A device control apparatus
By acquiring eye images and performing perspective correction, and adjusting light output in real time, the problem of uneven illumination caused by eyeball deflection in myopia control equipment has been solved, improving the safety and reliability of the equipment and ensuring the accuracy and safety of illumination.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- BEIJING TONGREN HOSPITAL AFFILIATED TO CAPITAL MEDICAL UNIV
- Filing Date
- 2026-06-08
- Publication Date
- 2026-07-07
AI Technical Summary
During use, existing myopia control devices may cause uneven light distribution due to eyeball deflection, potentially leading to light damage. Furthermore, beam deviation may cause the fovea to be mislit, affecting the efficacy and safety of the treatment.
The image acquisition module acquires eye images, the parameter acquisition module extracts pupil and light spot parameters, and the homography matrix is used for perspective correction. The control module adjusts the light output in real time to ensure that the light spot acts on the preset area.
It improves the operational safety and reliability of lighting equipment, avoids light deviation and insufficient dosage caused by eye position deviation, reduces the risk of potential phototoxic damage, and ensures the specificity and effectiveness of lighting.
Smart Images
Figure CN122342902A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of myopia prevention and control technology, and in particular to a device control apparatus. Background Technology
[0002] In the current field of myopia prevention and control among children and adolescents, low-intensity red light therapy based on the principle of photobiomodulation (PBM) has gradually moved from the laboratory to clinical and home applications. Most existing mainstream devices use semiconductor light-emitting diodes (LEDs) of specific wavelengths (such as 650nm) as the treatment light source, aiming to slow the progression of myopia by improving choroidal blood perfusion and inhibiting excessive axial elongation.
[0003] Medical research indicates that, to balance efficacy and safety, the effective target area for red light therapy is not the fovea centralis, but rather the surrounding ring area. However, the effectiveness and safety of this treatment are highly dependent on the uniformity of energy distribution and the accuracy of the light projection. If the child's eyeball deflects significantly during treatment, the distribution of incident light in the fundus will change. This could not only result in insufficient light dose to the target area due to partial obstruction of the treatment beam by the iris, reducing efficacy, but also potentially lead to misdirection of the ring-shaped light spot to the fovea centralis, causing potential photodamage risks. Summary of the Invention
[0004] The purpose of this application is to provide a device control apparatus to improve the operational safety and reliability of myopia correction devices. The specific technical solution is as follows: This application provides a device control apparatus, including: The image acquisition module is used to acquire an image of the eye formed by the reflection of the ring light output by the lighting device; The parameter acquisition module is used to process the eye image to obtain the user's pupil parameters and the spot parameters corresponding to the ring spot; wherein, the ring spot is a spot formed by the reflection of ring light through the cornea; The parameter correction module is used to perform perspective correction on the pupil parameters and the light spot parameters based on a pre-set homography matrix to obtain the corresponding corrected pupil parameters and corrected light spot parameters. The condition determination module is used to determine whether the ring light meets the preset control conditions based on the corrected pupil parameters and the corrected light spot parameters. The control module is used to control the output state of the light output unit of the lighting device when it is determined that the ring light does not meet the preset control conditions.
[0005] The solution provided in this application acquires pupil and annular light spot parameters by collecting eye reflection images and performs perspective correction using a homography matrix. This eliminates geometric deviations caused by the shooting angle and accurately reflects the true positional relationship of the eye. Furthermore, by determining whether preset control conditions are met based on the corrected geometric parameters, and adjusting the output state of the light output unit in real time if not, it ensures that the annular light output by the illumination device always acts on the preset annular region of the fundus. This avoids illumination deviation, insufficient irradiation dose, or abnormal light spot irradiation of the fovea due to eye position displacement, thereby significantly reducing illumination and improving the operational safety and reliability of the illumination device. Attached Figure Description
[0006] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below.
[0007] Figure 1 This is a schematic diagram of the structure of a device control apparatus provided in an embodiment of this application. Detailed Implementation
[0008] The technical solutions in the embodiments of this application will now be described with reference to the accompanying drawings.
[0009] Figure 1 This is a schematic diagram of the structure of a device control apparatus provided in an embodiment of this application, as shown below. Figure 1 As shown, the device control unit 100 may include: an image acquisition module 110, a parameter acquisition module 120, a parameter correction module 130, a condition determination module 140, and a control module 150.
[0010] Image acquisition module 110 is used to acquire an eye image formed by the reflection of the ring light output by the lighting device.
[0011] In this embodiment, ring light refers to low-intensity light emitted by an illumination device (such as a device for photobiological regulation) and distributed in a circular pattern, used to irradiate the user's eyes for myopia prevention. In this example, the ring light can be ring-shaped red light.
[0012] An eye image refers to a visual image captured by the image acquisition unit of a lighting device, which includes the user's eye and the light spot formed by the reflection of a ring of light through the cornea.
[0013] When using a light-emitting device to irradiate a user for myopia prevention, the device projects a ring of light of a preset wavelength onto the user's eyes. The ring of light is reflected after hitting the user's cornea, forming a corresponding reflected light spot in the user's eye area. The image acquisition module 110 can acquire real-time images of the eye, including the user's eye, pupil, and corneal reflected light spot, thus completing the image data acquisition.
[0014] In this embodiment, the illumination device can use 650nm red light as an imaging light source (in this example, LED (Light-Emitting Diode) light source is preferred), and the system does not need to introduce a supplementary lighting unit, thus avoiding the unknown biological risks that may be caused to myopia prevention and control from the source.
[0015] The parameter acquisition module 120 is used to process the eye image to obtain the user's pupil parameters and the spot parameters corresponding to the ring spot; wherein, the ring spot is a spot formed by the reflection of ring light through the cornea.
[0016] Pupil parameters refer to the point set parameters used to characterize the outline of a user's pupil.
[0017] A ring-shaped light spot refers to a bright spot formed in the eye image after the ring of light emitted by the lighting device is reflected by the surface of the user's cornea. It has an inner contour and an outer contour.
[0018] Spot parameters refer to the point set parameters used to characterize the inner and outer contours of annular spot.
[0019] After obtaining the eye image, the parameter acquisition module 120 sequentially performs grayscale conversion and filtering / denoising preprocessing on the acquired eye image to suppress environmental and image noise. An edge detection algorithm is then used to extract contour information from the image, obtaining the outer edge point set of the user's pupil, as well as the point sets corresponding to the inner and outer contours of the annular light spot. The point set of the pupil contour is extracted as the pupil parameter. The point sets of the inner and outer contours of the annular light spot are also extracted as the light spot parameter. This implementation process will be described in detail in the following embodiments, and will not be repeated here.
[0020] The parameter correction module 130 is used to perform perspective correction on the pupil parameters and the light spot parameters based on a pre-set homography matrix to obtain the corresponding corrected pupil parameters and corrected light spot parameters.
[0021] The homography matrix refers to the inherent parameters obtained by calibrating the lighting equipment before it leaves the factory. It is used to correct perspective distortion of the geometric parameters detected in the image and eliminate positional and dimensional deviations caused by the shooting angle. The implementation process of constructing the homography matrix will be described in detail in the following embodiments, and will not be repeated here.
[0022] Perspective correction refers to the process of converting geometric parameters in an image plane into corresponding parameters in the real physical plane based on the homography matrix.
[0023] Corrected pupil parameters refer to the point set parameters that reflect the outline of the pupil after perspective correction.
[0024] Corrected spot parameters refer to the point set parameters that reflect the inner and outer contours of the spot after perspective correction.
[0025] After obtaining the pupil parameters and light spot parameters, the homography matrix calibrated and stored by the lighting equipment can be called through the parameter correction module 130 to correct the point set parameters of the pupil contour and the point set parameters of the inner and outer contours of the light spot, respectively. Then, the corrected pupil parameters and light spot parameters are determined based on the corrected point sets. The specific correction process will be described in detail in the following embodiments, and will not be repeated here.
[0026] The condition determination module 140 is used to determine whether the ring light meets the preset control conditions based on the corrected pupil parameters and the corrected light spot parameters.
[0027] Preset control conditions refer to the geometric constraints set to ensure the safety and effectiveness of illumination, used to determine whether the position and size of the ring light relative to the pupil are within a safe and effective range.
[0028] After obtaining the corrected pupil parameters and the corrected light spot parameters, the condition determination module 140 can calculate the actual distance between the center point of the corrected pupil and the center point of the corrected light spot based on these parameters. Then, combining the inner ring radius, outer ring radius, and preset safety radius, it determines whether the relative position and size of the pupil region and the annular light spot meet the preset geometric constraints for safe and effective illumination. Finally, based on the determination result, it can be determined whether the current annular light meets the preset control conditions for safe and effective illumination. This implementation process will be described in detail in the following embodiments, and will not be repeated here.
[0029] The control module 150 is used to control the output state of the light output unit of the lighting device when it is determined that the ring light does not meet the preset control conditions.
[0030] The light output unit may include: a ring light source, a prompt sound output device for the lighting equipment, etc.
[0031] Controlling the output state of the light output unit can include: outputting prompts to instruct the user to adjust their eye position, or controlling the light output unit to reduce the output power of the light source corresponding to the ring light, or directly controlling the light output unit to stop outputting red light.
[0032] When it is determined that the ring light does not meet the preset control conditions, indicating that the current eye position is offset and the illumination position deviates from the safe and effective area, the output state of the light output unit of the illumination device can be controlled by the control module 150. For example, it can output prompt information to instruct the user to adjust their eye position. Alternatively, it can immediately perform a light source adjustment operation, either reducing the output power of the light source or directly controlling the light source to stop outputting red light, until the user's eye position returns to a state that allows the ring light to meet the preset control conditions again, and then restore normal illumination output. The implementation process of controlling the light output unit will be described in detail in the following embodiments, and will not be repeated here.
[0033] This embodiment acquires pupil and annular light spot parameters by collecting eye reflection images and performs perspective correction using a homography matrix. This eliminates geometric deviations caused by the shooting angle and accurately reflects the true positional relationship of the eye. Furthermore, by determining whether preset control conditions are met based on the corrected geometric parameters, and adjusting the output state of the light output unit in real time if not, it ensures that the annular light output by the lighting device always acts on the preset annular region of the fundus. This avoids illumination deviation, insufficient irradiation dose, or abnormal light spot irradiation of the fovea due to eye position deviation, thereby significantly reducing illumination and improving the operational safety and reliability of the lighting device.
[0034] Furthermore, from a clinical perspective, by eliminating infrared supplemental lighting and implementing strict projection area monitoring, especially active avoidance of the fovea, this application helps ensure the specificity and effectiveness of the myopia control effect of the illumination device, reducing the risk of ineffective irradiation and potential phototoxic damage, and making the illumination process more in line with medical safety standards. From an engineering and user perspective, the high-precision offline homography calibration based on SVD (Singular Value Decomposition) (described in the following embodiments) ensures that each device has undergone accurate geometric calibration at the factory. This guarantees the measurement consistency and stability of the device throughout its entire life cycle, eliminating the need for any form of daily calibration or professional settings by the user. This simplified operating logic lowers the barrier to entry for home users. At the same time, LED light sources themselves have better spot uniformity and human eye tolerance compared to lasers, and combined with real-time alignment detection and feedback, this helps to create a safe and easy-to-use process for children to maintain long-term.
[0035] Next, the implementation process for obtaining the light spot parameters and pupil parameters will be described in detail.
[0036] In this embodiment, the parameter acquisition module 120 may include: a point set acquisition unit and a spot parameter acquisition unit, wherein, The point set acquisition unit is used to perform edge detection on the eye image using an edge detection algorithm to obtain the first point set of the user's pupil contour, and the second point set of the inner contour and the third point set of the outer contour of the ring-shaped light spot; The light spot parameter acquisition unit is used to use the first point set as the pupil parameter, and the second point set and the third point set as the light spot parameter.
[0037] In this embodiment, the edge detection algorithm is an image processing algorithm used to extract a set of target contour pixels from an eye image, including but not limited to adaptive thresholding and Canny edge detection.
[0038] The first point set refers to the set of points consisting of multiple pixel coordinates on the pupil outline in an eye image.
[0039] The second point set is a set of points consisting of multiple pixel coordinates on the inner contour of the ring-shaped spot in the eye image.
[0040] The third point set is a set of points consisting of multiple pixel coordinates on the outer contour of the ring-shaped spot in the eye image.
[0041] After obtaining the eye image and processing it through grayscale conversion and Gaussian filtering, the point set acquisition unit can use an edge detection algorithm to extract the contour edge pixels in the preprocessed eye image. Pixels belonging to the pupil edge are classified into the first point set, pixels belonging to the inner edge of the ring spot are classified into the second point set, and pixels belonging to the outer edge of the ring spot are classified into the third point set. For example, after edge detection of a single frame of the eye image, a total of 120 continuous edge points in the pupil region are extracted, forming the first point set. Additionally, 100 inner contour edge points and 110 outer contour edge points of the ring spot are extracted, forming the second and third point sets, respectively.
[0042] After obtaining the first point set, the second point set, and the third point set, the spot parameter acquisition unit can use the first point set as the pupil parameter and the second and third point sets as the spot parameters.
[0043] This application embodiment accurately extracts the inner and outer contour point sets of the pupil and the ring-shaped light spot through edge detection, which can stably and accurately obtain the contour parameters of the pupil and the light spot, providing a data foundation for subsequent acquisition of corrected pupil parameters and corrected light spot parameters.
[0044] Next, the implementation process of parameter correction and obtaining the corrected pupil parameters and corrected spot parameters will be described in detail.
[0045] In one implementation of this application, the parameter correction module 130 may include: a point set correction unit, a first fitting unit, a second fitting unit, a first parameter acquisition unit, and a second parameter acquisition unit, wherein, The point set correction unit is used to perform perspective correction on the first point set, the second point set, and the third point set using the homography matrix, respectively, to obtain the corresponding first corrected point set, the second corrected point set, and the third corrected point set; The first fitting unit is used to fit the first set of correction points to obtain the pupil fitting ellipse corresponding to the pupil contour. The second fitting unit is used to fit the second set of correction points and the third set of correction points respectively to obtain the inner contour fitting ellipse of the inner contour and the outer contour fitting ellipse of the outer contour of the annular light spot. The first parameter acquisition unit is used to determine the pupil center point and effective pupil radius of the pupil outline based on the pupil fitting ellipse, and to use the pupil center point and effective pupil radius as the corrected pupil parameters; the effective pupil radius is the minor axis radius of the pupil fitting ellipse. The second parameter acquisition unit is used to determine the center point, inner ring radius, and outer ring radius of the annular light spot based on the inner contour fitting ellipse and the outer contour fitting ellipse, and to use the center point, inner ring radius, and outer ring radius as the correction light spot parameters; the inner ring radius is the minor axis radius of the inner contour fitting ellipse, and the outer ring radius is the major axis radius of the outer contour fitting ellipse.
[0046] In this embodiment, after obtaining the first point set, the second point set, and the third point set, the point set correction unit can use the homography matrix to perform perspective correction on the first point set, the second point set, and the third point set respectively, so as to obtain the first corrected point set corresponding to the first point set, the second corrected point set corresponding to the second point set, and the third corrected point set corresponding to the third point set.
[0047] In practical implementation, due to the fixed tilt angle between the camera and the optical axis of the eye, the set of contour points processed directly on the image plane contains severe perspective distortion. To reduce this error, the coordinate points on the image plane are mapped to a predefined reference plane coordinate system. This transformation is based on homogeneous coordinate representation, which introduces an additional dimension into the two-dimensional coordinates, making the perspective projection process a linear matrix multiplication form, and restoring it to two-dimensional coordinates after normalization.
[0048] For each point in the three contour point sets obtained in the previous step The system performs the following homography transformation: 1. Convert two-dimensional Cartesian coordinates Convert to 3D homogeneous coordinates ; 2. Obtain its homogeneous coordinates on the world reference plane through matrix multiplication. The homogeneous coordinates can be obtained by multiplying the three-dimensional homogeneous coordinates by the homography matrix, that is: ;in, It is a homography matrix; 3. Restore homogeneous coordinates to two-dimensional Cartesian coordinates through normalization operations. : , .
[0049] After performing this transformation on all contour points, the system obtains three new sets of contour points in the world coordinate system that have had perspective distortion eliminated: the first set of correction points, the second set of correction points, and the third set of correction points.
[0050] The pupil fitting ellipse is a standard ellipse obtained by fitting an ellipse to the corrected pupil contour point set. It is used to characterize the shape, position and size of the pupil.
[0051] After obtaining the first set of correction points, the first fitting unit can use the least squares method to perform ellipse fitting on all pixel coordinate points in the first set of correction points, so that the overall deviation between the fitted ellipse and the first set of correction points is minimized. This ellipse is the pupil fitting ellipse corresponding to the pupil contour.
[0052] The inner contour fitting ellipse is a standard ellipse obtained by fitting the inner contour point set of the corrected annular spot to an ellipse. It is used to characterize the shape, position and size of the inner ring of the spot.
[0053] The outer contour fitting ellipse is a standard ellipse obtained by fitting the corrected outer contour point set of the annular spot to an ellipse. It is used to characterize the shape, position and size of the outer ring of the spot.
[0054] After obtaining the second and third correction point sets, the second fitting unit can fit the second and third correction point sets respectively to obtain the inner contour fitting ellipse of the inner contour and the outer contour fitting ellipse of the outer contour of the ring-shaped light spot. Specifically, after obtaining the second and third correction point sets, the same ellipse fitting method as for the pupil contour can be used to perform least-squares ellipse fitting on the second and third correction point sets respectively, obtaining the inner contour fitting ellipse matching the inner contour of the light spot and the outer contour fitting ellipse matching the outer contour of the light spot. That is, by fitting the point sets corresponding to the inner and outer contours of the light spot respectively, two ellipses with high concentricity are obtained, representing the shape and size of the inner and outer rings of the light spot respectively.
[0055] After obtaining the pupil fitting ellipse, the first parameter acquisition unit can determine the pupil center point and effective pupil radius of the pupil contour based on the pupil fitting ellipse, and use the pupil center point and effective pupil radius as the correction pupil parameters, where the effective pupil radius is the minor axis radius of the pupil fitting ellipse.
[0056] After obtaining the inner and outer contour fitting ellipses, the second parameter acquisition unit can determine the center point, inner ring radius, and outer ring radius of the annular light spot based on these ellipses, and use these parameters as correction parameters. The inner ring radius can be the minor axis radius of the inner contour fitting ellipse, and the outer ring radius can be the major axis radius of the outer contour fitting ellipse. In practical applications, the center Cin of the inner ring ellipse and the center Cout of the outer ring ellipse can be calculated separately. To improve robustness, the final center Cs of the annular light spot is defined as the average of the two: Cs = (Cin + Cout) / 2. Simultaneously, the effective inner ring radius Rin (minor axis) (i.e., the inner ring radius) and the effective outer ring radius Rout (major axis) (i.e., the outer ring radius) are calculated separately.
[0057] This application embodiment uses homography matrix for precise perspective correction, eliminating position and size distortion caused by shooting angle, accurately obtaining correction parameters that reflect the true physical state of the eye, and screening core parameters to form a standardized parameter set, providing reliable and accurate data support for subsequent preset control condition judgment, effectively improving the accuracy of illumination position and size detection, and ensuring the safety and effectiveness of illumination.
[0058] Next, the implementation process for calculating the homography matrix will be described in detail.
[0059] In one implementation of this application, the device control apparatus 100 may further include: an image acquisition module, a control point formation module, an equation system construction module, a parameter vector acquisition module, and a homography matrix acquisition module, wherein, The image acquisition module is used to acquire the calibration image corresponding to the calibration plate after placing a calibration plate with known physical dimensions at the calibration position of a simulated human cornea; The control point forming module is used to perform corner detection on the calibration image and obtain the pixel coordinates of each world coordinate point on the calibration board in the calibration image to form multiple sets of mutually matching coordinate control points. The equation system construction module is used to establish linear equations with homography matrix parameters as unknowns based on each set of matched coordinate control points, and to construct corresponding overdetermined linear equation systems based on multiple sets of coordinate control points. The parameter vector acquisition module is used to apply normalization constraints to the parameters based on the scale invariance of the homography matrix, and to perform singular value decomposition on the coefficient matrix of the overdetermined linear equation system using the least squares algorithm. The vector corresponding to the minimum singular value in the decomposition result is taken as the optimal parameter vector. The homography matrix acquisition module is used to reconstruct the optimal parameter vector into the homography matrix.
[0060] In this embodiment, the calibration board refers to a planar board with known physical dimensions (the distance between each corner point and the size of the board body are clearly defined) used to establish the correspondence between world coordinates and image pixel coordinates, providing a benchmark for homography matrix calibration.
[0061] The calibration position of the simulated human cornea refers to the preset placement position of the calibration plate during factory calibration, which is consistent with the distance and angle of the image acquisition module of the light-illuminating device in a real-world usage scenario, simulating the relative position of the human cornea and the light-illuminating device.
[0062] A calibration image is an image captured by the image acquisition module of a lighting device when it is aligned with a calibration board placed at a calibration position. The image contains the complete outline of the calibration board and all its corner points.
[0063] After the lighting equipment is manufactured but before leaving the factory, it undergoes a dedicated factory calibration process. A checkerboard calibration board with preset specifications (e.g., 10×8 grid, 10mm side length per grid) and known physical dimensions is selected and fixedly placed in a preset calibration position. This position simulates the relative distance and angle between the human cornea and the lighting equipment during actual use, ensuring that the calibration scenario is consistent with the actual use scenario. The image acquisition module of the lighting equipment is activated to acquire images of the calibration board, ensuring that the acquired calibration image is clear and complete, fully displaying all corners of the calibration board without blurring, occlusion, excessive distortion, or other issues. After acquisition, the calibration image is stored in the temporary storage unit of the lighting equipment for subsequent processing. For example, a 10×8 grid calibration board is selected, with each grid having a side length of 10mm and clearly defined physical dimensions. It is placed at a calibration position (simulating the position of the human cornea) 30cm away from the image acquisition module and perpendicular to the acquisition lens. The acquisition module is then activated to acquire one clear calibration image. The image clearly shows the 80 corner points of the grid, without obstruction or blurring. This calibration image is stored for subsequent corner point detection.
[0064] Corner detection refers to the image processing operation of processing the acquired calibration image to identify and extract preset corner points (such as the intersection points of a checkerboard calibration board).
[0065] The world coordinate point is the coordinate of a preset corner point on the calibration plate in the real physical space (world coordinate system). Its specific value is predetermined based on the known physical dimensions of the calibration plate and is a fixed reference coordinate.
[0066] The pixel coordinates are the world coordinates on the calibration board, and the corresponding pixel position coordinates in the acquired calibration image are determined by the pixel distribution pattern of the image acquisition module.
[0067] Coordinate control points are matching pairs consisting of a world coordinate point and its corresponding pixel coordinate point, used to construct equations for solving the homography matrix.
[0068] After obtaining the calibration image, the control point formation module can call a preset corner detection algorithm (such as the Harris corner detection algorithm) to process the stored calibration image. First, the image undergoes grayscale conversion and filtering / denoising preprocessing to suppress noise interference with corner detection. Then, the corner detection algorithm identifies and extracts the pixel coordinates of all corners on the calibration board, marking the specific location of each corner in the image. Simultaneously, based on the known physical dimensions of the calibration board, the world coordinates corresponding to each corner are determined (the origin of the world coordinate system is pre-set as the lower left corner of the calibration board, the X-axis is along the horizontal direction of the calibration board, and the Y-axis is along the vertical direction of the calibration board). Finally, the world coordinates of each corner are mapped one-to-one with their corresponding pixel coordinates, forming multiple sets of mutually matching coordinate control points. Each set of control points contains one world coordinate point and one corresponding pixel coordinate point, ensuring accurate matching. For example, corner detection is performed on the acquired calibration image. After preprocessing, the pixel coordinates of 80 corner points are extracted using the Harris algorithm (e.g., the pixel coordinates of one corner point are (250, 320)). Based on the 10mm size of each grid on the calibration board, the world coordinates corresponding to this corner point are determined to be (200mm, 300mm). The world coordinates are matched with the pixel coordinates to form a set of coordinate control points. All 80 corner points are processed in sequence to finally form 80 sets of mutually matching coordinate control points.
[0069] The parameters of the homography matrix are the elements that make up the homography matrix. They are the unknowns in solving the homography matrix and determine the perspective transformation effect of the homography matrix.
[0070] A linear equation is a first-order equation established with the homography matrix parameters as unknowns and based on the correspondence of a set of coordinate control points. It is used to characterize the transformation relationship between world coordinates and pixel coordinates.
[0071] Overdetermined linear equation systems are linear equation systems formed by establishing linear equations from multiple sets of coordinate control points, resulting in a number of equations greater than the number of unknowns. They are used to improve the accuracy of homography matrix solutions.
[0072] The equation construction module can establish linear equations with homography matrix parameters as unknowns for each set of matched coordinate control points (world coordinate points and pixel coordinate points), based on the fundamental relationship of homography transformation. Two linear equations are established for each coordinate control point to ensure that the equations accurately represent the perspective transformation relationship between world coordinates and pixel coordinates. This process is repeated for all coordinate control points to establish corresponding linear equations. Since the number of coordinate control points (multiple sets) far exceeds the number of homography matrix parameters, all linear equations are combined to form an overdetermined linear equation system. This system has more equations than unknowns, effectively reducing noise interference and improving the accuracy of homography matrix solution. For example, 80 sets of coordinate control points can be selected, and two linear equations with homography matrix parameters as unknowns can be established for each set, resulting in 160 linear equations for 80 sets of control points. Since the homography matrix has only 8 degrees of freedom, these 160 linear equations are combined to construct the corresponding overdetermined linear equation system, used to solve for homography matrix parameters, etc.
[0073] Scale invariance is an inherent property of homography matrices, meaning that the product of a homography matrix and any non-zero constant represents the same perspective transformation effect. This leads to uncertainty in its parameters, requiring constraints to determine a unique solution.
[0074] Normalization constraints are constraints imposed on the parameters of a homography matrix to address the parameter uncertainty caused by the scale invariance of the homography matrix, ensuring that the parameters have unique and definite values.
[0075] The least squares algorithm is a numerical method for solving overdetermined linear equations. It obtains the optimal estimate of unknowns by minimizing the equation solving error.
[0076] The coefficient matrix is a matrix composed of the coefficients of unknowns in an overdetermined system of linear equations, and it is the object of singular value decomposition.
[0077] Singular value decomposition (SVD) is a matrix decomposition method that decomposes the coefficient matrix into the product of three specific matrices, and is used to find the optimal solution to the least squares problem.
[0078] The minimum singular value is the singular value with the smallest value among the set of singular values obtained after singular value decomposition. The vector corresponding to this singular value is the optimal solution vector of the overdetermined linear equation system.
[0079] The optimal parameter vector is the homography matrix parameter vector obtained by solving the overdetermined linear equations system, which minimizes the solution error and contains all elements of the homography matrix.
[0080] First, considering the scale invariance of the homography matrix (the product of the homography matrix and any non-zero constant represents the same perspective transformation), the solution to the overdetermined linear equation system is not unique. Therefore, the parameter vector acquisition module can apply a normalization constraint (usually a parameter vector magnitude of 1) to the homography matrix parameters to ensure a unique parameter solution. Then, the coefficient matrix of the overdetermined linear equation system is extracted, and the least squares algorithm is used to perform singular value decomposition on the coefficient matrix, decomposing it into the product of three specific matrices, resulting in a set of singular values (arranged in descending order). The column vector corresponding to the smallest singular value in the decomposition result is selected; this vector is the optimal parameter vector that minimizes the solution error of the overdetermined linear equation system, containing all parameter elements of the homography matrix. For example, applying a normalization constraint of parameter vector magnitude of 1 to the constructed overdetermined linear equation system eliminates the uncertainty of the solution caused by scale invariance. Extract the coefficient matrix of the system of equations and perform singular value decomposition on it to obtain 160 singular values (arranged in descending order). Select the singular value with the smallest value and extract the column vector corresponding to the singular value. This vector is the optimal parameter vector, which contains 9 elements (corresponding to the 9 elements of the homography matrix) and has a vector magnitude of 1.
[0081] The homography matrix acquisition module extracts all elements from the optimal parameter vector and, following the fixed structure of a homography matrix (3×3 matrix), sequentially fills the elements of the parameter vector into the corresponding positions of the matrix, completing the reconstruction from the parameter vector to the 3×3 homography matrix. After reconstruction, the homography matrix is verified to confirm its accurate perspective transformation between world coordinates and pixel coordinates, ensuring the transformation error is within a preset allowable range. Once verified, the homography matrix is permanently stored in the non-volatile memory of the lighting device as an inherent parameter for subsequent perspective correction steps. For example, the extracted optimal parameter vector (containing 9 elements) is reconstructed by sequentially filling the 9 elements into the first, second, and third rows of the 3×3 matrix. This matrix can then be verified by substituting any set of coordinate control points. If the pixel coordinates obtained through matrix transformation have an error of less than 1 pixel compared to the actually acquired pixel coordinates, the verification is successful, and the homography matrix is stored in the non-volatile memory of the lighting device for later use.
[0082] This application embodiment employs a one-time calibration process before shipment, establishing coordinate correspondences using a calibration plate of known dimensions, constructing an overdetermined linear equation system and applying constraints, solving for optimal parameters through singular value decomposition, and finally reconstructing an accurate homography matrix. This ensures that the homography matrix can effectively eliminate distortion caused by the image shooting angle, providing reliable inherent parameter support for subsequent perspective correction of pupil and light spot parameters, improving the accuracy of geometric parameter detection of lighting equipment, and ensuring the safety and effectiveness of myopia prevention lighting.
[0083] Next, the process of determining whether the ring light meets the preset control conditions will be described in detail.
[0084] In one implementation of this application, the condition determination module 140 may include: a distance acquisition unit, a first acquisition unit, a second acquisition unit, and a condition determination unit, wherein, The distance acquisition unit is used to calculate the absolute value of the difference between the center point of the pupil and the center point of the light spot to obtain the target distance; The first acquisition unit is used to calculate the difference between the inner ring radius of the light spot and the preset safety radius to obtain the first radius difference; wherein, the preset safety radius is used to characterize the radius related to the physiological size of the fovea of the human eye, and serves as the radius parameter of a circular region centered on the pupil center; The second acquisition unit is used to calculate the difference between the effective radius of the pupil and the outer ring radius of the light spot to obtain the second radius difference. The condition determination unit is used to determine whether the ring light meets the preset control conditions based on the target distance, the first radius difference, and the second radius difference.
[0085] In this embodiment, the distance acquisition unit first acquires the coordinates of the pupil center point and the light spot center point after perspective correction. Both coordinates reflect the actual physical positions. The coordinate differences between the two center points in the horizontal and vertical directions are calculated separately, and the absolute values of these differences are taken to eliminate the influence of positive and negative signs. Then, based on the absolute values of the horizontal and vertical differences, the straight-line distance between the two center points is calculated using the geometric distance calculation formula. This straight-line distance is the target distance. The magnitude of the target distance directly reflects the degree of offset between the pupil center and the light spot center; the greater the offset, the greater the target distance.
[0086] The first acquisition unit can extract the inner ring radius of the light spot obtained after perspective correction (the minor axis radius of the fitted ellipse of the corrected inner contour), and simultaneously call the preset safety radius pre-stored in the illumination device (this radius is preset according to the physiological size of the fovea of the human eye and is a fixed value). The value of the inner ring radius of the light spot is subtracted from the value of the preset safety radius to obtain the difference, which is the first radius difference. The sign of the first radius difference reflects whether the inner ring area of the light spot can completely accommodate a circular area centered on the pupil center and with the preset safety radius as its radius, providing a basis for determining whether myopia control illumination will mistakenly illuminate the fovea.
[0087] Simultaneously, the second acquisition unit can extract the effective pupil radius (minor axis radius of the pupil-fitted ellipse) and the outer ring radius of the light spot (major axis radius of the outer contour-fitted ellipse) obtained after perspective correction. Subtracting the outer ring radius of the light spot from the effective pupil radius yields the difference, which is the second radius difference. The magnitude of this second radius difference reflects the matching relationship between the outer ring size of the annular light spot and the pupil size, used to determine whether the annular light can cover the target illumination area around the pupil, ensuring sufficient illumination.
[0088] The judgment logic corresponding to the preset control conditions is as follows: the target distance must be simultaneously less than or equal to the first radius difference and the second radius difference. The condition determination unit can substitute the calculated target distance, the first radius difference, and the second radius difference into the preset judgment criteria and compare the magnitudes of the three. If the target distance is less than or equal to the first radius difference and the second radius difference, then the ring light is determined to meet the preset control conditions and can continue normal illumination. If neither of the above conditions is met, then the ring light is determined to not meet the preset control conditions, and the output state of the light output unit needs to be adjusted.
[0089] This application embodiment, by accurately calculating the center offset distance between the pupil and the light spot, the size difference between the light spot and the safe range, and the size difference between the light spot and the pupil, combined with preset judgment criteria, can quickly and accurately determine whether the ring light meets the preset control conditions. This effectively avoids the problem of ring light misilluminating the fovea or insufficient illumination caused by eye position deviation, providing a precise judgment basis for the output status control of the subsequent light output unit, and further improving the operational safety and reliability of the lighting equipment.
[0090] Next, the process of determining that the ring light does not meet the preset control conditions will be described in detail.
[0091] In one implementation of this application, the control module 150 may include: a size relationship acquisition unit and a preset control condition determination unit, wherein, A size relationship acquisition unit is used to acquire the size relationship between the target distance and the difference between the first radius and the difference between the second radius; A preset control condition determination unit is used to determine that the ring light does not meet the preset control conditions when the target distance is greater than the first radius difference or the target distance is greater than the second radius difference.
[0092] In this embodiment, the size relationship acquisition unit can access three parameters—target distance, first radius difference, and second radius difference—calculated earlier, ensuring that all three parameters are accurate values after perspective correction, without calculation errors or missing parameters. Subsequently, two sets of comparison relationships are established: one between the target distance and the first radius difference, and the other between the target distance and the second radius difference. The values are compared one by one: first, the target distance is compared with the first radius difference, and the relationship is recorded (target distance is greater than, equal to, or less than the first radius difference). Then, the target distance is compared with the second radius difference, and the relationship is recorded again. Finally, the two sets of comparison results are integrated to obtain the complete size relationships between the target distance and the first and second radius differences, providing a direct basis for determining subsequent preset control conditions.
[0093] Based on the aforementioned size relationship, preset control conditions can be used to determine whether the unit detection simultaneously satisfies two conditions: "target distance is less than or equal to the first radius difference" and "target distance is less than or equal to the second radius difference." Both conditions must be met simultaneously; neither can be omitted. If, after judgment, either condition is not met or both conditions are not met, it indicates that the projected ring light has exceeded the safe distance and illuminated the fovea centralis, and / or has illuminated outside the pupil, thus failing to achieve the expected myopia control effect.
[0094] In this embodiment, during device operation, the system's real-time alignment evaluation and closed-loop control logic employs a composite alignment strategy that is "center-avoidance and fully effective." Its core objective is to precisely cover the peripheral retinal area using a ring-shaped light spot while strictly avoiding any illumination on the fovea of the macula. To this end, the system first loads the homography matrix H pre-stored in non-volatile memory and then performs real-time alignment evaluation on the pupil center. , center of the ring-shaped light spot Perspective correction is performed using geometric parameters obtained from fitting the inner and outer contours. Through this transformation, the system obtains the corrected pupil center in a distortion-free "world" coordinate system. Center of light spot And the effective pupil radius, which characterizes the size of both. inner ring radius of the light spot With outer ring radius .
[0095] Based on these precise geometric parameters, the control system establishes a set of criteria that include both safety and effectiveness constraints. The first is the safety criterion, the "center avoidance principle," which aims to ensure that the pupil center (corresponding to the fovea in the fundus) is always within the "hollow" safety zone of the ring-shaped light spot. A safety radius related to the physiological dimensions of the fovea is also established. The system needs to ensure that the pupil center... Center of the circle A circular area with radius is completely contained within the area centered on the light spot. Center of the circle Within the hollow inner ring region with radius [missing information].
[0096] This geometric relationship can be transformed into the Euclidean distance between the two center points. Must meet: or .
[0097] At this point, a security threshold can be defined. Alignment must meet the following requirements. The second criterion is the "full-ring pass-through principle," which aims to ensure that the entire ring beam passes completely through the pupil and projects onto the fundus, avoiding insufficient illumination due to obstruction. This geometric relationship can be transformed into the distance between two center points. Must meet: or .
[0098] At this point, a validity threshold can be defined. Alignment must meet the following requirements. .
[0099] Alignment is deemed valid only when both the safety criterion and the validity criterion are met. Therefore, the final control condition is: This means that the ring light does not meet the preset control conditions at this time.
[0100] This application embodiment, by accurately comparing the target distance with the difference between the two radii, clearly sets dual-condition judgment logic, which can quickly and accurately identify abnormal ring light projection scenarios, promptly determine the situation where the preset control conditions are not met, and provide accurate trigger signals for the subsequent output state adjustment of the light output unit. This effectively avoids safety risks and insufficient light intensity caused by abnormal lighting position and size, and further improves the judgment accuracy and operational safety of the lighting equipment.
[0101] Next, the implementation process of output state control of the optical output unit will be described in detail.
[0102] In one implementation of this application, the control module 150 may include: a prompt output unit, a power reduction unit, and a stop output unit, wherein, The prompt output unit is used to output prompt information for eye position adjustment; A power reduction unit is used to control the optical output unit to reduce the output power of the ring light; The stop output unit is used to control the light output unit to stop outputting red light.
[0103] In this embodiment, when it is determined that the ring light does not meet the preset light control conditions, an eye position adjustment prompt can be generated. In this example, the prompt can be a light prompt, an sound prompt, or a combination of light and sound, etc. This embodiment does not limit the form of the eye position adjustment prompt.
[0104] After generating the eye position adjustment prompt information, the prompt output unit can output the eye position adjustment prompt information to prompt the user to adjust the eye position until the user's eye position makes the ring light meet the preset control conditions.
[0105] When the system determines that the ring light does not meet the preset control conditions (such as eye position deviation, abnormal spot position, etc.), and it is not necessary to completely stop the illumination, the light source power adjustment process is initiated. The power reduction unit can send a power reduction command to the light output unit corresponding to the ring light. This command includes a preset low power threshold (this threshold is the minimum effective power set to ensure eye safety, avoiding light damage while retaining a weak indication function). Upon receiving the command, the light output unit immediately adjusts the output current, reducing the current value to match the preset low power threshold, thereby reducing the output power of the light source. After the power reduction, the system monitors the light source output power in real time to ensure it remains stable within the preset low power range, while continuously acquiring eye images and monitoring eye position status, providing a basis for subsequent power restoration or further adjustment. This step does not require completely shutting down the light source; safety protection is achieved solely through power reduction, balancing safety with the convenience of subsequent illumination restoration.
[0106] When the system determines that the ring light does not meet the preset control conditions and there is a high risk of light damage (such as severe spot deviation or excessive eye position deviation), it initiates the light source stop output procedure. The stop output unit can send a stop output command to the light output unit. Upon receiving the command, the light output unit can cut off the output current, stopping power supply to the light source. The light source immediately stops emitting ring light after power failure, completely terminating illumination of the user's eyes. Simultaneously, the system triggers an audible and visual alarm to remind the user or caregiver to adjust their eye position and continuously monitors the eye image. Once the system detects that the eye position has recovered and the ring light once again meets the preset control conditions, it can resume normal light source output according to preset commands, ensuring the safety of the illumination.
[0107] Following the content of the above embodiments, the distance calculated in real time can be... Combined with this dynamically calculated threshold Comparison: like If the alignment is accurate and effective, the LED will maintain the set illumination power.
[0108] like If the light spot's inner ring is about to illuminate the fovea (violating the safety criterion), it is determined that a deviation has occurred. This may mean that the outer ring of the light spot is about to illuminate the fovea (violating the validity criterion), or that the outer ring of the light spot is partially blocked by the pupil, resulting in insufficient illumination (violating the validity criterion). The system will immediately issue an audible and visual alarm, or the control circuit will momentarily reduce or pause the LED output until the user adjusts their eye position to restore the alignment to the required level.
[0109] This application embodiment uses three differentiated output state control methods for the light output unit. When the ring light does not meet the preset control conditions, it can flexibly select the output eye position adjustment prompt information according to the risk level, reduce the power or stop the output. It can not only retain the convenience of light restoration in low-risk scenarios, but also completely avoid the risk of light damage in high-risk scenarios, quickly block unsafe light states, provide dual protection for myopia prevention and control safety, and further improve the operational safety and reliability of the light equipment.
[0110] In the above embodiments, implementation can be achieved entirely or partially through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented entirely or partially as a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid state disk (SSD)).
[0111] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.
[0112] The various embodiments in this specification are described in a related manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions of the method embodiments.
[0113] The above description is merely a preferred embodiment of this application and is not intended to limit the scope of protection of this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application are included within the scope of protection of this application.
Claims
1. A device control system, characterized in that, include: The image acquisition module is used to acquire an image of the eye formed by the reflection of the ring light output by the lighting device; The parameter acquisition module is used to process the eye image to obtain the user's pupil parameters and the spot parameters corresponding to the ring spot; wherein, the ring spot is a spot formed by the reflection of ring light through the cornea; The parameter correction module is used to perform perspective correction on the pupil parameters and the light spot parameters based on a pre-set homography matrix to obtain the corresponding corrected pupil parameters and corrected light spot parameters. The condition determination module is used to determine whether the ring light meets the preset control conditions based on the corrected pupil parameters and the corrected light spot parameters. The control module is used to control the output state of the light output unit of the lighting device when it is determined that the ring light does not meet the preset control conditions.
2. The apparatus according to claim 1, characterized in that, The parameter acquisition module includes: The point set acquisition unit is used to perform edge detection on the eye image using an edge detection algorithm to obtain the first point set of the user's pupil contour, and the second point set of the inner contour and the third point set of the outer contour of the ring-shaped light spot; The light spot parameter acquisition unit is used to use the first point set as the pupil parameter, and the second point set and the third point set as the light spot parameter.
3. The apparatus according to claim 2, characterized in that, The parameter correction module includes: The point set correction unit is used to perform perspective correction on the first point set, the second point set, and the third point set using the homography matrix, respectively, to obtain the corresponding first corrected point set, the second corrected point set, and the third corrected point set; The first fitting unit is used to fit the first set of correction points to obtain the pupil fitting ellipse corresponding to the pupil contour. The second fitting unit is used to fit the second set of correction points and the third set of correction points respectively to obtain the inner contour fitting ellipse of the inner contour and the outer contour fitting ellipse of the outer contour of the annular light spot. The first parameter acquisition unit is used to determine the pupil center point and effective pupil radius of the pupil outline based on the pupil fitting ellipse, and to use the pupil center point and effective pupil radius as the corrected pupil parameters; the effective pupil radius is the minor axis radius of the pupil fitting ellipse. The second parameter acquisition unit is used to determine the center point, inner ring radius, and outer ring radius of the annular light spot based on the inner contour fitting ellipse and the outer contour fitting ellipse, and to use the center point, inner ring radius, and outer ring radius as the correction light spot parameters; the inner ring radius is the minor axis radius of the inner contour fitting ellipse, and the outer ring radius is the major axis radius of the outer contour fitting ellipse.
4. The apparatus according to claim 1, characterized in that, The device further includes: The image acquisition module is used to acquire the calibration image corresponding to the calibration plate after placing a calibration plate with known physical dimensions at the calibration position of a simulated human cornea; The control point forming module is used to perform corner detection on the calibration image and obtain the pixel coordinates of each world coordinate point on the calibration board in the calibration image to form multiple sets of mutually matching coordinate control points. The equation system construction module is used to establish linear equations with homography matrix parameters as unknowns based on each set of matched coordinate control points, and to construct corresponding overdetermined linear equation systems based on multiple sets of coordinate control points. The parameter vector acquisition module is used to apply normalization constraints to the parameters based on the scale invariance of the homography matrix, and to perform singular value decomposition on the coefficient matrix of the overdetermined linear equation system using the least squares algorithm. The vector corresponding to the minimum singular value in the decomposition result is taken as the optimal parameter vector. The homography matrix acquisition module is used to reconstruct the optimal parameter vector into the homography matrix.
5. The apparatus according to claim 3, characterized in that, The condition determination module includes: The distance acquisition unit is used to calculate the absolute value of the difference between the center point of the pupil and the center point of the light spot to obtain the target distance; The first acquisition unit is used to calculate the difference between the inner ring radius of the light spot and the preset safety radius to obtain the first radius difference; wherein, the preset safety radius is used to characterize the radius related to the physiological size of the fovea of the human eye, and serves as the radius parameter of a circular region centered on the pupil center; The second acquisition unit is used to calculate the difference between the effective radius of the pupil and the outer ring radius of the light spot to obtain the second radius difference. The condition determination unit is used to determine whether the ring light meets the preset control conditions based on the target distance, the first radius difference, and the second radius difference.
6. The apparatus according to claim 5, characterized in that, The control module includes: A size relationship acquisition unit is used to acquire the size relationship between the target distance and the difference between the first radius and the difference between the second radius; A preset control condition determination unit is used to determine that the ring light does not meet the preset control conditions when the target distance is greater than the first radius difference or the target distance is greater than the second radius difference.
7. The apparatus according to claim 1, characterized in that, The control module includes: The prompt output unit is used to output prompt information for eye position adjustment; A power reduction unit is used to control the optical output unit to reduce the output power of the ring light; The stop output unit is used to control the light output unit to stop outputting red light.