A subjective optometry prediction method combining wavefront aberration and optical signals

Abstract

The present invention proposes a subjective optometry prediction method that combines wavefront aberration with optical signals. This method measures the wavefront aberration data of the eyeball through a wavefront sensor, obtains the biological parameters of the cornea, lens, and retina through multi-field scanning with a galvanometer system, and after screening high-credibility data points, uses the NURBS algorithm to construct a three-dimensional model of the eyeball with a gradient refractive index; optimizes the corneal stress distribution using finite element analysis, it iteratively corrects the model through deformation energy functions, and dynamically compensates for eye movement errors based on eye tracking data; generates multi-field angle vision chart images through ray tracing simulation, and predicts subjective optometry results in combination with an image quality assessment algorithm. The present invention integrates high-precision modeling, dynamic error suppression, and intelligent simulation technologies to significantly improve the accuracy, personalized adaptability, and dynamic scene adaptability of optometry predictions, providing an efficient and reliable solution for refractive correction.

Description

Technical Field

[0001] The present invention relates to the technical field of medical optometry, and in particular to a subjective optometry prediction method combining wavefront aberration with optical signals. Background Art

[0002] A computer refractometer is a digitized objective optometry device, and its accuracy has a good guiding role in myopia prevention and control and vision correction. However, the current conventional computer refractometer only measures optometry data in the direction of the visual axis, and the optometry data in the direction of the visual axis alone has no direct matching relationship with the subjective optometry data. For vision screening of adolescents, in order to obtain objective optometry and subjective optometry reference data, computer refractometer and eye chart tests must be performed at the same time, which requires a lot of manpower and material resources, and the data is not comprehensive enough. Since the eyeball is not a strictly symmetrical structure, the refractive power of different parts is different, so a refractive power measurement plan for multiple field of view angles and a subjective optometry prediction for the corresponding field of view angles are required.

[0003] Wavefront aberrations are divided into low-order wavefront aberrations and high-order wavefront aberrations. Among them, low-order wavefront aberrations are often used to correct the effects of myopia or hyperopia on patients. However, high-order wavefront aberrations including coma, spherical aberration and trefoil aberration can also affect the patient's visual quality and weaken the visual performance of the eye structure. Accurate measurement of the subject's high-order aberrations can more accurately correct the subject's vision.

[0004] As for the current status of vision screening, most of them require subjective and objective optometry to be carried out simultaneously, which not only consumes a lot of manpower and material resources, but also has a single detection field of view, weak data correlation, and is easily affected by the status during the test. Summary of the Invention

[0005] In response to the above problems, the present invention proposes a subjective optometry prediction system that combines wavefront aberration with optical signals to solve the problem that computer optometry and biometry cannot be performed simultaneously. It also provides optometry data in different field of view angle directions and corresponding subjective optometry prediction data.

[0006] The specific scheme of the present invention is as follows:

[0007] A subjective optometry prediction method combining wavefront aberration with optical signals comprises the following steps:

[0008] S1, the wavefront aberration data of the subject's eye is measured by a wavefront sensor;

[0009] S2, controls the galvanometer system to scan and measure different areas of the subject's eyes to obtain multi-field biological parameters;

[0010] S3, screening the wavefront aberration data and multi-field biological parameters, extracting data points with a credibility greater than a credibility threshold, and constructing a three-dimensional model of the eyeball using the NURBS algorithm;

[0011] S4, assigning gradient refractive index to the three-dimensional model, simulating light paths at different viewing angles, and generating a simulated image of the eye chart corresponding to each viewing angle;

[0012] S5, uses an image quality assessment algorithm to evaluate the simulated image and outputs a predicted subjective optometry result.

[0013] Furthermore, the method also includes calculating corneal stress distribution in combination with finite element analysis, establishing a deformation energy function, and converging the energy function by iteratively optimizing control points and weight parameters to correct the three-dimensional model.

[0014] By combining finite element analysis with calculations of corneal stress distribution and the establishment of a deformation energy function, we can more accurately simulate corneal deformation under actual stress conditions, making the constructed 3D model more realistic and reliable. Optimizing control points and weight parameters to achieve convergence of the energy function can further improve the model's accuracy, providing a more accurate basis for subsequent subjective refraction predictions and enhancing the credibility of subjective refraction results.

[0015] Furthermore, the method also includes collecting eye tracking data in real time, updating the coordinates of NURBS control points through affine transformation, and dynamically compensating for eye movement errors.

[0016] Real-time eye tracking data collection and the use of affine transformations to update the coordinates of NURBS control points dynamically compensate for eye movement errors during the measurement process. This ensures that the constructed 3D eye model maintains high accuracy even when the eye moves involuntarily, avoiding data deviations caused by eye movement, and ensuring that the final subjective refraction prediction results more accurately reflect the subject's actual vision.

[0017] Furthermore, the S1 specifically includes:

[0018] The laser light source emits the detection light to the subject's eyeball, which is reflected by the retina to form a distorted wavefront;

[0019] The distorted wavefront is divided into sub-wavefronts by using a micro-array lens, and the spot offset of the sub-wavefronts is detected by a CCD;

[0020] The Southwell reconstruction model and Zernike polynomials are used to reconstruct the wavefront phase of the spot offset to obtain low-order and high-order aberration parameters.

[0021] Using laser light sources, micro-array lenses, and CCDs, combined with the Southwell reconstruction model and Zernike polynomials, wavefront phase reconstruction can accurately capture low- and high-order aberration parameters of the subject's eye. Compared to traditional measurement methods, this method provides a more comprehensive and detailed description of wavefront aberrations, providing richer and more accurate aberration information for subsequent vision correction and subjective refraction prediction, helping to improve the accuracy and personalization of refraction.

[0022] Furthermore, S2 specifically includes: controlling the incident light angle by galvanometer deflection, collecting optical coherence signals from the center, upper side, lower side, nasal side, and temporal side of the cornea, and extracting the corneal curvature, lens position, and retinal distance parameters of each area.

[0023] By controlling the incident light angle through galvanometer deflection and collecting optical coherence signals from multiple areas of the cornea, more comprehensive multi-field biological parameters can be obtained. This helps to more accurately understand the refractive properties of different parts of the eye, providing more sufficient data support for building an accurate three-dimensional eye model, thereby ensuring that subjective optometry predictions at different field angles are more consistent with the subject's actual visual conditions.

[0024] Furthermore, the method further includes generating a personalized correction solution based on the wavefront aberration data:

[0025] The correction surface is fitted based on the corneal multi-point measurement data to optimize the lens's spherical curvature, cylindrical curvature and astigmatism angle to minimize the wavefront aberration after correction.

[0026] A personalized correction plan is generated based on wavefront aberration data, taking into account multi-point corneal measurement data and fitting the correction curve to optimize lens parameters such as spherical curvature, cylindrical curvature, and astigmatism angle. This enables the design of corrective lenses to better fit the actual structure and aberrations of the subject's eyeball, achieving more precise vision correction, effectively reducing residual wavefront aberrations after correction, improving the subject's visual quality, and reducing visual discomfort after fitting the glasses.

[0027] Furthermore, the S5 specifically includes:

[0028] According to the edge detection algorithm, the letter area of ​​each row in the image is calculated and the ROI is extracted;

[0029] Perform histogram equalization and adaptive binarization processing on the ROI area;

[0030] Assess image quality using the sharpness metric.

[0031] Using an edge detection algorithm to extract the ROI, followed by histogram equalization and adaptive binarization, effectively removes background interference, enhances image contrast and clarity, and makes the letter areas in the simulated image more distinct and legible. Evaluating image quality through the clarity metric more accurately determines the smallest letter a subject can distinguish at different viewing angles, leading to more precise predictions of subjective eye examination results and improved accuracy of eye examination predictions.

[0032] Furthermore, the updating of the NURBS control point coordinates by affine transformation to dynamically compensate for eye movement errors specifically includes the following steps:

[0033] Define the global coordinate system and the local coordinate system;

[0034] Convert the control points into homogeneous coordinates and establish a dynamic change matrix;

[0035] Update control point coordinates in real time to perform motion prediction compensation.

[0036] Defining the global coordinate system and the local coordinate system and establishing a dynamically changing matrix, as well as updating the control point coordinates in real time for motion prediction compensation, can more accurately correct the errors caused by eye movement, further improve the accuracy of constructing the three-dimensional eye model, and ensure the reliability and accuracy of the subjective optometry prediction results.

[0037] Furthermore, the construction of the three-dimensional eye model by the NURBS algorithm specifically includes:

[0038] Extract boundary feature point sets of the anterior and posterior surfaces of the cornea and the anterior and posterior surfaces of the lens;

[0039] The feature point set is spatially gridded, and the grid density is positively correlated with the local curvature;

[0040] The principal curvature direction in the grid is determined by principal component analysis, and initial control points are inserted along the curvature gradient direction;

[0041] Construct a two-variable NURBS surface expression;

[0042] The cumulative chord length method is used to generate non-uniform node vectors to complete the reconstruction of the three-dimensional model of the eyeball surface.

[0043] Extracting boundary feature points from surfaces like the cornea and lens and rasterizing them, along with principal component analysis to determine the principal curvature directions and insert initial control points, allows for more accurate simulation of the complex morphology and curvature of the eye's surface. Model reconstruction using the cumulative chord length method to generate non-uniform node vectors ensures a 3D eye model that more realistically reflects the subject's eye geometry, providing a more accurate model foundation for subsequent light path simulation and subjective refraction prediction, and enhancing the reliability and accuracy of prediction results.

[0044] Compared with existing technologies, the present invention offers the following advantages: By integrating wavefront aberration measurement with multi-field biometric scanning, combining it with a NURBS algorithm to construct a gradient refractive index three-dimensional eye model, and employing finite element analysis to optimize corneal stress distribution, the present invention achieves high-precision eye modeling and biomechanical property restoration. Multi-angle scanning with a galvanometer and dynamic eye tracking compensation effectively suppresses eye movement errors, improving dynamic detection reliability. Ray tracing generates multi-field-angle simulated eye chart images, combining image quality assessment algorithms and Zernike aberration separation technology to predict subjective refraction results and generate personalized correction solutions (such as optimizing spherocylinder parameters and astigmatism angles). Furthermore, by utilizing non-uniform node vector generation, principal component analysis curvature interpolation, and adaptive image processing algorithms, the present invention balances computational efficiency with data robustness. This solution, which deeply integrates objective biometrics with subjective visual simulation, offers significant advantages in refraction accuracy, personalized fit, and dynamic scene adaptability, providing a prediction model for refractive correction that more closely reflects the real visual experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the embodiments of the present drawings or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present drawings. For ordinary technicians in this field, other drawings can be obtained based on the structures shown in these drawings without paying any creative work.

[0046] Figure 1 is a flow chart of the method of the present invention;

[0047] Figure 2 Measure point map for 3D reconstruction;

[0048] Figure 3 Schematic diagram of light deflection in the human eye;

[0049] Figure 4 The predicted subjective optometry image and the corrected subjective optometry image;

[0050] Figure 5 This is a schematic diagram of the system functions. DETAILED DESCRIPTION

[0051] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is described and illustrated below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention. Based on the embodiments provided by the present invention, all other embodiments obtained by those of ordinary skill in the art without creative work are within the scope of protection of the present invention.

[0052] Explanation of terms:

[0053] Subjective optometry: It is an examination achieved through interactive communication between the optometrist and the customer. It requires the customer's subjective feelings to judge the end point of the examination, which corresponds to objective examinations such as retinoscopy and computer optometry.

[0054] Hartmann-Shack wavefront sensor: Hartmann-Shack wavefront sensor, a device used to detect light wavefront.

[0055] Southwell reconstruction model: A mathematical model used in structural mechanics or numerical analysis, particularly in stress analysis or displacement calculation of complex structures. The core idea of ​​this model is to gradually approximate the solution to the problem through iterative relaxation, a classic numerical calculation method.

[0056] Zernike polynomials: These are a complete set of infinite polynomials in two variables that are continuously orthogonal within the unit circle. It's important to note that Zernike polynomials are orthogonal only within the continuous region within the unit circle; they are not orthogonal at discrete coordinates within the unit circle. Aberrations in optical systems are often described using power series expansions. Because Zernike polynomials share the same form as aberration polynomials observed in optical testing, they are often used to describe wavefront characteristics.

[0057] Mooney-Rivlin equation: It is a common equation used to describe elastic materials. It was proposed by Mooney and Rivlin. The Mooney-Rivlin equation can be used to describe materials with nonlinear large deformation behavior. The equation is based on the relationship between the stress tensor and the strain tensor, usually in binomial form.

[0058] The Levenberg-Marquardt algorithm is a method for least squares estimation of regression parameters in nonlinear regression. It was proposed by DW Marquardt in 1963, based on a 1944 paper by K. Levenberg. This method combines the steepest descent method with the linearization method (Taylor series). The steepest descent method is applicable to situations where parameter estimates are far from the optimal value at the beginning of the iteration, while the linearization method, the Gauss-Newton method, is applicable to the later stages of the iteration, when parameter estimates are close to the optimal value. Combining these two methods can quickly find the optimal value.

[0059] like Figure 1 As shown, the present invention provides a subjective optometry prediction method combining wavefront aberration with optical signals, which specifically includes the following steps:

[0060] S1, measuring the wavefront aberration data of the subject's eyeball through a wavefront sensor.

[0061] Specifically, a Hartmann-Shack wavefront sensor is used to measure the irregular wavefront aberration of the eye surface. The wavefront sensor includes a helium-neon laser, an acousto-optic modulator, a neutral density filter, a spatial filter, a CCD camera, a microarray lens, and a polarization beam splitter.

[0062] The wavefront aberration data is measured in the following way:

[0063] The laser light source emits the detection light to the subject's eyeball, which is reflected by the retina to form a distorted wavefront;

[0064] The distorted wavefront is divided into sub-wavefronts by using a micro-array lens, and the spot offset of the sub-wavefronts is detected by a CCD;

[0065] The Southwell reconstruction model and Zernike polynomials are used to reconstruct the wavefront phase of the spot offset to obtain low-order and high-order aberration parameters.

[0066] For an ideal wavefront, a standard dot matrix can be formed on the CCD. However, since the human eye is not an ideal wavefront, the reflected light waves have an inclination angle, and the focus will move accordingly. The amount of movement is proportional to the local slope of the actual wavefront.

[0067] Specifically, the Hartmann-Shack wavefront sensor calculates the offset of R phase points to be estimated, and uses the Southwell reconstruction model to process the slopes in the x and y directions respectively.

[0068] The slope in the x direction is: ;

[0069] Slope in the y direction: ;

[0070] get Equations:

[0071] ;

[0072] ;

[0073] Where W(x,y) represents the wavefront phase distribution;

[0074] Represents the slope in the x direction at position (i, j);

[0075] Represents the slope in the y direction at position (i, j);

[0076] represents the phase value at position (i, j);

[0077] P x 、P y is the pixel spacing, which is used to convert the slope into phase difference.

[0078] Phase reconstruction using Zernike polynomials:

[0079] ;

[0080] in, is the i-th Zernike expansion coefficient, reflecting the contribution of each order of aberration;

[0081] represents the i-th zernike polynomial;

[0082] imax represents the number of polynomials;

[0083] represents the reconstructed wavefront phase distribution.

[0084] Furthermore, the present invention can also generate a corresponding optimal optometry solution based on the wavefront aberration data, and simulate appropriate glasses fitting data through a program.

[0085] Specifically, the formula for simulating the front surface of the eyeglass lens is:

[0086] ;

[0087] Where Z1 represents the first zernike polynomial;

[0088] is the spherical curvature;

[0089] represents the square of the radial distance, indicating that the front surface is a rotationally symmetric optical surface (such as a spherical or aspherical surface);

[0090] Application scenario: lenses used to correct myopia or hyperopia, spherical curvature The same in any direction.

[0091] The formula for simulating the back surface of the eyeglass lens is:

[0092] ;

[0093] Among them, Z2 represents the second zernike polynomial;

[0094] is the cylindrical curvature;

[0095] This formula only contains y 2 , indicating that the rear surface is a non-rotationally symmetric cylinder or torus (such as a cylindrical mirror), and the cylindrical curvature Acts only in the y direction;

[0096] Application scenario: Lenses used to correct astigmatism (collapse), with cylindrical curvature independently controlled in different directions.

[0097] According to the actual astigmatism introduction angle θ and the above parameters, the wavefront aberration of the lens can be obtained, the optimal spherical cylinder can be obtained, and the most suitable lens for the subject can be calculated. 、 And θ, where θ is obtained by converting the (x, y) in the above Cartesian coordinate system to the polar coordinate system, so that the vision correction effect after fitting the glasses reaches the best state.

[0098] Considering that the change in pupil light volume caused by the change in the optometry darkroom environment and the actual outdoor environment will affect the actual eyeglass fitting effect, the present invention adds a conversion function for large pupil Zernike coefficient and small pupil Zernike coefficient. Specifically, when measuring, the pupil diameter is 6mm~7mm and is converted to the Zernike coefficient of a normal pupil of 4mm. The relevant formula is:

[0099] ;

[0100] in, is the phase correction factor, which is related to the phase symmetry of the Zernike polynomials and ensures the consistency of the aberration phase when the pupil is zoomed;

[0101] is the coefficient of the nth-order Zernike polynomial under large pupil;

[0102] is the coefficient of the nth order zernike polynomial under small pupil;

[0103] is the coefficient of the n+2ith order zernike polynomial under large pupil;

[0104] n is the radial order;

[0105] m is the angular order;

[0106] N is the maximum Zernike order corresponding to a large pupil;

[0107] Indicates traversing all high-order aberration orders;

[0108] : a constant determined by n and i;

[0109] : a constant determined by n and j;

[0110] : A constant determined by n and j.

[0111] S2, controls the galvanometer system to scan and measure different areas of the subject's eyes to obtain multi-field biological parameters.

[0112] Specifically, such as Figure 2 As shown in the figure, the galvanometer is driven by voltage to precisely control the incident light angle, and multi-point optical coherence signals are collected from the corneal center (N1), superior side (N2), inferior side (N3), nasal side (N4), and temporal side (N5). Using the central area image and signal as the baseline, parameters such as corneal curvature, lens position, retinal distance, corneal thickness, and lens thickness are synchronously extracted from each area.

[0113] Specifically, the galvanometer system obtains multi-point data from the N2 to N5 areas by scanning the optical signals of the eye axis in five directions, and then calculates three-dimensional structural parameters such as the distance from the posterior surface of the cornea to the anterior surface of the lens, and the distance from the posterior surface of the lens to the retina, and finally fits the eye model that includes the corneal morphology, lens morphology and retinal morphology.

[0114] S3, screen the wavefront aberration data and multi-field biological parameters, extract data points with credibility greater than the credibility threshold, and construct a three-dimensional model of the eyeball using the NURBS algorithm.

[0115] Specifically, first establish the initial reference surface:

[0116] The optical coherence signal at point N1, the center of the pupil, is measured to extract three-dimensional structural parameters such as the corneal starting point and the corneal fitting surface. With point N1 as the origin of the polar coordinate system, the initial reference point in polar coordinates is established by combining the initial angle of the galvanometer and the signal position.

[0117] Establishing an accurate 3D eye model of a subject requires accurate measurement data. Considering the influence of the subject's eye movements and muscle reactions on the measurement data, it is necessary to use the kernel density estimation (KDE) method to extract data points in the measurement data group whose credibility is greater than the credibility threshold. That is, in the coordinate system where the data is stored, the KDE method is used to find points in the distribution data whose probability density estimate is greater than the credibility threshold. The corresponding formula in the polar coordinate system is:

[0118] ;

[0119] in, is the angular coordinate of the i-th sample point;

[0120] is the radial coordinate of the i-th sample point;

[0121] Indicates the point below the polar coordinates The probability density estimate at ;

[0122] g represents the total number of sample points involved in density estimation;

[0123] and It is the radial and angular bandwidth parameter, which controls the range of the kernel function;

[0124] and It is a radial and angular kernel function used to calculate the probability density of data points and filter high-confidence data.

[0125] After screening out high-credibility data, the data is classified according to its position angle, and data points on the same surface but different positions (cornea, lens, retina) are selected for surface fitting.

[0126] The three-dimensional model of the eyeball is constructed using the NURBS algorithm, specifically including:

[0127] Extract boundary feature point sets of the anterior and posterior surfaces of the cornea and the anterior and posterior surfaces of the lens;

[0128] The feature point set is spatially gridded, and the grid density is positively correlated with the local curvature;

[0129] The principal curvature direction in the grid is determined by principal component analysis, and the initial control points are inserted along the curvature gradient direction. .

[0130] Construct a two-variable NURBS surface expression:

[0131] ;

[0132] in, are control points that determine the shape of the surface;

[0133] The three-dimensional coordinate vector of a point on the surface when the parameters are (u, v);

[0134] is the weight parameter, which adjusts the influence of the control point on the surface;

[0135] 、 are the B-spline basis functions in the u and v directions, defined by the node vectors, which control the local continuity of the surface, where p and q are the polynomial degrees in the corresponding directions;

[0136] I is the upper limit of the control point index in the u direction;

[0137] J is the upper limit of the control point index in the v direction.

[0138] The cumulative chord length method is used to generate non-uniform node vectors to complete the reconstruction of the three-dimensional model of the eyeball surface.

[0139] Its node vector Satisfy the non-uniformity condition:

[0140] ;

[0141] in, It is a parameterized ordered point sequence, ensuring that the node distribution matches the point cloud density;

[0142] Represents the Euclidean distance between the i-th point and the i+1-th point;

[0143] represents the sum of the distances between all adjacent pairs of points;

[0144] is the number of adjacent point pairs in the point sequence;

[0145] h represents the maximum value of the control point index.

[0146] Furthermore, the corneal stress distribution is calculated by combining finite element analysis and the deformation energy function is established:

[0147] ;

[0148] Among them, k1 and k2 are the main curvatures, reflecting the local curvature of the cornea;

[0149] A represents the overall area of ​​the corneal surface;

[0150] Represents the total bending energy of the corneal surface;

[0151] Indicates how well the model fits the measured biological data;

[0152] G k is the measured point;

[0153] Predict points for the model, i.e. points on the surface;

[0154] α and β are weight coefficients, α is used to control The effect on the total energy, β is used to control Impact on total energy;

[0155] Optimize control points through iteration and weight parameters Make the energy function converge and correct the three-dimensional model.

[0156] Specifically, the finite element model Mooney-Rivlin constitutive equation was used to define the cornea as a hyperelastic material, and the Levenberg-Marquardt algorithm was used to minimize the energy function until the convergence threshold was less than .

[0157] Accordingly, the optical signals corresponding to the data points in the N2~N5 area are mapped to the polar coordinate system, and combined with the N1 point benchmark data to construct a complete eye model including the irregular curvature of the posterior corneal surface.

[0158] S4, assign gradient refractive index to the three-dimensional model and simulate the light path at different viewing angles, such as Figure 3 As shown, a simulated image of the vision chart corresponding to each field of view angle is generated.

[0159] Specifically, set the myopia simulation and hyperopia simulation, the myopia chart size is , the size of the hyperopia chart is ; In the simulation software, the hyperopia chart is set at 5m in front of the eye under examination, and the myopia chart is set at 25cm in front of the eye under examination.

[0160] S5, uses the image quality assessment algorithm to evaluate the simulated image and outputs the predicted subjective optometry results, such as Figure 4 shown.

[0161] Extract ROI: Calculate the letter area of ​​each row in the image based on the edge detection algorithm, excluding background interference;

[0162] Perform histogram equalization and adaptive binarization processing on the ROI area to eliminate uneven brightness and enhance the contrast of letter boundaries;

[0163] Image quality is evaluated using clarity quantification metrics, image recognizability is evaluated using the structural similarity index (SSIM) or peak signal-to-noise ratio (PSNR), and the predicted subjective visual acuity value is output.

[0164] Furthermore, there is eye movement interference during the measurement process, so the coordinates of the NURBS control points are updated through affine transformation to dynamically compensate for eye movement errors. The specific implementation method is as follows:

[0165] Define the global eye coordinate system and the cornea local coordinate system. The mapping relationship between the two is as follows:

[0166] ;

[0167] Where R is Rotation matrix, calculated from eye tracking data;

[0168] T is The translation vector is determined by the displacement of the pupil center.

[0169] Convert the control points into homogeneous coordinates and establish a dynamic change matrix.

[0170] Specifically, the eye movement changes are obtained based on real-time data, where the rotation components are 、 、 , the translation component is 、 、 , combining the rotation component and the translation component, the dynamic change matrix is ​​established as:

[0171] ;

[0172] Update control point coordinates in real time to perform motion prediction compensation.

[0173] Specifically, the updated control points are:

[0174] ;

[0175] in, For the old control point;

[0176] is the new control point.

[0177] Accordingly, a motion prediction compensation algorithm is added, and the first-order linear extrapolation is used to predict the transformation matrix at the next moment:

[0178] ;

[0179] in, To predict the smoothing factor, adjust the prediction sensitivity;

[0180] Represents the change of affine matrix over time, which is matrix;

[0181] is the affine transformation matrix at time t;

[0182] The preset affine transformation matrix for the next moment.

[0183] like Figure 5 As shown, the specific implementation process of the present invention is as follows:

[0184] After the system is started, it uses dynamic image feedback and position sensors to synchronously track the subject's jaw support movement and measure the three-dimensional movement of the body to locate the position of the binocular pupils; after obtaining the eye image signal, the image center and the pupil center signal are overlapped, and the optical coherence measurement device emits weak light to lock the front surface signal of the cornea. The host computer controls the depth direction movement through the feedback signal until the front surface signal reaches the preset range.

[0185] After the signal positioning of the anterior corneal surface is completed, the optical coherence device stops emitting light, and the near-infrared LED of the wavefront aberration measurement system emits detection light. The diffusely reflected light from the retina forms a distorted wavefront, which is focused by the lens group and the microarray lens and then received by the CCD. The point array offset is analyzed, and a system of differential equations related to the wavefront slope is constructed to solve the slope of each wavefront. The expansion coefficients are calculated based on the combination of Zernike polynomials, and the wavefront aberration parameters are reconstructed.

[0186] After the wavefront measurement is completed, the optical coherence device emits light again to obtain backscattered signals from various surfaces of the eyeball and extract axial reference data (axial length, lens thickness, anterior chamber depth, corneal thickness, etc.); the corneal vertex is used as the origin of the polar coordinate system (r-axis), and the motor positioning is synchronized with the signal from the anterior corneal surface to establish a polar coordinate reference; the scanning galvanometer scans the superior, inferior, nasal, and temporal sides of the cornea in turn, and the polar coordinate depth is calculated by the optical signal distance difference. The angle information is determined by combining the galvanometer angle and the distance from the corneal vertex to the CCD to construct an initial model of the corneal surface; the galvanometer scans along the boundary points in steps, filling the polar coordinate system point cloud data to complete the refined modeling of the cornea; the signals such as corneal thickness and lens thickness are integrated to fit the curves of the posterior corneal surface, lens, and retina; finite element analysis is used to calculate corneal stress in real time, dynamically update surface coordinate points, synchronously fit each tissue structure, and assign refractive index (cornea 1.376, aqueous humor / vitreous 1.336, lens 1.416).

[0187] The host computer generates a personalized three-dimensional model of the eyeball, locates each tissue structure through feature recognition, and completes the refractive index assignment; the built-in simulation algorithm generates multi-angle light focusing images, and calls the myopia (115mm×105mm, 25cm) / hyperopia (787mm×1092mm, 5m) vision chart; performs edge detection and contrast normalization on the simulated image, identifies the minimum distinguishable visual mark based on the structural similarity index (SSIM), and outputs the predicted visual acuity value.

[0188] It should be noted that the present invention is not limited to the above-mentioned embodiments. The above-mentioned embodiments are merely examples, and any embodiments having substantially the same structure and effect as the technical concept within the scope of the technical solution of the present invention are all included in the technical scope of the present invention. In addition, without departing from the scope of the present invention, other embodiments that can be conceived by those skilled in the art and that combine some of the constituent elements in the embodiments are also included in the scope of the present invention.

Claims

1. A subjective optometry prediction method combining wavefront aberration and optical signal, characterized in that: The following steps are involved: S1, the wavefront aberration data of the subject's eye is measured by a wavefront sensor; S2, controls the galvanometer system to scan and measure different areas of the subject's eyes to obtain multi-field biological parameters; S3, screening the wavefront aberration data and multi-field biological parameters, extracting data points with a credibility greater than a credibility threshold, and constructing a three-dimensional model of the eyeball using the NURBS algorithm; The method of constructing the three-dimensional eyeball model by using the NURBS algorithm specifically includes: Extract boundary feature point sets of the anterior and posterior surfaces of the cornea and the anterior and posterior surfaces of the lens; The feature point set is spatially gridded, and the grid density is positively correlated with the local curvature; The principal curvature direction in the grid is determined by principal component analysis, and initial control points are inserted along the curvature gradient direction; Construct a two-variable NURBS surface expression; The cumulative chord length method is used to generate non-uniform node vectors to complete the reconstruction of the three-dimensional model of the eyeball surface; S4, calculate the corneal stress distribution by combining finite element analysis, establish the deformation energy function, and make the energy function converge by iteratively optimizing the control points and weight parameters to correct the three-dimensional model; S5, assigning gradient refractive index to the three-dimensional model, simulating light paths at different viewing angles, and generating a simulated image of the eye chart corresponding to each viewing angle; S6, uses an image quality assessment algorithm to evaluate the simulated image and outputs a predicted subjective optometry result.

2. The subjective optometry prediction method combining wavefront aberration and optical signal according to claim 1, characterized in that: The method also includes collecting eye tracking data in real time, updating the coordinates of NURBS control points through affine transformation, and dynamically compensating for eye movement errors.

3. The subjective optometry prediction method combining wavefront aberration and optical signal according to claim 1, characterized in that: Said S1 specifically includes: The laser light source emits the detection light to the subject's eyeball, which is reflected by the retina to form a distorted wavefront; The distorted wavefront is divided into sub-wavefronts by using a micro-array lens, and the spot offset of the sub-wavefronts is detected by a CCD; The Southwell reconstruction model and Zernike polynomials are used to reconstruct the wavefront phase of the spot offset to obtain low-order and high-order aberration parameters.

4. The subjective optometry prediction method combining wavefront aberration and optical signal according to claim 1, characterized in that: The S2 specifically includes: controlling the incident light angle by galvanometer deflection, collecting optical coherence signals from the center, upper side, lower side, nasal side, and temporal side of the cornea, and extracting the corneal curvature, lens position, and retinal distance parameters of each area.

5. The subjective optometry prediction method combining wavefront aberration and optical signal according to claim 1, characterized in that: The method further includes generating a personalized correction plan based on the wavefront aberration data: The correction surface is fitted based on the corneal multi-point measurement data to optimize the lens's spherical curvature, cylindrical curvature and astigmatism angle to minimize the wavefront aberration after correction.

6. The subjective optometry prediction method combining wavefront aberration with optical signals according to claim 1, characterized in that: The S5 specifically includes: According to the edge detection algorithm, the letter area of ​​each row in the image is calculated and the ROI is extracted; Perform histogram equalization and adaptive binarization processing on the ROI area; Assess image quality using the sharpness metric.

7. The subjective optometry prediction method combining wavefront aberration with optical signals according to claim 2, characterized in that: The method of updating the coordinates of the NURBS control points by affine transformation to dynamically compensate for eye movement errors specifically includes the following steps: Define the global coordinate system and the local coordinate system; Convert the control points into homogeneous coordinates and establish a dynamic change matrix; Update control point coordinates in real time to perform motion prediction compensation.

Images