Method and device for geometric characterization of optical objectives to be manufactured by stacking several optical elements

Abstract

The present invention relates to a method (400) for optical characterization of a target optical objective to be manufactured by stacking a plurality of optical elements. The method (400) includes a characterization stage (402) comprising the following steps: - determining (404) an assembly (JA) that includes at least one characteristic parameter of the optical element for each element; and - providing (406) an estimated geometry set (JGE) as a function of the assembly set by a geometry characterization model, the estimated geometry set containing data related to at least one geometric parameter of at least one optical interface of the stack, the geometry characterization model being pre-trained with a training set called a training database, the training set being based on optical objectives with the same architecture as the target objective. This invention also relates to an apparatus for geometrically characterizing optical objectives by implementing such a geometric characterization method. This invention also relates to a method and system for manufacturing optical objectives by implementing such a characterization method or apparatus.

Description

Technical Field

[0001] This invention relates to a method for geometric characterization of an optical objective, particularly before or during the manufacture of the optical objective, in order to estimate at least one indicator related to the quality of the optical objective (once assembled). This invention also relates to an apparatus for carrying out such a method for geometric characterization. Furthermore, this invention relates to a method and system for manufacturing optical objectives that carries out such a geometric characterization method or apparatus.

[0002] The technical field of this invention is the qualitative characterization of optical objectives, particularly before or during their manufacture. Prior Technology

[0003] An optical objective is constructed by stacking optical elements (in a cylindrical form, in a given order, and according to the stacking direction, also referred to as the Z-axis). The optical elements of an optical objective (hereinafter also referred to as an objective) may be optical lenses, which are converging, diverging, aspherical, or other complex shapes as appropriate, or sleeves, also referred to as "spacers" or "SOMA®", or beam splitters or filters.

[0004] Optical lenses are used in various electronic devices, such as cameras for video and still photography, smartphones, etc., to image scenes, or as light sources to project patterns or illuminate scenes.

[0005] Once assembled, the optical objectives are tested to determine performance data related to their function, primarily verifying whether the objective meets quality standards. Currently, various techniques exist for testing optical objectives after manufacturing, such as measuring MTF ("modulation transfer function") values.

[0006] One limitation of current characterization techniques is that optical objectives must be manufactured before they can be characterized. Therefore, when an optical objective is deemed unsatisfactory, it is typically scrapped, resulting in a loss of time and resources.

[0007] Therefore, current technology cannot improve the production efficiency of optical objectives because such technology is involved in the very late stages of optical objective manufacturing.

[0008] The purpose of this invention is to overcome at least one of the disadvantages mentioned above.

[0009] Another objective of this invention is to provide a less time-consuming and more efficient solution for characterizing optical objectives.

[0010] Another object of the present invention is to provide a solution for characterizing optical objectives, which makes it possible to improve the development of methods for manufacturing optical objectives and their components.

[0011] Another object of the present invention is to provide a solution for characterizing optical objectives, thereby making it possible to improve the efficiency of optical objective production. Summary of the Invention

[0012] This invention proposes a method for geometrically characterizing a target optical objective lens to be manufactured by stacking a plurality of optical elements to achieve at least one of the aforementioned objectives. The method includes a characterization stage comprising the following steps: - Determine a set referred to as an assembly set, which for at least one optical element includes a set referred to as an individual set, which contains at least one parametric characteristic of the optical element; and - By means of a geometric characterization model, a data set, referred to as the estimated geometric set, is provided as a function of the assembly set. The estimated geometric set contains data related to at least one geometric parameter of at least one optical interface of the stack. The geometric characterization model is pre-trained with a training set, referred to as the training database, which is composed of optical objectives with the same architecture as the target objective.

[0013] Therefore, the method according to the invention makes it possible to estimate at least one geometric parameter related to at least one, and particularly each, optical interface of the target optical objective, which is obtained based on at least one characteristic parameter of at least one (particularly each) optical element constituting the target optical objective. In other words, by means of the method according to the invention, it is possible to obtain an estimated index related to the optical objective before manufacturing it. This estimated index provided by the method according to the invention can be used, for example, by comparing it with a threshold value or a range of values ​​to estimate the quality or functional performance of the target objective (once manufactured), in order to determine whether the target optical objective (once manufactured) will perform well. Therefore, even before the manufacture of the target objective begins, or during its manufacture, it is possible to obtain an opinion on the performance of the target objective. The method according to the invention makes it possible to increase the manufacturing efficiency of optical objectives, which is defined as the number of satisfactory optical objectives divided by the total number of optical objectives manufactured. The method according to the invention also makes it possible to reduce the time and cost losses associated with producing optical objectives whose performance will not be satisfactory.

[0014] In addition, compared with current characterization techniques (such as characterization techniques by measuring MTF values), the characterization of optical objectives based on the parameters of the optical elements constituting the optical objective is faster and less time-consuming.

[0015] First, this invention makes it possible to provide an indication of the quality of an optical objective based on the optical elements that constitute it. Therefore, during the design phase of an optical objective, its composition can be adjusted, for example, by replacing one optical element with another or by modifying the configuration of the optical system. In fact, the characteristics of each optical component of the objective can be known before manufacturing begins, or even during the design phase of the optical objective, or even during the design phase of each optical element. Therefore, this invention makes it possible to obtain an indication of the quality of the optical objective very early in the manufacturing process.

[0016] At least one data item of the estimated geometric set, and in particular the estimated geometric set, can be compared with at least one predetermined threshold or range of values ​​to obtain an indication of the functional quality or functional performance of the target optical objective.

[0017] This indication may be a classification of the target optical objective, such as "good" or "bad". Alternatively or additionally, this indication may be any other type of information.

[0018] The optical elements that make up an optical objective are stacked according to a stacking direction (hereinafter also referred to as the Z-axis, or the axis of the optical objective). The plane perpendicular to the Z-axis (that is, the plane along which the optical elements extend) is hereinafter referred to as the XY plane.

[0019] In this application, the geometric characterization model (hereinafter also referred to as the characterization model) provides information related to the geometry of the optical interface of the objective lens in this application (also referred to as the "geometric parameters" of the optical interface).

[0020] The term "geometric parameters of an optical interface" refers to, for example, without loss of generality: - The position of the optical interface on the Z-axis inside the objective lens; - The position of the vertex of the optical interface, especially in the XY plane and / or - The position of the vertices of the optical interface, especially according to the Z-axis, - The tilt angle (skew and / or tilt) of the optical interface relative to the Z-axis - In the XY plane, the optical interface or optical element is eccentric relative to the Z-axis.

[0021] When each of these optical objectives is designed to contain the same optical elements (which are stacked in the same manner as designed), the two optical objectives have the same architecture.

[0022] In this application, the term "embedded optical interface" of an optical objective refers to an interface that is only visible within the optical objective, or that can be accessed through at least one other optical interface of the objective. The at least one other optical interface that is visible to the embedded interface may be an optical interface of the same optical element, or an optical interface of an optical element other than the embedded interface.

[0023] The estimated geometry set may contain information related to at least one geometric parameter of at least one embedded interface of the stack.

[0024] The estimated geometry set may contain information related to at least one geometric parameter of at least one non-embedded interface of the stack.

[0025] According to a specific example, at least one individual set of optical elements (hereinafter referred to as JI) may include any combination of at least one of the following parameters: - At least one optical parameter of the optical element; - At least one geometric parameter of the optical element; - At least one manufacturing parameter of the optical element.

[0026] For example, at least one optical parameter of an optical element may be one of the following optical parameters: - Refractive index, - An Abbe number - …etc.

[0027] For example, at least one geometric parameter of an optical element may be one of the following geometric parameters: - The geometry of one (especially each) optical interface of an optical element. - The center of curvature of one (especially each) optical interface of an optical element. - The position of the vertex of one (especially each) optical interface of an optical element. - The optical element has at least one thickness along its circumference. - The inner diameter and / or outer diameter of the optical element, - The concentricity or eccentricity of the optical interface of an optical element. - The surface roughness of at least one (particularly each) interface of an optical element. - …etc.

[0028] For example, at least one of the following manufacturing parameters may be a manufacturing parameter of an optical element: - Molding temperature, - One-molding pressure, - One-time molding time, - Used for molding one of the polymers, - …etc.

[0029] At least one parameter of the optical element may be provided by the supplier or manufacturer.

[0030] These parameters can be incorporated into the optical element specifications during the design phase of the optical element, or measured during or after the manufacturing process of the optical element.

[0031] Such parameters can be any of the parameters listed above.

[0032] Alternatively or additionally, at least one parameter of an optical element may be calculated based on a digital model of the element.

[0033] In fact, there are various software tools available for simulation and design, such as "OSLO®" or "Zemax®", which make it possible to digitally model optical elements so as to determine at least one characteristic of the optical element by simulating the digital modeling of the optical element.

[0034] Such parameters can be any of the parameters listed above.

[0035] Alternatively or otherwise, at least one parameter of an optical element can be measured by a measuring device.

[0036] For example, at least one geometric parameter can be obtained by optical profilometry or mechanical profilometry. In fact, such instruments make it possible to detect the shape of optical elements, the thickness of optical elements, the position of vertices, roughness, etc.

[0037] For example, at least one geometric parameter can be obtained by an optical interferometer in point mode or full-field mode.

[0038] Such parameters can be any of the parameters listed above.

[0039] Depending on the specific instance, the estimated geometry set may include some or all of the estimated data of the original optical measurements obtained from the stack of optical elements of the target optical objective.

[0040] For example, the geometry set may include estimated confocal measurements or estimated optical interferometry measurements performed on the optical element stack. In other words, the estimated geometry set may include estimated optical measurements without performing such optical measurements. Therefore, the method according to the invention makes it possible to first facilitate and accelerate the development phase of the method for manufacturing optical objectives, and secondly improve the monitoring and implementation of production in the second case.

[0041] The geometry set may specifically include estimated optical measurements generated only from one side or one side of the optical element stack, without flipping the stack. Therefore, the estimated geometry set contains estimated data related to at least one optical objective, and particularly to each optical interface (including each embedded optical interface of the optical element stack).

[0042] Based on a specific example, the estimated geometry set may include estimated data of at least one confocal measurement performed on the optical element stack of the target optical objective (preferably from one side of the stack).

[0043] Conventionally, confocal measurement is performed using an apparatus comprising a first opening (or aperture) imaged onto the surface to be measured via a focusing lens. This opening is illuminated by a light beam originating from a light source and then guided toward the surface to be measured. As the beam is reflected by the surface, it is redirected toward the focusing lens and then toward a second lens positioned in front of a detection element, thus forming a conjugate image of the illuminated point on the measured surface. The advantage of this configuration is reduced depth of field, making it easier to distinguish objects (or surfaces) from each other. To perform detection, the confocal measurement system is displaced relative to the object being measured. Maximum intensity is detected on the detection element when the surface enters the focal point defined by the focusing lens. A particular configuration of the confocal measurement system uses chromatic lenses to focus and image a beam of light from a multi-color light source. Therefore, different wavelengths define different focal points along the optical axis of the lens. Detection using a spectrometer of the reflected light makes it possible to identify the reflected wavelength and infer information about the height (or distance) between the lens and the measured surface. This configuration makes it possible to eliminate or reduce the displacement of the confocal measurement system. Specifically, when the confocal measurement system is displaced along a plane perpendicular to the illumination axis of the surface to be measured, surface topography information can be obtained.

[0044] Based on a specific example, the estimated geometry set may contain estimated data of at least one optical interferometry performed on the optical element stack of the target optical objective (preferably from one side of the stack).

[0045] Conventionally, optical interferometric measurements are performed using an optical interferometer that emits a low-coherence light source. This light source emits a beam called the measurement beam along the stacked optical elements, and more particularly along the Z-axis. The measurement beam illuminates the stacked optical elements by focusing in the XY plane at a more or less wider measurement point, and then travels through the stacked optical elements, particularly along the stacked direction, and sequentially through each optical interface. At each optical interface, a portion of the beam is reflected, forming a reflected beam. This reflected beam is then captured by a sensor located on the same side as the emitting element, and is characterized by optical interference with a reference beam that also originates from the light source. The term "coherence region" refers to the area on the sensor where interference between the measurement beam and the reference beam can form. The coherence region can be shifted by changing the difference in the length of the optical paths between the two beams (e.g., by modifying the optical length of one or both beams). Optical interferometers enable the selective detection of interference signals at each interface (i.e., each surface located within the coherence region) at the level where the coherence region is located. Preferably, the coherence length of the light source is adjusted to be shorter than the minimum optical distance between two adjacent interfaces of the optical element. Therefore, for each measurement, a single interface is located within the coherence region, and thus, the acquired interference signal contains only the effect of that single interface, or originates only from that single interface. Interferometric measurements are performed based on the field of view determined by the measuring components of the interferometer.

[0046] Depending on the specific example, the interferometer can operate in point mode by being configured to detect point interference signals at a point in the field of view or in a point detector. The estimated geometry set may be or may contain the interference signal or interferogram, which is a function of the intensity signal and the displacement of the coherent region along the z-axis. For example, the interference signal can be considered as a series of interferometric rays associated with various optical interfaces.

[0047] Alternatively or additionally, the interferometer may include an interferometer (referred to as a full-field interferometer) configured to detect full-field interference signals in the field of view and, for example, represented by a detection element, as a 2D image (interferometric image).

[0048] Therefore, the interface to be measured can be imaged based on the field of view in a single measurement or by means of a scanning beam.

[0049] In a specific implementation example, the measurement signal can be formed by point interference signals associated with the pixels of the detection element, the intensity of which is detected based on the displacement along the Z-axis of the coherence region.

[0050] According to one example, the interferometer may include an interferometer sensor with a Michelson interferometer. According to another example, the interferometer may include an interferometer sensor with a Mach-Zehnder interferometer.

[0051] Depending on the specific example, point-mode interferometers and full-field interferometers can be combined.

[0052] Depending on the specific instance, the estimated geometry set may contain part or all of the estimated data of the original optical interference values ​​obtained from the stack of optical elements of the target optical objective.

[0053] In other words, under this condition, the geometric characterization model obtains the set of components at the input and provides an estimated geometric set at the output, which includes the estimated original optical interferometry data.

[0054] The estimated raw optical interferometry data may include the following for each optical interface: - [In point mode]: At least one interference ray at the measurement point, the position and amplitude of which depend on the geometric distance of the optical interface relative to a reference point / plane. Specifically, the reference point / plane is a known reference interface that emits the measurement wave or forms part of the measurement device. Therefore, the position of the interference ray makes it possible to determine the distance of the associated optical interface at the measurement point. - [In full-field mode]: Acquiring a series of 2D interference signals for multiple optical path differences makes it possible to obtain shape information of the optical interface. This sequence can be acquired in different ways depending on the analytical technique employed. Multiple 2D interference signals can be acquired specifically using phase-shifting interferometry or vertical scanning interferometry. According to another specific example, which is by no means limiting, the interference signals can be processed using digital holographic computing methods. Typically, the 2D interference signals include amplitude information and phase information. Images associated with these amplitude information and images associated with these phase information can be constructed based on the interference signals.

[0055] For example, the estimated geometry set may include an estimate of the measured interference signal.

[0056] For example, the estimated geometry set may contain estimated raw data about at least one interference ray, which represents the position and, if applicable, the amplitude of the interference ray.

[0057] According to another example, the estimated geometry set may contain estimated raw data that represents amplitude and / or phase images associated with the interferometric image.

[0058] The following text provides examples of raw data for point mode interferometry with reference to Figure 3b, and examples of raw data for full-field mode interferometry with reference to Figures 3c to 3f.

[0059] Depending on the specific instance, the estimated geometry set may contain an estimate of at least one geometric parameter of the optical interface of the target objective lens.

[0060] For example, the estimated geometry set may include, for at least one optical interface or optical element in a stack: - At least one estimated position value of the optical interface or element of the objective lens; - At least one estimated thickness value for the optical elements of the objective lens; - In the XY plane, at least one estimated eccentricity value of at least one optical interface or element relative to the Z-axis, or relative to the center position of another interface; or - At least one estimated tilt angle value for at least one optical interface or element relative to the Z-axis, or relative to another interface; - At least one interface, for example, at least one estimated topography or shape profile value relative to a reference plane or axis, from which, for example, eccentricity or tilt angle values ​​are inferred.

[0061] The position of the Z-axis of the optical interface can be used to determine the position of the interference ray corresponding to that interface.

[0062] The thickness of the optical element along the Z-axis can be determined by calculating the distance between the interference rays corresponding to each element in the optical interface.

[0063] The position of an optical interface relative to the Z-axis can be determined by performing several optical interferometric measurements in the central region of a special optical objective. By observing the position of the Z-axis along the interference rays associated with the interface during these measurements, the position of the vertex of the optical interface can be determined. The position of the vertex of the optical interface makes it possible to determine the position of the interface relative to the Z-axis in the XY plane, and thus determine its eccentricity relative to the Z-axis.

[0064] In another instance, the position of the interface relative to the Z-axis can be obtained, for example, by detecting the interference image of the interface in the central region of the optical objective and analyzing the image and / or analyzing the associated image of amplitude or phase, particularly to obtain the contour of the surface and the position of the apex of the optical interface.

[0065] The position of an optical element relative to the Z-axis can be determined as a function of the position of its optical interface.

[0066] The tilt angle of an optical interface relative to the Z-axis can be determined by performing several optical interferometric measurements, specifically in the peripheral region of the optical objective. By following the position of the interference ray associated with the interface along the Z-axis during these measurements, the position of the interface at its edge level can be determined according to the axis, thus making it possible to determine the tilt angle of the interface relative to the Z-axis.

[0067] The tilt angle of an optical element relative to the Z-axis can be determined as a function of the tilt angle of its optical interface.

[0068] In addition to or replacing the position of the interference rays, it is also possible to determine each of these geometric parameters by using the amplitude of the interference rays.

[0069] Depending on the specific instance, at least one training set may contain: - At least one set of assemblies, referred to as the training set of assemblies, is obtained from an optical objective, referred to as the training optical objective, which has the same architecture as the target objective. - At least one geometry set, referred to as the training geometry set, is obtained from the training optical objective.

[0070] Of course, each training set of components (each training geometry set) contains data with the same properties as the reference sets of components (each estimation geometry set). Therefore, all the properties described above for the reference sets of components (each reference estimation geometry set) apply to the training sets of components (each applicable to the training geometry set).

[0071] Depending on the specific instance, for at least one training set, the training set components can be obtained partially or entirely through simulation.

[0072] For example, during the design phase of an optical element that forms part of a stack of optical objectives, its architecture can be modeled by representing the optical interface (especially the optical interface of a lens) using analytical formulas and by digitally indicating its spacing. Theoretical values ​​for the refractive index and Abbe number of the materials involved can also be given. These theoretical values ​​can then be input into commercially available optical design software, such as those mentioned above, to simulate and define the characteristic parameters of the optical element.

[0073] Depending on the specific example, for at least one training set, the training set components can be obtained partially or entirely through measurement.

[0074] In this situation, at least one parameter of the training assembly set is measured, for example in a manner similar to that described in the reference assembly set above.

[0075] According to a particularly advantageous specific example, during the manufacture of objectives in the same batch as the target objective, at least one training set can be obtained by forming a portion of the batch of objectives based on training objectives.

[0076] In other words, in this scenario, the training database is obtained partially or entirely from optical objectives, which are part of the same batch as the target optical objectives and are pre-manufactured. Therefore, the characterization model is more accurate, and it makes it possible to perform more precise geometric characterization.

[0077] The term "objectives from the same batch" refers to objective lenses that originate from the same conceptual architecture (identical design) to achieve similar optical performance. Furthermore, these objective lenses may also share common manufacturing characteristics, such as originating from the same production line, being produced using the same machines, and having similar production cycles.

[0078] In this situation, a first portion of optical objectives manufactured from the same batch is used to form a training database. Specifically, for each optical objective from this first portion of the batch, a training set is constructed, comprising: - Training assembly set: the latter contains at least one parameter of at least one optical element, measured or provided by a supplier or obtained through simulation; and - At least one training geometry set: the latter contains at least one data item related to at least one geometric parameter of at least one (particularly each) optical interface of the stack, the at least one geometric parameter being measured, for example by confocal measurement or by interferometry, or obtained by simulation based on the digital modeling of the optical objective in question.

[0079] Therefore, the first manufactured optical objectives from a batch make it possible to form a training database. This database is used to train a geometry characterization model. Once trained, the geometry characterization model is used to characterize subsequent optical objectives from that batch.

[0080] Depending on the specific instance, for at least one training set, the training geometry set can be obtained partially or entirely through measurement.

[0081] In this case, the training geometry set can be measured by at least one confocal measurement and / or at least one interferometric measurement (as described above) to provide raw data or values ​​related to at least one geometric parameter of at least one optical interface of the optical stack.

[0082] Depending on the specific instance, for at least one training set, the training geometry set can be obtained partially or entirely through simulation.

[0083] For example, during the design phase of an optical objective, its architecture can be modeled by representing the optical interface (especially the lens's optical interface) using analytical formulas and digitally indicating its spacing. Theoretical values ​​for the refractive index and Abbe number of the materials involved can also be provided. These theoretical values ​​can then be input into commercially available optical design software to simulate and optimize the objective's parameters by calculating theoretical performance. Therefore, optical transfer properties can be calculated quite accurately by simulating optical ray propagation at different points in the scene to be viewed and at different associated points on the detection area.

[0084] Therefore, the training geometry set can be obtained through simulation. Thus, the training database can be constructed, partially or entirely, through simulation, which is faster and requires less effort and resources.

[0085] Depending on the specific example, the geometric characterization model may include: - Neural networks, particularly regression neural networks, and even more particularly deep learning CNN ("convolutional neural network") neural networks. Such neural networks can contain no limit to at least about ten layers, for example, about twenty layers; - A polynomial linear regression model; - A Gaussian equation, obtained by a least squares method; or - A statistical analysis method.

[0086] According to a specific example, the method according to the present invention may include training a geometric characterization model using a training database.

[0087] The training phase can be implemented using a training database containing numerous training sets denoted as JE1 to JEk. Each training set JEi contains: - At least one training assembly set JAA i, which is obtained by measurement or simulation based on a training optical objective (once trained), the architecture of which is the same as that of the target optical objective, the target optical objective being characterized by a geometrically characterized model; - At least one training geometry set JGA i, which is obtained by measurement or simulation based on the training optical objective, the architecture of which is the same as that of the target optical objective, which will be characterized by a geometry characterization model.

[0088] When the feature model is a neural network, especially a CNN, the training phase may include repeated training steps multiple times.

[0089] The training steps may include a testing step. This testing step includes the step of providing a training set (e.g., JAA 1) at the input of the neural network during its operation. The neural network provides an estimated training geometry set (denoted as JGA 1 e) at its output.

[0090] During the steps following the testing step, the error E1 between set JGA ie and the training geometry set JGA 1 of the training set JE 1 can be calculated. For example, the calculated error E1 can be the Euclidean distance or cosine distance between set JGA 1e and set JGA 1.

[0091] The testing steps can be repeated for each training set JE1 to JEk to obtain k error values ​​E1 to Ek respectively associated with each training set JE1 to JEk.

[0092] Then, the training steps may include, for example, calculating the total error (denoted as EG) of all training sets JE1 to JEk by summing the k errors JE1 to JEk obtained.

[0093] Then, the training steps may include a feedback step, during which the coefficients or weights of the CNN neural network can be updated, for example, through an error gradient backpropagation algorithm.

[0094] The training steps can be repeated several times until the total error EG no longer changes during several (e.g., 5) consecutive iterations of the training steps. When this is the case, the CNN neural network can be considered fully trained and the training phase can be terminated.

[0095] Alternatively, or in addition to what has just been described, the neural network may be trained using the first part of the training dataset (e.g., JE1 to JEj) and validated using the second part of the training dataset (e.g., JEj+1 to JEk). If the output of the obtained neural network is sufficiently close to the expected value, the learning can be considered acceptable. Otherwise, other training sets may be presented, or the network topology may be modified (number of layers, number of neurons per layer, etc.) until satisfactory learning is achieved.

[0096] Of course, functional characterization models are not limited to neural networks.

[0097] According to the alternative, the functional characterization model may include or may be a correlation search method between the training set JAA i and the training geometry set JGA i of each training set JE i, for example by regression methods.

[0098] Depending on the specific instance, the relevance search can be performed using the least squares method. This involves establishing a hypothetical polynomial relationship between each JE_i and sets JAA_i and JGA_i. The least squares method then makes it possible to find an optimal set of polynomial coefficients that minimizes the error between the output computed from the obtained polynomial and JGA_is.

[0099] According to another aspect of the present invention, an apparatus for geometrically characterizing a target optical objective lens to be manufactured by stacking multiple optical elements is provided, the apparatus comprising: - At least one component for determining a set referred to as an assembly set, the assembly set including, for at least one optical element, a set referred to as an individual set, the individual set including at least one characteristic parameter of the optical element; and - a geometry characterization model which is pre-trained with a database known as a training database of a training set composed of resorting to an optical objective configuration identical to the architecture of that target objective, in order to provide a data set known as an estimated geometry set as a function of the set of components containing data related to at least one geometric parameter of at least one optical interface of the stack.

[0100] The characterization device visual situation contains any combination of at least one of the characteristics described above with reference to the geometrical characterization method according to the present invention, which is not repeated in detail here for the sake of brevity.

[0101] Specifically, the characterization model can be integrated in computer processing modules such as processors, chips, computers, tablets, servers, etc., whether dedicated or not.

[0102] According to another state of the invention, a method for fabricating a batch of optical objective lenses is presented, the method comprising a second manufacturing phase comprising at least one repetition of the step of manufacturing the optical objective lens of the batch, the step comprising the following operations: - Based on individual sets of each of the optical elements for the optical objective, the set of combinations is judged for the optical objective, and - Characterizing the objective lens by resorting to a characterization method according to the present invention.

[0103] The obtained at least one and especially the values ​​of each estimated geometry set may be compared with at least one geometric value or a series of geometric values ​​in order to determine whether the estimated quality of the optical objective is satisfactory.

[0104] If the estimated quality of the optical objective is satisfactory, the optical objective can be fabricated.

[0105] If the estimated quality of the optical objective is unsatisfactory, the optical objective may withstand at least one other test or not be manufactured.

[0106] Alternatively or in addition, if the estimated quality of the optical objective is not satisfactory, the optical objective may be reworked to improve its estimated quality. For example, at least one optical element of the optical objective lens may be repositioned (e.g., pivoting according to the axis of rotation), or replaced with another optical element.

[0107] Therefore, even before producing the optical assembly, or even before starting the manufacture of the optical objective, it may be possible to verify or not verify the assembly of optical elements forming the objective. This makes it possible to improve the efficiency of manufacturing method development based on inferred information about the optical components, and to improve the efficiency of producing optical objectives using such methods based on well-obtained predictive data before the objective is fully completed. Thus, among other things, it may also make it possible to reduce losses in both the assemblies and the optical components of the objective that could point to a favorable outcome.

[0108] Changes in at least one data item of the geometry set can be monitored over time, or estimated quality can be inferred based on the estimated geometry set. This change can be used to monitor the manufacturing of the same batch of objectives. For example, when a change indicates a significant deviation or reduction in the functional quality of the batch of optical objectives, an alert can be generated, allowing for readjustment of the manufacturing process.

[0109] Advantageously, the manufacturing method according to the invention may include a first manufacturing stage, prior to a second manufacturing stage, comprising at least one repetition of the step of manufacturing the batch of optical objectives, the step comprising the following operations: - Based on the individual set of optical elements of the optical objective, a set of components is determined for the optical objective, and - The optical elements are stacked to form the optical objective. - Measurements of one of the geometric sets of this optical objective. - The training set is stored in the training database, and the training set is formed by the following: ■ This assembly set, and ■ This measurement geometry set.

[0110] This first manufacturing stage makes it possible to build a training library for training geometrically characterized models used during the second manufacturing stage. Simple Explanation of the Diagram

[0111] Other advantages and features will become apparent when examined in conjunction with the accompanying drawings, which are not limiting examples: - [Figure 1] is a schematic representation of a non-limiting specific example of an optical element that can be used to manufacture an optical objective lens; - [Figure 2] is a diagrammatic representation of a non-limiting specific example of an optical objective lens that can be characterized by the present invention; - [Figures 3a] to [Figures 3f] are schematic representations of non-limiting specific examples of optical interferometry that can be implemented in this invention; - [Figure 4] is a schematic diagram illustrating a non-limiting specific example of the method for geometric characterization of optical objectives according to the present invention. - [Figure 5] is a schematic diagram of a non-limiting specific example of a device for geometric characterization of an optical objective according to the present invention. - [Figure 6] is a diagrammatic representation of a non-limiting specific instance that can be used in the training phase of training a geometric characterization model in this invention; and - [Figure 7] is a schematic representation of a non-limiting specific example of a method for manufacturing an optical objective according to the present invention.

[0112] It is fully understood that the following description will be in no way a limiting example. Variations of the invention are conceivable, particularly those containing only selections of the features described below, isolated from the other features described, if such selection is sufficient to provide a technical advantage or to distinguish the invention from the state of the prior art. Such selection includes at least one (preferred functional) feature without structural details, or only such part if only a portion of the structural details is sufficient to provide a technical advantage or to distinguish the invention from the state of the prior art.

[0113] Specifically, if there is no technical disagreement on the combination of all variations and all specific instances described, then they can be combined together.

[0114] In the figures and in the description below, elements common to several figures retain the same reference numerals. Implementation

[0115] Figure 1 is a schematic representation of a non-limiting specific example of an optical element that can be used to manufacture an optical objective lens.

[0116] The optical element 100 of Figure 1 can be used with at least one other optical element to manufacture an optical objective. An example of an optical objective will be described by way of a non-limiting example with reference to Figure 2.

[0117] Optical element 100 may be a lens, beam splitter, etc. In the following text, and without loss of versatility, optical element is considered to be a lens.

[0118] For example, the optical lens 100 can be manufactured by injection molding. Injection molding generally follows these sequential steps: - Inject polymer and fill the mold, - Pressurization, - Maintain pressure, - Cooling, and - Demolding.

[0119] While common, lens manufacturing methods using injection can be volatile and introduce errors in the lens's characteristic parameters.

[0120] Lens 100 can be characterized by a set of parameters called individual set JI.

[0121] An individual set JI may contain at least one parameter related to the process of manufacturing the lens, referred to as a manufacturing parameter. An individual set may contain any combination of at least one of the following manufacturing parameters: - Molding temperature, denoted as TM; - Molding pressure, denoted as PM; - Molding duration, denoted as DM; - Polymers used for molding are denoted as POM.

[0122] The values ​​of these parameters are typically determined by the lens manufacturer. These values ​​may be referred to as input data items in the manufacturing process, or measured during the lens manufacturing process.

[0123] Furthermore, lens 100 has a given geometry. It includes two interfaces 102 1 and 102 2, each interface also having a given geometry itself. Therefore, lens 100 has a shape characterized by at least one geometric parameter associated with the shape of the lens. Alternatively or additionally, individual set JI may contain any combination of at least one of the following geometric parameters: - The geometry of each of the optical interfaces 102 1 and 102 2; - The centers of curvature of each of the optical interfaces 102 1 and 102 2 are denoted as CC1 and CC2, respectively; - The positions of the vertices (denoted as A1 and A2) of each of the optical interfaces 102 1 and 102 2; - The thickness of lens 100 along its periphery is denoted as H1 and H2; - The inner diameters (D11 and D21, respectively) and / or outer diameters (D12 and D22, respectively) of each of interfaces 102 1 and 102 2; - Concentricity or eccentricity values ​​of interfaces 102.1 and 102.2 - Surface roughness of each of optical interfaces 102 1 and 102 2; - …etc.

[0124] The value of at least one geometric parameter may be provided by the manufacturer. Alternatively or additionally, the value of at least one geometric parameter may be measured, for example, by optical or mechanical profilometry. Alternatively or additionally, the value of at least one geometric parameter may be determined by simulating a digital model based on lens 100. Alternatively or additionally, the value of at least one geometric parameter may be measured, for example, by optical interferometry.

[0125] Furthermore, lens 100 possesses optical characteristics because it is an optical element. Therefore, it is characterized by at least one optical parameter. Thus, alternatively or additionally, individual sets JI may include at least one of the following optical parameters: - The refractive indices of each of the optical interfaces 102 1 and 102 2 are denoted as I1 and I2, respectively; - Abbe number, denoted as Ab. - …etc.

[0126] The value of at least one optical parameter may be provided by the manufacturer. Alternatively or additionally, the value of at least one optical parameter may be determined, for example, by optical measurement.

[0127] Therefore, based on the non-restrictive concrete instance, the individual set JI of the lens can be written as: JI={TM,PM,DM,PO; CC1,CC2,A1,A2,H1,H2,D11,D12,D21,D22; I1,I2,Ab} Generally, an individual set may contain an M1 parameter, where M1≥1, and preferably M1≥2.

[0128] This individual set of lenses can be used, wherein an individual set of at least one other optical element of an optical objective forms a set called an assembly set (denoted as JA). If an optical objective contains N optical elements, the assembly set JA may contain N×M1 parameters and may correspond to a matrix comprising N columns and M1 rows.

[0129] Of course, at least one set of two optical elements may contain the same number of parameters or different numbers of parameters.

[0130] Figure 2 is a diagrammatic representation of a non-limiting specific example of an optical objective lens that can be characterized by the present invention.

[0131] Optical objectives are used to focus an image of a scene onto an image plane, and are typically constructed using a CMOS camera (also known as a "CMOS imager system," or CIS). These optical objectives are generally composed of stacked optical elements, including any combination of optical elements such as lenses, spacers, and opaque rings.

[0132] During the manufacture of an optical objective, the individual optical elements of the objective are individually selected and stacked with other optical elements in an assembly sleeve in a given order. The stack and sleeve are then securely fastened together using known techniques, such as bonding.

[0133] In Figure 2, and by way of a non-limiting example only, the optical objective 200 includes four lenses 202 to 208 stacked along the stacking direction 210 (also referred to as the Z-axis) in the tube 212. At least two of the lenses 202 to 208 may be separated from each other by an aperture space (referred to as an "air gap") or by a spacer sleeve or ring (also referred to as a spacer).

[0134] For example, at least one of lenses 202 to 208 may be lens 100 of Figure 1.

[0135] Each of lenses 202 to 208 includes two interfaces in the stacking direction 210, namely, an upstream interface and a downstream interface. Therefore, lens 202 has an upstream interface 214 1 and a downstream interface 214 2, lens 204 has an upstream interface 214 3 and a downstream interface 214 4, lens 206 has an upstream interface 214 5 and a downstream interface 214 6, and lens 208 has an upstream interface 214 7 and a downstream interface 214 8.

[0136] Therefore, for the optical objective 200 in Figure 2, the assembly set JA can contain individual sets for each of the optical elements 202 to 208, denoted as JI 1 to JI 4 respectively, such that JA = {JI ​​1;JI 2;JI 3;JI 4}

[0137] The assembly set JA can be determined even before the lenses of the stacked optical objectives 200 are assembled, provided that the optical elements constituting the optical objectives are known. In some specific instances, the assembly set can be determined at the design stage of each optical element constituting the optical objectives. Therefore, in order to optimize the quality of the optical objectives, the optical elements may be adjusted at the design stage or at the manufacturing stage.

[0138] Furthermore, for optical objective 200, and generally for any optical objective containing a stack of optical elements, it may be determined that a data set (referred to as a geometry set, hereinafter denoted as JG) contains data relating to at least one geometric parameter of the stack, and particularly to at least one geometric parameter of each optical interface 214 1 to 214 8.

[0139] Such a geometry set JG may contain data related to any of the following geometric parameters: - At least one position value of at least one interface 214 1 to 214 8, or at least one optical element 202 to 208 of objective lens 200; - At least one thickness value of the optical elements 202 to 208 of the objective lens 200; - In the XY plane, at least one interface 214 1 to 214 8, or at least one optical element 202 to 208, has at least one eccentricity relative to the Z-axis, or relative to the center position of another interface; or - At least one interface 214 1 to 214 8, or at least one optical element 202 to 208, having at least one tilt angle value relative to the Z-axis, or relative to another interface; - At least one topographic or shape profile value of at least one optical interface 214 1 to 214 8.

[0140] Generally, a geometry set may contain M² geometric parameters for each optical interface of an optical objective, where M² ≥ 1 and preferably M² ≥ 2. If an optical objective contains N optical elements, the assembly set JG may contain 2N × M² parameters and may correspond to a matrix comprising 2N columns and M² rows. Of course, the assembly set may contain the same number of geometric parameters for at least two optical interfaces, or a different number of geometric parameters for at least two optical interfaces.

[0141] The geometric set JG can directly contain the values ​​of geometric parameters. Conventionally, these values ​​can be measured from one side or one side of the optical objective 200 by optical interference or by confocal measurement, so as to avoid rotating it.

[0142] Alternatively, the geometry set JG may contain raw measurement data, such as, for example, optical interferometric measurements or confocal measurements.

[0143] Figure 3a is a diagrammatic representation of a non-limiting specific example of an optical interferometry measurement that can be implemented in this invention.

[0144] Optical interferometry is performed by an optical interferometer or interferometer 300, illustrated in height in Figure 3a. The instrument 300 includes a light source 302 and an interferometer sensor 304. The source 302 emits a coherent light beam 306 (referred to as the measurement beam) in a plane XY perpendicular to direction 210 at the measurement point or in the direction of the optical element stack according to the field of view 308. The measurement beam 306 then travels through the optical element stack, specifically along the Z-axis 210, and sequentially through each optical interface 214i. At each optical interface 214i, a portion 310i of the measurement beam 306 is reflected, such that: - Beam 310 1 is reflected by interface 214 1, - etc - Beam 310 8 is reflected by interface 214 8.

[0145] Each reflected beam 310i of the measurement beam 306 is then captured by a sensor 304, which is also optically connected to the emission source 302. When this reflected beam 310i and the reference beam 312 (also originating from the light source 302) are recombined on the sensor 304, an interference signal is generated, with the difference in the paths of the two corresponding beams being less than the coherence length 302 of the emission source. Specifically, for each reflected beam 310i, depending on the implemented illumination and detection mode, the sensor 304 provides an interference ray or interference image, referred to as the principal ray, at an optical distance corresponding to the position of the interface relative to the emission source 302 or any other predetermined reference. Of course, in addition to the beam 3101 reflected by the first interface 214 1 encountered by the measurement beam 306, a portion of each of the other reflected beams 310 2 to 310 8 may be reflected in another direction as they pass through the preceding interface, which can produce multiple reflected beams (not shown) captured by the sensor 304. These multiple reflected beams produce interference rays (called secondary rays), or secondary images, that typically have lower amplitudes.

[0146] Optical interferometry can be performed using a measurement beam from an interferometer illuminated by a low-coherence light source. For this purpose, the optical interferometer has a positioning member for relatively positioning the coherence region of the interferometer 304 at the interface level to be measured. The interface to be measured can be an embedded interface, i.e., one of the interfaces within an optical element. To reach such an embedded interface, the measurement beam must therefore pass through other interfaces of the optical objective. The interferometer makes it possible to selectively detect interference signals at each interface (i.e., each surface located within the coherence region) at the level where the coherence region is located, because the coherence length of the light source is adjusted to be shorter than the minimum optical distance between two adjacent optical interfaces of the optical objective. Therefore, preferably, for each measurement, a single interface is located within the coherence region.

[0147] Interference measurements can be performed based on the field of view determined by the measuring components of the instrument. Therefore, measurements can be performed across the entire field or by scanning the field of view.

[0148] Digital processing components can be configured to generate shape information items or geometric parameters of the interface measured by the field of view based on interference signals.

[0149] For example, examples of interference devices that can be used in the context of this invention are described in the document. As described in [WO2020 / 245511 A1].

[0150] Figure 3b provides a partial graphical representation of the raw measurement data obtained from optical interferometry (such as the optical interferometry described with reference to Figure 3a).

[0151] In this implementation example, measurement point illumination is used, and the coherent region is displaced along the optical axis Z 210 by means of a displacement member.

[0152] Therefore, as described with reference to Figure 3a, each interferometric measurement provides raw data 320. Raw data 320 includes principal rays 322i, each principal ray corresponding to an optical interface. For example, principal ray 322i is obtained for interface 214i, principal ray 322i is obtained for interface 214i, etc. (interface 214i is not shown in the example shown in Figure 3b).

[0153] The original data 320 also includes secondary rays corresponding to multiple reflections and associated with interfaces 214 2 to 214 8.

[0154] The optical position of each ray is given on the x-axis, and the normalized amplitude of each ray is given on the y-axis.

[0155] Figures 3c to 3f are partial illustrations of another example of raw measurement data obtained from optical interferometry (such as the optical interferometry described with reference to Figure 3a).

[0156] In the examples shown in Figures 3c to 3f, illumination according to the full-field pattern was used and the coherence region was located for measuring the surface of the embedded lens. Figure 3c shows the interference signal detected during digital holographic microscopy (DHM) measurement. Figures 3d and 3e show the amplitude and phase images calculated based on the interference signal (in this folded example), respectively. Figure 3f is an image of the surface topography of the embedded lens obtained based on phase information.

[0157] As indicated above, depending on the specific instance, the geometric set may contain part or all of the original measurement data, that is: - For example, in the lighting configuration based on the measurement points: ■ The position of each main ray, and the amplitude, depending on the situation, or ■ The position of each ray (primary ray or secondary ray), and, depending on the situation, its amplitude. - For example, in lighting configurations based on field of view: ■ Interference image detected by interferometer 204 ■ Amplitude images associated with interferometric images and / or phase images associated with interferometric images, or ■ Topographic images associated with interferometric images.

[0158] Depending on the specific example, the geometry set may contain geometric parameter values ​​related to the optical interface of the objective lens, rather than the original measurement data obtained from optical interferometry. - The position along the Z-axis of each interface 214i or each optical element 302 to 308. These distance values ​​can be obtained based on raw data from a single interferometric measurement. In this case, the geometry set contains the distance value for each interface (each optical element, respectively); - Eccentricity relative to the Z-axis or to other interfaces in the XY plane of each interface 214i or each optical element 202 to 208. For example, such eccentricity values ​​can be inferred based on the interface image associated with the optical interface. In this case, the geometry set contains two (signed) distance values ​​(one along the X-axis and the other along the Y-axis) for each interface of the optical element. - The tilt angle relative to the Z-axis, or the tilt angle of each optical interface 214i or other optical interfaces of each optical element 202 to 208. These tilt angle values ​​can be inferred based on several interferometric measurements performed at different measurement points (particularly in the peripheral region of the optical objective). In this case, the geometry set contains two angle values ​​(one relative to the X-axis and the other relative to the Y-axis) for each interface of the optical element.

[0159] Referring to Figure 3a, it should be understood that measuring the geometry of an optical objective can be time-consuming and requires the use of large optical interferometers. Furthermore, measuring the geometry of an optical objective necessitates a temporary halt to the manufacture of that objective during the time spent performing the measurement, which constitutes a slowdown or potential time loss in the production of optical objectives.

[0160] This invention proposes to estimate the set of geometric parameters for optical objectives (referred to as target optical objectives). To this end, the set of components, as described with reference to Figures 1 and 2, is given at the input of a pre-trained geometric characterization model, which provides an estimated set of geometric parameters (denoted as JGE) at the output.

[0161] Figure 4 is a schematic diagram illustrating a non-limiting specific example of the method for geometric characterization of optical objectives according to the present invention.

[0162] Method 400 of Figure 4 can be used for the geometric characterization of an optical objective containing several lenses (such as, for example, optical lens 200 of Figure 2, but not limited thereto).

[0163] Method 400 includes a stage 402 of characterizing an optical objective (referred to as a target objective) during its manufacture.

[0164] The characterization stage 402 includes step 404 of determining the assembly set based on the individual set JI of each optical element in the stack of optical elements constituting the target optical objective, as described above. The individual set JI of each optical element may include one or more parameters associated with that optical element.

[0165] For each optical element, at least one parameter can be determined as described above based on information provided by the manufacturer or by measurement or simulation.

[0166] This step 404 provides an assembly set JA for the optical element stack of the optical objective lens, the assembly set containing at least one of the stack and, in particular, at least one characteristic parameter of each optical element.

[0167] Characterization phase 402 includes step 404, during which the assembly set JA determined for the target optical objective is provided to a pre-trained geometric characterization model. In response, the geometric characterization model provides an estimated geometry set JGE for the target objective.

[0168] JGE may contain one or more values. Preferably, JGE contains several values.

[0169] The geometry set JGE may contain at least one stacked, and in particular, geometric parameter values ​​for each optical interface, such as, for example, examples of geometric parameters (alignment, eccentricity, etc.) described with reference to Figures 3a to 3f.

[0170] The estimated geometry set JGE may include estimated raw values ​​of optical interferometric measurements or confocal measurements of at least one stacked, and in particular each optical interface, such as those described, for example, with reference to Figures 3a to 3f (interference signals, interference rays, etc.).

[0171] Depending on the circumstances, method 400 may include a stage 420 of training a geometric characterization model using a training database containing several training sets, the training sets being obtained by measurement or simulation based on objectives with the same architecture as the target objective. An example of a non-restrictive training stage is described below with reference to Figure 6.

[0172] Therefore, method 400 allows for the geometric characterization of a target optical objective by estimation using a pre-trained geometric characterization model, without performing measurements on the stack of optical elements of the optical objective, or even before manufacturing the optical objective begins. Thus, there may be indicators related to the quality of the optical objective before manufacturing it and deciding whether to manufacture it. For example, it may be possible to avoid manufacturing optical objectives whose estimated quality is unsatisfactory.

[0173] In addition, method 400 allows for the implementation of a design of experiment (DOE) to analyze the parameters of components in order to classify them based on the obtained geometric characterization results, so as to, for example, reject certain components very early in manufacturing, match categories between different optical components, or apply inherent corrections (modification of spacing, lens rotation, etc.) to this category.

[0174] Figure 5 is a schematic diagram illustrating a non-limiting specific example of a device for geometric characterization of an optical objective according to the present invention.

[0175] The device 500 can be used to characterize at least one optical objective (such as, for example, optical objective 200 in FIG2), particularly before or during the manufacture of the optical objective.

[0176] The device 500 includes a module 502 for determining the assembly based on the characteristics of each optical element constituting the target optical objective.

[0177] Module 502 may include: - An optical or mechanical profile measuring instrument 504; and / or - An optical interferometer 506 operating in either point mode or full-field mode, such as, for example, the optical interferometer 300 in Figure 3a. Used for measuring at least one geometric parameter of at least one (particularly each) optical element provided to constitute an optical objective.

[0178] Alternatively or additionally, module 502 may include a computerization unit 508, such as a processor or calculator, configured to: - Retrieves characteristic parameter values ​​of each optical element provided to constitute the optical objective from the database; and / or - Calculate at least one characteristic of the optical element by simulation and based on the modeling of at least one optical element.

[0179] Module 502 may alternatively or additionally include a user interface 510 that allows an operator to manually input values ​​for at least one characteristic parameter provided to constitute at least one optical element of an optical objective.

[0180] Module 502 provides an assembly JA for the stack of optical elements provided to form an optical objective lens.

[0181] Device 500 also includes a characterization module 512 that executes a geometric characterization model 514. This geometric characterization model takes at input a set of components JA provided by module 502 and at output an estimated geometry set JGE. The geometric characterization model 514 can be a computerized program or application and is presented in the following form: - Neural networks, especially regressive neural networks, and even more specifically deep learning CNN neural networks, or - A polynomial linear regression model, - A Gaussian equation, obtained by the least squares method. - A statistical analysis method, - …etc.

[0182] The geometric characterization module 512 can be any computing module or any computerized module that executes the characterization model 514, such as a server, computer, tablet computer, processor, calculator, electronic chip, etc.

[0183] Figure 6 is a diagrammatic representation of a non-limiting specific instance that can be used in the training phase of training a geometrically characterized model in this invention.

[0184] The training phase 600 in Figure 6 can be used to train the method for geometric characterization of optical objectives according to the present invention, or the geometric characterization module used in, for example, method 400 in Figure 4, when the training module is a neural network.

[0185] The neural network used can be a CNN (for "convolutional neural network"). It is important to note that the number of layers in the neural network varies depending on the number of data items in the set of combinations provided at the input of the neural network and the number of data items in the set of geometry required at the output.

[0186] The training phase of the 600 series is performed using training database 602, which contains numerous training datasets, denoted as JE1 to JEk. Each training set JEi contains, for example, training optical objectives: - At least one training set of JAA i; and - Training geometry set, JGA i.

[0187] The training optical objective has the same architecture as the target optical objective that is expected to be characterized by a geometric characterization model.

[0188] Each training assembly set JAA i can be obtained, for example, as described above with reference to Figures 1 and 2, that is, by determining the individual sets of optical elements constituting the training optical objective, and then determining the training assembly set based on the individual sets.

[0189] Each training geometry set JGA i can be obtained, for example, by measuring the training optical objectives, such as by optical interference as described above with reference to Figures 3a to 3f, or by simulation based on a digital model of the training optical objectives.

[0190] Training phase 600 includes training step 604.

[0191] Training step 604 includes testing step 606.

[0192] Test step 606 includes step 608, during which a training set (e.g., JAA 1) is given at the input of the neural network (e.g., JE 1). The neural network gives an estimated training geometry set (denoted as JGA 1 e) at the output.

[0193] During step 610 of test step 606, the error E1 between set JGA ie and set JGA 1 can be calculated. For example, the calculated error E1 can be the Euclidean distance or cosine distance between set JGA 1e and set JGA 1.

[0194] Test step 606 can be repeated for each training set JE1 to JEk to obtain k error values ​​E1 to Ek respectively associated with each training set JE1 to JEk.

[0195] Then, training step 604 may include step 612, for example, calculating the total error (EG) of all training sets JE1 to JEk by summing the obtained k errors JE1 to JEk.

[0196] Then, training step 604 may include feedback step 614, during which the coefficients or weights of the CNN neural network may be updated, for example, by using an error gradient backpropagation algorithm.

[0197] The training step 604 can be repeated several times until the total error EG no longer changes during several (e.g., 5) consecutive repetitions. When this is the case, the CNN neural network can be considered fully trained and the training phase can be terminated.

[0198] Alternatively, or in addition to what has just been described, the neural network may be trained using the first part of training dataset 602 (e.g., JE1 to JEj) and validated using the second part of training dataset 602 (e.g., JEj+1 to JEk). If the output of the obtained neural network is sufficiently close to the expected value, the learning can be considered acceptable. Otherwise, other training sets may be presented, or the topology of the network may be modified (number of layers, number of neurons per layer, etc.) until satisfactory learning is obtained.

[0199] Of course, geometric characterization models are not limited to neural networks.

[0200] According to the alternative, the geometry characterization model may include or may be a correlation search method between the training set of components JAA i and the training geometry set JGA i of each training set JE i, for example by regression methods.

[0201] Depending on the specific instance, relevance search can be performed using the least squares method. This can be achieved by establishing a hypothetical polynomial relationship between each JE i and sets JAA i and JGA i. The least squares method then makes it possible to find an optimal set of polynomial coefficients that minimizes the error between the output computed from the obtained polynomial and JGA is.

[0202] Figure 7 is a schematic representation of a non-limiting specific example of a method for manufacturing an optical objective lens according to the present invention.

[0203] Method 700 may include a first stage 702 of manufacturing, during which a first portion of a batch of objectives is manufactured. This first portion includes a plurality of optical objectives. During this stage 702, the optical objectives are manufactured in step 704, and then a training set is determined and stored in step 706 to form a training database, such as, for example, training database 602.

[0204] Then, during step 708, the geometric characterization model is trained using a training database, for example by implementing training phase 600 in Figure 6.

[0205] Method 700 may then include a second stage 710 of manufacturing, during which the remaining objectives of the batch are manufactured.

[0206] This stage 710 includes step 712 for selecting the optical elements that will constitute the optical objectives for each optical objective.

[0207] During step 713 , the set of assemblies is judged for stacking of selected optical components.

[0208] During step 714, preferably implemented prior to stacking optical components, using the geometry characterization model obtained in step 708, the optical objective to be fabricated is characterized by means of the method of functional characterization according to the present invention, and in particular by the method 400 of FIG.

[0209] Then, during step 716, the obtained at least one and especially the respective estimated geometric set values ​​can be compared with at least one range of geometric values ​​in order to determine whether the expected quality of the optical objective is satisfactory.

[0210] If the estimated geometry set demonstrates satisfactory quality of the optical objective, the fabrication of the optical objective may be initiated or continued during step 718 .

[0211] If the estimated geometry set demonstrates unsatisfactory quality, the fabrication of the optical objective lens can be cancelled. Alternatively, if the estimated geometry set demonstrates unsatisfactory quality, the configuration of the optical objective can be reworked. For example, at least one optical element of a replaceable optical objective lens.

[0212] Of course, the invention is not limited to the instances just described.

[0213] 100: Optical components 102 1: Interface / Optical Interface 102 2: Interface / Optical Interface 200: Optical objective 202: Lens 204: Lens 206: Lens 208: Lens 210: Stack direction 212: tube 214 1: Upstream interface 214 2: Downstream interface 214 3: Upstream interface 214 4: Downstream interface 214 5: Upstream interface 214 6: Downstream interface 214 7: Upstream interface 214 8: Downstream Interface 300: Interference equipment 302: Light source / optical components 304: Interference sensor / optical element 306: Beam / Optical Components 308: Field of View / Optical Elements 310 1: Beam 310 8: Beam 312: Reference Beam 320: Original Data 322 1: Main Ray 322 2: Main Ray 322 3: Main Ray 322 4: Main Ray 322 5: Main Ray 322 6: Main Ray 322 7: Main Ray 400: Method 402: Characterization Phase 404: Steps 406: Steps 420: Stage 500: Device 502: Module 504: Optical / Mechanical Profilometry Instruments 506: Optical Interferometers 508: Computerized Unit 510: User Interface 512: Specialized Module 514: Geometric Characterization Model 600: Training Phase 602: Training Database 604: Training Steps 606: Test Procedure 608: Steps 610: Steps 612: Steps 614: Steps 700: Method 702: Phase One 704: Steps 706: Steps 708: Steps 710: Second Phase 712: Steps 713: Steps 714: Steps 716: Steps 718: Steps CC1: Center of Curvature CC2: Center of Curvature D11:Inner diameter D12: Outer diameter D21:Inner diameter D22: Outer diameter H1: Thickness H2: Thickness JA: Assembly Set JGE: Estimating Geometric Sets

Claims

1. A method (400) for geometric characterization of a target optical objective (200) to be manufactured by stacking a plurality of optical elements (202 to 208), the method (400) comprising a characterization stage (402) including the following steps: The determination (404) is referred to as a set of assembly sets (JA), which includes a set of individual sets for each of at least two optical elements (202 to 208) of the plurality of optical elements; and by means of a geometry characterization model (514), a data set (406) referred to as an estimated geometry set (JGE) is provided as a function of the assembly set (JA), which includes data related to at least one geometric parameter of the stacked at least two optical interfaces (2141 to 2148), the geometry characterization model being pre-trained with a training set (JE1-JEk) of a database (602) referred to as a training database, which is composed of optical objectives with the same architecture as the target objective.

2. The method (400) as claimed in claim 1, wherein at least one individual set of an optical element (202 to 208) includes any combination of at least one of the following parameters: at least one optical parameter of the optical element (202 to 208); at least one geometric parameter of the optical element (202 to 208); at least one manufacturing parameter of the optical element (202 to 208).

3. The method (400) as described in claim 1 or 2, wherein at least one parameter of an optical element (202 to 208) is: provided by one of the suppliers of the optical element (202 to 208), measured by a measuring device, or calculated based on a digital modeling of the optical element (202-208).

4. The method (400) as described in claim 1 or 2, wherein the estimated geometry set (JGE) contains estimates of at least one geometric parameter of one of the optical interfaces (2141 to 2148) of the target optical objective (200).

5. The method (400) as described in claim 1 or 2, wherein the estimated geometry set (JGE) contains a portion or all of the estimated data of the original optical measurements (320) obtained from the stacking of the optical elements (202 to 208) of the target optical objective (200).

6. The method (400) as described in claim 1 or 2, wherein at least one training set (JE1 to JEk) comprises: at least one assembly set, referred to as the training assembly set, obtained from one of the optical objectives, referred to as the training optical objective, which is configured to be the same as the architecture of the target optical objective (200), and at least one geometry set, referred to as the training geometry set, obtained from the training optical objective.

7. The method (400) as described in claim 1 or 2, wherein the training database (602) contains at least one training set (JE1 to JEk) obtained from a training objective during the manufacture of objectives in the same batch as the target objective, the training objective forming part of the batch of objectives.

8. The method (400) as described in claim 1 or 2, wherein the geometric characterization model (514) comprises: a neural network, particularly a regression neural network, and even more particularly a deep learning CNN neural network; a multinomial linear regression model; a Gaussian equation obtained by a least squares method; or a statistical analysis method.

9. The method (400) as described in claim 1 or 2, comprising a stage (420; 600) of training the geometric property model with the training database (602).

10. An apparatus (500) for geometry characterization of a target optical objective (200) to be manufactured by stacking a plurality of optical elements (202 to 208), the apparatus (500) comprising: at least one component (502) for determining a set called an assembly set (JA), the assembly set comprising, for each of at least two of the plurality of optical elements (202 to 208), a set called an individual set, the individual set comprising at least one physical parameter characteristic of the optical element (202 to 208); and a geometry characterization model (514) pre-trained with a database (602) of a training set (JE1 to JEk) of optical objectives with the same architecture as the target optical objective (200) to provide a dataset called an estimated geometry set (JGE) as a function of the assembly set, the estimated geometry set comprising data relating to at least one geometric parameter of at least two optical interfaces (2141 to 2148) of the stack.

11. A method (700) for manufacturing a batch of optical objectives, comprising a second manufacturing stage (710) comprising at least one repetition of a step of manufacturing one of the optical objectives of the batch, the step comprising: determining (713) an assembly set for the optical objective based on an individual set of optical elements of the optical objective, and characterizing (714; 400) the objective by a characterization method (400) as claimed in any of claims 1 to 9.

12. The method (700) as described in claim 11, further comprising a first manufacturing stage (702) prior to the second manufacturing stage (710), comprising multiple repetitions of a step of manufacturing an optical objective from the batch of objectives, the step comprising the following operations: determining an assembly set for the optical objective based on an individual set of optical elements of the optical objective, and stacking the optical elements to form the optical objective, measuring a geometry set with respect to the optical objective, and storing (706) a training set (JEi) in a training database (602) formed by the assembly set and the measured geometry set.

Images