Calibrate to obtain the internal parameters of the forward-looking camera, eye-tracking camera, and eyeball, as well as the rigid positional relationship between the forward-looking camera, eye-tracking camera, and eyeball; Step 2: Calculate and obtain the unit vector of the direction of the gaze center based on the eye movement data, and then construct a monocular visual field pyramid (visual cone) with the calibrated monocular eyeball center as the vertex. Step 3: Transform the directions of the four visual cone edges in the monocular coordinate system to the forward-looking camera coordinate system using preset extrinsic parameters (rotation matrix, translation vector) and preset coordinate transformation formula. After transformation, the directions of the four visual cone edges in the forward-looking camera coordinate system are obtained. The four edges correspond to the four corners of the monocular field of view. Step 4: Using the coordinates of the center of the monocular eyeball in the camera coordinate system as the starting point of the ray, and the direction of the monocular visual cone as the ray direction, calculate the distance coefficient from the ray to the imaging plane using a preset formula, solve for the intersection point of each ray with the imaging plane, and obtain 4 pixel coordinates. Step 5: The area enclosed by these 4 pixel coordinates is the imaging range of the forward-facing camera that the user is actually looking at.

2. The method for calculating the gaze imaging range of a forward-looking camera based on eye tracking according to claim 1, characterized in that, This module is used to set the total horizontal and vertical viewing angles with the center of the eyeball as the vertex, and to obtain the unit vectors of the center of vision, the horizontal rightward unit vector of the eyeball, and the vertical downward unit vector of the eyeball; the boundary calculation module, connected to the vision construction module, is used to calculate the direction vectors of the four edges of the visual cone of the human eye, and the calculation formula is as follows: ,in Indicates the overall horizontal viewing angle, Indicates the total vertical viewing angle, This represents the unit vector in the direction of the line of sight center.

3. The method according to claim 1, characterized in that, The coordinate transformation step specifically involves: using a rotation matrix... Transform the four edge direction vectors to the coordinate system of the image acquisition device to obtain the corresponding edge direction vectors in the device coordinate system, i.e.: ,in It is the direction of rotation. These are the directions of the four visual cone edges in a monocular coordinate system.

4. The method for calculating the gaze imaging range of a forward-looking camera based on eye tracking according to claim 1 or 3, characterized in that, The rigid positional relationship between the forward-looking camera, the eye-tracking camera, and the eyeball mentioned in step one is precisely determined through the calibration process. Specifically, the eye-tracking camera collects eyeball feature points, and the forward-looking camera collects calibration plate feature points. Combining the internal parameters of the two, the relative positional parameters between the forward-looking camera coordinate system, the eye-tracking camera coordinate system, and the eyeball coordinate system are calculated. The relative positional parameters include translation parameters and rotation parameters. The translation parameters and rotation parameters correspond to the translation vector T and rotation matrix R in claim 3, respectively, forming a rigid constraint relationship among the three. This rigid positional relationship remains unchanged during the calculation of the imaging range.

5. The method for calculating the gaze imaging range of a forward-looking camera based on eye tracking according to claim 1, characterized in that, The four pixel coordinates mentioned in step four (i.e., the four points within the field of view of the forward-looking camera image) are obtained through the following projection calculation steps: First, using the formula... Calculate the distance coefficient from the ray to the imaging plane. (Verify the validity of the intersection point to ensure it lies within the imaging plane), then substitute it into the pixel coordinate formula. ,in Focal length in the horizontal direction The focal length in the vertical direction of the forward-looking camera. Use the coordinates of the principal point. Calculate the coordinates for each of the four view frustum edges.

6. The method for calculating the gaze imaging range of a forward-looking camera based on eye tracking according to claim 1, characterized in that, To further improve computational accuracy and better align with the binocular vision characteristics of the human eye, a binocular superposition optimization method is adopted, specifically including:

1. Using the center of the left eye and the center of the right eye as vertices respectively, calculate the imaging range of the left eye and the imaging range of the right eye respectively according to the method of the second to fourth steps in claim 1, and obtain 4 pixel coordinates of the left eye and 4 pixel coordinates of the right eye respectively. These pixel coordinates correspond to the projection positions of the four corners of the left and right eye visual fields on the imaging plane, respectively, and are not the imaging points of the human eye itself, and are used to define the boundaries of the respective fixation visual fields of the left and right eyes.

2. The imaging range of the left eye and the imaging range of the right eye are superimposed, and the intersection area of ​​the two is taken as the imaging range of the forward-looking camera that the user is actually looking at.

Images