Method, device, equipment and storage medium for predicting lens manufacturing related information

Abstract

This application discloses a method, apparatus, device, and storage medium for predicting lens manufacturing-related information. Based on a pre-trained mapping model, this application determines the mapping relationship between first and second information related to lens manufacturing. The first information is the manufacturing information of the lens elements constituting the lens, and the second information is the performance information of the lens assembled from the lens elements. Based on selected first information, the mapping model predicts the corresponding second information; conversely, based on selected second information, the mapping model predicts the corresponding first information. This application allows for the prediction of lens performance information based on the mapping relationship between lens manufacturing information and lens performance information before the lens elements are assembled into a lens. This facilitates the identification of lenses that do not meet performance requirements, improving lens yield. Furthermore, it allows the identification of suitable lenses based on lens performance information.

Description

Technical Field

[0001] This application generally relates to the technical field of assembling optical lenses into optical lenses. More specifically, this application relates to a method, apparatus, device, and storage medium for predicting information related to lens manufacturing. Background Technology

[0002] In the field of optical imaging for mobile devices, manufacturers of mobile phones, handheld cameras, drones and other terminals have a continuous pursuit of improving image quality and reducing costs. As one of the core components of imaging, improving the manufacturing precision and yield of lenses is the key to achieving the above goals.

[0003] Lenses are typically composed of multiple aspherical lens elements assembled together. Aspherical lenses have many parameters, such as thickness and aspherical coefficient. Even a slight error in any of these parameters can degrade the performance of the assembled lens, resulting in substandard lenses and lower manufacturing yield. In mass production, the overall lens manufacturing yield gradually decreases as the number of lens elements increases. Due to limitations in cost and efficiency, current methods cannot inspect the lenses manufactured in large quantities; substandard products are only discovered after assembly, leading to a low lens manufacturing yield.

[0004] Therefore, how to quickly and cost-effectively inspect lenses before assembling them to improve lens manufacturing yield has become an urgent problem to be solved. Summary of the Invention

[0005] This application provides a method, apparatus, device, and storage medium for predicting lens manufacturing-related information, in order to partially solve the aforementioned problems existing in the prior art.

[0006] The embodiments of this application adopt the following technical solutions:

[0007] In a first aspect, this application provides a method for predicting lens manufacturing-related information, the method comprising: determining a mapping relationship between first information and second information related to lens manufacturing based on a pre-trained mapping model, wherein the first information is manufacturing information of the lens elements constituting the lens, and the second information is performance information of the lens assembled from the lens elements; predicting corresponding second information based on selected first information using the mapping model, or predicting corresponding first information based on selected second information using the mapping model.

[0008] Optionally, the mapping model is pre-trained as follows: for each lens manufacturing mold, obtain several lenses manufactured by that lens manufacturing mold as sample lenses; obtain the manufacturing information of each sample lens; assemble one or more sample lenses into a lens as a sample lens, and determine the sample performance information of the sample lens; based on the manufacturing information of each sample lens and the mapping model to be trained, predict the performance information to be optimized of the sample lens; and train the mapping model to be trained according to the sample performance information and the performance information to be optimized.

[0009] Optionally, a number of lenses manufactured by the lens manufacturing mold are obtained as sample lenses, specifically including: obtaining a number of lenses manufactured by the lens manufacturing mold as candidate lenses; for each candidate lens, obtaining the actual manufacturing information of the candidate lens and the standard manufacturing information of the candidate lens when designing the candidate lens; determining the difference between the actual manufacturing information and the standard manufacturing information as the difference corresponding to the candidate lens; and selecting a sample lens from the number of candidate lenses based on the difference corresponding to each candidate lens.

[0010] Optionally, based on the selected first information, the corresponding second information is predicted using the mapping model, specifically including: selecting at least one lens manufacturing mold as a designated mold from a plurality of lens manufacturing molds; obtaining the manufacturing information of the lens manufactured by the designated mold as the selected first information; and predicting the performance information of the lens assembled from the lens manufactured by the designated mold according to the mapping model and the selected first information.

[0011] Optionally, among a plurality of lens manufacturing molds, at least one lens manufacturing mold is selected as a designated mold, specifically including: based on the mapping model, for each lens manufacturing mold, determining the overlap range between the performance information of the lens assembled by the lens manufactured by that lens manufacturing mold and the performance information of the lens assembled by the lens manufactured by at least one other lens manufacturing mold; and selecting the designated mold among the lens manufacturing molds according to the determined overlap range.

[0012] Optionally, based on the mapping model and the selected first information, predicting the performance information of the lens assembled from the lens manufactured by the specified mold specifically includes: using the lens manufactured by the specified mold as the specified lens; obtaining a lens surface topography image of the specified lens; determining a feature image of the specified lens based on the lens surface topography image; and predicting the performance information of the lens assembled from the specified lens based on the feature image and the mapping model.

[0013] Optionally, determining the feature image of the specified lens based on the lens surface topography map specifically includes: determining the height distribution map of the specified lens based on the lens surface topography map; and determining the feature image of the specified lens based on the height distribution map.

[0014] Optionally, determining the feature image of the specified lens based on the lens surface topography image specifically includes: obtaining the standard manufacturing information of the specified lens when designing the specified lens; determining the actual difference information corresponding to the specified lens based on the lens surface topography image and the standard manufacturing information; and determining the feature image of the specified lens based on the actual difference information.

[0015] Optionally, determining the feature image of the specified lens based on the lens surface topography image specifically includes: performing Fourier transform processing on the lens surface topography image; and determining the feature image of the specified lens based on the processed lens surface topography image.

[0016] Optionally, determining the feature image of the specified lens based on the lens surface topography image specifically includes: inputting the lens surface topography image into a feature extraction model to obtain the feature image of the specified lens output by the feature extraction model.

[0017] Optionally, the method further includes: adjusting the selected first information; predicting second information corresponding to the adjusted first information using the mapping model based on the adjusted first information; determining the difference between the second information corresponding to the selected first information and the second information corresponding to the adjusted first information; and selecting information from the lens performance information for processing based on the determined difference.

[0018] Optionally, based on the selected second information, the corresponding first information is predicted using the mapping model, specifically including: the selected second information includes the standard performance information of the target lens when designing the target lens; based on the standard performance information, the manufacturing information of the lens required for assembling the target lens is predicted using the mapping model.

[0019] Optionally, the method further includes: selecting at least one lens manufacturing mold from among a plurality of lens manufacturing molds, based on predicted first information and first information of lenses manufactured by each lens manufacturing mold, to obtain the target lens assembled from lenses manufactured by the selected lens manufacturing mold.

[0020] Optionally, the manufacturing information includes mold manufacturing information and / or mold usage information for manufacturing lenses. The mold manufacturing information includes at least one of the following: mold cavity surface roughness, machining tolerance, contour error, and flow channel structure parameters, venting structure parameters, and water channel structure parameters connected to the mold cavity. The mold usage information includes at least one of the following: barrel temperature, nozzle temperature, mold temperature, heating gradient, material type, material batch, injection pressure, holding pressure, injection speed, molding time, holding time, cooling time, and annealing time. The performance information includes at least one of the following: resolution, modulation transfer function (MTF), illuminance, aberrations, stray light, ghosting, depth of field, surface defects, maximum principal ray angle, focal length, aperture, field of view, and maximum image height.

[0021] In a second aspect, this application provides an apparatus for predicting lens manufacturing-related information. The apparatus includes: a determining module, configured to determine a mapping relationship between first information and second information related to lens manufacturing based on a pre-trained mapping model, wherein the first information is manufacturing information of the lens elements constituting the lens, and the second information is performance information of the lens assembled from the lens elements; and a predicting module, configured to predict corresponding second information based on selected first information using the mapping model, or to predict corresponding first information based on selected second information using the mapping model.

[0022] In a third aspect, this application provides a computer-readable storage medium, characterized in that the storage medium stores a computer program, which, when executed by a processor, implements the above-described method for predicting lens manufacturing related information.

[0023] In a fourth aspect, this application provides an electronic device including a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the program to implement the aforementioned method for predicting lens manufacturing related information.

[0024] The above-described technical solutions adopted in the embodiments of this application can achieve the following beneficial effects: This application, based on a pre-trained mapping model, determines the mapping relationship between first and second information related to lens manufacturing. The first information is the manufacturing information of the lens elements constituting the lens, and the second information is the performance information of the lens assembled from the lens elements. Based on selected first information, the mapping model predicts the corresponding second information; conversely, based on selected second information, the mapping model predicts the corresponding first information. Through this application, before the lens elements are assembled into a lens, the lens performance information can be predicted based on the mapping relationship between lens manufacturing information and lens performance information. This allows for the identification of lenses that do not meet performance requirements, preventing substandard lenses from entering subsequent processes before assembly. This solves the problem in existing technologies where substandard lens performance can only be detected after assembly, thus improving lens manufacturing yield.

[0025] Furthermore, based on the mapping relationship between lens manufacturing information and lens performance information, this application can determine the conditions that the lens manufacturing information needs to meet before manufacturing the lens, in order to manufacture a lens that meets the requirements and thus improve the lens manufacturing yield. Attached Figure Description

[0026] The above and other objects, features, and advantages of exemplary embodiments of this application will become readily understood by reading the following detailed description with reference to the accompanying drawings. In the drawings, several embodiments of this application are illustrated by way of example and not limitation, and the same or corresponding reference numerals denote the same or corresponding parts, wherein: Figure 1 A schematic diagram of the mold flow analysis simulation results of the optical lens injection molding process provided in this application embodiment; Figure 2 A schematic diagram of the cross-section of the optical lens and a schematic diagram of the lens contour scanning results are provided for embodiments of this application; Figure 3 A flowchart illustrating a method for predicting lens manufacturing-related information provided in this application embodiment; Figure 4 A flowchart illustrating another method for predicting lens manufacturing-related information provided in this application embodiment; Figure 5 A schematic diagram of the structure of an apparatus for predicting lens manufacturing-related information provided in an embodiment of this application; Figure 6 A schematic diagram of an electronic device provided in an embodiment of this application. Detailed Implementation

[0027] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this application. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0028] The following explains the terms related to the manufacture of optical lenses and the performance of optical lenses in this application.

[0029] I. Lens Structure and Geometric Parameter Terminology Center thickness: refers to the thickness at the geometric center of the lens, and is a core structural parameter of the lens.

[0030] Aspherical coefficient: A coefficient used to define the surface profile of an aspherical lens, determined by the aspherical equation to account for the curvature variation of the lens surface.

[0031] Outer diameter: The maximum radial dimension of the lens, used to determine the assembly dimensions between the lens and the lens barrel.

[0032] Sagittal height: The vertical distance from the reference plane at the edge of the lens surface to the highest point of the lens surface, representing the degree of curvature of the lens surface.

[0033] Fastening diameter: The radial dimension of the lens used for fastening and positioning with the lens barrel or spacer, ensuring the radial positioning accuracy of the lens.

[0034] Support position: The position where the lens is axially supported and limited within the lens barrel, used to control the assembly position of the lens along the optical axis.

[0035] Eccentricity: The radial offset of the geometric center of the lens relative to the optical axis of the optical system, which is an assembly error.

[0036] Tilt: The tilt angle of the lens surface normal relative to the optical axis, which is an assembly error.

[0037] II. Lens Surface Precision and Quality Terminology Peak-to-Valley (PV): The difference between the highest and lowest points on the lens surface, representing the accuracy of the lens surface shape.

[0038] Arithmetic Mean Roughness (Ra): The arithmetic mean of the microscopic irregularities on the surface of a lens, characterizing the smoothness of the surface.

[0039] Root Mean Square (RMS): The root mean square value of surface profile error, used to more accurately characterize surface quality.

[0040] Lens contour error: the deviation between the actual surface of the lens and the ideal design surface shape.

[0041] Aspherical equations: mathematical expressions used to accurately describe the surface shape of aspherical lenses.

[0042] III. Mold-related terminology (compression molding / injection mold) Mold material: The material used to manufacture lens forming molds, which affects the mold life and forming accuracy.

[0043] Mold coating: A functional film layer applied to the surface of a mold cavity to improve demolding performance, wear resistance, and surface quality.

[0044] Mold cavity inner surface roughness: The surface smoothness of the mold forming cavity directly determines the arithmetic mean roughness Ra of the lens.

[0045] Machining tolerance: Allowable dimensional deviations during mold machining.

[0046] The contour error of the mold cavity refers to the overall geometric deviation between the actual inner surface contour shape of the mold cavity and the theoretical design reference contour, which includes radial deviation of the contour, surface undulation, curvature deviation and local concave and convex deformation.

[0047] Damage caused by mold use: wear, scratches, deformation and other damage caused by repeated molding.

[0048] The runner structure parameters connected to the mold cavity refer to the geometric and structural dimensional parameters of the gating system in the injection mold, from the main runner and branch runners to the gating system before entering the lens cavity. These parameters include all structural features such as runner cross-sectional shape, runner diameter / cross-sectional length and width, runner length, runner taper, corner fillets, branch layout, runner slope, and gate connection dimensions. These parameters are used to control melt flow rate, flow velocity, shear rate, flow balance, pressure loss, and temperature holding capacity. They are key pre-construction structural parameters that directly determine the filling pattern, residual stress, and surface replication accuracy inside the mold cavity.

[0049] Flow channel length and shape: The size and shape of the material flow channels in an injection or molding system.

[0050] Nozzle shape: The structural shape of the injection molding machine nozzle affects the stability of material injection.

[0051] Gate shape and size: The structure and size of the entrance for material to enter the mold cavity, which affects the filling effect and appearance quality.

[0052] Venting structure parameters: These include structural parameters such as the depth, width, length, spacing, number of venting stages, and pressure relief channel dimensions of the venting grooves at the corresponding positions of the mold cavity. These parameters are used to promptly discharge trapped air inside the mold cavity and volatile gases generated by the resin during the injection molding process.

[0053] Venting structure location: The structural location in the mold used to vent gas from the mold cavity to avoid defects such as air bubbles and insufficient glue.

[0054] Water channel structural parameters: These include geometric parameters such as the diameter of the cooling water channels, the distance between the water channels and the mold cavity, the water channel layout, the water channel direction, the bend radius, and the symmetrical arrangement of the water channels. These parameters refer to the geometric dimensions and layout characteristics of the cooling water channels arranged around the mold cavity in the injection mold. They are used to control the cooling rate, temperature field uniformity, heat dissipation efficiency, and thermal balance of the mold and mold cavity. They are the core structural parameters for regulating the shrinkage, deformation, residual stress, and surface accuracy of the lens during injection molding.

[0055] The positional distribution of water channels relative to runners and gates: The layout of the mold cooling water channels affects the uniformity of cooling and molding efficiency.

[0056] IV. Injection Molding / Compression Molding Process Terminology Barrel temperature: The heating temperature of the injection molding machine barrel affects the molten state of the material.

[0057] Nozzle temperature: The temperature at the nozzle position of the injection molding machine.

[0058] Mold temperature: The forming temperature of the mold cavity affects the lens precision and internal stress.

[0059] Temperature gradient: The rate of temperature change during the heating process.

[0060] Material type: Model of optical plastic or optical glass.

[0061] Material batches: Different production batches of the same type of optical material can affect consistency.

[0062] Injection pressure: The pressure that propels material into the mold cavity during injection molding.

[0063] Holding pressure (also known as pressure holding pressure): The amount of pressure maintained inside the mold cavity after filling is completed.

[0064] Injection speed: The volume or rate at which material is injected into the mold cavity per unit time.

[0065] Molding time: The total time required to complete one molding cycle.

[0066] Holding time: The duration of the holding phase.

[0067] Cooling time: The time it takes for the lens to cool and solidify inside the mold.

[0068] Annealing time: The heat treatment time used to eliminate internal stress in the lens.

[0069] V. Lens Optical Performance Terminology Modulation Transfer Function (MTF): A core indicator for evaluating the image sharpness, clarity, and resolution of a lens.

[0070] Resolution: The MTF value characterizes the imaging resolution at different spatial frequencies.

[0071] Aberrations: Optical deviations that affect image quality, including spherical aberration, coma, astigmatism, field curvature, chromatic aberration, distortion, etc.

[0072] Color difference: Color deviation caused by inconsistent focusing of light of different wavelengths.

[0073] Optical distortion: is one of the inherent monochromatic aberrations of an optical system. It refers to the relative deviation between the actual principal ray image height and the ideal paraxial image height at different field of view positions. It is used to characterize the degree of bending deformation that occurs after an optical lens images a straight object on the object side, including barrel distortion and pincushion distortion.

[0074] TV distortion is a quantitative indicator of the overall shape distortion of an imaging system on the image plane, representing the geometric deformation caused by the inconsistent magnification of the image edges relative to the center.

[0075] Field curvature: The ideal imaging surface is a plane, but in reality it is a curved surface, which leads to blurred edges in the field of view.

[0076] Illuminance: Generally refers to relative illuminance, the ratio of the brightness at the edge of the image plane to the brightness at the center, which characterizes the uniformity of brightness.

[0077] Stray light: Interference light generated when scattered or reflected light, which is not required for imaging, reaches the image plane.

[0078] Ghosting: A false image formed by multiple reflections between lenses.

[0079] Depth of field: The range of object distances from which a lens can produce a clear image.

[0080] Maximum Chief Ray Angle Max (CRAmax): The maximum angle between the chief ray and the optical axis.

[0081] Effective Focal Length (EFL): This refers to the axial distance between the principal plane of an optical system and the point where parallel incident rays converge.

[0082] F-number: denoted as f / #, it is the aperture value of a lens, which is the ratio of the effective focal length (EFL) of the optical system to the entrance pupil diameter, and represents the light transmission capability.

[0083] Field of view (FOV): refers to the maximum range of a scene that an optical lens can clearly image, usually expressed as diagonal field of view, horizontal field of view, and vertical field of view.

[0084] Maximum image height (ImgH): The maximum effective imaging height of the image plane, which refers to the radial height of the image plane corresponding to the maximum field of view in the optical system, that is, the vertical distance from the center of the optical axis to the outermost edge of the effective imaging area.

[0085] Defective appearance: Visual defects such as stains, cracks, bubbles, scratches, and missing glue appear on the lens surface.

[0086] In existing technologies, lenses are typically composed of multiple aspherical lens elements assembled together. Currently, aspherical lenses are usually manufactured using conventional plastic processing techniques such as injection molding. During manufacturing, unavoidable errors occur in mold processing, process errors, and coating errors due to the low-reflectivity coating on the lenses. This leads to significant deviations between the actual optical performance of the aspherical lenses and their theoretical design, making it difficult to achieve ideal performance and manufacturing yield in the assembled lenses. For example, these errors can cause inaccuracies in various parameters of the aspherical lenses, such as thickness, aspherical coefficient, outer diameter, sagitta, snapping diameter, and mounting position. Even an error of only 1-2 micrometers in any parameter, such as eccentricity, tilt, PV, or Ra, can degrade the specific field-of-view performance of the assembled lens, resulting in defective products. Furthermore, since a lens is assembled from multiple aspherical lens elements, the various errors within these elements can have a cumulative coupling effect, collectively affecting the overall imaging performance of the lens.

[0087] In mass production scenarios involving millions of units, lens manufacturing yield decreases with the increasing number of lens elements, and the fluctuations become more pronounced. For example, it's common for some batches to have a yield of only 50%, while others plummet to 10% or even lower. Limited by production costs and inspection efficiency, current technologies cannot perform real-time inspection of mass-produced lenses; defective lenses are only detected after assembly, making it impossible to predict lens defects beforehand and prevent the production of defective lenses. Therefore, there is an urgent need for a novel, low-cost technology that can quickly detect lens errors before assembly to improve lens manufacturing yield.

[0088] In view of this, embodiments of this application provide a method for predicting lens manufacturing-related information. This method uses a pre-trained mapping model to determine the mapping relationship between lens manufacturing information and lens performance information. Based on the lens manufacturing information, the mapping model predicts the corresponding lens performance information; or, based on the performance information of a selected lens, the mapping model predicts the corresponding lens manufacturing information. On one hand, this application can predict the performance information of a lens assembled from lenses based on their manufacturing information after lens manufacturing but before the lenses are assembled into a lens. This allows for prediction of whether a lens is defective before assembly, thereby improving lens manufacturing yield. On the other hand, this application can also, after determining lens performance information (e.g., designing standard lens performance information), predict the lens manufacturing information required for manufacturing the lens based on that performance information, thereby manufacturing the lens according to the lens manufacturing information and improving lens manufacturing yield.

[0089] The specific embodiments of this application will now be described in detail with reference to the accompanying drawings.

[0090] Figure 1 This is a schematic diagram of the mold flow analysis simulation results of the optical lens injection molding process provided in the embodiments of this application.

[0091] like Figure 1 As shown, Figure 1 Figures (a)-(d) in the figure fully illustrate the melt filling process. Figure 1 Figure (e) in the diagram illustrates the key optical analysis results. Figure 1 Figures (a)-(d) in the figure are dynamic process simulation diagrams of the melt filling stage. They show the entire process of the plastic melt entering the cavity from the gate and gradually filling the lens mold in chronological order. Among them, region 01 is the unfilled area (also known as the cavity), and region 02 is the filled area. In the injection filling stage, after the molten resin enters the lens cavity from the gate, the melt front gradually advances from the side of the gate to the opposite side. Figure 1 Figure (a) shows the melt distribution in the initial stage of filling. As soon as the melt enters the mold, it moves from the right gate to the left, forming a distinct "semi-circular" melt front. Figure 1 Figure (b) shows the melt continuing to expand, with the leading edge gradually widening and flowing towards the center of the lens; Figure 1 Figure (c) shows that the melt almost fills most of the area, leaving only a small cavity, and the trend of melt merging at the gate can be seen; Figure 1 Figure (d) shows the filling process nearing completion, with only a tiny "final filling point" (a risk point for trapped gas) remaining opposite the gate. This filling process simulation clearly reflects the melt flow path, the shape of the leading edge, and the distribution at the end of the filling process, and can be used to assess filling balance, risk of trapped gas, and location of weld line formation. Figure 1Figure (e) shows the simulated residual stress distribution cloud map of the lens after molding, which is the core optical result of the lens mold flow analysis. The center of the lens is the low stress area, the edge of the lens is the medium stress area (near the gate and parting surface), and the area near the gate is the high stress area. The contour lines represent the stress distribution gradient. The denser the area, the more drastic the stress change and the more severe the birefringence, which directly affects the optical uniformity and imaging quality of the lens.

[0092] Figure 2 These are schematic diagrams of the optical lens cross-section and lens contour scanning results provided for embodiments of this application. Figure 2 As shown, Figure 2 Figure (a) shows a microscopic tomographic image of the lens cross-section. Figure 2 Figure (b) is Figure 2 The outline scan curve of the lens in the lens circled in area 20 of Figure (a).

[0093] Figure 2 Figure (a) shows the cross-section of an optical lens. The lens comprises multiple optical lenses and a supporting structure, allowing clear observation of the cross-sectional contours of each lens, the interlayer fit, and forming defects within the lens and at the interfaces. Taking the circled area 20 as an example, a three-dimensional contour scan of the lens is performed, and the detection results are as follows. Figure 2 As shown in Figure (b).

[0094] Figure 2 Figure (b) shows the curve obtained by contour scanning of the optical effective surface of the circled area 20. The horizontal axis is the scan position coordinate X (unit: mm), and the vertical axis is the surface height coordinate Y (unit: mm). This curve reflects the actual surface contour of the lens surface along the scanning direction. The curve shows a linear trend, corresponding to the theoretical design slope of the lens. The arrows on the curve mark several local fluctuations, which correspond to the surface deviations in the circled area (i.e., abnormal fluctuations compared to its aspherical equation). This indicates that there are errors between the actual molded lens surface and the theoretically designed surface. These errors are all generated during the lens injection molding process, showing abnormal characteristics at the interface between the lens's optical effective surface and the supporting structure. These characteristics are local surface deviations or other interface defects caused by uneven melt flow or holding pressure during the molding process, which will directly affect the optical imaging quality of the lens. These errors can be characterized by various methods such as peak-to-valley value PV, root mean square error RMS, and surface roughness error Ra, and cause the lens's aberrations, MTF resolution, and other performance indicators to fail to meet the standards, thus producing defective lenses.

[0095] The discrepancy between the actual surface of the formed lens and the theoretically designed surface can be caused by various manufacturing information during the lens manufacturing process, such as mold manufacturing information and mold usage information. The optical imaging quality of the lens directly affects the performance information of the lens.

[0096] In actual production, various delays and unevenness may occur in the injection filling wavefront. These are related to the inherent errors in the processing of each mold cavity, such as mold material, mold coating, surface roughness of the mold cavity, processing tolerance, contour error, and mold use damage. They are also related to other inherent structures of the mold cavity, such as runner length and shape, nozzle shape, gate shape and size, venting structure location, and the positional distribution of water channels relative to the runner and gate. Furthermore, they are related to variable process parameters used when using the mold, such as barrel temperature, nozzle temperature, mold temperature, heating gradient, material type, material batch, injection pressure, holding pressure, injection speed, molding time, holding pressure time, cooling time, and annealing time. Due to the large number and randomness of inherent and variable parameters, and the inability to stop the machine for real-time analysis and detection during mass production, it is impossible to link these adjustable mold parameters with the final performance of the product. Even if the product is defective, it is impossible to analyze the specific cause of the problem and which parameters can be adjusted to improve the problem.

[0097] The method for predicting lens manufacturing information provided in this application employs artificial intelligence (AI) techniques. It trains an AI model using measured surface processing defects and product performance to continuously improve the molding process in real-time without affecting production. Furthermore, it predicts and identifies potentially defective lenses before assembly, ultimately improving product yield. Specifically, this application establishes a mapping relationship between lens manufacturing information and lens performance information. Based on this mapping relationship, it predicts lens performance information given lens manufacturing information, and pre-determines whether the lens meets standards. This effectively detects lenses that do not meet lens standards before assembling them into a lens, thereby improving lens manufacturing yield. Alternatively, based on the mapping relationship, it predicts the manufacturing information of the lenses to be assembled into the lens given lens performance information, adjusting relevant parameters during lens manufacturing to produce lenses that meet performance requirements, thus improving lens manufacturing yield.

[0098] Figure 3 A flowchart illustrating a method for predicting lens manufacturing-related information provided in this application embodiment may specifically include the following steps: S310: Based on a pre-trained mapping model, determine the mapping relationship between first information and second information related to lens manufacturing, wherein the first information is the manufacturing information of the lens elements constituting the lens, and the second information is the performance information of the lens assembled from the lens elements; S330: Based on the selected first information, use the mapping model to predict the corresponding second information, or based on the selected second information, use the mapping model to predict the corresponding first information.

[0099] In step S310, the first information is the manufacturing information of the lens constituting the lens. In some embodiments of this application, the first information (i.e., manufacturing information) includes mold manufacturing information and / or mold usage information when manufacturing the lens. The mold manufacturing information includes at least one of the following: mold cavity surface roughness, machining tolerance, contour error, and flow channel structure parameters, venting structure parameters, and water channel structure parameters connected to the mold cavity. The mold usage information includes at least one of the following: barrel temperature, nozzle temperature, mold temperature, heating gradient, material type, material batch, injection pressure, holding pressure, injection speed, molding time, holding pressure time, cooling time, and annealing time. The second information is the performance information of the lens assembled from the lens. The second information (i.e., performance information) includes at least one of the following: resolution, modulation transfer function (MTF), illuminance, aberration, stray light, ghosting, depth of field, appearance defects, maximum principal ray angle, focal length, aperture, field of view, and maximum image height.

[0100] For explanations of the terms related to optical lenses and optical lenses in the first and second information, please refer to the above content, which will not be repeated here.

[0101] In some embodiments of this application, the mapping model may include a neural network model or other machine learning models. The neural network model may be a convolutional neural network model with residual connections, including several convolutional layers for extracting core features, pooling layers for compressing the extracted features, a non-linear activation function, a series of fully connected layers for output dimensionality reduction, and a softmax layer for output. Ultimately, it transforms the surface information corresponding to the input manufacturing information into a series of high-dimensional vector sequences that can be processed by machines. The neural network model may also be a ViT structure model, composed of multiple transformer encoders. The transformer encoder includes parallel multi-head self-attention weighting, a multilayer perceptron, layer normalization, and residual connection mechanisms, thereby transforming image patches into vector sequences.

[0102] This application can select different types of machine learning models as mapping models. The selected machine learning model can transform the surface information of the lens into a series of feature vectors, extract key features from the surface information, and realize the mapping relationship between the lens manufacturing information and the lens performance information based on the mapping model in this application. The specific details of the pre-trained mapping model in this application will be described in detail below.

[0103] In step S330, firstly, based on the manufacturing information of the selected lens, this application can use a mapping model to predict the performance information of the lens composed of the selected lens.

[0104] To further explain, this application can select at least one lens manufacturing mold as a designated mold from a plurality of lens manufacturing molds; obtain the manufacturing information of the lens manufactured by the designated mold as the selected first information; and predict the performance information of the lens assembled from the lens manufactured by the designated mold based on the mapping model and the selected first information.

[0105] In the actual lens injection molding process, multiple lens manufacturing molds can be used. Since the mold information of each lens manufacturing mold is different—for example, the surface roughness of the mold cavity varies—the surface profile information of the lenses injection molded from these molds will differ. This application can predict the performance information of lenses assembled into lenses using a mapping model based on the mold manufacturing information and / or mold usage information. Considering the difference in mold information for each lens manufacturing mold, in some embodiments of this application, for each lens manufacturing mold, a mapping model corresponding to that mold is trained based on the mold information of that mold and the information of the lenses manufactured by that mold, thus ensuring that each lens manufacturing mold has its own corresponding mapping model.

[0106] Furthermore, this application can pre-manufacture a mold for each lens and, based on the mapping model corresponding to the lens manufacturing mold, predict the performance information of the lens assembled from the lenses manufactured by that mold. In some embodiments of this application, the lens performance information can be a specific value or a range of values. Taking the lens resolution as an example, the resolution of the lens assembled from the lenses manufactured by that mold can be predicted to be a specific value or a range of values ​​using the mapping model corresponding to the lens manufacturing mold.

[0107] In some embodiments of this application, when selecting a designated mold from a plurality of lens manufacturing molds, the performance information of the lens assembled by the lens manufactured by that lens manufacturing mold and the performance information of the lens assembled by the lens manufactured by at least one other lens manufacturing mold can be determined based on a mapping model for each lens manufacturing mold; and the designated mold is selected from each lens manufacturing mold according to the determined overlap range.

[0108] Specifically, since this application can obtain the lens performance information corresponding to each lens manufacturing mold in advance, it can predict the lens performance information of the lens manufactured by the lens manufacturing mold and the lens manufactured by at least one other lens manufacturing mold based on the lens performance information corresponding to each lens manufacturing mold and the lens performance information corresponding to the lens of the lens manufacturing mold and at least one other lens manufacturing mold. In this way, the performance information of the lens corresponding to several lens manufacturing mold combinations can be obtained. Following the above description, in some embodiments of this application, when the lens performance information is a certain numerical range, the performance information of the lens corresponding to each lens manufacturing mold is also a numerical range. For each lens manufacturing mold combination, the overlapping range of the performance information of the lenses corresponding to all lens manufacturing molds in the lens manufacturing mold combination can be determined as the performance information of the lens corresponding to the lens manufacturing mold combination based on the lens performance information of each lens manufacturing mold in the combination. For example, when the performance information includes a resolution numerical range, the resolution numerical range corresponding to the lens manufacturing mold combination can be determined as the overlapping range of the lens performance information of the lens manufacturing mold combination based on the resolution numerical range corresponding to each lens manufacturing mold in the combination. This allows us to obtain the overlap range of lens performance information corresponding to each lens manufacturing mold combination.

[0109] When selecting a specific mold, one or more lens manufacturing mold combinations can be selected based on the overlap range of lens performance information corresponding to each lens manufacturing mold combination, according to actual needs. The lens manufacturing mold in the selected mold combination is then used as the specific mold. For example, based on the overlap range of lens performance information corresponding to the mold combination, the standard performance information of the lens during design can be compared, and the mold combination with the largest overlap range with the standard performance information can be selected. The lens manufacturing mold in the selected mold combination is then used as the specific mold. Continuing with the above example, in some embodiments of this application, when the performance information includes a resolution value range, based on the resolution value range corresponding to each lens manufacturing mold combination and the resolution value range in the standard performance information, the overlap range between the resolution value range corresponding to each lens manufacturing mold combination and the resolution value range in the standard performance information can be determined separately. This facilitates the selection of the mold combination with the largest overlap range with the resolution value range in the standard performance information, thereby determining the specific mold.

[0110] In one embodiment of this application, the lens manufacturing mold may include multiple cavities. A corresponding mapping model can be trained for each cavity, thus establishing a unique mapping model for each cavity of each lens mold. For example, for the first lens in a lens, its mold A has four cavities, and a unique mapping model A1, A2, A3, and A4 is established for each cavity. For the second lens, it similarly has mapping models B1-B4. If evaluated individually, it can be considered that A1 has better performance and consistency among A1-A4, and B2 has better performance and consistency among B1-B4. However, the combination of A1 and B2 may not have the highest yield. Rather, the combination of A3 and B4 may have better performance indicators than the combination of A1 and B2 because it achieves mutual compensation for certain defects. That is, local optimality is not global optimality. Therefore, the performance prediction results generated by each cavity should be aggregated to maximize the overlap between the performance prediction distribution ranges of each cavity, thereby improving the yield. Based on this, A1-A4 and B1-B4 are paired in a way that maximizes the overall yield, in order to predict how to achieve the overall optimal result without wasting any cavity.

[0111] Table 1 shows the yield rate of manually assembled mold cells.

[0112]

[0113] In Table 1, manual mold cavity matching can only be done with limited experience. Although the yield of some combinations is as high as 80%, other combinations may only be 30%, and the overall yield (referring to the average yield of each combination) is only about 55%. In mass production, a large number of defective products of the A1+B1 and A4+B4 combinations are wasted.

[0114] Table 2 shows the yield of the cavity combination predicted using the mapping model.

[0115]

[0116] In Table 2, after using the mapping model for prediction, although the highest combination yield decreased from 80% to 65%, the overall yield (referring to the average yield of each combination) reached 60%, an improvement of 5%. In the case of a lens containing multiple lenses, and each lens may have 8 or 16 cavities, the mapping model can predict the mass production situation from a small amount of trial production data, saving trial production costs and improving the overall yield.

[0117] After selecting a specific mold, the manufacturing information of the lens manufactured by the specified mold can be obtained as the selected first information. That is, the mold manufacturing information and / or mold usage information when manufacturing the lens by the specified mold can be used as the first information. Based on the mapping model corresponding to the specified mold and the first information, the performance information of the lens assembled from the lens manufactured by the specified mold can be predicted.

[0118] Specifically, this application may use a lens manufactured by a specified mold as a specified lens; obtain a lens surface topography image of the specified lens; determine a feature image of the specified lens based on the lens surface topography image; and predict the performance information of a lens assembled from the specified lens based on the feature image and a mapping model.

[0119] In the actual lens injection molding process, selecting a lens manufactured by a specific mold as the specified lens allows for the acquisition of the surface profile information of the specified lens.

[0120] The surface profile information may include a lens surface topography map and a height distribution map of the specified lens. The specified lens is measured to obtain its surface topography map, and the height distribution map is determined based on this map.

[0121] In one embodiment of this application, measurements are taken on the actual injection-molded lens to acquire its surface morphology. Interference or confocal methods can be used to measure the lens. The interference method involves interfering a reference laser with the laser reflected from the lens under test to observe changes in the circular interference fringes in the interference pattern. When measuring aspherical lenses, aspherical wavefront compensation is applied to the reference laser, and the resulting interference fringe changes can be directly converted into a height distribution map of the lens. The confocal method involves scanning the lens surface point-by-point with a focused laser, collecting the reflected light corresponding to each scanning point through pinholes on the conjugate surface to calculate the physical height of the lens surface and outputting a height distribution map. The obtained height distribution map can then be mapped to a Zernike polynomial using conventional methods such as optical design software. The coefficients of the Zernike polynomial are then decomposed into aberrations such as spherical aberration, astigmatism, and coma, which are observed in actual products. Reducing these aberrations is crucial for improving lens manufacturing yield.

[0122] In some embodiments of this application, the surface profile information may further include a difference map corresponding to a specified lens. When determining the difference map corresponding to a specified lens, standard manufacturing information of the specified lens during its design can be obtained. Based on the lens surface topography image and the standard manufacturing information, the actual difference information corresponding to the specified lens, i.e., the difference map corresponding to the specified lens, is determined. Specifically, the lens height distribution image can be compared with the curve of the aspherical equation exported from the optical design software, and the difference between the actual surface profile data and the theoretical surface profile data can be used to form a difference map, thereby improving the prediction accuracy of the mapping model.

[0123] In some embodiments of this application, the surface shape information can also be processed. For example, Fourier transform processing can be performed on the surface shape information to convert it to the frequency domain, thereby achieving noise reduction and image size compression. Based on the above description, this application can perform Fourier transform processing on lens surface topography maps, height distribution maps, difference maps, etc., to achieve the effect of noise reduction and image size compression.

[0124] In one embodiment of this application, for each surface profile or Fourier-transformed surface profile, manufacturing information used during the manufacturing process is recorded. This includes inherent mold parameters that remain unchanged during production, such as cavity surface roughness, machining tolerances, contour errors, and flow channel structure parameters, venting structure parameters, and water channel structure parameters (such as the length, width, cross-sectional shape, distance to the cavity, and relative orientation to the cavity). It also includes mold usage parameters that have adjustable degrees of freedom during production, such as barrel temperature, nozzle temperature, mold temperature, heating gradient, material type, material batch, injection pressure, holding pressure, injection speed, molding time, holding time, cooling time, and annealing time. In summary, a distribution range of the lens surface profile information can be established for each combination of manufacturing information combinations; that is, a correspondence between manufacturing information and lens surface profile information can be established.

[0125] In one embodiment of this application, for each combination of manufacturing information, multiple lens surface information, such as a lens surface topography map, should be scanned to extract common manufacturing errors from the multiple surface information.

[0126] In another embodiment of this application, priority is given to measuring lenses that are prone to manufacturing errors, such as those with multiple inversions in shape, thin thickness and large diameter, or uneven thickness distribution. Lenses with uniform shape and thickness that are less prone to manufacturing errors do not need to be measured.

[0127] After obtaining the surface shape information of a specified lens, the feature image of the specified lens can be determined based on the surface shape information.

[0128] Specifically, a lens surface topography image can be input into a feature extraction model to obtain a feature image of the specified lens output by the feature extraction model. In some embodiments of this application, the input information of the feature extraction model can be any one or a combination of surface type information. For example, the input information can be a lens surface topography image, a height distribution map, a difference map, or surface type information after Fourier transform, or a combination of a lens surface topography image and a difference map, etc. The output information of the feature extraction model is the feature image of the specified lens; that is, the feature extraction model extracts the feature vector of the surface type information of the specified lens to obtain the feature image of the specified lens. The feature extraction model in this application can be different types of machine learning models, such as neural network models, as long as it can extract the surface type information of the lens to obtain the feature image.

[0129] After determining the feature image of a specified lens, the performance information of the lens assembled from the specified lens can be predicted based on the feature image and the mapping model.

[0130] Specifically, the mapping relationship between the manufacturing information of the specified lens and the performance information of the lens assembled from the specified lens can be obtained through a mapping model based on the feature image of the specified lens, thereby determining the performance information of the lens assembled from the specified lens.

[0131] Based on the above, this application can predict the performance information of the lens before it is assembled into a complete lens. By using a preset performance threshold, it can determine whether the lens meets the requirements, i.e., whether it is up to standard. If the lens meets the requirements, it can proceed to subsequent processes to produce a compliant lens, thereby improving the lens manufacturing yield. Conversely, if the lens does not meet the requirements, it means the assembled lens is substandard and cannot be proceeded to subsequent processes to avoid producing a non-compliant lens, thus also improving the lens manufacturing yield.

[0132] In addition, in one embodiment of this application, the selected first information can be adjusted, and based on the adjusted first information, a mapping model is used to predict the second information corresponding to the adjusted first information; the difference between the second information corresponding to the selected first information and the second information corresponding to the adjusted first information is determined; and information is selected from the lens performance information for processing according to the determined difference.

[0133] In some embodiments of this application, selecting information from the lens performance information may include at least one performance information to be prioritized for adjustment or at least one performance information to be avoided for adjustment. Further, in other embodiments, the at least one performance information to be prioritized for adjustment may be at least one performance information that contributes the most to differences in lens performance. In still other embodiments, the at least one performance information to be avoided for adjustment may be at least one performance information that contributes the least to differences in lens performance.

[0134] Specifically, this application utilizes a mapping model to predict the performance information of one or more lenses that require priority adjustment based on their correlation probability, thereby identifying key factors affecting lens manufacturing yield. For example, the lens performance information can be sorted from highest to lowest correlation probability, and one or more lenses located at the beginning of the sequence can be selected as performance information to be adjusted first. Furthermore, the mapping model can also be used to predict the performance information of one or more lenses that should be avoided from adjustment based on their correlation probability, thereby identifying key factors affecting the rate of change (i.e., stability) of yield. Following the previous example, one or more lenses located at the end of the sequence can be selected as performance information to avoid adjustment, ensuring that highly sensitive variables prone to random errors remain constant over the long term.

[0135] In some embodiments of this application, one or more pieces of information in the first information can be adjusted, such as adjusting the temperature or pressure in the mold usage information. Based on the manufacturing information of the adjusted lens, a mapping model is used to predict the performance information of the lens obtained by assembling the adjusted lens. The difference between the predicted lens performance information before and after adjustment is determined. Since the lens performance information includes various specific lens parameters, the difference for each specific lens parameter can be determined. The lens parameters are then sorted according to the differences, for example, sorted from largest to smallest difference. The first one or more lens parameters in the sequence are those that need priority adjustment, while the last one or more lens parameters in the sequence are those that should be avoided.

[0136] Secondly, this application can also predict the manufacturing information of the lenses that make up the lens by using a mapping model based on the performance information of the selected lens.

[0137] Specifically, in general, the scenario of first obtaining lens performance information and then predicting lens manufacturing information is common in target lens design. When designing a target lens, its performance information—specifically, its standard performance information—can be determined. Then, based on this standard performance information, a mapping model is used to predict the manufacturing information of the lens required for assembling the target lens. This outputs a range of manufacturing information that meets certain performance requirements. These performance requirements include MTF (Mean Transform Factor), aberrations, etc., which are defined by specific standards during lens manufacturing and have predetermined threshold ranges. Therefore, when manufacturing the target lens, from several lens manufacturing molds, at least one mold can be selected based on the predicted manufacturing information of the lens required for assembling the target lens mapped from the standard performance information, and the manufacturing information of lenses manufactured using each mold. For example, the mold with the largest overlap between the manufacturing information of lenses manufactured using each mold and the predicted manufacturing information can be selected, resulting in a target lens assembled from lenses manufactured using the selected mold.

[0138] This application allows for the selection of a mold from the mold combination that has the greatest overlap with the lens performance information before lens production, based on the mapping model corresponding to each lens manufacturing mold. Compared to other unselected molds, the lens produced by the selected mold is more likely to meet the performance requirements of the assembled lens, resulting in a higher compliance rate and thus a higher lens manufacturing yield.

[0139] In one embodiment of this application, Figure 4 A flowchart illustrating another method for predicting lens manufacturing information provided in this application embodiment is shown below. Figure 4 As shown, the specific steps may include: S410: When injection molding the lens, collect the lens surface morphology image and manufacturing information; S420: After assembling the lens, collect performance information; S430: Extract feature vectors of the lens surface morphology image under a set of manufacturing information through a neural network; S440: Machine learn the mapping relationship between feature vectors and performance information to form an artificial intelligence detection model; S450: Use the artificial intelligence detection model to predict the range of manufacturing information required to meet the performance requirements of the lens.

[0140] Specifically, in some embodiments of this application, surface shape information and manufacturing information are collected from multiple lenses manufactured using a first injection molding mold; performance information is collected from one or more lenses assembled into a lens; feature vectors of the lens surface topography map associated with the manufacturing information are extracted through a neural network; the mapping relationship between the feature vectors and the performance information is machined to form a first artificial intelligence detection model; and the distribution range of the manufacturing information corresponding to the performance information meeting a preset threshold is predicted using the first artificial intelligence detection model.

[0141] In step S420, after the lens is assembled into a lens, its performance information is measured, including the MTF function reflecting resolution, relative illumination, aberrations (including chromatic aberration, optical distortion, TV distortion, field curvature, etc.), stray light, ghosting (referring to the phenomenon that light rays that are not required for imaging reach the image plane due to reflection and scattering within the lens), depth of field, appearance defects (referring to the presence of visual defects such as blemishes and cracks), maximum principal ray angle CRAmax, focal length EFL, aperture f / #, field of view FOV, maximum image height ImgH, and other specifications. These indicators may fail to meet the standards due to lens manufacturing defects. Through steps S410 and S420, a database of lens surface type information and lens performance information under certain manufacturing information has been established. However, because the causal relationship between the two is complex, it cannot be analyzed by theoretical methods or simple numerical simulations. Artificial intelligence methods are needed to learn and establish the correlation between the two.

[0142] In step S440, machine learning methods are used to train an AI model by combining the feature vectors and performance information of the lens surface topography obtained in the previous steps. This includes using convolutional neural networks, deep networks based on attention mechanisms, random forests, gradient boosting trees (GBDT), and other commonly used supervised or unsupervised learning methods to establish a data model that correlates lens surface topography information with performance information such as MTF and aberrations. This model can be used to predict the range of performance information changes mapped by the surface topography information under certain manufacturing information. Compared to directly establishing a correlation between manufacturing information and performance information, using actual detected lens manufacturing defect data can filter out more irrelevant variables, improve processing efficiency and accuracy, and help determine whether the final performance of the lens is mainly affected by this manufacturing information. If the result is negative, it indicates that the decrease in yield is due to other randomness in the assembly, testing, or manufacturing process, and the mold is not the bottleneck for improving yield. If the result is positive, it indicates that the manufacturing information and the final performance show a clear correlation. For inherent parameters, this can be corrected by mold modification, and for variable process parameters, adjustments can be made at any time when yield problems occur.

[0143] Because the causal relationship between lens aspherical processing defects and lens performance cannot be calculated by conventional means except by establishing a large artificial intelligence model, this application can identify the types of defects that affect lens yield through the above content, and provide early warnings to avoid the manufacturing information range corresponding to these defect types, thereby improving manufacturing yield.

[0144] In some embodiments of this application, multiple lenses are formed in the first cavity of a first injection molding mold. A second artificial intelligence detection model corresponding to the second cavity of a second injection molding mold is coupled with a first artificial intelligence detection model to predict the performance information of the assembled lenses formed in the first and second cavities. This application establishes a unique mapping model for each cavity using big data to transform locally optimal combinations into overall yield optimization.

[0145] This application also provides a method for training a mapping model, specifically: for each lens manufacturing mold, obtain several lenses manufactured by the lens manufacturing mold as sample lenses; obtain manufacturing information for each sample lens; assemble one or more sample lenses into a lens as a sample lens, and determine the sample performance information of the sample lens; predict the performance information to be optimized for the sample lens based on the manufacturing information of each sample lens and the mapping model to be trained; train the mapping model to be trained according to the sample performance information and the performance information to be optimized. Specifically, when obtaining sample lenses, several lenses manufactured by the lens manufacturing mold can be obtained as candidate lenses; for each candidate lens, obtain the actual manufacturing information of the candidate lens and the standard manufacturing information of the candidate lens when designing the candidate lens; determine the difference between the actual manufacturing information and the standard manufacturing information as the difference corresponding to the candidate lens; and select a sample lens from several candidate lenses according to the difference corresponding to each candidate lens.

[0146] Specifically, when selecting sample lenses, it is preferable to measure lenses that have multiple inversions in their shape, are thin but have a large diameter, or have uneven thickness distribution, which are prone to manufacturing errors. Lenses with uniform shape and thickness, which are less prone to manufacturing errors, do not need to be measured.

[0147] After assembling sample lenses into sample lenses, the performance information of the sample lenses is collected as sample performance information for training the mapping model. The manufacturing information of the sample lenses is mapped to their surface profile information, and feature images of the surface profile information are extracted. Based on these feature images, the mapping model predicts the performance information to be optimized for the sample lenses. The mapping model is trained with the goal of minimizing the difference between the sample lens performance information and the predicted performance information to be optimized, ensuring that the trained model can determine the mapping relationship between the lens manufacturing information and the lens performance information. Therefore, this application employs supervised learning to train the mapping model, enabling it to better determine the mapping relationship between the surface profile information corresponding to the manufacturing information and the performance information.

[0148] Based on the method for predicting lens manufacturing information shown in the above embodiments, this application also provides a schematic diagram of the structure of an apparatus for predicting lens manufacturing information, as shown below. Figure 5 As shown.

[0149] Figure 5 This is a schematic diagram of a device for predicting lens manufacturing-related information provided in an embodiment of this application. The device includes: a determining module 500, used to determine a mapping relationship between first information and second information related to lens manufacturing based on a pre-trained mapping model, wherein the first information is manufacturing information of the lens elements constituting the lens, and the second information is performance information of the lens assembled from the lens elements; and a predicting module 520, used to predict the corresponding second information based on the selected first information using the mapping model, or to predict the corresponding first information based on the selected second information using the mapping model.

[0150] Optionally, the device further includes: a training module 540; the training module 540 is specifically configured to: for each lens manufacturing mold, obtain several lenses manufactured by the lens manufacturing mold as sample lenses; obtain manufacturing information for each sample lens; assemble one or more sample lenses into a lens as a sample lens, and determine the sample performance information of the sample lens; predict the performance information to be optimized for the sample lens based on the manufacturing information of each sample lens and the mapping model to be trained; and train the mapping model to be trained according to the sample performance information and the performance information to be optimized.

[0151] Optionally, the training module 540 is specifically used to: acquire a plurality of lenses manufactured by the lens manufacturing mold as candidate lenses; for each candidate lens, acquire the actual manufacturing information of the candidate lens and the standard manufacturing information of the candidate lens when designing the candidate lens; determine the difference between the actual manufacturing information and the standard manufacturing information as the difference corresponding to the candidate lens; and select a sample lens from the plurality of candidate lenses according to the difference corresponding to each candidate lens.

[0152] Optionally, the prediction module 520 is specifically configured to: select at least one lens manufacturing mold as a designated mold from a plurality of lens manufacturing molds; obtain manufacturing information of the lens manufactured by the designated mold as the selected first information; and predict the performance information of the lens assembled from the lens manufactured by the designated mold based on the mapping model and the selected first information.

[0153] Optionally, the prediction module 520 is specifically used to, based on the mapping model, determine, for each lens manufacturing mold, the overlap range between the performance information of the lens assembled by the lens manufactured by that lens manufacturing mold and the performance information of the lens assembled by the lens manufactured by at least one other lens manufacturing mold; and select the designated mold among the lens manufacturing molds according to the determined overlap range.

[0154] Optionally, the prediction module 520 is specifically used to: use a lens manufactured by the specified mold as a specified lens; obtain a lens surface topography image of the specified lens; determine a feature image of the specified lens based on the lens surface topography image; and predict the performance information of a lens assembled from the specified lens based on the feature image and the mapping model.

[0155] Optionally, the prediction module 520 is specifically used to determine the height distribution map of the specified lens based on the lens surface topography map; and to determine the feature image of the specified lens based on the height distribution map.

[0156] Optionally, the prediction module 520 is specifically used to: acquire the standard manufacturing information of the specified lens when designing the specified lens; determine the actual difference information corresponding to the specified lens based on the lens surface topography image and the standard manufacturing information; and determine the feature image of the specified lens based on the actual difference information.

[0157] Optionally, the prediction module 520 is specifically used to perform Fourier transform processing on the lens surface topography image; and determine the feature image of the specified lens based on the processed lens surface topography image.

[0158] Optionally, the prediction module 520 is specifically used to input the lens surface topography image into the feature extraction model to obtain the feature image of the specified lens output by the feature extraction model.

[0159] Optionally, the device further includes: an adjustment module 560; the adjustment module 560 is specifically used to: adjust the selected first information; based on the adjusted first information, use the mapping model to predict the second information corresponding to the adjusted first information; determine the difference between the second information corresponding to the selected first information and the second information corresponding to the adjusted first information; and select information from the lens performance information for processing according to the determined difference.

[0160] Optionally, the prediction module 520 is specifically used to: include the selected second information as standard performance information of the target lens when designing the target lens; and, based on the standard performance information, predict the manufacturing information of the lens required for assembling the target lens by means of the mapping model.

[0161] Optionally, the device further includes an assembly module 580; the assembly module 580 is specifically used to select at least one lens manufacturing mold from among a plurality of lens manufacturing molds, based on predicted first information and first information of the lenses manufactured by each lens manufacturing mold, to obtain the target lens assembled from the lenses manufactured by the selected lens manufacturing mold.

[0162] Optionally, the manufacturing information includes mold manufacturing information and / or mold usage information for manufacturing lenses. The mold manufacturing information includes at least one of the following: mold cavity surface roughness, machining tolerance, contour error, and flow channel structure parameters, venting structure parameters, and water channel structure parameters connected to the mold cavity. The mold usage information includes at least one of the following: barrel temperature, nozzle temperature, mold temperature, heating gradient, material type, material batch, injection pressure, holding pressure, injection speed, molding time, holding time, cooling time, and annealing time. The performance information includes at least one of the following: resolution, modulation transfer function (MTF), illuminance, aberrations, stray light, ghosting, depth of field, surface defects, maximum principal ray angle, focal length, aperture, field of view, and maximum image height.

[0163] This application also provides a computer-readable storage medium storing a computer program that can be used to execute the method for predicting lens manufacturing information provided in the above embodiments.

[0164] This application also provides a computer program product, which includes a computer program or computer-executable instructions, and is stored in a computer-readable storage medium. A processor of a computer device reads the computer program from the computer-readable storage medium and executes the computer program, causing the computer device to perform the method for predicting lens manufacturing related information provided in this application.

[0165] Based on the method for predicting lens manufacturing information shown in the above embodiments, this application also proposes... Figure 6 The diagram shows a schematic structure of the electronic device 600. Figure 6 At the hardware level, the electronic device 600 includes a processor 610 and a memory 620, and may also include an internal bus, network interface, memory, and other hardware required for the business. The processor 610 reads the corresponding computer program from the memory 620 into memory and then runs it to implement the method for predicting lens manufacturing information as described in the above embodiments.

[0166] Of course, in addition to software implementation, this application does not exclude other implementation methods, such as logic devices or a combination of hardware and software, etc. In other words, the execution subject of the following processing flow is not limited to each logic unit, but can also be hardware or logic devices.

[0167] In the 1990s, improvements to a technology could be clearly distinguished as either hardware improvements (e.g., improvements to the circuit structure of diodes, transistors, switches, etc.) or software improvements (improvements to the methodology). However, with technological advancements, many methodological improvements today can be considered direct improvements to the hardware circuit structure. Designers almost always obtain the corresponding hardware circuit structure by programming the improved methodology into the hardware circuit. Therefore, it cannot be said that a methodological improvement cannot be implemented using hardware physical modules. For example, a Programmable Logic Device (PLD) (such as a Field Programmable Gate Array (FPGA)) is such an integrated circuit whose logic function is determined by the user programming the device. Designers can program and "integrate" a digital system onto a PLD themselves, without needing chip manufacturers to design and manufacture dedicated integrated circuit chips. Furthermore, nowadays, instead of manually manufacturing integrated circuit chips, this programming is mostly implemented using "logic compiler" software. Similar to the software compiler used in program development, the original code before compilation must be written in a specific programming language, called a Hardware Description Language (HDL). There are many HDLs, such as ABEL (Advanced Boolean Expression Language), AHDL (Altera Hardware Description Language), Confluence, CUPL (Cornell University Programming Language), HDCal, JHDL (Java Hardware Description Language), Lava, Lola, MyHDL, PALASM, and RHDL (Ruby Hardware Description Language). Currently, the most commonly used are VHDL (Very-High-Speed ​​Integrated Circuit Hardware Description Language) and Verilog. Those skilled in the art should understand that by simply performing some logic programming on the method flow using one of these hardware description languages ​​and programming it into an integrated circuit, the hardware circuit implementing the logical method flow can be easily obtained.

[0168] The controller can be implemented in any suitable manner. For example, it can take the form of a microprocessor or processor and a computer-readable medium storing computer-readable program code (e.g., software or firmware) executable by the (micro)processor, logic gates, switches, application-specific integrated circuits (ASICs), programmable logic controllers, and embedded microcontrollers. Examples of controllers include, but are not limited to, the following microcontrollers: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20, and Silicon Labs C8051F320. A memory controller can also be implemented as part of the control logic of the memory. Those skilled in the art will also recognize that, in addition to implementing the controller in purely computer-readable program code form, the same functionality can be achieved by logically programming the method steps to make the controller take the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, and embedded microcontrollers. Therefore, such a controller can be considered a hardware component, and the means included therein for implementing various functions can also be considered as structures within the hardware component. Alternatively, the means for implementing various functions can be considered as both software modules implementing the method and structures within the hardware component.

[0169] The systems, devices, modules, or units described in the above embodiments can be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, a computer can be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smartphone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or any combination of these devices.

[0170] For ease of description, the above devices are described in terms of function, divided into various units. Of course, in implementing this specification, the functions of each unit can be implemented in one or more software and / or hardware components.

[0171] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0172] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0173] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0174] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0175] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0176] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0177] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0178] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0179] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0180] This application can be described in the general context of computer-executable instructions, such as program modules, that are executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform a specific task or implement a specific abstract data type. This specification can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0181] The various embodiments in this application are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0182] The above description is merely an embodiment of this application and is not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of this application should be included within the scope of the claims in this specification.

Claims

1. A method for predicting lens manufacturing-related information, characterized in that, The method includes: Based on a pre-trained mapping model, a mapping relationship between first information and second information related to lens manufacturing is determined, wherein the first information is the manufacturing information of the lens elements constituting the lens, and the second information is the performance information of the lens assembled from the lens elements. Based on the selected first information, the corresponding second information is predicted using the mapping model; or, based on the selected second information, the corresponding first information is predicted using the mapping model. Specifically, predicting corresponding second information based on selected first information using the mapping model includes: selecting at least one lens manufacturing mold as a designated mold from a plurality of lens manufacturing molds; obtaining manufacturing information of the lens manufactured by the designated mold as the selected first information; and predicting the performance information of the lens assembled from the lens manufactured by the designated mold according to the mapping model and the selected first information.

2. The method as described in claim 1, characterized in that, The mapping model is pre-trained as follows: For each lens manufacturing mold, obtain several lenses manufactured by that lens manufacturing mold as sample lenses; Obtain manufacturing information for each sample lens; One or more sample lenses are assembled into a lens as a sample lens, and the sample performance information of the sample lens is determined. Based on the manufacturing information of each sample lens and the mapping model to be trained, predict the performance information of the sample lens to be optimized. The mapping model to be trained is trained based on the sample performance information and the performance information to be optimized.

3. The method as described in claim 2, characterized in that, Obtain several lenses manufactured using the lens manufacturing mold as sample lenses, specifically including: Obtain several lenses manufactured by the lens manufacturing mold as candidate lenses; For each candidate lens, obtain the actual manufacturing information of the candidate lens and the standard manufacturing information of the candidate lens when designing the candidate lens; The difference between the actual manufacturing information and the standard manufacturing information is determined as the difference corresponding to the candidate lens; Based on the differences corresponding to each candidate lens, a sample lens is selected from several candidate lenses.

4. The method as described in claim 1, characterized in that, Among a number of lens manufacturing molds, at least one lens manufacturing mold is selected as the designated mold, specifically including: Based on the mapping model, for each lens manufacturing mold, the overlap range between the performance information of the lens assembled by the lens manufactured by that lens manufacturing mold and the performance information of the lens assembled by the lens manufactured by at least one other lens manufacturing mold is determined. Based on the determined overlap range, the specified mold is selected from among the various lens manufacturing molds.

5. The method as described in claim 1, characterized in that, Based on the mapping model and the selected first information, predict the performance information of the lens obtained by assembling lenses manufactured using the specified mold, specifically including: The lens manufactured using the specified mold shall be designated as the specified lens; Obtain a surface topography image of the specified lens; Based on the lens surface topography, determine the feature image of the specified lens; Based on the feature image and the mapping model, predict the performance information of the lens obtained by assembling the specified lens.

6. The method as described in claim 5, characterized in that, Based on the lens surface topography image, the characteristic image of the specified lens is determined, specifically including: Based on the lens surface topography, determine the height distribution map of the specified lens; Based on the height distribution map, the characteristic image of the specified lens is determined.

7. The method as described in claim 5, characterized in that, Based on the lens surface topography image, the characteristic image of the specified lens is determined, specifically including: Obtain the standard manufacturing information of the specified lens when designing the specified lens; Based on the lens surface topography diagram and the standard manufacturing information, determine the actual difference information corresponding to the specified lens; Based on the actual difference information, the feature image of the specified lens is determined.

8. The method as described in claim 5, characterized in that, Based on the lens surface topography image, the characteristic image of the specified lens is determined, specifically including: The surface topography of the lens is subjected to Fourier transform processing; Based on the processed lens surface morphology image, the characteristic image of the specified lens is determined.

9. The method as described in claim 5, characterized in that, Based on the lens surface topography image, the characteristic image of the specified lens is determined, specifically including: The surface topography of the lens is input into the feature extraction model to obtain the feature image of the specified lens output by the feature extraction model.

10. The method as described in claim 1, characterized in that, The method further includes: Adjust the selected first information, and based on the adjusted first information, use the mapping model to predict the second information corresponding to the adjusted first information; Determine the difference between the second information corresponding to the selected first information and the second information corresponding to the adjusted first information; Based on the identified differences, select information from the lens performance information for processing.

11. The method as described in claim 1, characterized in that, Based on the selected second information, the corresponding first information is predicted using the mapping model, specifically including: The selected second information includes the standard performance information of the target lens when designing the target lens; Based on the standard performance information, the manufacturing information of the lens required for assembling the target lens is predicted through the mapping model.

12. The method as described in claim 11, characterized in that, The method further includes: Among several lens manufacturing molds, at least one lens manufacturing mold is selected based on the first predicted information and the first information of the lenses manufactured by each lens manufacturing mold, to obtain the target lens assembled from the lenses manufactured by the selected lens manufacturing mold.

13. The method as described in claim 1, characterized in that, The manufacturing information includes mold manufacturing information and / or mold usage information when manufacturing lenses. The mold manufacturing information includes at least one of the following: mold cavity surface roughness, machining tolerance, contour error, and flow channel structure parameters, venting structure parameters, and water channel structure parameters connected to the mold cavity. The mold usage information includes at least one of the following: barrel temperature, nozzle temperature, mold temperature, heating gradient, material type, material batch, injection pressure, holding pressure, injection speed, molding time, holding pressure time, cooling time, and annealing time. The performance information includes at least one of the following: resolution, modulation transfer function (MTF), illuminance, aberration, stray light, ghosting, depth of field, appearance defects, maximum principal ray angle, focal length, aperture, field of view, and maximum image height.

14. An apparatus for predicting lens manufacturing-related information, characterized in that, The device includes: The determination module is used to determine the mapping relationship between first information and second information related to lens manufacturing based on a pre-trained mapping model, wherein the first information is the manufacturing information of the lens elements constituting the lens, and the second information is the performance information of the lens assembled from the lens elements. The prediction module is used to predict corresponding second information based on selected first information using the mapping model, or to predict corresponding first information based on selected second information using the mapping model. Specifically, the prediction module is used to: select at least one lens manufacturing mold as a designated mold from a plurality of lens manufacturing molds; obtain manufacturing information of the lens manufactured by the designated mold as the selected first information; and predict the performance information of the lens assembled from the lens manufactured by the designated mold based on the mapping model and the selected first information.

15. A computer-readable storage medium, characterized in that, The storage medium stores a computer program, which, when executed by a processor, implements the method described in any one of claims 1-13.

16. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the method described in any one of claims 1-13.

Images