A Method and System for Designing Globally Tunable Optical Surfaces Based on Nonlinear Optimization
By constructing a comprehensive evaluation system for imaging and stray light suppression using a nonlinear optimization method, the problem of the separation between imaging and stray light suppression in the design of optical surfaces for implantable ocular endoscopes is solved, enabling rapid and accurate optical surface design that meets the physiological and optical needs of the human eye.
Patent Information
- Application Number
- CN202611120728.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-07-27
- Publication Date
- 2026-11-17
AI Technical Summary
In existing optical surface designs for implantable ophthalmic endoscopes, imaging performance evaluation and stray light suppression performance evaluation are separated, making it impossible to simultaneously match the dual hard design indicators of clear imaging across the entire field of view and uniform and stable illumination across the entire aperture. Furthermore, traditional design methods cannot achieve integrated data description, resulting in long design time and high error risk.
A globally tunable optical surface design method based on nonlinear optimization is adopted. By constructing a comprehensive performance evaluation method that combines imaging evaluation and stray light suppression evaluation, a correlation is established using weighting factors. The surface coefficient vector is repeatedly adjusted to generate complete surface data containing discrete height distribution and continuous surface shape description. A closed-loop verification process with automatic weight correction is set up.
It achieves dual design specification matching of optical surfaces across the entire domain, shortens the iteration convergence cycle, improves data matching accuracy, and ensures the accuracy of optical simulation analysis and lens processing.
Smart Images

Figure CN123116368A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of numerical design technology for optical surfaces, and in particular to a method and system for designing globally tunable optical surfaces based on nonlinear optimization. Background Technology
[0002] In existing optical surface design systems for implantable ophthalmic lenses, the imaging performance evaluation system and the stray light suppression performance evaluation system operate independently and in isolation. There is no correlation or balancing mechanism between the two evaluation logics. When iteratively optimizing the surface, only imaging or stray light suppression indicators can be used as the core evaluation criteria. It is impossible to simultaneously balance and control the two types of constraint boundaries—wavefront aberration across the entire field of view and stray light suppression across the entire radial aperture—during the optimization process. The surface structure optimized in this way is prone to exhibiting a phenomenon where one dimension of the indicator meets the design tolerance range, while another dimension deviates from the preset constraint boundary. This makes it impossible to simultaneously match the dual hard design requirements of implantable ophthalmic lenses—clear imaging over a wide field of view and uniform and stable illumination across the entire aperture—resulting in a surface that is difficult to adapt to the physiological optical needs of the human eye.
[0003] Traditional fully adjustable optical surface optimization processes employ fixed weight ratios to establish comprehensive evaluation standards. The entire design process lacks a closed-loop verification and correction mechanism. If, after joint verification of the complete surface data, it does not fully meet the dual constraints of imaging and illumination, the evaluation weight parameters must be manually modified, and the entire process of building the evaluation system, iterative optimization of surface coefficients, and generating the complete surface must be executed from scratch. Multiple iterations significantly lengthen the overall design time, resulting in slow convergence of the surface to a compliant standard shape. Furthermore, traditional design methods cannot simultaneously output two types of integrated surface data: discrete height point distribution and a fully continuous surface description. These two types of data must be calculated and generated independently, leading to data discrepancies. This directly reduces the data matching accuracy in subsequent lens CNC machining and optical simulation verification, increasing the risk of errors in the molding and processing of ophthalmic endoscope surfaces. Summary of the Invention
[0004] This invention provides a method and system for designing globally tunable optical surfaces based on nonlinear optimization, in order to solve the problems mentioned in the background art.
[0005] To achieve the above objectives, the present invention provides a globally tunable optical surface design method based on nonlinear optimization, comprising: Pt.1. Obtain the optical design input parameters for implantable ophthalmic endoscopes, generate surface design constraints from the optical design input parameters, and construct the initial surface shape representation basis and the initial vector of surface coefficients. Pt.2. Based on the surface design constraints, establish a comprehensive performance evaluation method that includes an imaging evaluation part and a stray light suppression evaluation part, wherein the imaging evaluation part and the stray light suppression evaluation part are associated through a weighting factor; Pt.3. Based on the comprehensive performance evaluation method, the initial vector of the surface coefficients is repeatedly adjusted until the stopping condition is met and the optimal surface coefficient vector is output. Each adjustment is determined by constructing local approximation conditions to determine the adjustment direction and adjustment range. Pt.4. Based on the combination of the optimal surface coefficient vector and the initial surface shape representation basis, generate complete surface shape data containing discrete height distribution and continuous surface shape description; Pt.5. Based on the surface design constraints, perform joint verification on the complete surface data. If the verification passes, confirm and output the complete surface data. If the verification fails, adjust the weighting factor and repeat Pt.2 to Pt.4 until the verification passes.
[0006] In a preferred embodiment, the step of obtaining optical design input parameters for an implantable ocular endoscope, generating surface design constraints from the optical design input parameters, and constructing an initial surface shape representation basis and an initial vector of surface coefficients includes: Extract field of view range data from optical design input parameters, determine the effective field of view region based on the field of view range data, and divide the field of view position sequence within the effective field of view region; Based on the ideal imaging requirements corresponding to each field of view position in the field of view position sequence, a corresponding wavefront aberration tolerance boundary value is configured for each field of view position, and the wavefront aberration tolerance boundary values corresponding to all field of view positions are combined into an imaging wavefront constraint boundary. Extract stray light suppression aperture distribution requirements from the optical design input parameters, determine the effective stray light suppression evaluation aperture region based on the stray light suppression aperture distribution requirements, and divide the radial position sequence within the effective stray light suppression evaluation aperture region; Based on the stray light suppression index target value corresponding to each radial position in the radial position sequence, configure the stray light suppression index deviation tolerance range corresponding to each radial position, and combine the stray light suppression index deviation tolerance ranges corresponding to all radial positions into stray light suppression index constraint boundaries. The imaging wavefront constraint boundary and the stray light suppression index constraint boundary are merged to generate the surface design constraint conditions.
[0007] In a preferred embodiment, the step of obtaining optical design input parameters for an implantable ocular endoscope, generating surface design constraints from the optical design input parameters, and constructing an initial surface shape representation basis and an initial vector of surface coefficients includes: The effective aperture size is extracted from the optical design input parameters. The surface shape expression space range is determined based on the effective aperture size. A family of basis functions is selected within the surface shape expression space range. The basis components in the family of basis functions have different spatial variation scales. Arrange the basis components in the basis function family from low to high according to the spatial variation scale, and combine all the arranged basis components into an initial surface shape expression basis; Each base component in the initial surface shape expression base is assigned a corresponding initial weight value, and all initial weight values are arranged in the order of the base components in the initial surface shape expression base to generate an initial weight value sequence. The initial surface shape representation basis is associated with and stored with the initial weight value sequence to generate an initial vector of surface coefficients.
[0008] In a preferred embodiment, the step of establishing a comprehensive performance evaluation method based on the surface design constraints, comprising an imaging evaluation component and a stray light suppression evaluation component, wherein the imaging evaluation component and the stray light suppression evaluation component are correlated through a weighting factor, includes: The imaging wavefront constraint boundary is read from the surface design constraint conditions. The allowable wavefront deviation range corresponding to each field of view position is determined according to the imaging wavefront constraint boundary. The spatial surface shape corresponding to the surface coefficient vector is compared with the ideal wavefront shape at each field of view position. The wavefront deviation degree value at each field of view position is generated according to the comparison result. The imaging evaluation result is generated according to the wavefront deviation degree value at all field of view positions. The stray light suppression index constraint boundary is read from the surface design constraints. The allowable stray light suppression index deviation range corresponding to each radial position is determined according to the stray light suppression index constraint boundary. The surface orientation of the spatial surface shape corresponding to the surface coefficient vector at each radial position is compared with the stray light suppression target orientation. The stray light suppression index deviation value at each radial position is generated according to the comparison result. The stray light suppression evaluation result is generated according to the stray light suppression index deviation values at all radial positions. A preset weighting factor is invoked, which is used to adjust the proportion of the imaging evaluation result and the stray light suppression evaluation result in the comprehensive evaluation. The imaging evaluation result and the stray light suppression evaluation result are correlated according to the weighting factor to generate a comprehensive evaluation value. A comprehensive performance evaluation method is established based on the correspondence between the comprehensive evaluation value and the surface coefficient vector.
[0009] In a preferred embodiment, the initial vector of the surface coefficients is repeatedly adjusted based on the comprehensive performance evaluation method until the stopping condition is met, at which point the optimal surface coefficient vector is output. Each adjustment determines the adjustment direction and magnitude by constructing local approximation conditions, including: The comprehensive performance evaluation method is invoked, and the spatial surface shape corresponding to the current surface coefficient vector is used as input to determine the current evaluation value of the comprehensive performance evaluation method at the current surface coefficient vector. The values of each coefficient component in the current surface coefficient vector are changed, and the changes in the evaluation value of the comprehensive performance evaluation method after each change are monitored. Based on the changes in the evaluation value, the trend information of the change of the comprehensive performance evaluation method along the direction of each coefficient component is determined. Based on the current evaluation value, the trend information, and the curvature constraint matrix dynamically constructed based on the gradient information of the current evaluation function, a local approximation condition is established in the neighborhood of the current surface coefficient vector. The local approximation condition is expressed in the following form: ; In the formula, This represents the adjustment vector to be determined. This represents the current evaluation value of the comprehensive performance evaluation method at the current surface coefficient vector. This represents a first-order change vector composed of the aforementioned trend information. This represents the symmetric curvature matrix constructed at the current surface coefficient vector based on the local curvature information of the comprehensive performance evaluation method; the local curvature information is obtained by perturbing the first-order change vector using numerical difference, or by recursively obtaining it from historical iteration information using a quasi-Newton update formula. Under the local approximation conditions, the adjustment vector corresponding to the minimum value of the local approximation conditions is determined, and the adjustment vector is used as a joint representation of the adjustment direction and adjustment magnitude. Based on the adjustment vector, the current surface coefficient vector is updated to generate the updated surface coefficient vector; The updated surface coefficient vector is used as the current surface coefficient vector for the next adjustment, and this process is repeated until the stopping condition is met, at which point the optimal surface coefficient vector is output.
[0010] In a preferred embodiment, the step of repeatedly adjusting the initial vector of surface coefficients based on the comprehensive performance evaluation method until the stopping condition is met and the optimal surface coefficient vector is output, wherein each adjustment determines the adjustment direction and adjustment magnitude by constructing local approximation conditions, further includes: After an adjustment is completed and an updated surface coefficient vector is generated, the comprehensive performance evaluation method is invoked. The spatial surface shape corresponding to the updated surface coefficient vector is used as input to determine the updated evaluation value of the comprehensive performance evaluation method at the updated surface coefficient vector. Read the degree of deviation between the updated evaluation value and the historical evaluation value recorded after the last adjustment, and use the degree of deviation as the change in the evaluation value; The change in the evaluation value is compared with a preset allowable change threshold. When the change in the evaluation value is less than or equal to the preset allowable change threshold, the stopping condition is determined to be met. After determining that the stopping condition is met, the updated surface coefficient vector is output as the optimal surface coefficient vector.
[0011] In a preferred embodiment, generating complete surface shape data containing discrete height distribution and continuous surface shape description based on the combination of the optimal surface coefficient vector and the initial surface shape representation basis includes: Retrieve the optimal surface coefficient vector and the initial surface shape representation basis, and pair each optimal coefficient in the optimal surface coefficient vector with the corresponding basis component in the initial surface shape representation basis to generate weighted surface shape components corresponding to each basis component; Within the surface shape representation space, a sampling position sequence is determined, and all weighted surface shape components are superimposed at each sampling position to generate superposition height values at each sampling position. The superposition height values at all sampling positions are then arranged into a discrete height distribution. Based on the superimposed height values at each sampling location in the discrete height distribution, establish a correspondence table between sampling locations and superimposed height values; The preset continuous interpolation rule is invoked, and the height change trend between adjacent sampling positions in the correspondence table is used as the interpolation basis. Supplementary height values are inserted between adjacent sampling positions to generate a supplementary height value sequence covering the entire surface expression space. The superimposed height value and the supplementary height value sequence are merged to generate a full-caliber height value distribution; Based on the full-caliber height value distribution, a surface shape description relationship between position coordinates and height values is established. This surface shape description relationship is used as the continuous surface shape description, and the discrete height distribution and the continuous surface shape description are used together as complete surface shape data.
[0012] In a preferred embodiment, the continuous surface shape description is formed in the following manner: Read the sampling position and superposition height value in the corresponding relationship table, and use each sampling position as a support point, and use the superposition height value at each sampling position as the height constraint value at the support point; The preset radial base interpolation method is invoked, and the position coordinates and height constraint values of all support points are used as input conditions to configure a corresponding radial base shape for each support point. The radial base shape takes the height constraint value at the support point and decays with distance at positions away from the support point. Based on the radial base morphology corresponding to all support points, a transition height value is generated between each adjacent support point. The transition height value is jointly determined by the radial base morphology corresponding to each adjacent support point, and the transition height values contributed by the radial base morphology on both sides are equal at the midpoint of the adjacent support points. The height constraint values at all support points and the transition height values between all adjacent support points are merged to form the surface shape description relationship, which is then used as a continuous surface shape description.
[0013] In a preferred embodiment, the step of jointly verifying the complete surface data based on the surface design constraints, and confirming the output of the complete surface data if the verification passes, and adjusting the weighting factor and repeating Pt.2 to Pt.4 until the verification passes, includes: When the complete surface data fails the joint verification, the imaging verification result and stray light suppression verification result generated during the joint verification process are read. The imaging verification result is compared with the imaging constraint boundary of the optical design input parameters to determine the degree of non-compliance of the imaging part. The stray light suppression verification result is compared with the stray light suppression constraint boundary of the optical design input parameters to determine the degree of non-compliance of the stray light suppression part. The degree of dissatisfaction of the imaging part is compared with the degree of dissatisfaction of the stray light suppression part. Based on the comparison result, the dominant dissatisfaction dimension is determined. The dominant dissatisfaction dimension is the one with the larger degree of dissatisfaction value between the imaging dimension and the stray light suppression dimension. The adjustment direction of the weighting factor is determined based on the dominant non-satisfied dimension. When the dominant non-satisfied dimension is the imaging dimension, the weighting factor is adjusted in the direction of increasing the proportion of the imaging evaluation part. When the dominant non-satisfied dimension is the stray light suppression dimension, the weighting factor is adjusted in the direction of increasing the proportion of the stray light suppression evaluation part. Based on the difference between the degree of dissatisfaction corresponding to the dominant dissatisfaction dimension and the degree of dissatisfaction of another dimension, the adjustment step size of the weighting factor is determined. The adjustment step size is positively correlated with the difference. The current weighting factor is updated according to the adjustment direction and the adjustment step size to generate the updated weighting factor. The updated weighting factor is fed back to the establishment stage of the comprehensive performance evaluation method. The comprehensive performance evaluation method is re-established with the updated weighting factor, and the iterative adjustment, the generation of the complete surface data, and the joint verification are executed again in sequence until the complete surface data passes the joint verification.
[0014] To address the aforementioned problems, this invention also provides a globally tunable optical surface design system based on nonlinear optimization, the system comprising: The parameter basis construction module is used to obtain the optical design input parameters of the implantable ophthalmic endoscope, generate surface design constraints from the optical design input parameters, and construct the initial surface shape expression basis and the initial vector of surface coefficients. The comprehensive evaluation modeling module is used to establish a comprehensive performance evaluation method that includes an imaging evaluation part and a stray light suppression evaluation part based on the surface design constraints. The imaging evaluation part and the stray light suppression evaluation part are associated through a weighting factor. The surface coefficient optimization module is used to repeatedly adjust the initial vector of the surface coefficients based on the comprehensive performance evaluation method until the stopping condition is met, and then output the optimal surface coefficient vector. Each adjustment determines the adjustment direction and adjustment range by constructing local approximation conditions. The complete surface shape generation module is used to generate complete surface shape data containing discrete height distribution and continuous surface shape description based on the combination of the optimal surface coefficient vector and the initial surface shape expression basis; The constraint verification iteration module is used to jointly verify the complete surface data based on the surface design constraints. If the verification passes, the complete surface data is output. If the verification fails, the weighting factor is adjusted and the comprehensive evaluation modeling module is repeated to the complete surface generation module until the verification passes.
[0015] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention constructs a unified and interconnected comprehensive performance evaluation system, establishing a quantitative correlation between the imaging evaluation module and the stray light suppression evaluation module based on weighting factors. During the entire process of surface coefficient iterative optimization, the wavefront aberration tolerance boundary and the stray light suppression index deviation tolerance boundary at each field of view position are simultaneously constrained. The spatial morphology of the surface is subjected to bidirectional synchronous equilibrium control. During the iteration process, the two core optical indicators of full-domain imaging sharpness and full-aperture illumination uniformity are continuously balanced. The final generated optical surface can fully meet the dual design specifications of implantable ophthalmic endoscopes for field of view imaging and aperture illumination, and is fully adapted to the physiological optical imaging needs of human eyes after implantation.
[0016] 2. This invention sets up a closed-loop joint verification process with automatic weight correction capability. After the complete surface data completes the constraint verification, it can automatically quantify and decompose the degree of deviation in the imaging and illumination dimensions. Based on the magnitude of the deviation, it can autonomously determine the step size and direction of the weighting factor adjustment. Without manual intervention, it can automatically cycle through the entire process of comprehensive evaluation construction, surface coefficient optimization, and complete surface generation until the verification meets the standards. It simultaneously outputs an integrated dataset of discrete sampling height distribution and global interpolation continuous surface description, which greatly shortens the iterative convergence cycle of compliant surfaces, ensures that the surface coordinates and height numerical data are consistent and unbiased, and effectively improves the data matching accuracy of subsequent optical simulation analysis and lens precision CNC machining. Attached Figure Description
[0017] Figure 1 This is a flowchart illustrating a globally tunable optical surface design method based on nonlinear optimization, provided in an embodiment of the present invention. Figure 2 This is a functional block diagram of a globally tunable optical surface design system based on nonlinear optimization, provided in an embodiment of the present invention.
[0018] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation
[0019] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
[0020] This application provides a method for designing a fully tunable optical surface based on nonlinear optimization. The execution entity of this method includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the method provided in this application: a server, a terminal, etc. In other words, the method for designing a fully tunable optical surface based on nonlinear optimization can be executed by software or hardware installed on a terminal device or a server device. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster. The server can be an independent server or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDN), and big data and artificial intelligence platforms.
[0021] Reference Figure 1 The diagram shown is a flowchart illustrating a globally tunable optical surface design method based on nonlinear optimization according to an embodiment of the present invention. In this embodiment, the globally tunable optical surface design method based on nonlinear optimization includes: Pt.1. Obtain the optical design input parameters for implantable ophthalmic endoscopes, generate surface design constraints from the optical design input parameters, and construct the initial surface shape representation basis and the initial vector of surface coefficients. In this embodiment of the invention, the step of obtaining optical design input parameters for an implantable ocular endoscope, generating surface design constraints from the optical design input parameters, and constructing an initial surface shape representation basis and an initial vector of surface coefficients includes: Extract field of view range data from optical design input parameters, determine the effective field of view region based on the field of view range data, and divide the field of view position sequence within the effective field of view region; Based on the ideal imaging requirements corresponding to each field of view position in the field of view position sequence, a corresponding wavefront aberration tolerance boundary value is configured for each field of view position, and the wavefront aberration tolerance boundary values corresponding to all field of view positions are combined into an imaging wavefront constraint boundary. Extract stray light suppression aperture distribution requirements from the optical design input parameters, determine the effective stray light suppression evaluation aperture region based on the stray light suppression aperture distribution requirements, and divide the radial position sequence within the effective stray light suppression evaluation aperture region; Based on the stray light suppression index target value corresponding to each radial position in the radial position sequence, configure the stray light suppression index deviation tolerance range corresponding to each radial position, and combine the stray light suppression index deviation tolerance ranges corresponding to all radial positions into stray light suppression index constraint boundaries. The imaging wavefront constraint boundary and the stray light suppression index constraint boundary are merged to generate the surface design constraint conditions.
[0022] The process of obtaining optical design input parameters for implantable ocular endoscopes, generating surface design constraints from these parameters, and constructing an initial surface shape representation basis and initial vectors for surface coefficients includes: The effective aperture size is extracted from the optical design input parameters. The surface shape expression space range is determined based on the effective aperture size. A family of basis functions is selected within the surface shape expression space range. The basis components in the family of basis functions have different spatial variation scales. Arrange the basis components in the basis function family from low to high according to the spatial variation scale, and combine all the arranged basis components into an initial surface shape expression basis; Each base component in the initial surface shape expression base is assigned a corresponding initial weight value, and all initial weight values are arranged in the order of the base components in the initial surface shape expression base to generate an initial weight value sequence. The initial surface shape representation basis is associated with and stored with the initial weight value sequence to generate an initial vector of surface coefficients.
[0023] Retrieve the optical design input parameter file containing the complete optical design requirements for implantable intraocular lenses. Read the field of view range data, which records the coverage area of the field of view, from the file. Use the minimum and maximum field of view values recorded in the data as boundaries to divide the complete effective field of view area. Divide the effective field of view area evenly according to a fixed division rule with equal angular intervals. Record each division point separately to form an ordered sequence of field of view positions.
[0024] Retrieve the preset standard imaging technology document for implantable ophthalmic endoscopes, extract the standard ideal imaging requirements for each field of view position sequence point in the document, assign a unique wavefront aberration allowable boundary value to each field of view position according to the corresponding rule of matching a single limit value with a single field of view position, and integrate and store the wavefront aberration allowable boundary values of all field of view position points in the order of the field of view position sequence. After integration, a complete set of imaging wavefront constraint boundaries is formed.
[0025] Retrieve the optical design input parameter file containing the complete optical design requirements for implantable intraocular lenses. Read the stray light suppression aperture distribution requirements that record the light transmission aperture brightness allocation standard from the file. Use the minimum and maximum aperture radial dimensions recorded in the requirements as boundaries to divide the complete effective stray light suppression evaluation aperture region. Divide the effective stray light suppression evaluation aperture region evenly according to a fixed division rule of equal radial length intervals. Record each division point separately to form an ordered radial position sequence.
[0026] Retrieve the preset standard illumination technology document for implantable ophthalmic endoscopes, extract the target value of the standard stray light suppression index corresponding to each radial position sequence point in the document, assign a unique stray light suppression index deviation tolerance range to each radial position according to the corresponding rule of matching a single radial point with a single interval, and integrate and store the stray light suppression index deviation tolerance ranges of all radial position points in the order of the radial position sequence. After integration, a complete set of stray light suppression index constraint boundaries is formed.
[0027] Perform a complete data merging operation on the complete set of imaging wavefront constraint boundaries and the complete set of stray light suppression index constraint boundaries. All limit values, intervals, and point sequence information recorded in the two sets of boundary data are synchronously included in the same set of constraint files. After the integration is completed, complete and usable surface design constraint conditions are generated.
[0028] The aperture size data of the complete surface space range is retrieved and recorded. The spatial boundary defined by this data is used as the surface expression space range. Within this range, a family of basis functions containing multi-scale spatial variation characteristics is selected as the basic expression carrier. All basis components within the family of basis functions are arranged in an orderly manner according to the scale of spatial fluctuation changes from small to large. All the arranged basis components are uniformly collected to form the initial surface expression basis.
[0029] According to the arrangement order of the basis components inside the initial surface shape representation basis, a fixed initial weight value is assigned to each type of basis component. All assigned initial weight values are sorted and recorded in the order of basis arrangement, forming an ordered sequence of initial weight values.
[0030] A one-to-one binding and storage relationship is established between the initial surface shape representation basis and the initial weight value sequence. After the two types of data are bound and stored synchronously, a complete and usable initial vector of surface coefficients is generated.
[0031] The optical design input parameters of all optical design standards for implantable intraocular lenses can be fully read and stored. The effective aperture size of the light-transmitting area boundary of the lens is extracted from the parameters. The maximum radial length and axial thickness limit recorded by the effective aperture size are used as fixed boundaries to lock the complete surface shape expression space. Within this space, a predefined family of multiple types of basis functions is selected as the surface shape description carrier. Each type of basis component within the basis function family corresponds to a unique and fixed scale of spatial undulation change. Different basis components are matched with non-overlapping scale intervals.
[0032] The unique spatial variation scale values corresponding to all basis components within the basis function family are identified one by one. All basis components are sorted and arranged in an orderly manner according to a fixed sorting rule from smallest to largest scale value. All basis components that have been arranged in an orderly manner are uniformly collected into the same set of data. After collection, a complete set of initial surface shape expression basis is obtained.
[0033] Retrieve the preset base weight benchmark library, and match the fixed initial weight values in the benchmark library for each type of base component according to the arrangement order of the base components inside the initial surface shape expression base. Strictly follow the base arrangement order of the initial surface shape expression base to organize and record all the matched initial weight values one by one, and form an ordered and complete sequence of initial weight values after recording one by one.
[0034] Establish a corresponding storage rule that binds each basis component in the initial surface shape representation basis to the corresponding initial weight value in the initial weight value sequence. According to the binding rule, the two sets of data are synchronously stored in the same set of structured data carrier to complete the association storage operation. After all association storage operations are completed, a complete and usable initial vector of surface coefficients is generated.
[0035] Pt.2. Based on the surface design constraints, establish a comprehensive performance evaluation method that includes an imaging evaluation part and a stray light suppression evaluation part, wherein the imaging evaluation part and the stray light suppression evaluation part are associated through a weighting factor; In this embodiment of the invention, the step of establishing a comprehensive performance evaluation method that includes an imaging evaluation component and a stray light suppression evaluation component based on the surface design constraints, wherein the imaging evaluation component and the stray light suppression evaluation component are associated through a weighting factor, includes: The imaging wavefront constraint boundary is read from the surface design constraint conditions. The allowable wavefront deviation range corresponding to each field of view position is determined according to the imaging wavefront constraint boundary. The spatial surface shape corresponding to the surface coefficient vector is compared with the ideal wavefront shape at each field of view position. The wavefront deviation degree value at each field of view position is generated according to the comparison result. The imaging evaluation result is generated according to the wavefront deviation degree value at all field of view positions. The stray light suppression index constraint boundary is read from the surface design constraints. The allowable stray light suppression index deviation range corresponding to each radial position is determined according to the stray light suppression index constraint boundary. The surface orientation of the spatial surface shape corresponding to the surface coefficient vector at each radial position is compared with the stray light suppression target orientation. The stray light suppression index deviation value at each radial position is generated according to the comparison result. The stray light suppression evaluation result is generated according to the stray light suppression index deviation values at all radial positions. A preset weighting factor is invoked, which is used to adjust the proportion of the imaging evaluation result and the stray light suppression evaluation result in the comprehensive evaluation. The imaging evaluation result and the stray light suppression evaluation result are correlated according to the weighting factor to generate a comprehensive evaluation value. A comprehensive performance evaluation method is established based on the correspondence between the comprehensive evaluation value and the surface coefficient vector.
[0036] The constraint dataset containing the complete storage surface design constraints is retrieved. A complete set of imaging wavefront constraint boundaries is extracted from this dataset. The wavefront aberration tolerance boundary values matched at each field of view within the imaging wavefront constraint boundaries are extracted. The upper and lower limits of the wavefront aberration tolerance boundary values corresponding to each point are used as fixed standards to define the allowable wavefront deviation range specific to each field of view. The spatial surface shape generated by the current surface coefficient vector mapping is retrieved. The actual wavefront shape presented by this spatial surface shape at each field of view is compared item by item with the ideal wavefront shape of the preset fixed standard at that field of view. The comparison operation uses the spatial distribution points of the wavefront shape at the same field of view as a unified reference. The difference between the two sets of wavefront shapes at each field of view is uniformly recorded to form the wavefront deviation degree value of the corresponding point. The wavefront deviation degree values generated at all field of view positions are uniformly summarized and integrated according to the predetermined arrangement order of the field of view position sequence. After summarizing and integrating, a complete set of imaging evaluation results is generated.
[0037] The complete constraint dataset of the stored surface design constraints is retrieved. From this dataset, a complete set of stray light suppression index constraint boundaries is retrieved. The allowable range of stray light suppression index deviation for each radial position within the stray light suppression index constraint boundary is extracted. The upper and lower limits of the range corresponding to each point are used as fixed standards to define the allowable stray light suppression index deviation range for each radial position. The spatial surface shape generated by the current surface coefficient vector mapping is retrieved. The surface spatial orientation angle corresponding to this spatial surface shape at each radial position is extracted. The extracted surface orientation angle is compared item by item with the stray light suppression target orientation angle of the preset fixed standard at the radial position. The comparison operation uses the aperture radial coordinate at the same radial position as a unified reference. The difference between the two sets of orientation angles at each radial position is uniformly recorded to form the stray light suppression index deviation value for the corresponding point. The stray light suppression index deviation values generated at all radial positions are uniformly summarized and integrated according to the predetermined arrangement order of the radial position sequence. After summarizing and integrating, a complete set of stray light suppression evaluation results is generated.
[0038] The system retrieves the weighting factors that are fixedly stored in the preset weighting repository. The weighting factors are pre-set with two sets of fixed numerical ranges that correspond to the allocation ratio benchmarks of the imaging evaluation results and stray light suppression evaluation results in the comprehensive evaluation system. The two sets of ratio benchmarks recorded in the weighting factors are used as allocation standards to perform corresponding ratio allocation processing on the complete set of imaging evaluation results and the complete set of stray light suppression evaluation results. After the allocation processing is completed, the two types of evaluation data that have been allocated ratios are uniformly merged. After the merging is completed, a unique corresponding comprehensive evaluation value is generated.
[0039] Establish a fixed mapping storage rule that corresponds one-to-one between surface coefficient vectors and comprehensive evaluation values. After inputting any set of surface coefficient vectors, record the entire standardized operation process of generating imaging evaluation results, stray light suppression evaluation results, and comprehensive evaluation values as a fixed execution link. Encapsulate and store this fixed execution link to form a comprehensive performance evaluation method that can be called repeatedly.
[0040] Pt.3. Based on the comprehensive performance evaluation method, the initial vector of the surface coefficients is repeatedly adjusted until the stopping condition is met and the optimal surface coefficient vector is output. Each adjustment is determined by constructing local approximation conditions to determine the adjustment direction and adjustment range. In this embodiment of the invention, the initial vector of the surface coefficients is repeatedly adjusted based on the comprehensive performance evaluation method until the stopping condition is met, at which point the optimal surface coefficient vector is output. Each adjustment determines the adjustment direction and magnitude by constructing local approximation conditions, including: The comprehensive performance evaluation method is invoked, and the spatial surface shape corresponding to the current surface coefficient vector is used as input to determine the current evaluation value of the comprehensive performance evaluation method at the current surface coefficient vector. The values of each coefficient component in the current surface coefficient vector are changed, and the changes in the evaluation value of the comprehensive performance evaluation method after each change are monitored. Based on the changes in the evaluation value, the trend information of the change of the comprehensive performance evaluation method along the direction of each coefficient component is determined. Based on the current evaluation value, the trend information, and the curvature constraint matrix dynamically constructed based on the gradient information of the current evaluation function, a local approximation condition is established in the neighborhood of the current surface coefficient vector. The local approximation condition is expressed in the following form: ; In the formula, This represents the adjustment vector to be determined. This represents the current evaluation value of the comprehensive performance evaluation method at the current surface coefficient vector. This represents a first-order change vector composed of the aforementioned trend information. This represents the symmetric curvature matrix constructed at the current surface coefficient vector based on the local curvature information of the comprehensive performance evaluation method; the local curvature information is obtained by perturbing the first-order change vector using numerical difference, or by recursively obtaining it from historical iteration information using a quasi-Newton update formula. The specific construction method of the symmetric curvature matrix H is as follows: Let the current coefficient vector be... The overall evaluation value is The first-order change vector is If the numerical difference method is used, the element in the i-th row and j-th column of H is approximated as follows: in, The preset small perturbation step size, and They are respectively the i-th and the i-th Unit vectors in each coordinate direction; if the quasi-Newton BFGS update method is used, then Based on the previous step The coefficient vector changes and gradient changes are recursively updated according to the BFGS correction formula to ensure... The positive definiteness and the accuracy of approximation of the current local curvature.
[0041] Under the local approximation conditions, the adjustment vector corresponding to the minimum value of the local approximation conditions is determined, and the adjustment vector is used as a joint representation of the adjustment direction and adjustment magnitude. Based on the adjustment vector, the current surface coefficient vector is updated to generate the updated surface coefficient vector; The updated surface coefficient vector is used as the current surface coefficient vector for the next adjustment, and this process is repeated until the stopping condition is met, at which point the optimal surface coefficient vector is output.
[0042] The method based on the comprehensive performance evaluation involves repeatedly adjusting the initial vector of the surface coefficients until a stopping condition is met, at which point the optimal surface coefficient vector is output. Each adjustment determines the direction and magnitude of the adjustment by constructing local approximation conditions. The method also includes: After an adjustment is completed and an updated surface coefficient vector is generated, the comprehensive performance evaluation method is invoked. The spatial surface shape corresponding to the updated surface coefficient vector is used as input to determine the updated evaluation value of the comprehensive performance evaluation method at the updated surface coefficient vector. Read the degree of deviation between the updated evaluation value and the historical evaluation value recorded after the last adjustment, and use the degree of deviation as the change in the evaluation value; The change in the evaluation value is compared with a preset allowable change threshold. When the change in the evaluation value is less than or equal to the preset allowable change threshold, the stopping condition is determined to be met. After determining that the stopping condition is met, the updated surface coefficient vector is output as the optimal surface coefficient vector.
[0043] The encapsulated comprehensive performance evaluation method is retrieved, and the spatial surface shape generated by the current surface coefficient vector mapping is completely sent into the standardized execution link corresponding to the comprehensive performance evaluation method. The entire standardized operation process of generating imaging evaluation results, stray light suppression evaluation results, and comprehensive evaluation values is completed in sequence. After the process is completed, the current evaluation value uniquely corresponding to the current surface coefficient vector is locked and completely stored.
[0044] Following the fixed arrangement order of the base components, a fixed-amplitude numerical modification operation is performed on each independent coefficient component within the current surface coefficient vector. After each numerical modification operation of a single coefficient component is completed, the comprehensive performance evaluation method is run once, and the corresponding evaluation value generated after the modification operation is recorded. The current evaluation value retained before the modification is compared with the evaluation value generated after a single modification operation to obtain the evaluation value change record corresponding to a single coefficient dimension. The evaluation value change records corresponding to all coefficient dimensions are collected and organized in an orderly manner according to the base arrangement order. After collection and organization, the change trend information covering all coefficient dimensions is obtained.
[0045] The system retrieves the current evaluation value, the fully collected trend information, and the curvature constraint parameters stored in advance. It then defines a fixed-size neighborhood data range centered on the data point where the current surface coefficient vector is located. The three types of retrieved data are used as the basic material to build local approximation conditions that fit this neighborhood data range. The trend information is uniformly regularized into a first-order change vector. Based on the local curvature information at the current surface coefficient vector, a symmetric curvature matrix is constructed using numerical difference or quasi-Newton update methods. Using the current evaluation value as the reference value, the first-order change vector as the first-order change reference, and the symmetric curvature matrix as the second-order change reference, a complete neighborhood response description rule is built. This set of neighborhood response description rules constitutes the local approximation conditions.
[0046] Using the neighborhood response description rule corresponding to the completed local approximation condition as the sole evaluation criterion, the neighborhood response output results corresponding to all feasible adjustment vectors are traversed, and the only set of adjustment vectors that can make the neighborhood response output result reach the lowest global value is selected. Each data item in this set of adjustment vectors carries two types of fixed information: the offset path and the offset scale of the corresponding coefficient dimension, and fully serves as the joint representation carrier of the adjustment direction and adjustment magnitude corresponding to this coefficient update operation.
[0047] Read the adjustment vector that carries the joint representation information of adjustment direction and adjustment range, and perform superposition and integration processing on the offset data of each corresponding dimension in the adjustment vector and the original values of the coefficient components at the same position in the current surface coefficient vector. After the superposition and integration processing is completed, a new and complete set of coefficient data is generated. This new and complete set of coefficient data is the updated surface coefficient vector.
[0048] The updated surface coefficient vector generated by the superposition and integration process replaces the original current surface coefficient vector as the input carrier for the new round of iteration. The entire standardized operation process of calling the comprehensive performance evaluation method, monitoring coefficient value changes, building local approximation conditions, adjusting vector solution, and updating coefficient vector is completely repeated. After each round of iteration, the system's preset stopping condition judgment threshold is called to complete the compliance verification. When the verification result matches the judgment threshold standard corresponding to the stopping condition, all iteration operations are terminated, and the updated surface coefficient vector generated in this round of iteration is marked as the optimal surface coefficient vector and output.
[0049] After running the comprehensive performance evaluation method, the output shows the complete comprehensive performance evaluation chain. It takes the spatial surface shape corresponding to the current surface coefficient vector as input and sequentially generates imaging evaluation results, stray light suppression evaluation results, and a comprehensive evaluation value. The final retained comprehensive evaluation value is fixed as... The corresponding numerical content.
[0050] Based on the changes in evaluation values obtained after modifying each coefficient component, the values of each coefficient in the surface coefficient vector are modified individually. After each modification, the comprehensive performance evaluation method is run to record the corresponding evaluation value. The single-dimensional change record is obtained by comparing the original current evaluation value with the single-dimensional change record. All dimensional change records are arranged and summarized in the basis order to form a complete first-order change vector. .
[0051] This represents the symmetric curvature matrix, used to represent the second-order variation characteristics of the comprehensive performance evaluation method within the current neighborhood. In this embodiment, Instead of fixed preset values, the evaluation function is constructed based on the actual shape of the evaluation function at the current surface coefficient vector. Specifically, in each iteration, the first-order change vector g is numerically perturbed using the finite central difference method to estimate the approximate value of the Hessian matrix of the objective function at the current point; or the BFGS (Broyden-Fletcher-Goldfarb-Shanno) quasi-Newton update formula is used to recursively derive the current iteration's evaluation function using the gradient and coefficient changes from the previous and current iterations. Matrix. The matrix is dynamically updated as the iteration process proceeds to ensure that the local approximation conditions always accurately reflect the performance fluctuations around the current surface coefficient vector.
[0052] The adjustment vector to be solved has no pre-stored fixed values. It is the output obtained after a complete deduction and solution based on local approximation conditions. The number of internal dimensions of the vector is completely equivalent to the number of basis components contained in the surface coefficient vector.
[0053] The complete set of descriptive relationships for local approximation conditions relies on the current evaluation value. First-order change vector Symmetric curvature matrix The three types of materials are used together to build the system. All materials are taken from the optical performance change data collected in the neighborhood of the current surface coefficient vector.
[0054] The complete set of local approximation conditions is used to fully characterize the complete correspondence between the changes in the comprehensive evaluation value after the surface coefficients are numerically modified within a fixed neighborhood around the current surface coefficient vector.
[0055] The entire set of local approximation conditions can quantify the linear change in comprehensive performance caused by a small change in a single coefficient. This change is represented by a first-order change vector. The complete load-bearing capacity directly reflects the increase or decrease in the overall imaging and illumination performance when each base factor is adjusted individually.
[0056] Since the H matrix in this embodiment is constructed or updated in real time based on the actual performance evaluation function at the current surface coefficient vector, it can accurately characterize the second-order nonlinear coupling law of the comprehensive performance evaluation method changing with the coefficient combination when multiple sets of coefficients are modified simultaneously. The local approximation conditions established in this way have a numerical trend of nonlinear terms that is consistent with the true curvature at the current point, so that the adjustment vector s obtained by solving the minimum value of this approximation model can effectively guide the optimization direction and step size of the surface coefficient vector.
[0057] The entire set of local approximation conditions are based on The baseline performance value of the neighborhood center point is used as a reference to uniformly measure all adjustment vectors. The corresponding overall performance levels are compared to achieve a unified standard.
[0058] A set of adjustment vectors can be uniquely determined based on the entire set of local approximation conditions. This set of adjustment vectors can bring the overall performance to the lowest value within the neighborhood, and simultaneously lock the offset path and offset scale that each coefficient needs to be adjusted, providing a standardized adjustment basis for updating the surface coefficient vector.
[0059] The entire set of local approximation conditions only applies to the local neighborhood defined by the current surface coefficient vector, accurately describing the performance changes after small-scale coefficient modifications, avoiding deviations caused by describing performance changes across the entire domain, and ensuring that each round of coefficient iteration adjustment operation has a fixed and reliable execution benchmark.
[0060] The updated surface coefficient vector generated after the adjustment in the previous iteration is retrieved. Based on this vector, the corresponding spatial surface shape is restored by matching the initial surface shape representation basis. The restored spatial surface shape is then completely sent into the standardized execution link corresponding to the comprehensive performance evaluation method. The entire standardized operation process of generating imaging evaluation results, stray light suppression evaluation results, and comprehensive evaluation values is completed in sequence. After all the processes are completed, the updated evaluation value of the updated surface coefficient vector is fixed and archived.
[0061] Retrieve the historical evaluation values archived and saved in the local storage unit after the end of the previous iteration operation, retrieve the updated evaluation values archived after the completion of the current iteration, and use the complete numerical records of the two sets of evaluation values as a reference to complete the difference comparison operation. The fixed difference obtained from the comparison of the two sets of values is uniformly marked as the evaluation value change and archived and saved separately.
[0062] The system retrieves the permissible variation threshold pre-written into the storage area. This threshold is a fixed judgment value set to adapt to the convergence standard of the implantable ophthalmic endoscope surface. After retrieving the complete value variation of the archived evaluation value, a numerical comparison operation is performed with the permissible variation threshold. The comparison operation uses the two sets of original complete values as the sole judgment basis. When the value corresponding to the value variation is less than or equal to the value corresponding to the permissible variation threshold, the system generates a fixed judgment mark to indicate that the current iteration process meets the stopping condition.
[0063] After the recognition system generates a fixed judgment mark that satisfies the stopping condition, it locks the updated surface coefficient vector produced in this iteration, adds a unique data identifier for the optimal surface coefficient vector to this set of coefficient vectors, and completely exports the coefficient vector with the added identifier from the iteration storage area to complete the output operation.
[0064] Pt.4. Based on the combination of the optimal surface coefficient vector and the initial surface shape representation basis, generate complete surface shape data containing discrete height distribution and continuous surface shape description; In this embodiment of the invention, generating complete surface data containing discrete height distribution and continuous surface description based on the combination of the optimal surface coefficient vector and the initial surface shape representation basis includes: Retrieve the optimal surface coefficient vector and the initial surface shape representation basis, and pair each optimal coefficient in the optimal surface coefficient vector with the corresponding basis component in the initial surface shape representation basis to generate weighted surface shape components corresponding to each basis component; Within the surface shape representation space, a sampling position sequence is determined, and all weighted surface shape components are superimposed at each sampling position to generate superposition height values at each sampling position. The superposition height values at all sampling positions are then arranged into a discrete height distribution. Based on the superimposed height values at each sampling location in the discrete height distribution, establish a correspondence table between sampling locations and superimposed height values; The preset continuous interpolation rule is invoked, and the height change trend between adjacent sampling positions in the correspondence table is used as the interpolation basis. Supplementary height values are inserted between adjacent sampling positions to generate a supplementary height value sequence covering the entire surface expression space. The superimposed height value and the supplementary height value sequence are merged to generate a full-caliber height value distribution; Based on the full-caliber height value distribution, a surface shape description relationship between position coordinates and height values is established. This surface shape description relationship is used as the continuous surface shape description, and the discrete height distribution and the continuous surface shape description are used together as complete surface shape data.
[0065] The formation method of the continuous surface shape description includes: Read the sampling position and superposition height value in the corresponding relationship table, and use each sampling position as a support point, and use the superposition height value at each sampling position as the height constraint value at the support point; The preset radial base interpolation method is invoked, and the position coordinates and height constraint values of all support points are used as input conditions to configure a corresponding radial base shape for each support point. The radial base shape takes the height constraint value at the support point and decays with distance at positions away from the support point. Based on the radial base morphology corresponding to all support points, a transition height value is generated between each adjacent support point. The transition height value is jointly determined by the radial base morphology corresponding to each adjacent support point, and the transition height values contributed by the radial base morphology on both sides are equal at the midpoint of the adjacent support points. The height constraint values at all support points and the transition height values between all adjacent support points are merged to form the surface shape description relationship, which is then used as a continuous surface shape description.
[0066] The optimal surface coefficient vector output after iterative convergence is synchronously retrieved from the fixed storage area and the initial surface representation basis constructed in the early stage. A matching channel is established for the two sets of data in the order of the basis components. Each optimal coefficient in the optimal surface coefficient vector is matched to the basis components in the same order in the initial surface representation basis. The single basis and coefficient combination carrier that has been matched is uniformly defined as the weighted surface component. All matched weighted surface components are uniformly collected and archived.
[0067] Based on the surface shape expression space range defined by the effective aperture size in the early stage, the point layout operation is completed by adopting a dual fixed division standard of equal radial interval and equal angular interval. All layout points are arranged in an orderly manner according to radial and angular to form a complete sampling position sequence. All archived weighted surface shape components are retrieved. Within the sampling position sequence, the height values of all weighted surface shape components corresponding to each independent sampling point are read and a unified superposition and aggregation operation is performed. The height value obtained by the aggregation of each sampling point is marked as the superposition height value. The superposition height values corresponding to all sampling points are strictly collected and organized in an orderly manner according to the arrangement order of the sampling position sequence. After the organization is completed, a set of discrete height distributions is formed.
[0068] Extract all sampling location coordinates and corresponding superimposed height values within the discrete height distribution. Construct a standardized two-dimensional storage form with the sampling location coordinates as the first storage field and the superimposed height value as the second storage field. Each row of data in the form corresponds to a set of independent sampling points and a unique height value. The completed standardized two-dimensional storage form is fixedly named the Correspondence Relationship Table.
[0069] The pre-stored continuous interpolation rules adapted to the surface fitting standard of ophthalmic endoscopes are retrieved. These rules use the coordinates and superimposed height values of two adjacent sets of sampling points in the correspondence table as all reference materials. The height change trend corresponding to the coordinate span between two sets of adjacent sampling points is extracted as the sole reference for the interpolation operation. Transition points are arranged at fixed subdivision intervals within the coordinate interval defined by two sets of adjacent sampling points. The supplementary height value corresponding to each transition point is calculated one by one. The supplementary height values generated by all transition points are collected in an orderly manner according to the coordinate arrangement order. After collection, a supplementary height value sequence that completely covers the entire surface expression space is generated.
[0070] Extract the superimposed height values corresponding to all sampling points within the discrete height distribution, extract the complete supplementary height value sequence generated by the interpolation operation, and sort and collect the two types of height values into the same height data set according to the coordinate size. This integrated height data set covering the entire spatial coordinate range is defined as the full-caliber height value distribution.
[0071] Read all coordinate points and matching height values within the full-caliber height value distribution. Establish a unified global mapping and storage rule using spatial coordinates as the index and height values as the corresponding output content. This set of global mapping and storage rules is fixedly named the surface description relationship. This surface description relationship is separately marked as a continuous surface description. Retrieve the previously generated discrete height distribution and the currently obtained continuous surface description, and synchronously store the two sets of data into the same standardized data storage package. After the data is stored and integrated, complete surface data is generated.
[0072] Retrieve the previously constructed correspondence table, extract the sampling position coordinates and the superimposed height values matching the sampling positions from each complete record in the table, designate each set of independent sampling position coordinates as a support point, and fix the superimposed height value bound to the sampling position as the exclusive height constraint value of the corresponding support point. All support point coordinates and their corresponding height constraint values are uniformly organized and archived to form a support point dataset.
[0073] The radial basis interpolation method, which is pre-fixed and adapted to the surface fitting standard of implantable ophthalmic endoscopes, is retrieved. The coordinates of all support points and the height constraint values of all support points in the archived support point dataset are completely sent into the standardized execution link corresponding to this interpolation method. According to the fixed correspondence rule of matching a single support point with a single surface shape carrier, a unique radial basis shape is assigned to each support point. The output value of the radial basis shape at the support point's own coordinate position is completely consistent with the support point's height constraint value. In the region where there is a coordinate offset between the radial basis shape and the support point coordinates, the output value of the radial basis shape is adjusted synchronously according to a fixed distance attenuation rule. The farther away from the support point coordinates, the lower the output value of the radial basis shape is.
[0074] Select pairs of adjacent support points from the support point dataset in sequence, read the radial base shape output rules of each pair of adjacent support points, and set up transition points at fixed intervals within the coordinate interval defined by the two pairs of adjacent support points. For each transition point, read the height value of the radial base shape output synchronously on both sides and complete the integration processing to obtain the transition height value of the corresponding point. Set a unified judgment standard for the midpoint of the line connecting the coordinates of the two pairs of adjacent support points. The radial base shape output value of the support point on the left at the midpoint is completely equal to the radial base shape output value of the support point on the right.
[0075] Collect all fixed height constraint values corresponding to all support points, collect all transition height values generated by transition points within all adjacent support point intervals, and uniformly integrate the two types of height values according to the ascending order of spatial coordinates to form a complete global coordinate and height mapping record set. This mapping record set is fixedly named surface description relationship, and the integrated surface description relationship is directly marked as continuous surface description for subsequent encapsulation and storage of complete surface data.
[0076] Pt.5. Based on the surface design constraints, perform joint verification on the complete surface data. If the verification passes, confirm and output the complete surface data. If the verification fails, adjust the weighting factor and repeat Pt.2 to Pt.4 until the verification passes.
[0077] In this embodiment of the invention, the step of jointly verifying the complete surface data based on the surface design constraints, and if the verification passes, confirming the output of the complete surface data; if the verification fails, adjusting the weighting factor and repeating Pt.2 to Pt.4 until the verification passes, includes: When the complete surface data fails the joint verification, the imaging verification result and stray light suppression verification result generated during the joint verification process are read. The imaging verification result is compared with the imaging constraint boundary of the optical design input parameters to determine the degree of non-compliance of the imaging part. The stray light suppression verification result is compared with the stray light suppression constraint boundary of the optical design input parameters to determine the degree of non-compliance of the stray light suppression part. The degree of dissatisfaction of the imaging part is compared with the degree of dissatisfaction of the stray light suppression part. Based on the comparison result, the dominant dissatisfaction dimension is determined. The dominant dissatisfaction dimension is the one with the larger degree of dissatisfaction value between the imaging dimension and the stray light suppression dimension. The adjustment direction of the weighting factor is determined based on the dominant non-satisfied dimension. When the dominant non-satisfied dimension is the imaging dimension, the weighting factor is adjusted in the direction of increasing the proportion of the imaging evaluation part. When the dominant non-satisfied dimension is the stray light suppression dimension, the weighting factor is adjusted in the direction of increasing the proportion of the stray light suppression evaluation part. Based on the difference between the degree of dissatisfaction corresponding to the dominant dissatisfaction dimension and the degree of dissatisfaction of another dimension, the adjustment step size of the weighting factor is determined. The adjustment step size is positively correlated with the difference. The current weighting factor is updated according to the adjustment direction and the adjustment step size to generate the updated weighting factor. The updated weighting factor is fed back to the establishment stage of the comprehensive performance evaluation method. The comprehensive performance evaluation method is re-established with the updated weighting factor, and the iterative adjustment, the generation of the complete surface data, and the joint verification are executed again in sequence until the complete surface data passes the joint verification.
[0078] After the joint verification process is completed, retrieve the imaging verification results and stray light suppression verification results. Retrieve the imaging constraint boundaries stored internally in the optical design input parameters. Compare the actual deviation data of each field of view wavefront recorded in the imaging verification results with the allowable limits of each field of view wavefront within the imaging constraint boundaries. All records exceeding the limits generated by the comparison are summarized and organized to obtain the degree of non-compliance of the complete imaging part. Retrieve the stray light suppression constraint boundaries stored internally in the optical design input parameters. Compare the actual deviation data of each radial position stray light suppression index recorded in the stray light suppression verification results with the allowable range of each radial position stray light suppression index within the stray light suppression constraint boundaries. All records exceeding the limits generated by the comparison are summarized and organized to obtain the degree of non-compliance of the complete stray light suppression part.
[0079] Retrieve the total sum of values corresponding to the degree of non-satisfaction of the complete imaging component and the total sum of values corresponding to the degree of non-satisfaction of the complete stray light suppression component. Perform a comparison operation between the two sets of total values. The dimension with the larger value is defined as the dominant non-satisfaction dimension. The dimension with the larger value can only be determined as a unique category between the imaging dimension and the stray light suppression dimension.
[0080] If the dominant element identified does not meet the dimensional classification criteria, the system pre-sets two sets of fixed adjustment rules to match the two dimensional categories respectively. When the dominant element does not meet the dimensional classification criteria and is determined to be an imaging dimension, the imaging proportion enhancement rule is activated. This rule is used to change the stored value corresponding to the proportion of imaging evaluation results allocated within the weighting factor. When the dominant element does not meet the dimensional classification criteria and is determined to be a stray light suppression dimension, the illumination proportion enhancement rule is activated. This rule is used to change the stored value corresponding to the proportion of stray light suppression evaluation results allocated within the weighting factor. The two sets of rules correspond to a unique fixed adjustment direction for the weighting factor.
[0081] Extract the total sum of dissatisfaction levels for the dominant dissatisfaction dimension and the total sum of dissatisfaction levels for another dimension. A fixed difference record is formed between the two sets of values. The system is pre-configured with multiple fixed adjustment step size levels. The difference record is matched with the corresponding level to complete the adjustment step size locking operation. The larger the value corresponding to the difference record, the larger the change magnitude corresponding to the adjustment step size of the matched level. According to the locked adjustment direction and locked adjustment step size, the stored values of the proportion of the two types of evaluations within the original weighted factor are changed synchronously. After the change is completed, a brand new updated weighted factor is generated.
[0082] The updated weighting factor is completely transferred to the establishment of the comprehensive performance evaluation method as new input material. Based on the updated weighting factor, the imaging evaluation results and stray light suppression evaluation results are re-executed in a standardized process to complete the construction of the new comprehensive performance evaluation method. The operation of repeatedly adjusting the surface coefficient vector, generating complete surface data, and jointly verifying complete surface data are executed in sequence. After each round of joint verification, compliance is determined by comparing with the surface design constraint boundary. The loop process is terminated when all indicators of the complete surface data match the constraint boundary standard. If any indicator of the complete surface data exceeds the constraint boundary range, the entire loop operation is repeated continuously.
[0083] like Figure 2 The diagram shown is a functional block diagram of a globally tunable optical surface design system based on nonlinear optimization provided in an embodiment of the present invention.
[0084] The globally tunable optical surface design system based on nonlinear optimization described in this invention can be installed in electronic devices. Depending on the functions implemented, the globally tunable optical surface design system based on nonlinear optimization may include a parameter basis construction module, a comprehensive evaluation modeling module, a surface coefficient optimization module, a complete surface shape generation module, and a constraint verification iteration module. The module described in this invention can also be called a unit, which refers to a series of computer program segments that can be executed by the processor of an electronic device and can perform a fixed function, stored in the memory of the electronic device.
[0085] In this embodiment, the functions of each module / unit are as follows: The parameter basis construction module is used to obtain optical design input parameters for implantable ophthalmic endoscopes, generate surface design constraints from the optical design input parameters, and construct an initial surface shape expression basis and an initial vector of surface coefficients. The comprehensive evaluation modeling module is used to establish a comprehensive performance evaluation method that includes an imaging evaluation part and a stray light suppression evaluation part based on the surface design constraints. The imaging evaluation part and the stray light suppression evaluation part are associated through a weighting factor. The surface coefficient optimization module is used to repeatedly adjust the initial vector of the surface coefficients based on the comprehensive performance evaluation method until the stopping condition is met, and then output the optimal surface coefficient vector. Each adjustment determines the adjustment direction and adjustment range by constructing local approximation conditions. The complete surface shape generation module is used to generate complete surface shape data containing discrete height distribution and continuous surface shape description based on the combination of the optimal surface coefficient vector and the initial surface shape expression basis; The constraint verification iteration module is used to jointly verify the complete surface data based on the surface design constraints. If the verification passes, the complete surface data is output. If the verification fails, the weighting factor is adjusted and the comprehensive evaluation modeling module is repeated to the complete surface generation module until the verification passes.
[0086] In the several embodiments provided by this invention, it should be understood that the disclosed methods and systems can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and other division methods may be used in actual implementation.
[0087] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment, depending on actual needs.
[0088] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional modules.
[0089] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.
[0090] This application embodiment can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.
[0091] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.
Claims
1. A method for designing globally tunable optical surfaces based on nonlinear optimization, characterized in that, The method includes: Pt.
1. Obtain the optical design input parameters for implantable ophthalmic endoscopes, generate surface design constraints from the optical design input parameters, and construct the initial surface shape representation basis and the initial vector of surface coefficients. Pt.
2. Based on the surface design constraints, establish a comprehensive performance evaluation method that includes an imaging evaluation part and a stray light suppression evaluation part, wherein the imaging evaluation part and the stray light suppression evaluation part are associated through a weighting factor; Pt.
3. Based on the comprehensive performance evaluation method, the initial vector of the surface coefficients is repeatedly adjusted until the stopping condition is met and the optimal surface coefficient vector is output. Each adjustment is determined by constructing local approximation conditions to determine the adjustment direction and adjustment range. Pt.
4. Based on the combination of the optimal surface coefficient vector and the initial surface shape representation basis, generate complete surface shape data containing discrete height distribution and continuous surface shape description; Pt.
5. Based on the surface design constraints, perform joint verification on the complete surface data. If the verification passes, confirm and output the complete surface data. If the verification fails, adjust the weighting factor and repeat Pt.2 to Pt.4 until the verification passes.
2. The method for designing a globally tunable optical surface based on nonlinear optimization as described in claim 1, characterized in that, The process of obtaining optical design input parameters for implantable ocular endoscopes, generating surface design constraints from these parameters, and constructing an initial surface shape representation basis and initial vectors for surface coefficients includes: Extract field of view range data from optical design input parameters, determine the effective field of view region based on the field of view range data, and divide the field of view position sequence within the effective field of view region; Based on the ideal imaging requirements corresponding to each field of view position in the field of view position sequence, a corresponding wavefront aberration tolerance boundary value is configured for each field of view position, and the wavefront aberration tolerance boundary values corresponding to all field of view positions are combined into an imaging wavefront constraint boundary. Extract stray light suppression aperture distribution requirements from the optical design input parameters, determine the effective stray light suppression evaluation aperture region based on the stray light suppression aperture distribution requirements, and divide the radial position sequence within the effective stray light suppression evaluation aperture region; Based on the stray light suppression index target value corresponding to each radial position in the radial position sequence, configure the stray light suppression index deviation tolerance range corresponding to each radial position, and combine the stray light suppression index deviation tolerance ranges corresponding to all radial positions into stray light suppression index constraint boundaries. The imaging wavefront constraint boundary and the stray light suppression index constraint boundary are merged to generate the surface design constraint conditions.
3. The globally tunable optical surface design method based on nonlinear optimization as described in claim 2, characterized in that, The process of obtaining optical design input parameters for implantable ocular endoscopes, generating surface design constraints from these parameters, and constructing an initial surface shape representation basis and initial vectors for surface coefficients includes: The effective aperture size is extracted from the optical design input parameters. The surface shape expression space range is determined based on the effective aperture size. A family of basis functions is selected within the surface shape expression space range. The basis components in the family of basis functions have different spatial variation scales. Arrange the basis components in the basis function family from low to high according to the spatial variation scale, and combine all the arranged basis components into an initial surface shape expression basis; Each base component in the initial surface shape expression base is assigned a corresponding initial weight value, and all initial weight values are arranged in the order of the base components in the initial surface shape expression base to generate an initial weight value sequence. The initial surface shape representation basis is associated with and stored with the initial weight value sequence to generate an initial vector of surface coefficients.
4. The method for designing a globally tunable optical surface based on nonlinear optimization as described in claim 1, characterized in that, The method for establishing a comprehensive performance evaluation system based on the surface design constraints, comprising an imaging evaluation component and a stray light suppression evaluation component, wherein the imaging evaluation component and the stray light suppression evaluation component are correlated through a weighting factor, includes: The imaging wavefront constraint boundary is read from the surface design constraint conditions. The allowable wavefront deviation range corresponding to each field of view position is determined according to the imaging wavefront constraint boundary. The spatial surface shape corresponding to the surface coefficient vector is compared with the ideal wavefront shape at each field of view position. The wavefront deviation degree value at each field of view position is generated according to the comparison result. The imaging evaluation result is generated according to the wavefront deviation degree value at all field of view positions. The stray light suppression index constraint boundary is read from the surface design constraints. The allowable stray light suppression index deviation range corresponding to each radial position is determined according to the stray light suppression index constraint boundary. The surface orientation of the spatial surface shape corresponding to the surface coefficient vector at each radial position is compared with the stray light suppression target orientation. The stray light suppression index deviation value at each radial position is generated according to the comparison result. The stray light suppression evaluation result is generated according to the stray light suppression index deviation values at all radial positions. A preset weighting factor is invoked, which is used to adjust the proportion of the imaging evaluation result and the stray light suppression evaluation result in the comprehensive evaluation. The imaging evaluation result and the stray light suppression evaluation result are correlated according to the weighting factor to generate a comprehensive evaluation value. A comprehensive performance evaluation method is established based on the correspondence between the comprehensive evaluation value and the surface coefficient vector.
5. The method for designing a globally tunable optical surface based on nonlinear optimization as described in claim 1, characterized in that, Based on the comprehensive performance evaluation method, the initial vector of the surface coefficients is repeatedly adjusted until the stopping condition is met, at which point the optimal surface coefficient vector is output. Each adjustment determines the adjustment direction and magnitude by constructing local approximation conditions, including: The comprehensive performance evaluation method is invoked, and the spatial surface shape corresponding to the current surface coefficient vector is used as input to determine the current evaluation value of the comprehensive performance evaluation method at the current surface coefficient vector. The values of each coefficient component in the current surface coefficient vector are changed, and the changes in the evaluation value of the comprehensive performance evaluation method after each change are monitored. Based on the changes in the evaluation value, the trend information of the change of the comprehensive performance evaluation method along the direction of each coefficient component is determined. Based on the current evaluation value, the trend information, and the curvature constraint matrix dynamically constructed based on the gradient information of the current evaluation function, a local approximation condition is established in the neighborhood of the current surface coefficient vector. The local approximation condition is expressed in the following form: ; In the formula, This represents the adjustment vector to be determined. This represents the current evaluation value of the comprehensive performance evaluation method at the current surface coefficient vector. This represents a first-order change vector composed of the aforementioned trend information. This represents the symmetric curvature matrix constructed at the current surface coefficient vector based on the local curvature information of the comprehensive performance evaluation method; the local curvature information is obtained by perturbing the first-order change vector using numerical difference, or by recursively obtaining it from historical iteration information using a quasi-Newton update formula. Under the local approximation conditions, the adjustment vector corresponding to the minimum value of the local approximation conditions is determined, and the adjustment vector is used as a joint representation of the adjustment direction and adjustment magnitude. Based on the adjustment vector, the current surface coefficient vector is updated to generate the updated surface coefficient vector; The updated surface coefficient vector is used as the current surface coefficient vector for the next adjustment, and this process is repeated until the stopping condition is met, at which point the optimal surface coefficient vector is output. The symmetric curvature matrix H is updated during each adjustment to match the true second-order variation characteristics of the comprehensive performance evaluation method within the neighborhood of the current surface coefficient vector.
6. The method for designing a globally tunable optical surface based on nonlinear optimization as described in claim 5, characterized in that, The method based on the comprehensive performance evaluation involves repeatedly adjusting the initial vector of the surface coefficients until a stopping condition is met, at which point the optimal surface coefficient vector is output. Each adjustment determines the direction and magnitude of the adjustment by constructing local approximation conditions. The method also includes: After an adjustment is completed and an updated surface coefficient vector is generated, the comprehensive performance evaluation method is invoked. The spatial surface shape corresponding to the updated surface coefficient vector is used as input to determine the updated evaluation value of the comprehensive performance evaluation method at the updated surface coefficient vector. Read the degree of deviation between the updated evaluation value and the historical evaluation value recorded after the last adjustment, and use the degree of deviation as the change in the evaluation value; The change in the evaluation value is compared with a preset allowable change threshold. When the change in the evaluation value is less than or equal to the preset allowable change threshold, the stopping condition is determined to be met. After determining that the stopping condition is met, the updated surface coefficient vector is output as the optimal surface coefficient vector.
7. The method for designing a globally tunable optical surface based on nonlinear optimization as described in claim 1, characterized in that, The process of generating complete surface shape data containing discrete height distribution and continuous surface shape description based on the combination of the optimal surface coefficient vector and the initial surface shape representation basis includes: Retrieve the optimal surface coefficient vector and the initial surface shape representation basis, and pair each optimal coefficient in the optimal surface coefficient vector with the corresponding basis component in the initial surface shape representation basis to generate weighted surface shape components corresponding to each basis component; Within the surface shape representation space, a sampling position sequence is determined, and all weighted surface shape components are superimposed at each sampling position to generate superposition height values at each sampling position. The superposition height values at all sampling positions are then arranged into a discrete height distribution. Based on the superimposed height values at each sampling location in the discrete height distribution, establish a correspondence table between sampling locations and superimposed height values; The preset continuous interpolation rule is invoked, and the height change trend between adjacent sampling positions in the correspondence table is used as the interpolation basis. Supplementary height values are inserted between adjacent sampling positions to generate a supplementary height value sequence covering the entire surface expression space. The superimposed height value and the supplementary height value sequence are merged to generate a full-caliber height value distribution; Based on the full-caliber height value distribution, a surface shape description relationship between position coordinates and height values is established. This surface shape description relationship is used as the continuous surface shape description, and the discrete height distribution and the continuous surface shape description are used together as complete surface shape data.
8. The method for designing a globally tunable optical surface based on nonlinear optimization as described in claim 7, characterized in that, The formation method of the continuous surface shape description includes: Read the sampling position and superposition height value in the corresponding relationship table, and use each sampling position as a support point, and use the superposition height value at each sampling position as the height constraint value at the support point; The preset radial base interpolation method is invoked, and the position coordinates and height constraint values of all support points are used as input conditions to configure a corresponding radial base shape for each support point. The radial base shape takes the height constraint value at the support point and decays with distance at positions away from the support point. Based on the radial base morphology corresponding to all support points, a transition height value is generated between each adjacent support point. The transition height value is jointly determined by the radial base morphology corresponding to each adjacent support point, and the transition height values contributed by the radial base morphology on both sides are equal at the midpoint of the adjacent support points. The height constraint values at all support points and the transition height values between all adjacent support points are merged to form the surface shape description relationship, which is then used as a continuous surface shape description.
9. The method for designing a globally tunable optical surface based on nonlinear optimization as described in claim 1, characterized in that, Based on the surface design constraints, the complete surface data is jointly verified. If the verification passes, the complete surface data is output. If the verification fails, the weighting factor is adjusted and Pt.2 to Pt.4 are repeated until the verification passes, including: When the complete surface data fails the joint verification, the imaging verification result and stray light suppression verification result generated during the joint verification process are read. The imaging verification result is compared with the imaging constraint boundary of the optical design input parameters to determine the degree of non-compliance of the imaging part. The stray light suppression verification result is compared with the stray light suppression constraint boundary of the optical design input parameters to determine the degree of non-compliance of the stray light suppression part. The degree of dissatisfaction of the imaging part is compared with the degree of dissatisfaction of the stray light suppression part. Based on the comparison result, the dominant dissatisfaction dimension is determined. The dominant dissatisfaction dimension is the one with the larger degree of dissatisfaction value between the imaging dimension and the stray light suppression dimension. The adjustment direction of the weighting factor is determined based on the dominant non-satisfied dimension. When the dominant non-satisfied dimension is the imaging dimension, the weighting factor is adjusted in the direction of increasing the proportion of the imaging evaluation part. When the dominant non-satisfied dimension is the stray light suppression dimension, the weighting factor is adjusted in the direction of increasing the proportion of the stray light suppression evaluation part. Based on the difference between the degree of dissatisfaction corresponding to the dominant dissatisfaction dimension and the degree of dissatisfaction of another dimension, the adjustment step size of the weighting factor is determined. The adjustment step size is positively correlated with the difference. The current weighting factor is updated according to the adjustment direction and the adjustment step size to generate the updated weighting factor. The updated weighting factor is fed back to the establishment stage of the comprehensive performance evaluation method. The comprehensive performance evaluation method is re-established with the updated weighting factor, and the iterative adjustment, the generation of the complete surface data, and the joint verification are executed again in sequence until the complete surface data passes the joint verification.
10. A globally tunable optical surface design system based on nonlinear optimization, characterized in that, The system for implementing the globally tunable optical surface design method based on nonlinear optimization as described in claim 1 includes: The parameter basis construction module is used to obtain the optical design input parameters of the implantable ophthalmic endoscope, generate surface design constraints from the optical design input parameters, and construct the initial surface shape expression basis and the initial vector of surface coefficients. The comprehensive evaluation modeling module is used to establish a comprehensive performance evaluation method that includes an imaging evaluation part and a stray light suppression evaluation part based on the surface design constraints. The imaging evaluation part and the stray light suppression evaluation part are associated through a weighting factor. The surface coefficient optimization module is used to repeatedly adjust the initial vector of the surface coefficients based on the comprehensive performance evaluation method until the stopping condition is met, and then output the optimal surface coefficient vector. Each adjustment determines the adjustment direction and adjustment range by constructing local approximation conditions. The complete surface shape generation module is used to generate complete surface shape data containing discrete height distribution and continuous surface shape description based on the combination of the optimal surface coefficient vector and the initial surface shape expression basis; The constraint verification iteration module is used to jointly verify the complete surface data based on the surface design constraints. If the verification passes, the complete surface data is output. If the verification fails, the weighting factor is adjusted and the comprehensive evaluation modeling module is repeated to the complete surface generation module until the verification passes.
