Method, program, and apparatus for adjusting optical systems

Abstract

To provide an optical system adjustment method that can improve optical performance across the entire range of focal lengths in a variable magnification optical system. [Solution] The method for adjusting the optical system includes the steps of: acquiring a first optical performance obtained by measuring the variable magnification optical system; inputting the first optical performance into a first classifier and extracting multiple samples from multiple optical system samples generated by superimposing a predetermined amount on the design reference value of the variable magnification optical system; acquiring information for a second classifier corresponding to each of the multiple samples; inputting the second optical performance of the multiple samples into a second classifier and acquiring information for the cause of the defect in the variable magnification optical system; and determining a correction amount or correction priority corresponding to the cause of the defect for each of the multiple zoom positions of the variable magnification optical system.

Description

Technical Field

[0001] The present invention relates to a method for estimating and correcting the cause of defects inside an optical system from the transmitted light of the optical system, and particularly to a method for improving the lens performance due to manufacturing errors.

Background Art

[0002] Conventionally, in an optical product having lenses of multiple patterns, a method for estimating the cause of defects inside the optical system has been proposed (see Patent Document 1). In recent years, a method for estimating the cause of lens defects by machine learning has been proposed.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] However, when the optical system is a zoom lens and there are members with different sensitivities or members with different amounts of eccentricity depending on the zoom position, it may be difficult to correct the defective portions simultaneously at all zoom positions.

[0005] An object of the present invention is to provide an adjustment method for an optical system capable of improving the optical performance at all focal lengths in the entire range of a zoom optical system.

Means for Solving the Problems

[0006] One aspect of the present invention is a method for adjusting an optical system, characterized by comprising the steps of: acquiring a first optical performance obtained by measuring a variable magnification optical system; inputting the first optical performance into a first classifier and extracting a plurality of samples from a plurality of optical system samples generated by superimposing a predetermined amount on the design reference value of the variable magnification optical system; acquiring information for a second classifier corresponding to each of the plurality of samples; inputting the second optical performance of the plurality of samples into the second classifier and acquiring information for the cause of a defect in the variable magnification optical system; and determining a correction amount or correction priority corresponding to the cause of a defect for each of the plurality of zoom positions of the variable magnification optical system. [Effects of the Invention]

[0007] According to the present invention, it is possible to provide an optical system adjustment method that can improve the optical performance across the entire range of focal lengths in a variable magnification optical system. [Brief explanation of the drawing]

[0008] [Figure 1] This figure shows the configuration of a wavefront aberration measuring device according to an embodiment of the present invention. [Figure 2] This is a diagram showing the configuration of a computer. [Figure 3] This is a conceptual diagram showing the input and output structure of the first classifier constructed using machine learning. [Figure 4] This is a conceptual diagram of the input and output in the first discriminator. [Figure 5] This is a conceptual diagram showing the input and output structure of a second classifier constructed using machine learning. [Figure 6] This is a conceptual diagram of the input and output in the second discriminator. [Figure 7] This flowchart shows a method for inspecting an optical system to estimate the cause of a defect. [Figure 8] This is a diagram illustrating the calculation method for the transmitted wavefront in Example 1. [Figure 9] This is a flowchart showing the method for constructing the first classifier in Example 1. [Figure 10] It is a flowchart showing a method for extracting similar error cases in Example 1. [Figure 11] It is a flowchart showing a method for constructing a second discriminator in Example 1. [Figure 12] It is a flowchart showing an optical system correction method. [Figure 13] It is a flowchart showing a method for constructing a first discriminator in Example 2. [Figure 14] It is a flowchart showing a method for extracting similar error cases in Example 2. [Figure 15] It is a flowchart showing a method for constructing a second discriminator in Example 2. [Figure 16] It is a diagram for explaining a method for calculating a transmitted wavefront in Examples 3 and 4. [Figure 17] It is a diagram for explaining a method for correcting mechanical parts and the effect in Example 3. [Figure 18] It is a flowchart showing a method for estimating an optical transfer function in Example 5. [Figure 19] It is a flowchart showing a method for constructing a second discriminator in Example 5. [Figure 20] It is a flowchart showing a method for constructing a first discriminator in Example 6. [Figure 21] It is a flowchart showing a method for constructing a second discriminator in Example 6. [Figure 22] It is a flowchart showing a method for constructing a first discriminator in Example 7. [Figure 23] It is a flowchart showing a method for constructing a second discriminator in Example 7.

Embodiments for Carrying Out the Invention

[0009] Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. In each figure, the same members are denoted by the same reference numerals, and duplicate explanations are omitted.

[0010] Figure 1 shows the configuration of a wavefront aberration measuring device (hereinafter referred to as the measuring device) 10 according to an embodiment of the present invention. Light sources 1 and 2 output light beams for measuring the transmitted wavefront. The optical system 120 is the object to be inspected and has multiple patterns of lenses. Each lens has so-called manufacturing errors and assembly errors, such as shape errors of the lens itself and eccentricity or tilt of the optical system. Wavefront sensors 4 and 5 measure (detect) the wavefront aberration of the light beam transmitted through the optical system 120. The personal computer 6 receives the acquired wavefront aberration via a USB cable. Although Figure 1 shows an example in which a Shack-Hartmann sensor is used for wavefront measurement, it is not limited to this and it is also possible to use a Fizeau interferometer or the like.

[0011] Figure 2 shows the configuration of personal computer 6. Personal computer 6 has a CPU (processing unit) 21 for performing calculations, a memory 22 for storing calculation results, a keyboard 23 for inputting information, a mouse 24 for selecting information, and a display 25 for displaying information.

[0012] Figure 3 is a conceptual diagram showing the input and output structure of the first classifier constructed using machine learning.

[0013] Optical system samples are used to create the collected data. An optical system sample is a sample in which a manufacturing error (a predetermined amount) is superimposed on the design standard value of the object under test. The object under test is the optical system that is the subject of defect analysis. The manufacturing error is the amount of error in the lens group (hereafter, lens group includes individual lenses) included in the optical system. For example, this includes the parallel eccentricity of the lens group, the tilt eccentricity of the lens group, the spacing between lens groups, the thickness of the lenses, and the shape of the lens surfaces. Multiple patterns of optical system samples are created by superimposing such manufacturing errors within an assumed range (a predetermined range) onto the design standard value. Preferably, the assumed range is at least 1 times and at most 5 times the manufacturing tolerance range of each component. A number is assigned to the optical system sample as identification information. This is called the optical system sample number and is used as input data.

[0014] Furthermore, the collected data is processed to create input data. Specifically, the optical performance of each optical system sample is calculated, and the obtained optical performance is used as input data. The optical performance includes information on wavefront aberration, ray aberration, and at least one of image plane information.

[0015] Specific algorithms that use input data for machine learning include nearest neighbors, Naise Bayes, decision trees, and support vector machines. Deep learning, which generates its own features and joint weighting coefficients for learning, is also an option. Appropriately, any of the above algorithms that are available can be applied to this embodiment. The machine learning unit creates a decision tree according to the input data input to the input layer and learns the trends of the input data so that it can output output data for the input data. In this way, the first classifier is constructed.

[0016] Preparing the input data for the estimation phase requires acquired data. The acquired data is the optical performance of the object under test, obtained by measuring the transmitted wavefront or transmitted light with the measuring device 10 shown in Figure 1.

[0017] Figure 4 is a conceptual diagram of the input and output in the first classifier. When acquired data is input to the first classifier, it outputs an optical system sample number. The output data contains identification information for the optical system sample, and by following the number, the eccentricity state and optical performance can be obtained. The output data is used to extract optical system samples that are similar to the acquired data.

[0018] Figure 5 is a conceptual diagram showing the input and output structure of the second classifier constructed using machine learning.

[0019] Creating the collected data requires calculating the relationship between optical performance and optical transfer function. The optical performance and optical transfer function calculated from the number of optical system samples necessary for calculating the relationship between optical performance and optical transfer function are used as input data. Specific algorithms that use the input data for machine learning include nearest neighbor method, Naise Bayes method, decision tree, and support vector machine. Deep learning, which generates its own features and joint weighting coefficients for learning, is also an option. Appropriately, any of the above algorithms that are available can be applied to this embodiment. The machine learning unit creates a decision tree according to the input data input to the input layer and learns the trend of the input data so that it can output output data for the input data. In this way, a second classifier is constructed.

[0020] Preparing the input data for the estimation phase requires acquired data. This acquired data is the calculated optical performance.

[0021] Figure 6 is a conceptual diagram of the input and output in the second classifier. When acquired data is input to the second classifier, it outputs an optical transfer function. The output data is used to determine the optical transfer function for the acquired data.

[0022] The following example demonstrates how to perform optical performance defect analysis using measurable optical performance, identify defective areas, and perform appropriate corrections across the entire zoom range. [Examples]

[0023] This embodiment describes a method for estimating and correcting the cause of deterioration (defect) in the axial optical performance of an object under test using axial wavefront aberration.

[0024] Figure 7 is a flowchart showing a method for inspecting the optical system to estimate the cause of a defect in this embodiment. The processing in this flowchart is executed by the CPU 21 according to a computer program.

[0025] In step S101, the CPU 21 creates optical system samples for each zoom position.

[0026] In step S102, the CPU 21 calculates the on-axis wavefront aberration (first optical performance) for each of the multiple optical system samples created in step S101.

[0027] The method for calculating wavefront aberration will be explained below with reference to Figure 8. Figure 8 is a diagram illustrating the method for calculating the transmitted wavefront. The on-axis light 130 is light that passes through the optical system 120 and is focused at the image point 140. Multiple patterns of light rays are tracked from an unillustrated object point (at infinity in Figure 8) to the image point 140, and the distribution of the optical path lengths of the light rays emitted from the optical system 120 becomes the transmitted wavefront 160. The difference between the ideal transmitted wavefront 150, which has no aberration, and the actual transmitted wavefront 160 is the wavefront aberration.

[0028] Returning to Figure 7, in step S103, the CPU 21 constructs a first classifier using the wavefront aberrations acquired in step S102. Referring to Figure 9, a method using decision trees to construct the first classifier will be explained. Figure 9 is a flowchart of the method for constructing the first classifier in this embodiment. In step S201, the CPU 21 inputs the wavefront aberrations of the optical system samples acquired in step S102. In step S202, the CPU 21 inputs the optical system sample number assigned to the optical system sample as identification information. In step S203, the CPU 21 associates the wavefront aberrations of the optical system samples input in steps S201 and S202 with the optical system sample numbers. In step S204, the CPU 21 determines whether input for all of the multiple patterns of optical system samples created has been completed. If the CPU 21 determines that input for all of the multiple patterns of optical system samples created has been completed, it executes the process in step S205; otherwise, it executes the process in step S201. In step S205, the CPU 21 constructs a number of decision trees. Thus, the first classifier is constructed. When wavefront aberration is input to the first classifier, it becomes possible to obtain optical system sample numbers that have optical performance tendencies similar to the input.

[0029] In this embodiment, a decision tree is used to construct the first classifier, but the method is not limited to this, as long as it is possible to obtain an optical system sample similar to the object under test from wavefront aberration.

[0030] Returning to Figure 7, in step S104, the CPU 21 acquires optical information of the object under test. The CPU 21 measures the transmitted wavefront of the object under test and measures (acquires) the wavefront aberration (first optical performance) of the object under test.

[0031] In step S105, the CPU 21 extracts similar error cases using the wavefront aberration of the object under test acquired in step S104 and the first classifier constructed in step S103. Similar error cases are optical system samples with optical system configurations similar to those of the object under test. Figure 10 is a flowchart illustrating the method for extracting similar error cases in this embodiment. In step S301, the CPU 21 inputs the wavefront aberration of the object under test acquired in step S104 into the first classifier. In step S302, the CPU 21 classifies training data similar to the input data using machine learning. In step S303, the CPU 21 outputs optical system samples from the classified training data. This makes it possible to extract similar error cases.

[0032] Returning to Figure 7, in step S106, the CPU 21 extracts the cause of the defect for each similar error case. The CPU 21 performs calculations using the manufacturing error of each similar error case and the aberration sensitivity of each lens group in the object under test to obtain the wavefront aberration (second optical performance) for each lens group of each similar error case. The CPU 21 constructs a second classifier using the obtained wavefront aberrations of the lens groups of the similar error cases.

[0033] Referring to Figure 11, a method for constructing a second classifier using a decision tree will be described. Figure 11 is a flowchart of the method for constructing the second classifier in this embodiment. It is necessary to numerically determine the relationship between wavefront aberration and optical transfer function in advance using a predetermined number of optical system samples from similar error cases. The predetermined number is sufficient to determine the relationship between wavefront aberration and optical transfer function, and is less than the number of similar error cases. In step S401, the CPU 21 inputs the wavefront aberration of the predetermined number of optical system samples used when determining the relationship between wavefront aberration and optical transfer function. In step S402, the CPU 21 inputs the optical transfer function (third optical performance) of the predetermined number of optical system samples used when determining the relationship between wavefront aberration and optical transfer function. In step S403, the CPU 21 links the wavefront aberration and optical transfer function input in steps S401 and S402. In step S404, the CPU 21 determines whether all the parameters used when determining the relationship between wavefront aberration and optical transfer function have been input. If the CPU 21 determines that all the parameters used to determine the relationship between wavefront aberration and optical transfer function have been input, it executes the process in step S405; otherwise, it executes the process in step S401. In step S405, the CPU 21 constructs a number of decision trees. In this way, a second classifier is constructed. By inputting wavefront aberration into the second classifier, the optical transfer function can be obtained. The wavefront aberration of each lens group of similar error cases is input into the second classifier, and its respective optical transfer function is estimated. From the results, the cause of the defect in each similar error case can be extracted.

[0034] In this embodiment, a decision tree is used to construct the second classifier, but the method is not limited to this, as long as it is possible to obtain the optical transfer function from the wavefront aberration. Also, in step S106, it is necessary to determine the relationship between the wavefront aberration and the optical transfer function in advance, but the method is not limited as long as the relationship can be determined.

[0035] Returning to Figure 7, in step S107, the CPU 21 first compares the optical performance (fourth optical performance) obtained by superimposing the manufacturing error corresponding to each defect cause acquired in step S106 onto the design standard value of the object under test, with the optical performance of the object under test. Then, the CPU 21 estimates the defect cause that has an optical performance tendency similar to that of the object under test as the defect. The defect described here is the defect cause of the object under test. Optical performance includes at least one of wavefront aberration, ray aberration, and image plane information. Image plane information includes image plane curvature.

[0036] The following describes an example of this embodiment. The object under test is assumed to be the same optical system 120 as the one shown in Figure 8, which consists of group 1 G1, group 2 G2, and group 3 G3 arranged in order from the object side to the image side.

[0037] In step S101, the CPU 21 creates optical system samples for each zoom position. Specifically, the CPU 21 superimposes a manufacturing tolerance within the expected range onto the design reference values ​​of the optical system to create multiple patterns of optical system samples. In this example, 3000 optical system samples were created for each zoom position to be adjusted. For clarity, in the following explanation, we will refer to the zoom positions as two locations, the W end and the T end. It is desirable to set the number of zoom positions to the number necessary to guarantee the performance of the optical system.

[0038] In step S102, the CPU 21 calculates the wavefront aberration of the optical system sample created in step S101.

[0039] In step S103, the CPU 21 associates each optical system sample number with the calculated wavefront aberration and constructs the first classifier. When the wavefront aberration is input to the first classifier, the optical system sample number is output, and the eccentricity state and optical performance of the output optical system sample can be obtained.

[0040] In step S104, the CPU 21 acquires optical information of the object being measured using the measurement device shown in Figure 1, which employs a Shack-Hartmann sensor. The acquired wavefront aberration at the T-end of the object being measured is shown in Table 1.

[0041] [Table 1]

[0042] In step S105, the CPU 21 extracts similar error cases using the wavefront aberration of the object under test obtained in step S104. The CPU 21 inputs the values ​​in Table 1 into the first classifier constructed in step S103. Table 2 shows the T-end information of the obtained similar error cases. The "Optical System Sample Number" column shows the optical system sample number of the extracted similar error case, and the "Classification Result Probability" column shows the similarity of each optical system sample to the input, with 1 being the maximum output. In this example, optical system sample number 2725 is shown to be the most similar to the optical system of the object under test among the created optical system samples.

[0043] [Table 2]

[0044] In step S106, the CPU 21 calculates the wavefront aberration for each lens group of each similar error case using the manufacturing error of each similar error case and the aberration sensitivity of each lens group of the object under test. The CPU 21 then constructs a second classifier to determine, using machine learning, how much each wavefront aberration affects the optical transfer function. At this time, it is necessary to numerically determine the relationship between wavefront aberration and optical transfer function in advance using a predetermined number of optical system samples from the similar error cases. The CPU 21 links the calculated wavefront aberration with the optical transfer function. This makes it possible to obtain the corresponding optical transfer function by inputting the wavefront aberration of any optical system sample from the similar error cases. The wavefront aberration for each lens group of each similar error case is input to the second classifier, and the extent to which each affects the optical transfer function is obtained. The lens group that has the greatest influence on the optical transfer function is estimated to be the cause of the defect in each similar error case. The causes of the T-end defects obtained from this are shown in Table 3.

[0045] [Table 3]

[0046] In step S107, the CPU 21 estimates the defects in the object under test. Specifically, the CPU 21 compares the optical performance of the object under test with the optical performance of the object under test, when the manufacturing error corresponding to each defect cause obtained in step S106 is superimposed one by one onto the design standard value of the object under test, and estimates the defect cause that has an optical performance tendency similar to that of the object under test as the defect. In this example, the trends of peripheral optical performance were compared. As a result, in this example, the parallel eccentricity of group 1 G1 is estimated to be the most defective.

[0047] By using the method of this embodiment, it becomes possible to estimate the cause of degradation of on-axis optical performance through analysis utilizing information on on-axis wavefront aberration, thereby reducing the man-hours required compared to the conventional method of manually swapping good and defective parts to find the cause of the defect.

[0048] Next, we will explain a method for correcting the entire zoom range using the causes of optical performance degradation determined by the method of this embodiment.

[0049] Perform the above steps at each zoom position where you want to adjust for the malfunction, and determine the correction points and correction amounts for each zoom position. The correction amount is calculated, for example, based on the sensitivity of the aberration to be corrected at the correction point in each zoom position.

[0050] Next, the correction process will be explained. Figure 12 is a flowchart showing the optical system correction method, illustrating the process of determining the fault location using the method described above in this embodiment and correcting the fault location.

[0051] In step S421, the CPU 21 acquires data on the faulty locations at each zoom position. The faulty locations are estimated in order of their impact on the fault, and this ranking will be used in a later stage.

[0052] In step S422, the CPU 21 identifies the first correction point in the list of faulty locations at each zoom position. The first correction point is the location where correcting the fault is most effective.

[0053] In step S423, the CPU 21 calculates the sensitivity of aberrations at each zoom position. When determining the sensitivity, it individually calculates the eccentricity, thickness, surface accuracy, etc., of the optical components that make up the optical system.

[0054] In step S424, the CPU 21 calculates the amount of correction needed to correct the defect based on the sensitivity calculated in step S423. For example, if the defect is lens tilt, it calculates the amount of tilt to correct that tilt. The resolution of the zoom position can be set to the number required for the examination. In the following explanation, for the sake of clarity, the zoom lens positions will be limited to two positions, the W end and the T end. When correcting a specific aberration at the defect location, the amount of correction needed at the defect location is determined so that the specific aberration becomes 0 (including approximately 0).

[0055] In step S425, the CPU 21 compares the correction amounts for the problematic areas at the W-end and T-end and determines the zoom position with the smallest correction amount. Here, we will explain the case where the sensitivity differs drastically between the W-end and T-end at the same location. For example, if the sensitivity at the W-end is 1 and the sensitivity at the T-end is approximately 0, the correction of that faulty area will only be effective at the W-end, thus not degrading the aberration at the T-end. If such an item is lower in the hierarchy of faulty areas, it is convenient to correct only the W-end, so even if the correction priority is low, such items may be prioritized. Furthermore, if faulty areas are concentrated in a single zoom position, ingenuity is required for correction. For example, if the purpose of fault adjustment in a variable magnification optical system is resolution, correcting a faulty zoom position may negatively affect other fault-free zoom positions. In such cases, one method is to set a standard and reduce the resolution of fault-free zoom positions to correct the faulty zoom position.

[0056] In step S426, the CPU 21 applies the amount of correction for the faulty area at the zoom position determined in step S425 to all zoom positions.

[0057] In step S427, the CPU 21 determines whether there are any undercorrected zoom positions in the variable magnification optical system after the correction amount for the faulty parts has been applied. If the CPU 21 determines that there are undercorrected zoom positions, it executes the process in step S429; otherwise, it executes the process in step S428.

[0058] In step S428, the CPU 21 completes the correction.

[0059] In step S429, the CPU 21 corrects the under-corrected zoom position using the next candidate correction location. The next candidate correction location is, for example, an item in the list of correction locations for the zoom position determined by defect estimation. At this time, the next candidate correction location is selected according to the purpose of the correction, and is not selected in order of the degree of impact from the defect location list.

[0060] In step S430, the CPU 21 identifies the next candidate correction point for the faulty area at each zoom position.

[0061] In step S431, the CPU 21 calculates the sensitivity of the identified correction points at each zoom position. When determining the sensitivity, it individually calculates the eccentricity, thickness, surface accuracy, refractive index, etc., of the optical components that make up the optical system. In addition, it calculates sensitivity items as needed, such as spherical aberration, coma aberration, field curvature, astigmatism, and distortion.

[0062] In step S432, the CPU 21 calculates the amount of correction required to correct the malfunction based on the sensitivity calculated in step S431.

[0063] In step S433, the CPU 21 compares the correction amounts for the cause at the W end and T end, and determines the zoom position with the smallest correction amount.

[0064] In step S434, the CPU 21 applies the amount of correction for the faulty area at the zoom position determined in step S433 to all zoom positions.

[0065] In step S435, the CPU 21 determines whether there are any undercorrected zoom positions in the variable magnification optical system after the correction amount for the faulty parts has been applied. If the CPU 21 determines that there are undercorrected zoom positions, it executes the process in step S436; otherwise, it executes the process in step S430.

[0066] In step S436, the CPU 21 completes the correction.

[0067] This section describes the case where corrections for the causative areas selected by the above algorithm are applied to the mechanical parts supporting the optical system. In typical mass production, mechanical parts are created by molding with a mold. Let's assume there are four desired mechanical parts. Let's call these mechanical parts A, B, C, and D. The correction effect is greatest for A, and decreases towards D. We will explain how to classify these parts by mold conditions and determine the correction locations. Mold corrections are complexly influenced by factors such as difficulty, cost, and schedule. A matrix of these factors is shown in Table 3-1.

[0068] [Table 3-1]

[0069] If there are no problems with correcting the mold, we will start with A, which has the greatest impact. For example, if it is determined that correcting mold A is unreasonable in terms of the schedule, then when correcting the remaining three locations B, C, and D, we will start with B, which has the greatest impact. Prioritization can also be determined from the perspective of the cost of repairing the mold. If repairing mold B is the most expensive, we will start with A, which has the greatest impact among A, C, and D. Prioritization can also be determined from the perspective of the precision required for repairing the mold. It is difficult to achieve precision in areas where the mold buildup direction or gaps where material shorts out are present. If repairing mold A is the most difficult, we will correct the remaining three locations B, C, and D. In that case, we will start with B, which has the greatest impact.

[0070] This algorithm is not limited to the items and events described. For example, by setting the difficulty level of mold correction in multiple stages, classifying cost contents and creating an equation that shows the relationship between difficulty and cost, or setting new parameters that take into account the relationship between schedule and delivery, it is possible to determine the order of correction of defective areas with detail and accuracy. For other items as well, depending on the priority setting, it is also possible to mechanically calculate the locations and amounts of correction required to fix defects, for example, following the flow in Figure 12. This is repeated while lowering the priority of the cause of defects at each zoom position or selecting locations where the desired correction can be performed, until it converges at the desired zoom position or across the entire zoom range. This makes it possible to perform optimal correction at the desired zoom position or across the entire zoom range of the variable magnification optical system. [Examples]

[0071] In Example 1, a method using wavefront aberration information to estimate the cause of degradation of the optical performance on the axis of an object under test, and a method for correcting the cause of that degradation were described. In this example, a method for estimating the cause of degradation of the optical performance on the axis of an object under test will be described using ray aberration information.

[0072] Figure 7 is a flowchart showing a method for inspecting the optical system to estimate the cause of a defect in this embodiment. The processing in this flowchart is executed by the CPU 21 according to a computer program.

[0073] Step S101 is the same as in Example 1 and is therefore omitted.

[0074] In step S102, the CPU 21 calculates the on-axis ray aberration (first optical performance) for each of the multiple optical system samples created in step S101. Ray aberration can be easily obtained by conversion from wavefront aberration. Therefore, it is preferable to calculate the wavefront aberration using the same calculation method as in Example 1 and then convert it to ray aberration. In this example, a method of obtaining ray aberration by conversion from wavefront aberration has been described, but this is not limited to this method as long as ray aberration can be obtained.

[0075] In step S103, the CPU 21 constructs a first classifier using the ray aberrations acquired in step S102. Referring to Figure 13, a method using decision trees to construct the first classifier will be explained. Figure 13 is a flowchart of the method for constructing the first classifier in this embodiment. In step S501, the CPU 21 inputs the ray aberrations of the optical system samples acquired in step S102. In step S502, the CPU 21 inputs the optical system sample number assigned to the optical system sample as identification information. In step S503, the CPU 21 associates the ray aberrations of the optical system samples input in steps S501 and S502 with the optical system sample numbers. In step S504, the CPU 21 determines whether input for all of the multiple patterns of optical system samples created has been completed. If the CPU 21 determines that input for all of the multiple patterns of optical system samples created has been completed, it executes the process in step S505; otherwise, it executes the process in step S501. In step S205, the CPU 21 constructs a number of decision trees. Thus, the first classifier is constructed. When ray aberration is input to the first classifier, it becomes possible to obtain the sample number of an optical system with optical performance characteristics similar to the input.

[0076] In this embodiment, a decision tree is used to construct the first classifier, but the method is not limited to this, as long as it is possible to obtain an optical system sample similar to the object under test from the optical aberration.

[0077] Returning to Figure 7, in step S104, the CPU 21 measures the optical information of the object under test. It performs a transmitted wavefront measurement of the object under test and obtains the ray aberration (first optical performance) of the object under test.

[0078] In step S105, the CPU 21 extracts similar error cases using the ray aberration of the object under test acquired in step S104 and the first classifier constructed in step S103. Similar error cases are optical system samples with an optical system configuration similar to that of the object under test. Figure 14 is a flowchart illustrating the method for extracting similar error cases in this embodiment. In step S601, the CPU 21 inputs the ray aberration of the object under test acquired in step S104 into the first classifier. In step S602, the CPU 21 classifies training data similar to the input data using machine learning. In step S603, the CPU 21 outputs optical system samples from the classified training data. This makes it possible to extract similar error cases.

[0079] Returning to Figure 7, in step S106, the CPU 21 extracts the cause of the defect for each similar error case. The CPU 21 performs calculations using the manufacturing error of each similar error case and the aberration sensitivity of each lens group in the object under test to obtain the ray aberration (second optical performance) for each lens group of each similar error case. The CPU 21 constructs a second classifier using the ray aberration of the lens groups of the similar error cases.

[0080] Referring to Figure 15, a method using a decision tree to construct the second classifier will be explained. Figure 15 is a flowchart of the method for constructing the second classifier in this embodiment. It is necessary to numerically determine the relationship between ray aberration and optical transfer function in advance using a predetermined number of optical system samples from similar error cases. The predetermined number is sufficient to determine the relationship between ray aberration and optical transfer function, and is less than the number of similar error cases. In step S701, the CPU 21 inputs the ray aberration of the predetermined number of optical system samples used to determine the relationship between ray aberration and optical transfer function. In step S702, the CPU 21 inputs the optical transfer function (third optical performance) of the predetermined number of optical system samples used to determine the relationship between ray aberration and optical transfer function. In step S703, the CPU 21 links the ray aberration and optical transfer function input in steps S701 and S702. In step S704, the CPU 21 determines whether all the parameters used to determine the relationship between ray aberration and optical transfer function have been input. If CPU21 determines that all the parameters used to determine the relationship between ray aberration and optical transfer function have been entered, it executes the process in step S705; otherwise, it executes the process in step S701. In step S705, CPU21 constructs a number of decision trees. In this way, a second classifier is constructed. By inputting ray aberration into the second classifier, the optical transfer function can be obtained. The ray aberration of each lens group of similar error cases is input into the second classifier, and its respective optical transfer function is estimated. From the results, the cause of the defect in each similar error case can be extracted.

[0081] In this embodiment, a decision tree is used to construct the second classifier, but the method is not limited to this, as long as it is possible to obtain the optical transfer function from the ray aberration. Also, in step S106, it is necessary to determine the relationship between the ray aberration and the optical transfer function in advance, but the method is not limited as long as the relationship can be determined.

[0082] Returning to Figure 7, in step S107, the CPU 21 first compares the optical performance (fourth optical performance) obtained by superimposing the manufacturing error corresponding to each defect cause acquired in step S106 onto the design standard value of the object under test, with the optical performance of the object under test. Then, the CPU 21 estimates the defect cause that has an optical performance tendency similar to that of the object under test as the defect. The defect described here is the defect cause of the object under test. Optical performance includes at least one of wavefront aberration, ray aberration, and image plane information. Image plane information includes image plane curvature.

[0083] The following describes an example of this embodiment. The object under test is assumed to be the same optical system 120 as the one shown in Figure 8, which consists of group 1 G1, group 2 G2, and group 3 G3 arranged in order from the object side to the image side.

[0084] In step S101, the CPU 21 creates optical system samples for each zoom position. Specifically, the CPU 21 superimposes a manufacturing tolerance within the expected range onto the design reference values ​​of the optical system to create multiple patterns of optical system samples. In this example, 3000 optical system samples were created for each zoom position to be adjusted. For clarity, in the following explanation, we will refer to the zoom positions as two locations, the W end and the T end. It is desirable to set the number of zoom positions to the number necessary to guarantee the performance of the optical system.

[0085] In step S102, the CPU 21 calculates the ray aberration of the optical system sample created in step S101.

[0086] In step S103, the CPU 21 associates each optical system sample number with the calculated ray aberration and constructs the first classifier. When the ray aberration is input to the first classifier, the optical system sample number is output, and the eccentricity state and optical performance of the output optical system sample can be obtained.

[0087] In step S104, the CPU 21 acquires optical information of the object being measured using the measurement device shown in Figure 1, which employs a Shack-Hartmann sensor. The acquired wavefront aberration at the T-end of the object being measured is shown in Table 4.

[0088] [Table 4]

[0089] In step S105, the CPU 21 extracts similar error cases using the ray aberration of the object under test acquired in step S104. The CPU 21 inputs the values ​​in Table 4 into the first classifier constructed in step S103. Table 5 shows the T-end information of the obtained similar error cases. The "Optical System Sample Number" column shows the optical system sample number of the extracted similar error case, and the "Classification Result Probability" column shows the similarity of each optical system sample to the input, with 1 being the maximum output. In this example, optical system sample number 2 is shown to be the most similar to the optical system of the object under test among the created optical system samples.

[0090] [Table 5]

[0091] In step S106, the CPU 21 calculates the ray aberration for each lens group of each similar error case using the manufacturing error of each similar error case and the aberration sensitivity of each lens group of the object under test. The CPU 21 then constructs a second classifier to determine, using machine learning, how much each of the acquired ray aberrations of similar error cases affects the optical transfer function. At this time, it is necessary to numerically determine the relationship between wavefront aberration and optical transfer function in advance using a predetermined number of optical system samples from similar error cases. The CPU 21 links the calculated ray aberrations with the optical transfer function. This makes it possible to obtain the corresponding optical transfer function by inputting the ray aberration of any optical system sample from the similar error cases. The ray aberrations for each lens group of each similar error case are input to the second classifier, and the extent to which each affects the optical transfer function is obtained. The lens group that has the greatest impact on the optical transfer function is estimated to be the cause of the defect in each similar error case. The causes of T-end defects obtained from this are shown in Table 6.

[0092] [Table 6]

[0093] In step S107, the CPU 21 estimates the defects in the object under test. Specifically, the CPU 21 compares the optical performance of the object under test with the optical performance of the object under test, when the manufacturing error corresponding to each defect cause obtained in step S106 is superimposed one by one onto the design standard value of the object under test, and estimates the defect cause that has an optical performance tendency similar to that of the object under test as the defect. In this example, the trends of peripheral optical performance were compared. As a result, in this example, the parallel eccentricity of group 2G2 is estimated to be the most defective.

[0094] Next, we will explain a method for correcting the entire zoom range using the causes of optical performance degradation determined by the method of this embodiment.

[0095] Perform the above steps at each zoom position where you want to adjust for the malfunction, and determine the correction points and correction amounts for each zoom position. The correction amount is calculated, for example, based on the sensitivity of the aberration to be corrected at the correction point in each zoom position.

[0096] Next, we compare the correction amount for the causative area at the zoom position where correction is first applied with the correction amounts at other zoom positions. Here, we assume that the zoom position where correction is first applied is the T-end, and the other zoom positions are the W-end. We compare the correction amount for the causative area at the T-end with the correction amount for the same area at the W-end. For example, if the correction amount at the T-end is greater than the correction amount at the W-end, we limit the correction amount at the T-end to the W-end. In this case, the correction amount at the T-end becomes insufficient, resulting in undercorrection. However, the correction amount at the W-end becomes optimal. At this point, we focus on, for example, the second item of the causative area at the T-end. In causative area estimation, we have been able to analyze multiple areas that require correction. Therefore, we use the ranking determined by that analysis and correct the undercorrection at the T-end using the correction item with the second greatest impact. If we apply the optimal correction amount at the T-end to the second greatest impact area at the T-end, similar or different aberrations may occur at the W-end. In that case, we focus on the item at the W-end that has low sensitivity at the T-end to the causative area. By using items based on estimated cause locations, correction items that have a significant impact at the W end but a small impact at the T end are used to correct for insufficient or excessive correction at the W end. This process is repeated while considering combinations of malfunction causes at each zoom position until the desired zoom position or the entire zoom range is achieved, thereby enabling optimal correction of the variable magnification optical system at the desired zoom position or the entire zoom range.

[0097] This section describes the case where corrections to the causative locations selected by the above algorithm are applied to the mechanical parts supporting the optical system. In typical mass production, mechanical parts are created by molding with a mold. Let's assume there are four desired mechanical parts. Let's call these mechanical parts A, B, C, and D. The correction effect is greatest for A, and decreases towards D. This section explains how to quantify this effect using mold correction methods to determine the correction locations. Mold correction generally involves machining and build-up. Build-up is difficult to achieve high precision with. Therefore, the difficulty of mold correction is represented by quantifying the time required, and this quantification allows for rational determination. For example, weights are assigned as follows, and each of A, B, C, and D is quantified accordingly.

[0098] Mold correction (adding material) ... 3 points Mold correction (shaving) ... 1 point Type correction (schedule) ... 1 point every 3 days An example of quantification is shown below.

[0099] A 10 points, B 3 points, C 5 points, D 7 points In this example, the higher the score, the more difficult the task, so starting with B is the most efficient approach. [Examples]

[0100] Examples 1 and 2 described the estimation of the causes of degradation in the optical performance of an object under test on its axis. This example describes a method for estimating the causes of degradation in the optical performance of an object under test off-axis using information on off-axis wavefront aberration.

[0101] Figure 7 is a flowchart showing a method for inspecting the optical system to estimate the cause of a defect in this embodiment. The processing in this flowchart is executed by the CPU 21 according to a computer program.

[0102] Step S101 is the same as in Example 1 and is therefore omitted.

[0103] In step S102, the CPU 21 calculates the off-axis wavefront aberration (first optical performance) for each of the multiple optical system samples created in step S101.

[0104] The method for calculating wavefront aberration will be explained below with reference to Figure 16. Figure 16 is a diagram illustrating the method for calculating the transmitted wavefront. Off-axis light 131 is light that passes through the optical system 121 and is focused at the image point 141. Multiple patterns of light rays are tracked from an unillustrated object point (at infinity in Figure 16) to the image point 141, and the distribution of the optical path lengths of the light rays emitted from the optical system 121 becomes the transmitted wavefront 161. The difference between the ideal transmitted wavefront 151, which has no aberration, and the actual transmitted wavefront 161 is the wavefront aberration.

[0105] Step S103 is the same as in Example 1 and is therefore omitted.

[0106] In step S104, the CPU 21 acquires optical information of the object under test. The CPU 21 measures the transmitted wavefront of the object under test and acquires the wavefront aberration (first optical performance) of the object under test.

[0107] Steps S105 to S107 are the same as in Example 1 and are therefore omitted.

[0108] The following describes an example of this embodiment. The object under test is assumed to be the same optical system 121 as the one shown in Figure 15, which consists of group 1 g1, group 2 g2, and group 3 g3 arranged in order from the object side to the image side.

[0109] In step S101, the CPU 21 creates optical system samples for each zoom position. Specifically, the CPU 21 superimposes a manufacturing tolerance within the expected range onto the design reference values ​​of the optical system to create multiple patterns of optical system samples. In this example, 3000 optical system samples were created for each zoom position to be adjusted. For clarity, in the following explanation, we will refer to the zoom positions as two locations, the W end and the T end. It is desirable to set the number of zoom positions to the number necessary to guarantee the performance of the optical system.

[0110] In step S102, the CPU 21 calculates the wavefront aberration of the optical system sample created in step S101.

[0111] In step S103, the CPU 21 associates each optical system sample number with the calculated wavefront aberration and constructs the first classifier. When the wavefront aberration is input to the first classifier, the optical system sample number is output, and the eccentricity state and optical performance of the output optical system sample can be obtained.

[0112] In step S104, the CPU 21 acquires optical information of the object being measured using the measuring device shown in Figure 1, which employs a Shack-Hartmann sensor. The acquired wavefront aberration at the T-end of the object being measured is shown in Table 7.

[0113] [Table 7]

[0114] In step S105, the CPU 21 extracts similar error cases using the wavefront aberration of the object under test acquired in step S104. The CPU 21 inputs the values ​​in Table 7 into the first classifier constructed in step S103. Table 8 shows the T-end information of the obtained similar error cases. The "Optical System Sample Number" column shows the optical system sample number of the extracted similar error case, and the "Classification Result Probability" column shows the similarity of each optical system sample to the input, with 1 being the maximum output. In this example, optical system sample number 2895 is shown to be the most similar to the optical system of the object under test among the created optical system samples.

[0115] [Table 8]

[0116] In step S106, the CPU 21 calculates the wavefront aberration for each lens group of each similar error case using the manufacturing error of each similar error case and the aberration sensitivity of each lens group of the object under test. The CPU 21 then constructs a second classifier using the acquired wavefront aberrations of similar error cases to determine, through machine learning, how much each affects the optical transfer function. At this time, it is necessary to numerically determine the relationship between wavefront aberration and optical transfer function in advance using a predetermined number of optical system samples from similar error cases. The CPU 21 links the calculated wavefront aberration with the optical transfer function. This makes it possible to obtain the corresponding optical transfer function by inputting the wavefront aberration of any optical system sample from similar error cases. The wavefront aberration for each lens group of each similar error case is input to the second classifier, and the extent to which each affects the optical transfer function is obtained. The lens group that has the greatest impact on the optical transfer function is estimated to be the cause of the defect in each similar error case. The causes of T-end defects obtained from this are shown in Table 9.

[0117] [Table 9]

[0118] In step S107, the CPU 21 estimates the defects in the object under test. Specifically, the CPU 21 compares the optical performance of the object under test with the optical performance of the object under test, when the manufacturing error corresponding to each defect cause obtained in step S106 is superimposed one by one onto the design standard value of the object under test, and estimates the defect cause that has an optical performance tendency similar to that of the object under test as the defect. In this example, the trends of optical performance on the axis were compared. As a result, in this example, the parallel eccentricity of group g3 is estimated to be the most defective.

[0119] Perform the above steps at each zoom position where you want to adjust for the malfunction, and determine the correction points and correction amounts for each zoom position. The correction amount is calculated, for example, based on the sensitivity of the aberration to be corrected at the correction point in each zoom position.

[0120] Next, we compare the correction amount for the causative area at the zoom position where correction is first applied with the correction amounts at other zoom positions. Here, we assume that the zoom position where correction is first applied is the T-end, and the other zoom positions are the W-end. We compare the correction amount for the causative area at the T-end with the correction amount for the same area at the W-end. For example, if the correction amount at the T-end is greater than the correction amount at the W-end, we limit the correction amount at the T-end to the W-end. In this case, the correction amount at the T-end becomes insufficient, resulting in undercorrection. However, the correction amount at the W-end becomes optimal. At this point, we focus on, for example, the second item of the causative area at the T-end. In causative area estimation, we have been able to analyze multiple areas that require correction. Therefore, we use the ranking determined by that analysis and correct the undercorrection at the T-end using the correction item with the second greatest impact. If we apply the optimal correction amount at the T-end to the second greatest impact area at the T-end, similar or different aberrations may occur at the W-end. In that case, we focus on the item at the W-end that has low sensitivity at the T-end to the causative area. By using items based on estimated cause locations, correction items that have a significant impact at the W end but a small impact at the T end are used to correct for insufficient or excessive correction at the W end. This process is repeated while considering combinations of malfunction causes at each zoom position until the desired zoom position or the entire zoom range is achieved, thereby enabling optimal correction of the variable magnification optical system at the desired zoom position or the entire zoom range.

[0121] This section describes the case where corrections to the cause of a problem, selected by the above algorithm, are applied to the mechanical components supporting the optical system. In this case, further considerations regarding the correction location may be necessary. This is the mechanical shape of the part supporting the lens. Figure 17 is a diagram illustrating the method and effect of correcting the mechanical components, and is a schematic diagram showing two lenses supported by mechanical components. The lens 17B on the right side of the figure has two parameters, the tilt of the receiving surface and the distance between the receiving surface and the lens 17A on the left side, determined by a single mechanical component element. In such a mechanical configuration, if this is the faulty area, it is possible to correct both the tilt and the distance of the lenses simultaneously, so a higher effect can be expected from the correction of this part. [Examples]

[0122] Example 3 described a method using wavefront aberration information to estimate the cause of degradation of the off-axis optical performance of a test object. In this example, a method for estimating the cause of degradation of the off-axis optical performance of a test object will be described using ray aberration information.

[0123] Figure 7 is a flowchart showing a method for inspecting the optical system to estimate the cause of a defect in this embodiment. The processing in this flowchart is executed by the CPU 21 according to a computer program.

[0124] Step S101 is the same as in Example 1 and is therefore omitted.

[0125] In step S102, the CPU 21 calculates the off-axis ray aberration (first optical performance) for each of the multiple optical system samples created in step S101. Ray aberration can be easily obtained by conversion from wavefront aberration. Therefore, it is preferable to calculate the wavefront aberration using the same calculation method as in Example 1 and then convert it to ray aberration. In this example, a method of obtaining ray aberration by conversion from wavefront aberration has been described, but this is not limited to this method as long as ray aberration can be obtained.

[0126] Step S103 is the same as in Example 2 and is therefore omitted.

[0127] In step S104, the CPU 21 measures the optical information of the object under test. It performs a transmitted wavefront measurement of the object under test and obtains the optical aberration (first optical performance) of the object under test.

[0128] Steps S105 to S107 are the same as in Example 2 and are therefore omitted.

[0129] The following describes an example of this embodiment. The object under test is assumed to be the same optical system 121 as the one shown in Figure 15, which consists of group 1 g1, group 2 g2, and group 3 g3 arranged in order from the object side to the image side.

[0130] In step S101, the CPU 21 creates optical system samples for each zoom position. Specifically, the CPU 21 superimposes a manufacturing tolerance within the expected range onto the design reference values ​​of the optical system to create multiple patterns of optical system samples. In this example, 3000 optical system samples were created for each zoom position to be adjusted. For clarity, in the following explanation, we will refer to the zoom positions as two locations, the W end and the T end. It is desirable to set the number of zoom positions to the number necessary to guarantee the performance of the optical system.

[0131] In step S102, the CPU 21 calculates the ray aberration of the optical system sample created in step S101.

[0132] In step S103, the CPU 21 associates each optical system sample number with the calculated wavefront aberration and constructs the first classifier. When the wavefront aberration is input to the first classifier, the optical system sample number is output, and the eccentricity state and optical performance of the output optical system sample can be obtained.

[0133] In step S104, the CPU 21 acquires optical information of the object being measured using the measurement device shown in Figure 1, which employs a Shack-Hartmann sensor. The acquired wavefront aberration at the T-end of the object being measured is shown in Table 10.

[0134] [Table 10]

[0135] In step S105, the CPU 21 extracts similar error cases using the wavefront aberration of the object under test obtained in step S104. The CPU 21 inputs the values ​​in Table 10 into the first classifier constructed in step S103. Table 11 shows the T-end information of the obtained similar error cases. The "Optical System Sample Number" column shows the optical system sample number of the extracted similar error case, and the "Classification Result Probability" column shows the similarity of each optical system sample to the input, with 1 being the maximum output. In this example, optical system sample number 555 is shown to be the most similar to the optical system of the object under test among the created optical system samples.

[0136] [Table 11]

[0137] In step S106, the CPU 21 calculates the ray aberration for each lens group of each similar error case using the manufacturing error of each similar error case and the aberration sensitivity of each lens group of the object under test. The CPU 21 then constructs a second classifier to determine, using machine learning, how much each of the acquired ray aberrations of similar error cases affects the optical transfer function. At this time, it is necessary to numerically determine the relationship between ray aberration and optical transfer function in advance using a predetermined number of optical system samples from similar error cases. The CPU 21 links the calculated ray aberration with the optical transfer function. This makes it possible to obtain the corresponding optical transfer function by inputting the ray aberration of any optical system sample from the similar error cases. The ray aberration for each lens group of each similar error case is input to the second classifier, and the extent to which each affects the optical transfer function is obtained. The lens group that has the greatest impact on the optical transfer function is estimated to be the cause of the defect in each similar error case. The causes of T-end defects obtained from this are shown in Table 12.

[0138] [Table 12]

[0139] In step S107, the CPU 21 estimates the defects in the object under test. Specifically, the CPU 21 compares the optical performance of the object under test with the optical performance of the object under test, when the manufacturing error corresponding to each defect cause obtained in step S106 is superimposed one by one onto the design standard value of the object under test, and estimates the defect cause that has an optical performance tendency similar to that of the object under test as the defect. In this example, the trends in optical performance on the axis were compared. As a result, in this example, the tilt of group 1g1 is estimated to be the most defective.

[0140] Next, we will explain a method for correcting the entire zoom range using the causes of optical performance degradation determined by the method of this embodiment.

[0141] Perform the above steps at each zoom position where you want to adjust for the malfunction, and determine the correction points and correction amounts for each zoom position. The correction amount is calculated, for example, based on the sensitivity of the aberration to be corrected at the correction point in each zoom position.

[0142] Next, we compare the correction amount for the causative area at the zoom position where correction is first applied with the correction amounts at other zoom positions. Here, we assume that the zoom position where correction is first applied is the T-end, and the other zoom positions are the W-end. We compare the correction amount for the causative area at the T-end with the correction amount for the same area at the W-end. For example, if the correction amount at the T-end is greater than the correction amount at the W-end, we limit the correction amount at the T-end to the W-end. In this case, the correction amount at the T-end becomes insufficient, resulting in undercorrection. However, the correction amount at the W-end becomes optimal. At this point, we focus on, for example, the second item of the causative area at the T-end. In causative area estimation, we have been able to analyze multiple areas that require correction. Therefore, we use the ranking determined by that analysis and correct the undercorrection at the T-end using the correction item with the second greatest impact. If we apply the optimal correction amount at the T-end to the second greatest impact area at the T-end, similar or different aberrations may occur at the W-end. In that case, we focus on the item at the W-end that has low sensitivity at the T-end to the causative area. By using items based on estimated cause locations, correction items that have a significant impact at the W end but a small impact at the T end are used to correct for insufficient or excessive correction at the W end. This process is repeated while considering combinations of malfunction causes at each zoom position until the desired zoom position or the entire zoom range is achieved, thereby enabling optimal correction of the variable magnification optical system at the desired zoom position or the entire zoom range. [Examples]

[0143] Examples 1 to 4 described methods for estimating the causes of deterioration in the optical performance of an object under test. This example describes a method for estimating the optical transfer function at the time of aberration occurrence from aberration information of the object under test.

[0144] Figure 18 is a flowchart showing the method for estimating the optical transfer function in this embodiment. The processing in this flowchart is executed by the CPU 21 according to the computer program.

[0145] In step S801, the CPU 21 calculates the relationship between aberration information and the optical transfer function. The aberration information includes at least one of wavefront aberration or ray aberration. Using the aberration information, the CPU 21 calculates how much the optical transfer function deteriorates when the aberration occurs.

[0146] In step S802, the CPU 21 constructs a second classifier. The method for constructing the second classifier using a decision tree will be explained below with reference to Figure 19. Figure 19 is a flowchart showing the method for constructing the second classifier in this embodiment. In step S901, the CPU 21 inputs aberration information. In step S902, the CPU 21 inputs the optical transfer function. In step S903, the CPU 21 links the aberration information input in steps S901 and S902 with the optical transfer function. In step S904, the CPU 21 determines whether all the parameters used to determine the relationship between the aberration information and the optical transfer function have been input. If the CPU 21 determines that all the parameters used to determine the relationship between the aberration information and the optical transfer function have been input, it executes the process in step S905; otherwise, it executes the process in step S901. In step S905, the CPU 21 constructs a number of decision trees. In this way, the second classifier is constructed. By inputting aberration information into the second classifier, the optical transfer function can be obtained.

[0147] In this embodiment, a decision tree is used to construct the second classifier, but the method is not limited to this, as long as it is possible to obtain optical system samples similar to the object under test from aberration information.

[0148] Furthermore, while this embodiment uses a decision tree to construct the second classifier, it is not limited to this method, as long as it is possible to obtain the optical transfer function from the aberration information.

[0149] In step S803, the CPU 21 inputs aberration information to the second discriminator.

[0150] In step S804, the CPU 21 estimates (obtains) the optical transfer function for the input estimated by the second discriminator.

[0151] In step S805, the CPU 21 outputs the optical transfer function for the input.

[0152] The following describes an example of this embodiment. The object under test is assumed to be the same optical system 120 as the one shown in Figure 8, which consists of group 1 G1, group 2 G2, and group 3 G3 arranged in order from the object side to the image side.

[0153] In step S801, the CPU 21 calculates the relationship between aberration information and the optical transfer function in the optical system for which the optical transfer function is to be determined.

[0154] In step S802, the CPU 21 constructs a second classifier to determine the optical transfer function. Specifically, the CPU 21 uses machine learning with a decision tree to link the aberration information of the optical system calculated in step S801 with the optical transfer function and learns the trend.

[0155] In step S803, the CPU 21 inputs aberration information of the optical system for which the optical transfer function is to be determined into the second classifier constructed in step S802. In this example, the aberration information shown in Table 13 was input. Here, the coma component is shown using the Zernike coefficient.

[0156] [Table 13]

[0157] In step S804, the CPU 21 obtains the optical transfer function for the input, which was estimated from a number of decision trees by the second classifier constructed in step S802.

[0158] In step S805, the CPU21 outputs the estimation results. An example of the output display is shown in Table 14. The estimated optical transfer function is displayed in the second row, in the format [[Optical transfer function in the meridional direction, Optical transfer function in the sagittal direction]], under the row "prediction". Multiplying the displayed value by 100 will change the unit to "%".

[0159] [Table 14]

[0160] By using the method of this embodiment, the optical transfer function can be estimated from aberration information. This enables the analysis of optical performance defects as shown in other embodiments.

[0161] Perform the above steps at each zoom position where you want to adjust for the malfunction, and determine the correction points and correction amounts for each zoom position. The correction amount is calculated, for example, based on the sensitivity of the aberration to be corrected at the correction point in each zoom position.

[0162] Next, we compare the correction amount for the causative area at the zoom position where correction is first applied with the correction amounts at other zoom positions. Here, we assume that the zoom position where correction is first applied is the T-end, and the other zoom positions are the W-end. We compare the correction amount for the causative area at the T-end with the correction amount for the same area at the W-end. For example, if the correction amount at the T-end is greater than the correction amount at the W-end, we limit the correction amount at the T-end to the W-end. In this case, the correction amount at the T-end becomes insufficient, resulting in undercorrection. However, the correction amount at the W-end becomes optimal. At this point, we focus on, for example, the second item of the causative area at the T-end. In causative area estimation, we have been able to analyze multiple areas that require correction. Therefore, we use the ranking determined by that analysis and correct the undercorrection at the T-end using the correction item with the second greatest impact. If we apply the optimal correction amount at the T-end to the second greatest impact area at the T-end, similar or different aberrations may occur at the W-end. In that case, we focus on the item at the W-end that has low sensitivity at the T-end to the causative area. By using items based on estimated cause locations, correction items that have a significant impact at the W end but a small impact at the T end are used to correct for insufficient or excessive correction at the W end. This process is repeated while considering combinations of malfunction causes at each zoom position until the desired zoom position or the entire zoom range is achieved, thereby enabling optimal correction of the variable magnification optical system at the desired zoom position or the entire zoom range. [Examples]

[0163] This embodiment describes a method for estimating the cause of optical performance degradation using image plane information obtained from transmitted light measurement results. The image plane information includes image plane curvature.

[0164] Figure 7 is a flowchart showing a method for inspecting the optical system to estimate the cause of a defect in this embodiment. The processing in this flowchart is executed by the CPU 21 according to a computer program.

[0165] Step S101 is the same as in Example 1 and is therefore omitted.

[0166] In step S102, the CPU 21 calculates image plane information (first optical performance) from the transmitted light of each of the multiple optical system samples created in step S101.

[0167] In step S103, the CPU 21 constructs a first classifier using the image plane information acquired in step S102. Referring to Figure 20, a method using a decision tree to construct the first classifier will be explained. Figure 20 is a flowchart of the method for constructing the first classifier in this embodiment. In step S351, the CPU 21 inputs the image plane information of the optical system sample acquired in step S102. In step S352, the CPU 21 inputs the optical system sample number assigned to the optical system sample as identification information. In step S353, the CPU 21 associates the image plane information of the optical system sample input in steps S351 and S352 with the optical system sample number. In step S354, the CPU 21 determines whether input for all of the multiple patterns of optical system samples created has been completed. If the CPU 21 determines that input for all of the multiple patterns of optical system samples created has been completed, it executes the process in step S355; otherwise, it executes the process in step S351. In step S355, the CPU 21 constructs a number of decision trees. This is how the first classifier is constructed. When image plane information is input to the first classifier, it becomes possible to obtain optical system sample numbers that have optical performance tendencies similar to the input.

[0168] In this embodiment, a decision tree is used to construct the first classifier, but the method is not limited to this, as long as it is possible to obtain optical system samples similar to the object under test from the image plane information.

[0169] Returning to Figure 7, in step S104, the CPU 21 acquires the optical information of the object under test. The CPU 21 performs a transmitted wavefront measurement of the object under test and acquires the image plane information (first optical performance) of the object under test.

[0170] In step S105, the CPU 21 extracts similar error cases using the image plane information of the object to be examined acquired in step S104 and the first classifier constructed in step S103.

[0171] In step S106, the CPU 21 extracts the cause of the defect for each similar error case. The CPU 21 performs calculations using the manufacturing error of each similar error case and the aberration sensitivity of each lens group in the object under test to obtain image plane information (second optical performance) for each lens group of each similar error case. The CPU 21 constructs a second classifier using the wavefront aberration of the lens groups of the obtained similar error cases.

[0172] Referring to Figure 21, a method for constructing a second classifier using a decision tree will be described. Figure 21 is a flowchart of the method for constructing the second classifier in this embodiment. It is necessary to numerically determine the relationship between image plane information and optical transfer function in advance using a predetermined number of optical system samples from similar error cases. The predetermined number is sufficient to determine the relationship between wavefront aberration and optical transfer function, and is less than the number of similar error cases. In step S361, the CPU 21 inputs the image plane information of the predetermined number of optical system samples used when determining the relationship between image plane information and optical transfer function. In step S362, the CPU 21 inputs the optical transfer function (third optical performance) of the predetermined number of optical system samples used when determining the relationship between image plane information and optical transfer function. In step S363, the CPU 21 links the image plane information and optical transfer function input in steps S361 and S362. In step S364, the CPU 21 determines whether all the parameters used when determining the relationship between image plane information and optical transfer function have been input. If the CPU 21 determines that all the parameters used to determine the relationship between the image plane information and the optical transfer function have been input, it executes the process in step S365; otherwise, it executes the process in step S361. In step S365, the CPU 21 constructs a number of decision trees. In this way, a second classifier is constructed. By inputting the image plane information into the second classifier, the optical transfer function can be obtained. The image plane information of each lens group for each similar error case is input into the second classifier, and the respective optical transfer functions are estimated. From the results, the cause of the defect for each similar error case can be extracted.

[0173] In this embodiment, a decision tree is used to construct the second classifier, but the method is not limited to this, as long as it is possible to obtain the optical transfer function from the image plane information. Also, in step S106, it is necessary to determine the relationship between wavefront aberration and the optical transfer function in advance, but the method is not limited as long as the relationship can be determined.

[0174] In step S107, the CPU 21 first compares the optical performance (fourth optical performance) obtained by superimposing the manufacturing error corresponding to each defect cause acquired in step S106 onto the design standard value of the object under test, with the optical performance of the object under test. Then, the CPU 21 estimates the defect cause that has an optical performance tendency similar to that of the object under test as the defect. The defect described here is the defect cause of the object under test. Optical performance includes at least one of wavefront aberration, ray aberration, and image plane information.

[0175] The following describes an example of this embodiment. The object under test is assumed to be the same optical system 120 as the one shown in Figure 8, which consists of group 1 G1, group 2 G2, and group 3 G3 arranged in order from the object side to the image side.

[0176] In step S101, the CPU 21 creates optical system samples for each zoom position. Specifically, the CPU 21 superimposes a manufacturing tolerance within the expected range onto the design reference values ​​of the optical system to create multiple patterns of optical system samples. In this example, 3000 optical system samples were created for each zoom position to be adjusted. For clarity, in the following explanation, we will refer to the zoom positions as two locations, the W end and the T end. It is desirable to set the number of zoom positions to the number necessary to guarantee the performance of the optical system.

[0177] In step S102, the CPU 21 calculates the image plane information of the optical system sample created in step S101.

[0178] In step S103, the CPU 21 associates each optical system sample number with the image plane information obtained through calculation and constructs the first classifier. When the image plane information is input to the first classifier, the optical system sample number is output, and the eccentricity state and optical performance of the output optical system sample can be obtained.

[0179] In step S104, the CPU 21 acquires image plane information of the object by measuring the transmitted light of the object. The acquired image plane information of the object is shown in Table 15.

[0180] [Table 15]

[0181] In step S105, the CPU 21 extracts similar error cases using the image plane information of the object under test acquired in step S104. The CPU 21 inputs the values ​​in Table 15 into the first classifier constructed in step S103. The obtained similar error cases are shown in Table 16. The "Optical System Sample Number" column shows the optical system sample number of the extracted similar error case, and the "Classification Result Probability" column shows the similarity of each optical system sample to the input, with 1 being the maximum output. In this example, optical system sample number 2617 is shown to be the most similar to the optical system of the object under test among the created optical system samples.

[0182] [Table 16]

[0183] In step S106, the CPU 21 performs calculations using the manufacturing error of each similar error case and the aberration sensitivity of each lens group in the object under test to obtain image plane information for each lens group of each similar error case. The CPU 21 uses the obtained image plane information of the similar error cases to construct a second classifier for machine learning to determine how much each affects the optical transfer function. At this time, it is necessary to numerically determine the relationship between image plane information and the optical transfer function in advance using a predetermined number of optical system samples from the similar error cases. The CPU 21 links the calculated image plane information with the optical transfer function. This makes it possible to obtain the corresponding optical transfer function by inputting the image plane information of any optical system sample from the similar error cases. The image plane information for each lens group of each similar error case is input to the second classifier to obtain how much each affects the optical transfer function. The lens group that has the greatest influence on the optical transfer function is estimated to be the cause of the defect in each similar error case. The defect causes obtained from this are shown in Table 17.

[0184] [Table 17]

[0185] In step S107, the CPU 21 estimates the defects in the object under test. Specifically, the CPU 21 compares the optical performance of the object under test with the optical performance of the object under test, when the manufacturing error corresponding to each defect cause obtained in step S106 is superimposed one by one onto the design standard value of the object under test, and estimates the defect cause that has an optical performance tendency similar to that of the object under test as the defect. In this example, the tilt of group 2G2 is estimated to be the most likely defect.

[0186] By performing analysis using image plane information with the method of this embodiment, the amount of work required for analysis is reduced, making it possible to efficiently estimate the cause of deterioration in the optical performance of the object under test.

[0187] Perform the above steps at each zoom position where you want to adjust for the malfunction, and determine the correction points and correction amounts for each zoom position. The correction amount is calculated, for example, based on the sensitivity of the aberration to be corrected at the correction point in each zoom position.

[0188] Next, we compare the correction amount for the causative area at the zoom position where correction is first applied with the correction amounts at other zoom positions. Here, we assume that the zoom position where correction is first applied is the T-end, and the other zoom positions are the W-end. We compare the correction amount for the causative area at the T-end with the correction amount for the same area at the W-end. For example, if the correction amount at the T-end is greater than the correction amount at the W-end, we limit the correction amount at the T-end to the W-end. In this case, the correction amount at the T-end becomes insufficient, resulting in undercorrection. However, the correction amount at the W-end becomes optimal. At this point, we focus on, for example, the second item of the causative area at the T-end. In causative area estimation, we have been able to analyze multiple areas that require correction. Therefore, we use the ranking determined by that analysis and correct the undercorrection at the T-end using the correction item with the second greatest impact. If we apply the optimal correction amount at the T-end to the second greatest impact area at the T-end, similar or different aberrations may occur at the W-end. In that case, we focus on the item at the W-end that has low sensitivity at the T-end to the causative area. By using items based on estimated cause locations, correction items that have a significant impact at the W end but a small impact at the T end are used to correct for insufficient or excessive correction at the W end. This process is repeated while considering combinations of malfunction causes at each zoom position until the desired zoom position or the entire zoom range is achieved, thereby enabling optimal correction of the variable magnification optical system at the desired zoom position or the entire zoom range. [Examples]

[0189] In this embodiment, a method for estimating defects at an arbitrary image height using both on-axis and off-axis aberration information is described in order to improve the accuracy of defect estimation. The aberration information described hereafter includes at least one of on-axis and off-axis wavefront aberration and on-axis and off-axis ray aberration.

[0190] Figure 7 is a flowchart showing a method for inspecting the optical system to estimate the cause of a defect in this embodiment. The processing in this flowchart is executed by the CPU 21 according to a computer program.

[0191] Step S101 is the same as in Example 1 and is therefore omitted.

[0192] In step S102, the CPU 21 calculates the aberration information (first optical performance) for each of the multiple optical system samples created in step S101.

[0193] In step S103, the CPU 21 constructs a first classifier using the aberration information acquired in step S102. Referring to Figure 22, a method using a decision tree to construct the first classifier will be explained. Figure 22 is a flowchart of the method for constructing the first classifier in this embodiment. In step S451, the CPU 21 inputs the aberration information of the optical system sample acquired in step S102. In step S452, the CPU 21 inputs the optical system sample number assigned to the optical system sample as identification information. In step S453, the CPU 21 associates the image plane information of the optical system sample input in steps S451 and S452 with the optical system sample number. In step S454, the CPU 21 determines whether input for all of the multiple patterns of optical system samples created has been completed. If the CPU 21 determines that input for all of the multiple patterns of optical system samples created has been completed, it executes the process in step S455; otherwise, it executes the process in step S451. In step S455, the CPU 21 constructs a number of decision trees. This is how the first classifier is constructed. When aberration information is input into the first classifier, it becomes possible to obtain optical system sample numbers that have optical performance tendencies similar to the input.

[0194] In this embodiment, a decision tree is used to construct the first classifier, but the method is not limited to this, as long as it is possible to obtain optical system samples similar to the object under test from aberration information.

[0195] Returning to Figure 7, in step S104, the CPU 21 performs a transmitted wavefront measurement at an arbitrary image height of the object under test and acquires aberration information (first optical performance) of the object under test.

[0196] In step S105, the CPU 21 extracts similar error cases using the aberration information of the object under test acquired in step S104 and the first classifier constructed in step S103.

[0197] In step S106, the CPU 21 extracts the cause of the defect for each similar error case. The CPU 21 performs calculations using the manufacturing error of each similar error case and the aberration sensitivity of each lens group in the object under test to obtain aberration information (second optical performance) for each lens group of each similar error case. The CPU 21 constructs a second classifier using the aberration information of the lens groups of the similar error cases that it has obtained.

[0198] Referring to Figure 23, a method for constructing a second classifier using a decision tree will be described. Figure 23 is a flowchart of the method for constructing the second classifier in this embodiment. It is necessary to numerically determine the relationship between aberration information and optical transfer function in advance using a predetermined number of optical system samples from similar error cases. The predetermined number is sufficient to determine the relationship between aberration information and optical transfer function, and is less than the number of similar error cases. In step S461, the CPU 21 inputs the wavefront aberrations of the predetermined number of optical system samples used when determining the relationship between aberration information and optical transfer function. In step S462, the CPU 21 inputs the optical transfer function (third optical performance) of the predetermined number of optical system samples used when determining the relationship between aberration information and optical transfer function. In step S463, the CPU 21 links the aberration information and optical transfer function input in steps S461 and S462. In step S464, the CPU 21 determines whether all the parameters used when determining the relationship between aberration information and optical transfer function have been input. If the CPU 21 determines that all the parameters used to determine the relationship between aberration information and the optical transfer function have been input, it executes the process in step S465; otherwise, it executes the process in step S461. In step S465, the CPU 21 constructs a number of decision trees. In this way, a second classifier is constructed. By inputting aberration information into the second classifier, the optical transfer function can be obtained. The aberration information of each lens group of similar error cases is input into the second classifier, and its respective optical transfer function is estimated. From the results, the cause of the defect in each similar error case can be extracted.

[0199] In this embodiment, a decision tree is used to construct the second classifier, but the method is not limited to this, as long as it is possible to obtain the optical transfer function from the aberration information. Also, in step S106, it is necessary to determine the relationship between the aberration information and the optical transfer function in advance, but the method is not limited as long as the relationship can be determined.

[0200] In step S107, the CPU 21 first compares the optical performance (fourth optical performance) obtained by superimposing the manufacturing error corresponding to each defect cause acquired in step S106 onto the design standard value of the object under test, with the optical performance of the object under test. Then, the CPU 21 estimates the defect cause that has an optical performance tendency similar to that of the object under test as the defect.

[0201] The following describes an example of this embodiment. The object under test is assumed to be the same optical system 120 as the one shown in Figure 8, which consists of group 1 G1, group 2 G2, and group 3 G3 arranged in order from the object side to the image side.

[0202] In step S101, the CPU 21 creates optical system samples for each zoom position. Specifically, the CPU 21 superimposes a manufacturing tolerance within the expected range onto the design reference values ​​of the optical system to create multiple patterns of optical system samples. In this example, 3000 optical system samples were created for each zoom position to be adjusted. For clarity, in the following explanation, we will refer to the zoom positions as two locations, the W end and the T end. It is desirable to set the number of zoom positions to the number necessary to guarantee the performance of the optical system.

[0203] In step S102, the CPU 21 calculates the aberration information of the optical system sample created in step S101.

[0204] In step S103, the CPU 21 associates each optical system sample number with the calculated aberration information and constructs a first classifier. When the aberration information is input to the first classifier, the optical system sample number is output, and the eccentricity state and optical performance of the output optical system sample can be obtained.

[0205] In step S104, the CPU 21 acquires optical information of the object under test. The acquired wavefront aberrations of the object under test are shown in Table 18.

[0206] [Table 18]

[0207] In step S105, the CPU 21 extracts similar error cases using the aberration information of the object under test obtained in step S104. The CPU 21 inputs the values ​​in Table 18 into the first classifier constructed in step S103. The information of the obtained similar error cases is shown in Table 19. The "Optical System Sample Number" column shows the optical system sample number of the extracted similar error case, and the "Classification Result Probability" column shows the similarity of each optical system sample to the input, with 1 being the maximum output. In this example, optical system sample number 529 is shown to be the most similar to the optical system of the object under test among the created optical system samples.

[0208] [Table 19]

[0209] In step S106, the CPU 21 performs calculations using the manufacturing error of each similar error case and the aberration sensitivity of each lens group in the object under test to obtain aberration information for each lens group of each similar error case. The CPU 21 uses the obtained aberration information of the similar error cases to construct a second classifier that uses machine learning to determine how much each affects the optical transfer function. At this time, it is necessary to numerically determine the relationship between aberration information and the optical transfer function in advance using a predetermined number of optical system samples from the similar error cases. The CPU 21 links the calculated aberration information with the optical transfer function. This makes it possible to obtain the corresponding optical transfer function by inputting the aberration information of any optical system sample from the similar error cases. The aberration information for each lens group of each similar error case is input to the second classifier to obtain how much each affects the optical transfer function. The lens group that has the greatest impact on the optical transfer function is estimated to be the cause of the defect in each similar error case. The defects obtained from this are shown in Table 20.

[0210] [Table 20]

[0211] In step S107, the CPU 21 estimates the defects in the object under test. Specifically, the CPU 21 compares the optical performance of the object under test with the optical performance of the object under test, when the manufacturing error corresponding to each defect cause obtained in step S106 is superimposed one by one onto the design standard value of the object under test, and estimates the defect cause that has an optical performance tendency similar to that of the object under test as the defect. In this example, the parallel eccentricity of group 2G2 is estimated to be the most likely defect.

[0212] By using the method of this embodiment, it becomes possible to estimate defects from more information, thereby improving the estimation accuracy.

[0213] Perform the above steps at each zoom position where you want to adjust for the malfunction, and determine the correction points and correction amounts for each zoom position. The correction amount is calculated, for example, based on the sensitivity of the aberration to be corrected at the correction point in each zoom position.

[0214] Next, we will explain the process of making corrections.

[0215] Next, we compare the correction amount for the causative area at the zoom position where correction is first applied with the correction amounts at other zoom positions. Here, we assume that the zoom position where correction is first applied is the T-end, and the other zoom positions are the W-end. We compare the correction amount for the causative area at the T-end with the correction amount for the same area at the W-end. For example, if the correction amount at the T-end is greater than the correction amount at the W-end, we limit the correction amount at the T-end to the W-end. In this case, the correction amount at the T-end becomes insufficient, resulting in undercorrection. However, the correction amount at the W-end becomes optimal. At this point, we focus on, for example, the second item of the causative area at the T-end. In causative area estimation, we have been able to analyze multiple areas that require correction. Therefore, we use the ranking determined by that analysis and correct the undercorrection at the T-end using the correction item with the second greatest impact. If we apply the optimal correction amount at the T-end to the second greatest impact area at the T-end, similar or different aberrations may occur at the W-end. In that case, we focus on the item at the W-end that has low sensitivity at the T-end to the causative area. By using items based on estimated cause locations, correction items that have a significant impact at the W end but a small impact at the T end are used to correct for insufficient or excessive correction at the W end. This process is repeated while considering combinations of malfunction causes at each zoom position until the desired zoom position or the entire zoom range is achieved, thereby enabling optimal correction of the variable magnification optical system at the desired zoom position or the entire zoom range. (Other examples) The present invention can also be realized by supplying a program that implements one or more of the functions of the above-described embodiments to a system or device via a network or storage medium, and by having one or more processors in the computer of that system or device read and execute the program. It can also be realized by a circuit (e.g., an ASIC) that implements one or more functions.

[0216] This embodiment includes the following methods and configurations. (Method 1) A step of obtaining a first optical performance obtained by measuring the variable magnification optical system, The steps include: inputting the first optical performance into a first discriminator and extracting multiple samples from multiple optical system samples generated by superimposing a predetermined amount onto the design reference value of the variable magnification optical system; The steps include obtaining information about a second classifier corresponding to each of the aforementioned multiple samples, The steps include inputting the second optical performance of the plurality of samples into the second discriminator to obtain information regarding the cause of the malfunction of the variable magnification optical system, A method for adjusting an optical system, characterized by comprising the step of determining a correction amount or correction priority corresponding to the cause of failure for each of the multiple zoom positions of the variable magnification optical system. (Method 2) The method for adjusting an optical system according to Method 1, further comprising the step of preparing the aforementioned plurality of optical system samples. (Method 3) The method for adjusting an optical system according to method 1 or 2, characterized in that the predetermined amount is a value that satisfies at least one predetermined range of parallel eccentricity of the lens group, inclined eccentricity of the lens group, spacing between lens groups, thickness of the lenses, and lens surface shape relative to the design reference value. (Method 4) The optical system adjustment method according to Method 3, characterized in that the predetermined range is in the range of 1 to 5 times the manufacturing tolerance range. (Method 5) The optical system adjustment method according to any one of methods 1 to 4, characterized in that the first discriminator is constructed by learning using the identification information of each of the plurality of optical system samples and the first optical performance of each of the plurality of optical system samples. (Method 6) The method for adjusting an optical system according to any one of methods 1 to 4, characterized in that the first optical performance includes at least one of aberration information and image plane information of the variable magnification optical system. (Method 7) The method for adjusting an optical system according to method 6, characterized in that the aberration information is wavefront aberration or ray aberration. (Method 8) The optical system adjustment method according to method 6 or 7, characterized in that the image plane information includes image plane curvature. (Method 9) The method for adjusting an optical system according to any one of methods 1 to 8, characterized in that the second discriminator is constructed by learning using the second optical performance and a third optical performance corresponding to the second optical performance. (Method 10) The optical system adjustment method according to Method 9, characterized in that, before the second discriminator is constructed, the relationship between the second optical performance and the third optical performance of a predetermined number of samples from the plurality of samples is determined. (Method 11) The method for adjusting an optical system according to method 9 or 10, characterized in that, in the step of obtaining information regarding the cause of a malfunction in the variable magnification optical system, the second optical performance of the plurality of samples is input to the second discriminator to obtain the third optical performance. (Method 12) The optical system adjustment method according to method 11, characterized in that, in the step of obtaining information regarding the cause of the malfunction of the variable magnification optical system, information for estimating the cause of the malfunction of the variable magnification optical system is obtained based on the third optical performance. (Method 13) The optical system adjustment method according to method 12, further comprising the step of estimating a malfunction of the variable magnification optical system by comparing the fourth optical performance of the plurality of samples with the optical performance of the variable magnification optical system, thereby estimating a malfunction of the variable magnification optical system from the causes of defects in the plurality of samples extracted based on the causes of defects in the variable magnification optical system. (Method 14) The optical system adjustment method according to method 13, characterized in that the fourth optical performance is the optical performance when the manufacturing error of each of the defects in the plurality of samples is superimposed on the design reference value individually. (Method 15) A method for adjusting an optical system according to any one of methods 9 to 14, characterized in that the third optical performance is an optical transfer function. (Method 16) The method for adjusting an optical system according to any one of the configurations of Method 1 to 15, characterized in that the second optical performance is calculated using the manufacturing error of the plurality of samples and the sensitivity to aberrations of each lens group included in the variable magnification optical system. (Method 17) The method for adjusting an optical system according to any one of the configurations of methods 1 to 16, characterized in that the second optical performance includes at least one of aberration information and image plane information of the variable magnification optical system. (Method 18) The optical system adjustment method according to method 17, characterized in that the aberration information is wavefront aberration or ray aberration. (Method 19) The optical system adjustment method according to method 16 or 17, characterized in that the image plane information includes image plane curvature. (Composition 1) A program characterized by causing a computer to execute the optical system adjustment method described in any one of methods 1 to 19. (Configuration 2) A light source that outputs a light beam for measuring the transmitted wavefront, Variable magnification optical system, A wavefront sensor for detecting the transmitted wavefront, It has a calculation unit that performs calculations using the output from the wavefront sensor, The apparatus is characterized in that the calculation unit acquires a first optical performance of the variable magnification optical system using the output from the wavefront sensor, inputs the first optical performance to a first classifier, extracts a plurality of samples from a plurality of optical system samples generated by superimposing a predetermined amount on the design reference value of the variable magnification optical system, acquires information for a second classifier corresponding to each of the plurality of samples, inputs the second optical performance of the plurality of samples to a second classifier, acquires information for the cause of defects in the variable magnification optical system, and determines a correction amount or correction priority corresponding to the cause of defects for each zoom position of the variable magnification optical system.

[0217] The embodiments described above are merely representative examples, and various modifications and changes can be made to each embodiment when implementing the present invention. [Explanation of Symbols]

[0218] light source 1,2 Wavefront sensors 4, 5 CPU (arithmetic unit) 21 Optical system 120

Claims

1. A step of obtaining a first optical performance obtained by measuring the variable magnification optical system, The steps include: inputting the first optical performance into a first discriminator and extracting multiple samples from multiple optical system samples generated by superimposing a predetermined amount onto the design reference value of the variable magnification optical system; The steps include obtaining information about a second classifier corresponding to each of the aforementioned multiple samples, The steps include inputting the second optical performance of the plurality of samples into the second discriminator to obtain information regarding the cause of the malfunction of the variable magnification optical system, A method for adjusting an optical system, characterized by comprising the step of determining a correction amount or correction priority corresponding to the cause of failure for each of the multiple zoom positions of the variable magnification optical system.

2. The method for adjusting an optical system according to claim 1, further comprising the step of preparing the plurality of optical system samples.

3. The method for adjusting an optical system according to claim 1 or 2, characterized in that the predetermined amount is a value that satisfies at least one predetermined range of parallel eccentricity of the lens group, inclined eccentricity of the lens group, spacing between lens groups, thickness of the lenses, and shape of the lens surface, relative to the design reference value.

4. The method for adjusting an optical system according to claim 3, characterized in that the predetermined range is in the range of 1 to 5 times the manufacturing tolerance range.

5. The method for adjusting an optical system according to claim 1 or 2, characterized in that the first classifier is constructed by learning using the identification information of each of the plurality of optical system samples and the first optical performance of each of the plurality of optical system samples.

6. The method for adjusting an optical system according to claim 1 or 2, characterized in that the first optical performance includes at least one of aberration information and image plane information of the variable magnification optical system.

7. The method for adjusting an optical system according to claim 6, characterized in that the aberration information is wavefront aberration or ray aberration.

8. The optical system adjustment method according to claim 6, characterized in that the image plane information includes image plane curvature.

9. The method for adjusting an optical system according to claim 1 or 2, characterized in that the second discriminator is constructed by learning using the second optical performance and a third optical performance corresponding to the second optical performance.

10. The method for adjusting an optical system according to claim 9, characterized in that, before the second discriminator is constructed, the relationship between the second optical performance and the third optical performance of a predetermined number of samples from the plurality of samples is determined.

11. The method for adjusting an optical system according to claim 9, characterized in that, in the step of obtaining information regarding the cause of a malfunction in the variable magnification optical system, the second optical performance of the plurality of samples is input to the second discriminator to obtain the third optical performance.

12. The optical system adjustment method according to claim 11, characterized in that, in the step of obtaining information regarding the cause of the malfunction of the variable magnification optical system, information for estimating the cause of the malfunction of the variable magnification optical system is obtained based on the third optical performance.

13. The optical system adjustment method according to claim 12, further comprising the step of estimating a malfunction of the variable magnification optical system by comparing the fourth optical performance of the plurality of samples with the optical performance of the variable magnification optical system, thereby estimating a malfunction of the variable magnification optical system from the causes of defects in the plurality of samples extracted based on the causes of defects in the variable magnification optical system.

14. The method for adjusting an optical system according to claim 13, characterized in that the fourth optical performance is the optical performance when the manufacturing error of each of the defects in the plurality of samples is superimposed on the design reference value individually.

15. The method for adjusting an optical system according to claim 9, characterized in that the third optical performance is an optical transfer function.

16. The method for adjusting an optical system according to claim 1 or 2, characterized in that the second optical performance is calculated using the manufacturing tolerance of the plurality of samples and the sensitivity to aberrations of each lens group included in the variable magnification optical system.

17. The method for adjusting an optical system according to claim 1 or 2, characterized in that the second optical performance includes at least one of the aberration information and image plane information of the variable magnification optical system.

18. The method for adjusting an optical system according to claim 17, characterized in that the aberration information is wavefront aberration or ray aberration.

19. The optical system adjustment method according to claim 17, characterized in that the image plane information includes image plane curvature.

20. A program characterized by causing a computer to execute the optical system adjustment method described in claim 1 or 2.

21. A light source that outputs a light beam for measuring the transmitted wavefront, Variable magnification optical system, A wavefront sensor for detecting the transmitted wavefront, It has a calculation unit that performs calculations using the output from the wavefront sensor, The apparatus is characterized in that the calculation unit acquires a first optical performance of the variable magnification optical system using the output from the wavefront sensor, inputs the first optical performance to a first classifier, extracts a plurality of samples from a plurality of optical system samples generated by superimposing a predetermined amount on the design reference value of the variable magnification optical system, acquires information for a second classifier corresponding to each of the plurality of samples, inputs the second optical performance of the plurality of samples to a second classifier, acquires information for the cause of defects in the variable magnification optical system, and determines a correction amount or correction priority corresponding to the cause of defects for each zoom position of the variable magnification optical system.

Images