Eye-tracking interaction method and system based on error-back guidance

By generating dynamic reverse visual cues and leveraging the smooth tracking and error correction capabilities of the human eye, the limitations of precision and accuracy in eye-tracking interaction technology are solved, achieving efficient and natural eye-tracking interaction suitable for extended reality.

CN121349306BActive Publication Date: 2026-11-10SOUTHEAST UNIV
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202511546035.X
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2025-10-28
Publication Date
2026-11-10
Estimated Expiration
2045-10-28

AI Technical Summary

Technical Problem

Existing eye-tracking interaction technologies have limitations in accuracy and precision, making it difficult to accurately select small-sized controls. Furthermore, prolonged use can lead to calibration failure, affecting the usability of the interaction. Existing improvement methods may increase the learning cost for users or disrupt the naturalness of the interaction.

Method used

By generating dynamic reverse visual cues, leveraging the smooth tracking and error correction capabilities of the human eye, the cues' positions are calculated, guiding users to adjust their gaze points to reduce errors. The offset is calculated using a PID controller, a fuzzy logic controller, or a machine learning model.

Benefits of technology

It enables precise one-time selection of micro-sized and densely packed targets, reducing the error rate, shortening task completion time, and making the interaction more natural and reducing cognitive load.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN121349306B_ABST
    Figure CN121349306B_ABST
Patent Text Reader

Abstract

This invention discloses an eye-tracking interaction method and system based on error-back guidance. The eye-tracking interaction method includes the following steps: obtaining the user's current gaze coordinates; calculating the error vector between the gaze coordinates and the center coordinates of controls surrounding the gaze; calculating the prompt position of a visual cue element based on the error vector; and generating a visual cue element at the prompt position. The visual cue element guides the user to adjust their gaze to reduce the error between the gaze coordinates and the center coordinates of controls surrounding the gaze. This invention, by dynamically generating error-back guidance visual cue elements, actively guides the user's gaze for precise reverse correction. This not only helps the user quickly lock onto the target but also effectively compensates for the system errors of the eye tracker and the user's physiological tremors, thereby achieving one-time accurate selection of micro-sized targets and densely packed targets, significantly reducing the error rate and shortening the task completion time.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of human-computer interaction technology, and in particular to a method and system for human-computer interaction using eye-tracking technology. Specifically, it is a method and system that guides the user's gaze by generating dynamic reverse visual cues to improve the accuracy and efficiency of the interaction. Background Technology

[0002] Eye-tracking interaction, as a natural and efficient human-computer interaction method, not only provides an alternative computer input channel for people with hand dysfunction, but also shows increasing application potential in the field of extended reality (XR). The most basic eye-tracking interaction method uses a person's gaze point as a cursor to interact with controls. However, due to the inherent limitations in accuracy and precision of eye-tracking technology, current eye-tracking interaction technologies struggle to accurately select small-sized controls. Furthermore, prolonged use of eye tracking often leads to calibration failures, further reducing the accuracy of eye tracking and affecting the usability of eye-tracking interaction.

[0003] Existing technologies such as dynamic control-to-display ratio, area cursors, and fisheye views can increase the interactive area of ​​controls, allowing for greater tolerance of errors. However, larger control areas can encroach on interface space, reduce information density, and potentially lead to ambiguity due to overlapping interactive areas. To circumvent the accuracy issues of eye tracking, some technologies employ gaze-independent interaction methods, such as blink commands, eye gesture commands, and smooth tracking, but these technologies support a limited number of commands. Furthermore, these technologies require manual definition of the mapping between eye movements and commands, increasing the learning cost for users and compromising the natural "what you see is what you get" nature of eye-tracking interaction. In recent years, with the development of artificial intelligence, machine learning algorithms have been increasingly used in eye-tracking interaction, extracting user eye movement features to identify user interaction intentions. However, their effectiveness depends on the training model method and also places higher demands on device performance.

[0004] Most of these technologies are based on the idea of ​​improving the machine's ability to recognize eye-tracking interaction intentions; however, they neglect the human's ability to actively correct errors. The current bottleneck of gaze-based eye-tracking interaction lies not only in the limited accuracy and precision of eye tracking and the difficulty in recognizing eye-tracking interaction intentions, but also in the fact that the interaction interface cannot naturally guide the human eye to actively correct errors. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide an eye-tracking interaction method and system based on error-back guidance. By dynamically generating reverse visual cues from eye-tracking errors, and utilizing the smooth tracking and error correction capabilities of the human eye, accurate and efficient eye-tracking interaction can be achieved.

[0006] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows: This invention first provides an eye-tracking interaction method based on error-back guidance, comprising the following steps: Get the coordinates of the user's current gaze point; Calculate the error vector between the gaze point coordinates and the center coordinates of the controls surrounding the gaze point; Based on the error vector, the prompting position of the visual cue element is calculated; A visual cue element is generated at the indicated location. The visual cue element guides the user to adjust their gaze point to reduce the error between the gaze point and the center coordinates of controls around the gaze point.

[0007] The step of calculating the cue position of the visual cue element based on the error vector includes: Based on the error vector, calculate the offset; Based on the obtained offset, the cue position of the visual cue element is calculated.

[0008] Based on the obtained offset, in the step of calculating the cue position of the visual cue element, the cue position of the visual cue element is:

[0009] in, for The prompting position of visual cues at all times; for The center coordinates of the time control; for Time offset.

[0010] Time offset The calculation is performed using a PID controller, a fuzzy logic controller, or a machine learning model.

[0011] The PID controller is based on the error vector. Calculate the output offset for:

[0012] in, , , It is an adjustable parameter. Between 0.4 and 1, Between 2 and 4 Between 0 and 0.1. =0.6, =3, =0.02. In the discrete-time system of a computer program, the integral term... By transferring the error in each frame With frame time The differential term is approximated by multiplication and summation. Calculate the current error Error compared to the previous frame The difference divided by the frame time This approximates the algorithm. This ensures stable execution of the algorithm in discrete-time systems.

[0013] The input features of the machine learning model are the error vector, eye movement speed, and acceleration of the current and several historical frames, and the output of the model is the offset. .

[0014] In the step of calculating the error vector between the gaze point coordinates and the center coordinates of the controls surrounding the gaze point, the calculated error vector is:

[0015] in, for Time error vector; for Constantly observe the coordinates of the point; for The center coordinates of the time control.

[0016] In the step of calculating the error vector between the gaze point coordinates and the center coordinates of the controls surrounding the gaze point, the controls surrounding the gaze point are those within a 3° viewing angle of the gaze point. This viewing angle can be converted into pixel length in the screen space coordinate system based on the screen pixel density and the distance between the eye and the screen.

[0017]

[0018] in, Indicates screen pixel density, Indicates the distance between the eyes and the screen. For the size of the viewpoint, For the screen below The pixel length corresponding to the viewpoint.

[0019] This invention also provides an eye-tracking interaction system based on error-back guidance, comprising the following core modules: The eye-tracking module is used to capture and output the user's gaze coordinate data in real time; An error calculation module is used to receive the gaze point coordinates and the position information of the target control from the eye tracking module, and to calculate the error vector between the gaze point coordinates and the position information of the target control. The intelligent control module dynamically calculates the offset and optimal position of the visual cue element based on the received error vector. The visual rendering module generates visual cue elements based on the optimal position of the visual cue elements received from the intelligent control module, and renders the generated visual cue elements onto the display interface.

[0020] The visual style of visual cue elements is configurable and includes, but is not limited to, one or a combination of the following: (1) Cursor form: A simple geometric shape, such as a circle, cross, or arrow.

[0021] (2) Copy form: A semi-transparent, outlined, or color-transformed copy of the target control.

[0022] (3) Guide line form: A line segment or curve that dynamically connects the visual cue element and the target control.

[0023] (4) Particle effect form: A particle effect system attracted or emitted by the position of the visual cue element.

[0024] The eye-tracking module specifically includes a near-infrared camera array, an infrared light source, and related image processing units integrated into the display device. It calculates the pupil-corneal reflection vector and ultimately outputs the two-dimensional gaze point coordinates in the screen coordinate system. For example, this module could be commercial eye tracker hardware and its driver software provided by manufacturers such as Tobii and EyeTribe.

[0025] The error calculation module can logically be a software function running on a central processing unit (CPU) or a graphics processing unit (GPU) that receives the gaze coordinates. and control position As input, and output error vector .

[0026] The intelligent control module, physically executed by the device's CPU or a dedicated microcontroller, logically executes control algorithms to calculate the offset and optimal position of visual cue elements. The control algorithm can be a PID controller, a fuzzy logic controller, or a machine learning model.

[0027] Taking a PID controller as an example, this PID controller is configured to perform the following calculations: ,in , , This is an adjustable control parameter. The offset is... .

[0028] The visual rendering module is typically executed by the GPU.

[0029] The interactive system may also include a parameter configuration module, providing an interface for users or system administrators to adjust parameters in the intelligent control module (such as those of the PID controller). , , (Value) or select different visual cue styles to achieve personalized customization and adaptive optimization of the system.

[0030] Compared with the prior art, the beneficial effects of the present invention are: The interaction is more precise and efficient. Existing technologies, such as traditional "gaze persistence" or "target magnification," are passive or compensatory interactions. They cannot correct the user's inherent gaze bias, requiring the user to adjust themselves or tolerate the error by magnifying the target. This invention dynamically generates visual cues that provide reverse guidance, transforming the user's gaze error into a clear moving target in real time and intuitively, actively guiding their gaze for precise reverse correction. This guidance method not only helps users quickly lock onto the target, but more importantly, it effectively compensates for the system errors of the eye tracker and the user's physiological tremors, thereby achieving one-time accurate selection of micro-targets (minimum size of 0.2° field of view) and densely packed targets, significantly reducing the error rate and shortening the task completion time.

[0031] The interaction method is more natural. Existing technologies often increase the cognitive load on users. For example, "gaze lingering" requires users to actively maintain their gaze, causing mental tension; "target zooming" changes the interface layout and may interfere with the user's visual search. This invention provides a continuous, guided visual feedback. Users do not need to consciously aim; they simply subconsciously follow the movement of visual cues. This "guided-follow" mode is more in line with the automatic processing mechanism of the human sensory-motor system, transforming the operation from conscious aiming to near-unconscious tracking, thereby reducing cognitive load from an ergonomic perspective. Attached Figure Description

[0032] Figure 1 This is a flowchart of the eye-tracking interaction method based on error-back guidance of the present invention; Figure 2 This is a schematic diagram of the module composition of the eye-tracking interaction system based on error-back guidance of the present invention; Figure 3 This is a flowchart of the eye-tracking interaction system based on error-back guidance of the present invention. Figure 4 This is a schematic diagram of the visual cue element style of the eye-tracking interaction system based on error-back guidance of the present invention; Figure 5 This is a diagram showing the interface state and core principle of the eye-tracking interaction system based on error-back guidance of the present invention. Figure 6 This is a comparison chart of the eye movement error response curve of the eye movement interaction system based on error back guidance of the present invention and the traditional eye movement error response curve without error back guidance. Detailed Implementation

[0033] The present invention will now be described in detail with reference to the accompanying drawings and preferred embodiments.

[0034] Example 1 This embodiment provides an eye-tracking interaction method based on error-backward guidance, such as... Figure 1 As shown, it includes the following steps: Get the coordinates of the user's current gaze point; Calculate the error vector between the gaze point coordinates and the center coordinates of the controls surrounding the gaze point; Based on the error vector, the prompting position of the visual cue element is calculated; A visual cue element is generated at the indicated location. The visual cue element guides the user to adjust their gaze point to reduce the error between the gaze point and the center coordinates of controls around the gaze point.

[0035] The step of calculating the cue position of the visual cue element based on the error vector includes: Based on the error vector, the offset is calculated; the offset The calculation is performed using a PID controller, a fuzzy logic controller, or a machine learning model.

[0036] PID controller based on error vector Calculate the output offset for:

[0037] in, , , It is an adjustable parameter. Between 0.4 and 1, Between 2 and 4 Between 0 and 0.1. =0.6, =3, =0.02. In the discrete-time system of a computer program, the integral term... By transferring the error in each frame With frame time The differential term is approximated by multiplication and summation. Calculate the current error Error compared to the previous frame The difference divided by the frame time This approximates the algorithm. This ensures stable execution of the algorithm in discrete-time systems.

[0038] Based on the obtained offset, the cue position of the visual cue element is calculated as follows:

[0039] in, for The prompting position of visual cues at all times; for The center coordinates of the time control; for Time offset.

[0040] The machine learning model takes error vectors, eye-tracking speed, and acceleration from the current frame and several historical frames as input, and outputs the offset. .

[0041] In the step of calculating the error vector between the gaze point coordinates and the center coordinates of the controls surrounding the gaze point, the calculated error vector is:

[0042] in, for Time error vector; for Constantly observe the coordinates of the point; for The center coordinates of the time control.

[0043] In the step of calculating the error vector between the gaze point coordinates and the center coordinates of controls surrounding the gaze point, the controls surrounding the gaze point are those within a 3° viewing angle of the gaze point. This viewing angle can be converted into pixel length in the screen space coordinate system based on the screen pixel density and the distance between the eye and the screen.

[0044]

[0045] in, Indicates screen pixel density, Indicates the distance between the eyes and the screen. For the size of the viewpoint, For the screen below The pixel length corresponding to the viewpoint.

[0046] Example 2 This embodiment provides an error-backward guided eye-tracking interaction system, implemented in a Unity environment that integrates the Tobii eye tracker. For the module composition of the error-backward guided eye-tracking interaction system, see [link to documentation]. Figure 2 ,include: The eye-tracking module is used to capture and output the user's gaze coordinate data in real time; An error calculation module is used to receive the gaze point coordinates and the position information of the target control from the eye tracking module, and to calculate the error vector between the gaze point coordinates and the position information of the target control. The intelligent control module dynamically calculates the offset and optimal position of the visual cue element based on the received error vector. The visual rendering module generates visual cue elements based on the optimal position of the visual cue elements received from the intelligent control module, and renders the generated visual cue elements onto the display interface.

[0047] The visual style of visual cue elements is configurable and includes, but is not limited to, one or a combination of the following forms (see...). Figure 4 ): (1) Cursor form: A simple geometric shape, such as a circle, cross, or arrow.

[0048] (2) Copy form: A semi-transparent, outlined, or color-transformed copy of the target control.

[0049] (3) Guide line form: A line segment or curve that dynamically connects the visual cue element and the target control.

[0050] (4) Particle effect form: A particle effect system attracted or emitted by the position of the visual cue element.

[0051] The eye-tracking module specifically includes a near-infrared camera array, an infrared light source, and related image processing units integrated into the display device. It calculates the pupil-corneal reflection vector and ultimately outputs the two-dimensional gaze point coordinates in the screen coordinate system. For example, this module could be commercial eye tracker hardware and its driver software provided by manufacturers such as Tobii and EyeTribe.

[0052] The error calculation module can logically be a software function running on a central processing unit (CPU) or a graphics processing unit (GPU) that receives the gaze coordinates. and control position As input, and output error vector .

[0053] The intelligent control module, physically executed by the device's CPU or a dedicated microcontroller, logically executes control algorithms to calculate the offset and optimal position of visual cue elements. The control algorithm can be a PID controller, a fuzzy logic controller, or a machine learning model.

[0054] Taking a PID controller as an example, this PID controller is configured to perform the following calculations: ,in , , This is an adjustable control parameter. The offset is... .

[0055] The visual rendering module is typically executed by the GPU.

[0056] The interactive system of this invention may further include a parameter configuration module, providing an interface for users or system administrators to adjust parameters in the intelligent control module (such as those of the PID controller). , , (Value) or select different visual cue styles to achieve personalized customization and adaptive optimization of the system.

[0057] The overall workflow of the eye-tracking interaction system based on error-back guidance of this invention can be found in [link to relevant documentation]. Figure 3 .

[0058] 1. System Initialization and Component Configuration In Unity, add the GazeGhost script component to UI controls (such as buttons) that require error-guided eye-tracking interactions. This component manages the entire guidance process. Additionally, an Image component can be optionally configured as a guide line to visually connect the gaze point to visual cues.

[0059] The following is an example of initializing key parameters: (1) triggerDistance: Set to 3.0. When the distance between the gaze point and the target is less than this threshold (in the form of a field of view, such as 3.0°), the system begins to activate the guidance.

[0060] (2) kp, ki, kd (PID control parameters): set to 0.6, 3 and 0.02 respectively to achieve a balance between response speed and stability.

[0061] (3) gazeSampleWindow: Set to 0.1 seconds to cache the most recent gaze data.

[0062] (4) gazeIDTThreshold (Gaze Distraction Threshold): Set to 1.0 (viewpoint) to determine whether the user is in a gaze state.

[0063] In the Start() method, the system initializes the gaze data queue gazeSamples and prepares the visualization materials.

[0064] 2. Dynamic generation and replacement of visual materials The `UpdateMaterials()` method is the core of achieving the duplicate visual effect. The system dynamically creates and replaces the materials of UI elements at runtime. This method includes the following steps: (1) For all Graphic components (such as Image, TextMeshPro text) under the target UI, the system creates a new instance of the original material.

[0065] (2) The system then finds and applies a special shader with the "(Ghost)" suffix to this new material. This shader is pre-written and can render a preset visual style based on the input parameters.

[0066] (3) The original materials are saved for restoration, and the newly generated visual cue element materials are stored in the ghostMaterials list and applied to UI elements. In this way, any complex UI style (including text and images) can automatically generate its corresponding copy form without the need to pre-create multiple copy objects.

[0067] (4) The visual cue elements can be configured with various visual styles, and specific embodiments include, but are not limited to, the following forms: a) Cursor format: The visual cue element is rendered as a simple geometric shape, such as a crosshair. This format is visually lightweight and suitable for providing concise guidance in dense user interfaces.

[0068] b) Copy Form: The visual cue element is rendered as a semi-transparent or outlined copy of the target control. This is the most intuitive guidance method, allowing the user to clearly see "where to look at the copy".

[0069] c) Guiding Line Form: In addition to the visual cues mentioned above, a dynamic guiding line is added, connecting the user's current gaze point to the visual cue. This form clearly reveals the spatial relationship between the visual cue and the user's current action, providing stronger directionality.

[0070] d) Particle Form: The ghost is represented as a visual area composed of a particle system. These particles can be emitted from or attracted to a preset position of the ghost, forming dynamic effects such as vortices and streamlines. This form, through rich dynamic visual effects, can achieve the guiding function in a more eye-catching and aesthetically pleasing way, and is especially suitable for scenarios that emphasize immersion, such as games and virtual reality.

[0071] 3. Real-time gaze signal processing and gaze state detection In the Update() loop of each frame, the system performs the following steps: (1) Signal Acquisition: Obtain the current screen coordinate input from an eye tracker (such as Tobii) or a mouse (as a backup). The system maintains a gaze sample queue, gazeSamples, within a time window (such as 0.1 seconds) and updates it continuously.

[0072] (2) Gaze state judgment: The I-DT (dispersion-time) algorithm is used to determine whether the user is in a gaze state. Specifically, the dispersion (i.e., half of the sum of the X-coordinate range and the Y-coordinate range) of all gaze points within the sampling window is calculated. When this dispersion is lower than a preset threshold (e.g., 1.0° viewing angle), it is determined that the user is gazing, and the system allows the activation of visual cue elements for guidance.

[0073] 4. Error Calculation and PID Intelligent Control When the user is in a gaze state and enters the trigger range, the system initiates an intelligent control loop. For interface states and core principles, please refer to [link to relevant documentation]. Figure 5 .

[0074] (1) Error Calculation: Calculate the vector error between the current target position and the latest gaze point, i.e. Figure 5 The error vector shown .

[0075] (2) PID control process: The system maintains an error queue of the most recent 3 frames. The discretization implementation of the PID controller is as follows: a) When the error of the first frame is collected, perform proportional (P) control: ghostOffset += kp * errors(0).

[0076] b) When the error of the second frame is collected, perform proportional-integral (PI) control: ghostOffset += kp *(errors(1) - errors(0)) + ki * Time.deltaTime * errors(1).

[0077] c) When the error of the 3rd frame or above is collected, execute the full PID control. The formula for calculating the derivative term (D) is: (kd / Time.deltaTime) * (errors(2) - 2 * errors(1) + errors(0)).

[0078] (3) Noise Filtering: To prevent the differential term from amplifying high-frequency noise, the calculated differential vector d is fed into an unscented Kalman filter (UKF) for smoothing, and the filtered result is then used in the final calculation. This ensures the smoothness and stability of the visual cue element's motion. For the error correction and smoothing convergence effects brought about by PID control and filtering, please refer to [link to relevant documentation]. Figure 6 The illustration.

[0079] 5. Updating and rendering visual cue elements The system uses the ghostOffset output by the PID controller (i.e.) Figure 5 In The system updates and renders the visual cue elements at their optimal positions, along with the configured visual styles.

[0080] (1) Visual cue element position: The visual cue element is rendered at the "target position - ghostOffset". The negative sign here is the key to achieving "reverse guidance". It ensures that the visual cue element appears on the side opposite to the direction of the error, thereby guiding the user to move their gaze point in the opposite direction to eliminate the error.

[0081] (2) Dynamic Transparency: The transparency (ghostAlpha) of visual cues is not fixed. It is derived through an AnimationCurve curve, which takes a normalized distance between visual cues (current visual cue offset distance / trigger distance) as input. This makes the visual cues more prominent when the error is large and fade out when the error is close to zero, providing a smooth visual transition.

[0082] (3) Material parameter passing: In the UpdateGhost() method, the calculated ghostOffset and ghostAlpha are passed to all visual cue element shaders. These shaders then render visual effects with corresponding offsets and opacities in real time based on these parameters.

[0083] (4) Visual guide lines: If guide lines are configured, their length and angle will be dynamically updated to point to the position of the visual cue element, and their transparency will also be synchronized with the visual cue element.

[0084] 6. State Reset When the user's gaze point moves out of the trigger distance or the gaze condition is no longer met, the system automatically clears the error queue, sets the transparency of the visual cue elements to zero, and restores the original material of the UI, thus exiting the guided state without interference.

[0085] Through the above embodiments, this invention successfully combines PID control theory, gaze recognition technology, and real-time shader rendering to create a dynamic, adaptive, and visually integrated eye-tracking guided interaction system. Those skilled in the art can implement this invention on the Unity platform and other similar platforms based on this description.

Claims

1. An eye-tracking interaction method based on error-backward guidance, characterized in that, Includes the following steps: Get the coordinates of the user's current gaze point; Calculate the error vector between the gaze point coordinates and the center coordinates of the controls surrounding the gaze point; Based on the error vector, the prompting position of the visual cue element is calculated; A visual cue element is generated at the cue location. The visual cue element is used to guide the user to adjust the gaze point to reduce the error between the gaze point and the center coordinates of the controls around the gaze point. The step of calculating the cue position of the visual cue element based on the error vector includes: Based on the error vector, calculate the offset; Based on the obtained offset, calculate the cue position of the visual cue element; Based on the obtained offset, in the step of calculating the cue position of the visual cue element, the cue position of the visual cue element is: in, for The prompting position of visual cues at all times; for The center coordinates of the time control; for Time offset The coordinates; In the step of calculating the error vector between the gaze point coordinates and the center coordinates of the controls surrounding the gaze point, the calculated error vector is: in, for Time error vector; for Constantly observe the coordinates of the point; for The center coordinates of the time control.

2. The eye-tracking interaction method based on error-back guidance according to claim 1, characterized in that, Time offset The calculation is performed using a PID controller, a fuzzy logic controller, or a machine learning model.

3. The eye-tracking interaction method based on error-back guidance according to claim 2, characterized in that, The PID controller is based on the error vector. Calculate the output offset for: in, , , It is an adjustable parameter. Between 0.4 and 1, Between 2 and 4 Between 0 and 0.

1.

4. The eye-tracking interaction method based on error-back guidance according to claim 3, characterized in that, =0.6, =3, =0.02。 5. The eye-tracking interaction method based on error-back guidance according to claim 2, characterized in that, The input features of the machine learning model are the error vector, eye movement speed, and acceleration of the current and several historical frames, and the output of the model is the offset. .

6. The eye-tracking interaction method based on error-back guidance according to any one of claims 1-5, characterized in that, In the step of calculating the error vector between the coordinates of the gaze point and the center coordinates of the controls around the gaze point, the controls around the gaze point are those whose distance from the gaze point is within 3° of the viewing angle.

7. An error-backward guided eye-tracking interaction system for implementing the eye-tracking interaction method according to any one of claims 1-6, characterized in that, It includes the following core modules: The eye-tracking module is used to capture and output the user's gaze coordinate data in real time; An error calculation module is used to receive the gaze point coordinates and the position information of the target control from the eye tracking module, and to calculate the error vector between the gaze point coordinates and the position information of the target control. The intelligent control module dynamically calculates the offset and optimal position of the visual cue element based on the received error vector. The visual rendering module generates visual cue elements based on the optimal position of the visual cue elements received from the intelligent control module, and renders the generated visual cue elements onto the display interface.

Citation Information

Patent Citations

  • User fixation point estimation and precision evaluation method based on visual tracking

    CN114816060A

  • Eye movement tracking system, eye movement tracking method, and eye movement tracking program

    CN115916032A