EP3807838A2 - Materialprüfung von optischen prüflingen - Google Patents
Materialprüfung von optischen prüflingenInfo
- Publication number
- EP3807838A2 EP3807838A2 EP19729717.9A EP19729717A EP3807838A2 EP 3807838 A2 EP3807838 A2 EP 3807838A2 EP 19729717 A EP19729717 A EP 19729717A EP 3807838 A2 EP3807838 A2 EP 3807838A2
- Authority
- EP
- European Patent Office
- Prior art keywords
- images
- optical
- defects
- variable
- defect
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M11/00—Testing of optical apparatus; Testing structures by optical methods not otherwise provided for
- G01M11/02—Testing optical properties
- G01M11/0221—Testing optical properties by determining the optical axis or position of lenses
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M11/00—Testing of optical apparatus; Testing structures by optical methods not otherwise provided for
- G01M11/02—Testing optical properties
- G01M11/0242—Testing optical properties by measuring geometrical properties or aberrations
- G01M11/0257—Testing optical properties by measuring geometrical properties or aberrations by analyzing the image formed by the object to be tested
- G01M11/0264—Testing optical properties by measuring geometrical properties or aberrations by analyzing the image formed by the object to be tested by using targets or reference patterns
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M11/00—Testing of optical apparatus; Testing structures by optical methods not otherwise provided for
- G01M11/02—Testing optical properties
- G01M11/0242—Testing optical properties by measuring geometrical properties or aberrations
- G01M11/0278—Detecting defects of the object to be tested, e.g. scratches or dust
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/50—Image enhancement or restoration using two or more images, e.g. averaging or subtraction
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/90—Dynamic range modification of images or parts thereof
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10141—Special mode during image acquisition
- G06T2207/10152—Varying illumination
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
Definitions
- Various embodiments of the invention generally relate to techniques for material testing optical specimens. Various embodiments of the invention relate in particular to techniques for material testing optical ones
- defects can occur in the manufacture of optical elements - for example lenses.
- defects include: scratches; inclusions; Material defects; matt surfaces; Schlieren; flaking; Drilling faults;
- optical elements are used as test objects in a manual
- SFS also has certain disadvantages and limitations. For example, it can often be complicated and time-consuming to obtain a sufficiently large number of images to carry out a reconstruction of the test object. Appropriate techniques can also be time consuming and prone to errors.
- Imaging system comprises, for each of at least one pose of the test object in relation to the imaging system: in each case actuation of the optical imaging system for capturing at least two images of the test object by means of variable-angle lighting and / or variable-angle detection.
- the procedure also includes, for each of the
- the result image shows a digital one
- the method further includes performing a material test based on the at least one result image that was generated based on the at least two images in the at least one pose.
- Carrying out the material test serves to assess the quality of the test object.
- the compliance with an optical standard can be checked, for example, on the basis of the detected defects.
- KNNs trained artificial neural network
- the entire optical effect of the optical test object can be assessed without resolving the influence of individual defects on the overall optical effect.
- Such an algorithm with machine learning can therefore receive the at least one result image as input. As an output, the machine learning algorithm
- the quality value could e.g. be binary, i.e. indicate "pass” or "fail” in connection with the fulfillment or non-fulfillment of a standard. Finer-resolution quality values would also be conceivable, for example quality values which show the deviation from the norm
- the algorithms described here with artificial intelligence or with machine learning can include at least one element from the group: a parametric mapping, a generative model, a
- Anomaly detector and a KNN.
- KNN examples include a deep KNN (DNN, Deep Neural Network), a time-delayed KNN, a folding KNN (CNN,
- LSTM Long short-term memory
- the captured images can also be called raw images.
- the optical device under test can be selected from the following group:
- the optical device under test can therefore be set up to deflect light, for example to refract or focus it.
- the optical test specimen can be made of glass or a polymer, for example.
- the optical test specimen can be at least partially transparent to visible light or infrared light or ultraviolet light.
- the optical device under test can have an optical effect, i.e. Break light.
- the pose of the test object in relation to the imaging system can, for example, be a position and / or an orientation of the test object in one by the imaging system designate the defined reference coordinate system. For example, that
- the pose of the test specimen can be changed, for example, by rotating the test specimen or by translating it.
- test object can be used with different parameters
- Illumination geometries are illuminated.
- the test object could be illuminated from different directions of illumination.
- An image of the at least two images can then be acquired when the test specimen is illuminated with the corresponding illumination geometry.
- An exemplary implementation of the variable-angle lighting is described, for example, in: US 2017/276 923 A1.
- An appropriately trained could be used for variable-angle lighting
- Lighting module can be used.
- a lighting module can be used which has a multiplicity of light sources. The different light sources can be controlled separately. Then, depending on which is activated
- Light source or activated light sources a different lighting geometry can be implemented. If two pictures with different
- Illumination geometries are associated, recorded, the digital contrast can be determined from a distance between the positions of the images of the test object in the two images.
- Another implementation for the lighting module comprises a light source and a mirror arrangement, for example a DMD (digital micromirror device).
- DMD digital micromirror device
- a different lighting geometry can then be activated depending on the position of the different mirrors.
- Illumination geometries are associated, recorded, the digital contrast can be determined from a distance between the positions of the images of the test object.
- a suitably designed detection optics can be provided for the angle-variable detection.
- the detection optics can be set up to image the
- An adjustable filter element which is arranged in a beam path of the detection optics that defines the image, can continue to be provided.
- a controller can be set up to control the adjustable filter element, to filter the spectrum of the beam path with different filter patterns, in order to enable angle-variable detection in this way.
- the controller can in particular be set up to capture images associated with different filter patterns.
- Different filter elements can be used, for example a filter wheel or a DMD or a liquid crystal filter or a movable filter plate or an amplitude filter element.
- the detection can be limited to certain angles of the light incident on the detector from the test specimen.
- Examples of the angle-variable detection are therefore based on an amplitude filtering of the imaging spectrum in or near a pupil plane of the detection optics of the optical device. This corresponds to the filtering of certain angles with which rays strike a sensor surface of the detector; i.e. individual beams are selectively transmitted according to their angle to the sensor surface. Different filter patterns are used; for each filter pattern, an associated image is captured by a detector. In the various described herein
- each filter pattern can define at least one translucent area which is surrounded by a non-translucent area.
- the translucent area could be spaced from the optical axis of a beam path defined by the detection optics, i.e.
- the translucent area is linear, i.e. that the corresponding filter pattern defines a line. Then only rays with angles in a narrowly limited angular range are let through.
- the filter patterns used can be converted into one another, for example, by translation along a vector.
- Filter pattern which has an off-axis, linear translucent area - a defocused test piece is shown shifted. If two images associated with different filter patterns are acquired, the digital contrast can be determined from a distance between the positions of the images of the test object. Appropriate digital contrasts can thus be generated by suitable processing of the various images, which were recorded by means of the variable-angle lighting and / or the variable-angle detection.
- the digital contrast can be selected from the following group, for example: virtual dark field; digital
- phase gradient contrast For example, using the techniques described in US 2017/276 923 A1, it may be possible to determine the phase gradient contrast. This means that opposite edges of the test object as a phase object can have different signs of the contrast.
- An - optionally quantitative - phase contrast could be determined, for example, by means of the so-called quantitative difference phase contrast technique.
- QDPC quantitative differential phase contrast technique
- a phase contrast as described in German patent application DE 10 2017 108 873.3, can also be obtained by appropriately using a transfer function when processing the images to generate the result image with the digital contrast.
- the transfer function can be a
- the transfer function can be suitable for predicting the at least one image for a specific lighting geometry and a specific test object.
- the transfer function can have a real-valued part and / or an imaginary part.
- the real-valued portion of the transfer function can correspond to a decrease in the intensity of the light as it passes through the test specimen.
- An amplitude object typically has significant attenuation of the light.
- the imaginary part of the transfer function can denote a shift in the phase of the light passing through the device under test.
- a phase object typically has a significant shift in the phase of the light. It can be different Techniques for determining the transfer function can be used. In one example, the transfer function could be determined based on an Abbe technique.
- a reference transfer function could be determined using an Abbe technique.
- the device under test can be separated into different spatial frequency components. Then an overlay of an infinite number of harmonic grids can model the test object.
- the light source can also be broken down into the sum of different point light sources. Another example concerns
- TCC transmission cross-coefficient matrix
- the frequencies that the optics transmit are limited to the area in which the TCC assumes values other than 0.
- a system with a high coherence factor or coherence parameter consequently has a larger area with TCC F 0 and is capable of higher ones
- the TCC typically contains all of the information in the optical system, and the TCC often also takes complex pupils such as e.g. B. in the Zernike phase contrast or triggered by aberrations.
- the TCC can enable the optical transmission function to be separated from the object transmission function. In some examples it would also be possible that the
- Transfer function is specified and no determination as TCC or according to Abbe has to be made.
- An exemplary technique is in Tian, Waller regarding Eq. 13 described. It shows how, based on a Tichonov regularization, a result image by means of an inverse Fourier transformation and based on the transfer function H * and also based on the Spatial frequency space representation of a combination I DPC of two images of the
- DUT can be determined with different lighting geometries:
- I DPC describes the spectral decomposition of a combination of two images IT and IB, which have different lighting geometries that are related to each other
- the lighting geometry does not have to be strictly semicircular.
- four LEDs could be used, which are arranged on a semicircle.
- Illumination directions are used, such as individual light emitting diodes. Furthermore, in Eq. 2 normalization to 1 instead of I T + I B , or to another value. Instead of offsetting IT and IB, others could
- an absorption component in particular, can be reduced on the basis of a real-value component of the transfer function.
- I DPC is proportional to the local increase in phase shift due to the device under test.
- Phase shift can be caused by a change in the thickness of the test object or the topography of the test object and / or by a change in the optical properties.
- two images I DPC 1 and I DPC , 2 can be determined, one with a pair of semicircular lighting geometries that are arranged up-down in a lateral plane perpendicular to the beam path ⁇ I DPC, 1), and one with a pair of semicircular ones Illumination geometries, which are arranged left-right in the lateral plane ⁇ I DPC, 2 ) ⁇
- both I DPC 1 and I DPC , 2 can be used to determine the Result image are taken into account, see summation index j in Eq. 1.
- a virtual dark field can be generated by suitable processing of the images.
- the virtual dark field is calculated, for example, as component by component
- Lighting geometries can calculate the results of this calculation
- variable-angle lighting and / or the variable-angle detection By using the variable-angle lighting and / or the variable-angle detection, it is therefore possible to provide the result image with a tailored contrast. This enables the defects to be made particularly visible. This enables the material testing to be carried out particularly reliably and robustly.
- variable-angle lighting and / or the variable-angle detection can be used particularly quickly, so that the measuring time per test specimen can be reduced.
- the material inspection is carried out fully automatically or semi-automatically. It could therefore be possible to carry out the material test automatically. Then, for example, the quality of the optical test objects could be evaluated automatically. In particular, the test object could also be classified according to a reference, such as an ISO standard, etc. For example, it would be possible to use a KNN, which - through suitable training or
- the reference measurement data can, for example, show images captured by means of variable-angle lighting and can be annotated with the corresponding categories “pass” and “fail”. In such a scenario, it may be unnecessary to identify and / or segment individual defects or to determine the influence of individual defects on the optical effect of the test object.
- performing the material check includes that
- the defect detection algorithm can be designed to provide - in addition to the mere detection of defects - additional information about detected defects.
- the additional information can describe properties of defects in qualitative and / or quantitative terms. For example, a location and / or shape and / or size of detected defects could be determined. Such additional information can be stored in an error card.
- the defect detection algorithm can generally be different
- Defect detection algorithm uses one or more of the following techniques: machine learning; Threshold value formation, for example directly or on filtered data or by means of differential imaging using reference images; statistical analysis;
- the defect detection algorithm can use a trained KNN, for example. There would be the option, for example, of the detected defects automatically to be divided into different classes using machine learning. It can be assigned to different defect classes. The size of the defects could be quantified using segmentation, for example. The result image can be segmented accordingly. For example, when evaluating the defects using machine learning techniques, the classification of the respective defect in relation to predefined defect classes, as well as the associated one
- Segmentation of the respective result image with regard to the position of the defect can be determined.
- Defect detection algorithm also recognizes defects that belong to defect classes that are unknown a priori, for example, an anomaly detector can be used.
- An anomaly detector can detect any deviations from a specification. The anomaly detector can thus detect defects, even if they are not associated with previously known or trained defect classes. Examples include "support vector machines", Bayesian networks, etc.
- the classes and sizes of the defects can be used to test whether a suitable standard has been met.
- an error map that is generated by the defect detection algorithm can be compared with a reference error map that maps the standard. For example, could - based on such
- Comparison - determine whether the test object is within or outside a tolerance. It could also be determined that a check for compliance with the appropriate standard is possible at all: if, for example, an uncertainty in connection with the defect detection algorithm is too great, the
- Anomaly detector is used, which determines whether defects or generally structures are present in the images for which no automatic checking is provided, for example because the associated defect classes are unknown.
- Another option involves replacing the manual visual inspection with a semi-automated visual inspection. This can be done by displaying the result image with the digital contrast on a screen to a user. The latter can carry out a comparison with the standard and, for example, perform a quality value of the material test depending on the comparison.
- one or more of the unprocessed raw images can be taken into account directly when carrying out the material test. This means that it may be possible to supplement the use of result images with the use of raw images.
- a defect detection algorithm could not only receive result images of the angle-variable lighting as input, but also raw images. For example, images that were captured using bright field lighting could be taken into account. As an alternative or in addition, images could also be taken into account that were recorded during dark field illumination.
- raw images can be obtained with even greater accuracy.
- a computing unit is set up to control an optical imaging system in order to use each of at least one pose of an optical test specimen in relation to the
- Imaging system each have at least two images of the test specimen using at least one of variable-angle lighting and variable-angle detection.
- Computing unit is also set up for each of the at least one pose: each processing of the corresponding at least two images to generate one
- the computing unit is also set up to use one based on the at least one result image that was generated based on the at least two images in the at least one pose of the test object
- the computing unit could include a graphics card with a large number of parallel processing pipelines, for example in contrast to a “central processing unit”, CPU.
- effects can be achieved that are comparable to the effects that can be achieved for a method for material testing as described above.
- FIG. 1 schematically illustrates an optical imaging system according to various examples.
- FIG. 2 schematically illustrates a controller for an optical imaging system according to various examples.
- FIG. 3 schematically illustrates an optical lighting module
- FIG. 4 schematically illustrates an illumination geometry of the angle variable
- FIG. 5 schematically illustrates an optical detection system of an optical imaging system according to various examples, the optical detection system being used for the variable angle
- FIG. 6 schematically illustrates filter patterns of the detection optics according to various examples.
- FIG. 7 is a flow diagram of an example method.
- FIG. 8 is a flow diagram of an exemplary method, wherein FIG. Illustrated 8 aspects related to image capture.
- FIG. 9 is a flow diagram of an exemplary method, wherein FIG. Illustrated 9 aspects related to image evaluation.
- FIG. 10 is a flow diagram of an exemplary method, wherein FIG. Illustrated 10 aspects related to material testing.
- FIG. 11 schematically illustrates an image acquisition and image evaluation for generating result images with digital contrast according to various examples, as well as a material test.
- FIG. 12 schematically illustrates an optical imaging system according to various examples.
- FIG. 13 schematically illustrates an optical imaging system according to various examples.
- FIG. 14 schematically illustrates an imaging optical system according to various examples.
- FIG. 15 schematically illustrates an optical imaging system according to various examples.
- FIG. 16 schematically illustrates an optical imaging system according to various examples.
- FIG. 17 schematically illustrates an imaging optical system according to various examples.
- connections and couplings between functional units and elements shown in the figures can also be implemented as an indirect connection or coupling.
- a connection or coupling can be implemented wired or wireless.
- Functional units can be implemented as hardware, software or a combination of hardware and software.
- FIG. 1 illustrates an exemplary optical imaging system 100 (hereinafter also referred to as optical device 100).
- optical device 100 for example, the optical
- Device 100 implement a light microscope, for example in transmitted light geometry.
- the optical device 100 it may be possible to display small structures of a test specimen or sample object fixed by a sample holder 113 in enlarged form.
- a detection optics 112 is set up to generate an image of the test object on a detector 114.
- the detector 114 can then be set up to acquire one or more images of the test object. Viewing through an eyepiece is also possible.
- a lighting module 111 is set up to the test specimen, which is on the
- Sample holder 113 is fixed to illuminate.
- this lighting could be implemented using Köhler's lighting.
- a condenser lens and a condenser aperture diaphragm are used. This leads to a particularly homogeneous intensity distribution of the light used for lighting in the plane of the test object.
- partially incoherent lighting can be implemented.
- the lighting module 111 could also be set up to illuminate the test object in dark field geometry.
- a controller 115 is provided to control the various components 111-114 of the optical device 100.
- the controller 115 could be set up to control a motor of the sample holder 113 by one
- controller 115 could be implemented as a microprocessor or microcontroller.
- the controller 115 could comprise an FPGA or ASIC, for example.
- FIG. 2 illustrates aspects relating to the controller 115.
- the controller 115 comprises a computing unit 501, for example a microprocessor, an FPGA or an ASIC.
- the computing unit 501 can also be a PC with a central computing unit and / or a computing unit which is designed for parallel processing, for example a graphics card.
- the controller 115 also includes a memory 502, for example a non-volatile memory. It is possible that program code is stored in the memory 502. The program code can be loaded and executed by the computing unit 501.
- the computing unit 501 can then be set up, based on the program code, for material testing techniques optical test specimen, to control the optical device for performing an angle-variable illumination or angle-variable detection, etc.
- the lighting module 111 is therefore set up to one
- Lighting module 111 different lighting geometries of the
- Illumination of the device under test can be implemented.
- the different illumination geometries can correspond to illumination of the test object from at least partially different directions of illumination.
- Lighting module 111 comprise a plurality of adjustable lighting elements which are set up to modify or emit light locally.
- the controller 115 can control the lighting module 111 or the lighting elements in order to implement a specific lighting geometry.
- FIG. 3 illustrates others
- FIG. 3 illustrates aspects related to the lighting module 111
- Lighting module 111 is set up for variable-angle lighting. In FIG.
- the lighting module 111 has a multiplicity of adjustable lighting elements 121 in a matrix structure.
- the matrix structure is oriented in a plane perpendicular to the beam path of the light (lateral plane;
- the adjustable lighting elements 121 could be implemented as light sources, for example as light emitting diodes. Then it would be possible, for example, that different light-emitting diodes with different light intensities light Emit lighting of the test object. This enables an illumination geometry to be implemented. In another implementation, that could be implemented as light sources, for example as light emitting diodes. Then it would be possible, for example, that different light-emitting diodes with different light intensities light Emit lighting of the test object. This enables an illumination geometry to be implemented. In another implementation, that could
- Illumination module 111 can be implemented as a spatial light modulator (Engl., SLM).
- SLM spatial light modulator
- the SLM can intervene in a spatially resolved manner
- FIG. 4 illustrates aspects related to an exemplary lighting geometry 300.
- the light intensity 301 provided for the various adjustable ones
- the illumination geometry 300 is dependent on the position along the axis XX 'and is therefore variable in angle. By selecting elements 121 more or less distant, the strength of the angle variable can be changed
- Lighting can be varied. By varying the direction between the
- Orientation of the angle-variable lighting can be changed.
- the imaging position of the optical test object varies depending on the illumination geometry 300. This change in the imaging position can be used to generate the digital contrast.
- variable-angle detection can also be used in a further example.
- the detection optics 112 is set up accordingly.
- FIG. 5 illustrates aspects related to the detection optics 112.
- FIG. 5 illustrates an embodiment of the detection optics 112 for angle-variable detection.
- FIG. 5 illustrates an exemplary implementation of the detection optics 112 with an adjustable one
- the filter element 119 is arranged in the region of a conjugate plane of the beam path 135, that is to say close to or at a pupil plane of the beam path 135.
- the (spatial frequency) spectrum of the beam path 135 is therefore filtered. In the spatial area, this corresponds to the selection of different angles 138, 139, under which the light falls along corresponding beams 131, 132 from the test specimen 150 onto the sensor surface 211.
- the filter element 119 also forms one
- the beam path 135 is implemented by lenses 202, 203.
- the beams 131, 132 of the beam path 135 are shown starting from a test specimen 150 arranged in a defocused manner through the detection optics 112 to the detector 114, that is to say in particular a sensor surface 211 of the detector 114.
- the rays 131 correspond to a filter pattern 380 which has a first translucent area 381 with a
- Translucent area 382 defined in the X direction.
- the device under test 150 is arranged out of focus and has a focus position 181 which is not equal to zero. Therefore, the rays fall on the at different angles 138, 139
- the images 151, 152 of the test object 150 caused by the rays 131, 132 are positioned on the sensor surface 211 at a distance 182 from one another.
- the distance 182 depends on the focus position 181:
- NA denotes a correction value of the oblique angle of incidence 138, 139.
- NA can be determined empirically or by beam path calculation, for example. The equation shown is an approximation. In some examples, it may also be possible for the angle of incidence 138, 139 to be dependent on the
- Focus position 181 i.e. to be taken into account by Dz. This dependency can be system-specific.
- FIG. 5 is a one-dimensional representation of filter patterns 380-1, 380-2. In some examples, however, filter patterns could also be used, the one
- FIG. 6 illustrates aspects related to example filter patterns 351, 352.
- the filter pattern 351 defines one line and the filter pattern 352 defines another line.
- the filter pattern 351 can be converted into the filter pattern 352 by translation along the vector 350.
- the lines of the filter patterns 351, 352 are parallel to one another along their entire lengths. Generally it would be it is also possible that lines are used which run parallel to one another only along part of their lengths, for example along at least 50% of their lengths or along at least 80% of their lengths. That way, a more flexible choice of
- Filter patterns 351, 352 can be guaranteed.
- the lines of the filter patterns 351, 352 are arranged off-axis with respect to the optical axis 130. This can maximize the length of vector 350; whereby the distance 182 of the images 151, 152 can be maximized. This allows the strength of the angle-variable detection to be adjusted.
- Orientation of the angle variable detection can be changed.
- FIG. 7 illustrates a flow diagram of an example method.
- the method of FIG. 7 are executed by the controller 115
- the computing unit 501 could load program code from the memory 502 and then the method according to FIG. 7 perform.
- the image is captured in block 1001.
- an optical imaging system such as optical device 100 of FIG. 1, can be controlled to capture one or more images.
- the optical imaging system for capturing at least two images of an optical test specimen can be controlled by means of at least one of angle-variable illumination and angle-variable detection.
- variable angle lighting can use several variables
- Illumination geometries include, each having at least one
- Illumination geometry is associated with one of the at least two images. Different lighting geometries can be implemented, for example, by activating different light sources.
- the angle-variable detection can be done by filtering a spectrum of a
- Beam path by providing appropriate filters near or in the
- Pupil level of a detection optics of the optical imaging system is Pupil level of a detection optics of the optical imaging system.
- a variation in the strength and / or orientation of the variable-angle lighting can be achieved by varying the light sources used or the ones used
- Illumination geometries take place. Accordingly, the strength and / or orientation of the angle-variable detection can be varied by varying the filter pattern used.
- the strength and / or the orientation of the variable-angle illumination and / or the variable-angle detection are varied, the strength and / or direction of a subsequently determined digital contrast can vary. By varying the direction of the digital contrast, certain defects can be made particularly visible.
- the image evaluation takes place in block 1002. This may include processing the at least two images captured in block 1001 to produce a digital contrast result image.
- the digital contrast can be selected, for example, from the following group: phase contrast; Phasengradientenkontrast; and virtual
- a material test is then carried out in block 1003.
- the material test can generally serve to assess the quality of the optical test object.
- the material test can also be carried out automatically in block 1003.
- a figure of merit can be determined, such as using a KNN or other machine learning algorithm.
- the KNN can receive one or more of the result images from block 1002 as an input card.
- the KNN can provide the quality value as an output.
- the quality value could, for example, specify whether the test object is "within a standard", "outside a standard” or - optionally - "indefinite".
- performing the material test can also be used to detect individual defects in the optical test object, for example using a defect detection algorithm for the automated detection of defects.
- Blocks 1001-1003 do not have to be processed sequentially; it would rather be e.g. it is possible to start image evaluation 1002 and possibly material testing in blocks 1002, 1003 already during the image acquisition from block 1001.
- FIG. 8 is a flow diagram of an example method.
- FIG. 8 shows an exemplary implementation of block 1001 from FIG. 7. This means that FIG. 8 shows details of the implementation of image acquisition.
- an orientation and / or a Z position and / or an XY position of the optical test specimen is set. This means that the test subject's pose is adjusted. This could be done, for example, by controlling a motorized sample holder 113 by the controller 115 (see FIG. 1). For example, the sample holder could rotate the optical specimen for orientation
- the sample holder 113 it would be possible for the sample holder 113 to move the optical test specimen along the beam path by the Z position - ie the focus position.
- the XY position - that is, the position perpendicular to the beam path - to be set.
- the user is asked via a human-machine interface to set a certain pose of the test object.
- test object remains stationary, but the detection optics are shifted relative to the test object (camera defocus).
- Expansion compared to the field of view of the optical imaging system includes the use of so-called "sliding windows".
- sliding windows images - for example of result images and / or raw images - can be processed, even if these images do not completely fit into a memory of the corresponding computing unit, for example into a level 1 or level 2 memory of the computing unit. Then it may be necessary to first save the corresponding image in an external RAM block and then to load and process individual pieces of the images according to the sliding windows.
- successive sliding windows of a sequence of sliding windows to contain an overlap area which reproduces the same area of the test object. It would therefore be possible to evaluate sequentially overlapping areas of an image - for example using an algorithm with machine learning then combine the results of this evaluation into an overall result, for example in order to obtain the error map as described here.
- Image edges each individual image from the camera can be transferred directly to the KNN as input.
- the acquisition of raw images and the generation of result images and (ii) the performance of the material test based on an algorithm with machine learning can be carried out at least partially in parallel, for example for different poses of the test object in relation to the imaging system and / or different sliding windows of a corresponding sequence. This allows the total time for image acquisition and evaluation in
- orientation of the optical test specimen can be particularly helpful when certain regions of the optical test specimen are only visible under a certain orientation; this would be conceivable, for example, for partially mirrored surfaces, etc. Some defects can only be found under a certain one
- Imaging modalities In general, the most varied
- Imaging modalities can be set, such as: color of the light used;
- Polarization of the light used Polarization of the light used; Strength of the angle variable lighting and / or angle variable detection; Orientation of the variable-angle lighting and / or variable-angle detection.
- Such a variation of the imaging modality can achieve that certain defects can be made particularly well visible, i.e. have a high contrast. For example, certain defects could only be visible with a certain orientation and strength of the variable-angle lighting and / or variable-angle detection. Accordingly, it would be possible for certain defects in the optical test specimen to be visible only with a specific color or polarization of the light used.
- Block 1013 then records at least two images with variable-angle lighting and / or variable-angle detection.
- the one or more imaging modalities set in block 1012 are used.
- the poses of the optical device under test set in block 1011 are used for the optical imaging system.
- the angle of illumination or the angle of detection for the different images is varied according to the set modality.
- two line patterns 351, 352 could be used; these have the same orientation (in the X direction) and also the same strength of the angle-variable detection (same distance from the main optical axis).
- block 1014 it is checked whether a further variation of the one or more imaging modalities should take place for the currently valid pose. If this is the case, block 1012 is executed again and the corresponding variation of the one or more imaging modalities is carried out. Then - for the newly set one or more imaging modalities - again in a further iteration of block 1013, at least two images by means of angle-variable illumination and / or
- angle-variable detection detected For example, in a further iteration compared to the line patterns 351, 352 from FIG. 6 line patterns rotated by 90 ° in the XY plane are used - this would correspond to a variation in the orientation of the angle-variable detection.
- the strength of the angle-variable detection could be achieved, for example, by shifting the translucent areas of the line patterns 351, 362 along the X axis.
- block 1015 is executed. In block 1015 it is checked whether a further pose should be set. If this is the case, block 1011 is executed again.
- FIG. 9 is a flow diagram of an example method.
- FIG. 9 relates to techniques which are used in connection with the image evaluation according to block 1002 from FIG. 7
- a current pose is selected from one or more predetermined poses for which images have been captured. For example, all of the poses shown in block 1011 of FIG. 8 have been activated, stored in a list and
- An imaging modality is then selected in block 1022.
- all of the one or more imaging modalities described in block 1012 of FIG. 8, are stored in a list, and are then selected in sequence in various iterations from block 1022.
- the respective at least two images which in block 1013 from FIG. 8 combined for the currently selected pose and for which one or more imaging modalities have been recorded.
- a respective result image is obtained in this way.
- the result image has a digital contrast.
- the result image can have different digital contrasts.
- pairs of images can be recorded, for example, each with complementary, linear ones
- Illumination directions are associated (this would correspond, for example, to the activation of all light sources 121 which are on a column or in a row of the matrix of the
- Illumination module 111 from FIG. 3 are arranged). Accordingly, one of the line-shaped filter patterns 351, 352 from FIG. 6 can be used, which can then be rotated by 90 ° in the XY plane.
- a difference could then be formed, e.g. according to
- left and right denote the images, which are each associated with a left or right-oriented semicircular lighting geometry or filter pattern and where praise and praise denote the images which are each associated with an upward or downward oriented semicircular lighting geometry or filter pattern.
- Filter patterns can be determined during the variable-angle detection.
- block 1025 it is then checked in block 1025 whether further images have been acquired for a further pose and, if necessary, a further iteration of block 1021 is carried out. If no more pose for selection image evaluation is complete. Then all images were processed and the corresponding result images are available.
- the material test can then be carried out. Corresponding examples are shown in FIG. 10 shown.
- FIG. 10 illustrates an example method.
- the method according to FIG. 10 illustrates an exemplary implementation of the material test according to block 1003 from FIG. 7th
- a defect detection algorithm for the automatic detection of defects is applied to the available result images.
- Defect detection algorithm considers relationships between the result images. This can correspond to a coherent evaluation of the registered result images. This enables 3-D defects to be recognized. To the
- defects can be recognized that have an extent across several result images.
- extensive defects can be detected, for example, in the Z direction (XY plane). If block 1031 is not executed, it would be possible to
- Defect detection algorithm to apply for each of the result images. This means that the defect detection algorithm can be executed multiple times and receives one result image as input for each execution.
- Defect detection algorithm can be particularly simple.
- the defect detection algorithm can issue an error card for each execution. If the defect detection algorithm is executed several times for different result images, several error cards are obtained. The error cards mark defects in the respective result picture
- Block 1033 is again an optional block. In particular, block 1033
- the defects marked by the error cards can be registered on a global coordinate system. Such can be of different
- Defect detection algorithm are recognized, related to each other. In this way, a global fault map can be obtained in which defects are marked in the entire area of the optical test specimen.
- Error cards such as a global error card, with one or more reference error cards.
- Reference error cards can be derived from an industry standard, for example.
- Reference error cards can, for example, indicate tolerances for certain defects.
- a quality value of the material test can be determined depending on the comparison. For example, it could be checked whether the identified defects remain within a tolerance range specified by the reference error card. It can So - in connection with the determination of the quality value - the test object is assigned to one of several predefined result classes - eg “in tolerance” / “out of tolerance” with regard to a suitable standard or other specifications.
- the output can also be "undetermined”: a manual visual inspection can then be suggested. This corresponds to a "red-yellow-green” principle.
- uncertainty related to the comparison can occur for several reasons. For example, a
- Anomaly detector used in connection with the defect detection algorithm indicate that there are unknown defects, e.g. from an algorithm trained with different defect classes - e.g. a KNN - cannot be recognized.
- the quality value can indicate, for example, whether the optical test specimen fulfills a certain industrial standard or not.
- block 1034 can take on different forms.
- the defect detection algorithm in block 1032 is based on a plurality of result images, all of which are based on a common global defect cards.
- the comparison in block 1034 could be between a single global error map generated by the corresponding one
- Defect detection algorithm is obtained, and the reference error map can be performed.
- these multiple error maps could each be compared to the reference error map or compared to multiple corresponding reference error maps.
- a relative arrangement of the various error cards could be taken into account by registering in block 1033 accordingly.
- the error maps in the various examples described herein may have different information content.
- defect card As a general rule: different information about the defects detected by the defect detection algorithm can be stored in the defect card.
- Defect detection algorithm for a detected defect is a corresponding one
- defect classes include: scratches; inclusions; Material defects; matt surfaces;
- defect classes are: dirt; dig; holes; cracks; splashes;
- (///) Another example of additional information that can be determined and output by the defect detection algorithm is the influence of the respective defect on the optical effect of the test object. For example, it may happen that some defects - even though they are comparatively large from a geometric point of view - have only a minor influence on the optical effect of the test object. On the other hand, in some scenarios, even small defects can have a major impact on the optical effect of the test object.
- the influence of the defect on the optical effect could describe a change in the optical effect qualitatively or quantitatively.
- the change in focal length for a lens implementing the device under test could be specified with regard to the sign of the change and optionally the magnitude of the change.
- Defect detection algorithm can be determined and output is an aggregated influence of all detected defects on the optical effect of the test object. This is based on the knowledge that it may sometimes be possible for two separate defects to intensify or weaken in total, depending on the type and arrangement of the defects.
- Defect detection algorithm output error card can contain different additional information about the defects in different examples.
- the comparison in block 1034 can take on different configurations.
- the reference error map could, for example, specify a maximum extent of defects and / or a maximum density of defects - regardless of the classification of the defect.
- the reference defect map could, for example, specify a maximum number of defects of a certain defect class.
- the reference error map can map these number of stages.
- these step numbers are defined over the edge length of the defect area, for example as a square. This allows a direct relationship between segmentation through the defect detection algorithm - that is, the number of pixels assigned to the defect - and the number of stages over the imaging scale.
- Defect classes the length and width can also be determined. This can include adapting algorithmic pattern shapes to the segmentation;
- the size of the scratches can be determined by means of
- Image processing methods - e.g. skeleton, medial axis transformation - can be determined.
- the position and extent of the defect in relation to the test area can be determined, in particular in the case of edge protrusions and layer offset. This can be done by means of an algorithm which detects the surroundings of the defect, for example the exact position of the edge of the test object.
- Another criterion that can be represented by the reference defect map and the defect map is the accumulation of defects in a certain percentage of the test area. This can correspond to a maximum density of defects.
- Defect detection algorithm can be used.
- the defect detection algorithm in an implementation it would be possible for the defect detection algorithm to include a trained KNN.
- the defect detection algorithm could include a convolution network.
- the KNN can be set up to act as an input card for a
- a KNN typically includes an input layer, one or more hidden layers, and an output layer.
- the output layer can comprise one or more neurons.
- a position and / or a classification of detected defects can be indicated in this way.
- the KNN could be set up to provide an output card - which then implements the error card or from which the error card can be derived - which indicates a position and a classification of the detected defects.
- Network can correspond to different entries in the output card.
- Examples include a threshold analysis of the digital contrast of the result images and a statistical analysis of the digital contrast. For example could be assumed that defects have a particularly strong digital contrast; in this way, the defects can be segmented using the threshold value analysis. From the form of the segmentation, for example, an assignment to the predefined defect classes could then take place as a classification. For example, statistical analysis could be used to determine a density of defects or an average distance between defects.
- FIG. 11 illustrates aspects related to techniques for material testing.
- Raw data 800 are obtained from image capture 1001. This raw data corresponds to images 890, which at
- test specimen 150 different modalities and / or different poses of the test specimen 150 with respect to the optical device 100 were detected.
- Z-stacks 801 of images 890 are acquired, each for different wavelengths 802-1, 802-2.
- Illumination geometry in the case of variable-angle illumination or the filter pattern in the case of variable-angle detection can be varied. For example, are in the example of FIG. 11 for a wavelength 802-1, 802-2, three images each with the same Z position (same stack position in the Z stack 801), at different angles of the
- variable-angle lighting / variable-angle detection
- XY stacks could also be detected.
- Result data 891 are now obtained from raw data 800.
- the result images 891 are formed by the appropriate combination of the different images 890 of the
- Images 890 are combined that have the same pose and the same imaging modality.
- the uppermost images which correspond to the wavelength 802-1
- the uppermost images are combined with one another in order to obtain a corresponding result image 891.
- a Z stack 801 of result images 891 is thereby obtained, in each case for each imaging modality 802-1, 802-2.
- the defect detection algorithm is then applied, 1032. This results in a Z stack 801 of error cards 895 which in each case marks the defects 901-903 in different Z planes.
- FIG. 11 also illustrates that a KNN 950 can be used as a defect detection algorithm in block 1032.
- the KNN 950 comprises an input layer (far left) with several neurons, hidden intermediate layers and an output layer (far right) with several neurons.
- sliding windows are used to display the information
- the KNN 950 can be trained in block 1052.
- To train the KNN 950 reference measurements for a large number of optical test objects and manual annotation of visible defects can be carried out by experts in block 1051.
- Semi-automated machine learning would also be conceivable. Training links the different layers, i.e. e.g. appropriate weights, etc.
- the various defects 901 -903 are then measured in block 1061. This can correspond to a segmentation of the marked defects 901 -903 in the different result images 891.
- Defects 901-903 are also classified. For example, in the scenario of FIG. 11 the different defects 901 -903 different
- Defect classes assigned (illustrated by the different filling).
- the defects 901-903, which are marked in the error cards 895, can also be registered on a global coordinate system, ie a relative positioning of the various defects 901-903 with respect to one another or with respect to a reference point can be determined.
- reference data can be taken into account in block 1062, which are obtained, for example, by a calibration measurement with a reference sample 910.
- the reference data can determine the absolute size and / or relative positioning of the different defects 901-903 to one another
- the processed error cards 895 can then be compared in block 1034 with a reference error card 896, which in the example of FIG. 11 from one
- a quality value can be determined in block 1035, which in the example of FIG.
- the KNN 950 is trained on the basis of
- Reference measurements In general, it would be possible, as an alternative or in addition to such reference measurements, for a repeated or continuous re-training of the KNNs 950 - or generally for defect detection algorithms that use machine learning - to be carried out. It is possible, for example, that the output of KNNs 950 or the result of the material test are checked repeatedly in the ongoing test process and new reference measurements are generated from them. This can be automated or at least partially monitored. With these newly generated
- Reference measurements can be re-trained regularly or for certain events (new standard introduced, change of measurement setup, change of the test object). That the KNN can generally be re-trained on the basis of a result of the material test. This can create a continuous
- Refinement or improvement of the detection accuracy of the KNN can be achieved.
- one of the following two goals (or both) can be achieved: (/) Improvement of the metric applied to the KNN, eg accuracy of the Segmentation, defect detection rate, number of false positive / false negative
- FIG. 12 illustrates aspects related to the optical device 100.
- the optical device 100 comprises an illumination module 111, which is set up for variable-angle illumination of an optical test object 150.
- Sample holder 113 can adjust the pose of the specimen 150 in relation to the optical
- Vary device 100 by rotation.
- the imaging takes place in transmitted light geometry.
- FIG. 13 illustrates aspects related to the imaging optical system.
- FIG. 13 illustrates a further exemplary implementation of the imaging system 100.
- the optical device 100 according to FIG. 13 basically corresponds to the optical device 100 according to FIG. 12 furnished. In FIG. 13 takes place
- the lighting module 111 has a central recess through which light reflected on the test specimen 150 can reach the detector 114.
- FIG. 14 illustrates aspects related to imaging optical system 100
- Example of FIG. 14 basically corresponds to the example of FIG. 13.
- a beam splitter 111 -5 is used to implement the incident light geometry.
- FIG. 15 illustrates aspects related to the imaging optical system 100.
- the example of FIG. 15 basically the example of FIG. 14.
- the lighting module 111-1 is set up for imaging in incident light geometry; while the lighting module 111-2 is set up for imaging in transmitted light geometry.
- FIG. 16 illustrates aspects related to the imaging optical system.
- the example of FIG. 16 basically corresponds to the example of FIG. 15. In the example of FIG.
- the lighting modules 111 -1, 111 -2 are not implemented by an array of light-emitting diodes; but by means of micromirror devices or DMDs 111-1B, 111-2B, in each case in combination with a single light source 111-1A, 111-2A.
- FIG. 17 illustrates aspects related to the imaging optical system 100.
- the lighting module 111 is not set up for variable-angle lighting and only comprises a single light source 111 -9.
- the sample holder 113 is in turn set up for rotation and Z displacement of the optical test specimen 150.
- the detection optics 112 is set up for angle-variable detection.
- Different filter patterns 380-1, 380-2 in the area of a pupil plane can be used to filter the spectrum of the beam path.
- the various examples described herein may use a correction lens.
- the correction lens can be any suitable lens.
- the correction lens can be any suitable lens.
- the beam path of the light from the illumination module 111 to the detector 114 may be arranged before or after the optical test specimen 150.
- the correction lens it may be possible to compensate for an optical effect of the optical test specimen 150, which can implement a lens, for example.
- the correction lens can have an optical effect that is complementary to the optical test specimen. This makes it possible to correct falsifications due to the refractive power of the optical specimen 15. This makes it possible to reduce falsifications in the images and result images due to the optical effect of the test object.
- (i) and (ii) are performed by separate algorithms.
- a defect detection algorithm could first be used to perform (i); then (ii) subsequently one
- Output of the defect detection algorithm can be compared with a reference error card. Based on the comparison, for example on the basis of a discrepancy between an error card as the output of the defect detection algorithm and the reference error card, (iii) the compliance with the standard can then be checked. In other words, the comparison, for example on the basis of a discrepancy between an error card as the output of the defect detection algorithm and the reference error card, (iii) the compliance with the standard can then be checked.
- Such a KNN can receive one or more of the result images as an input card and have an output card with two or three neurons that correspond to “in tolerance”, “outside tolerance” and optionally “indefinitely”.
- the techniques described herein can be integrated into a production process. For example, an optical imaging system, as described above, could be populated with an optical device under test by a robot arm; then the techniques described herein can be carried out.
- Adjust the focus position of the optical test object Z-stacks of images can be captured.
- the pose of the specimen can be varied with respect to the imaging optical system.
- material testing can be automated.
- Algorithms with machine learning can be used, for example as a defect detection algorithm or in another form.
- an image field of the algorithm can be expanded with machine learning, for example by a parallel atrous convolution and a normal convolution. In this way, size-limited inputs from machine learning algorithms can be adapted to the comparatively extensive raw images and / or
- the machine learning algorithm input would include a sequence of sliding windows. Different windows of the sequence can represent different areas of the at least one result image, with or without overlap. It is then possible that the algorithm output from a combination of results of a sequence of
- Results is determined.
- the sequence of results can correspond to the sequence of sliding windows. Such a technique makes it possible to expand the image field of the algorithm.
- a CNN with the name "U-net” can be used as an example for a KNN, see Ronneberger, Olaf, Philipp Fischer, and Thomas Brox. "U-net: Convolutional networks for biomedical image segmentation.” International Conference on Medical image computing and computer-assisted intervention. Springer, Cham, 2015. Such a CNN U-net can be used by using skip connections
- each raw image and / or each result image with digital contrast could be viewed as a separate channel.
- Such a multi-modal input can be different from a conventional "3-channel red / green-blue" input:
- the goal is the common processing of all Available image information, ie result images and / or raw images.
- Such collaborative processing can be done, for example, by direct
- Concatenation of all channels can be achieved.
Landscapes
- Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Geometry (AREA)
- Quality & Reliability (AREA)
- Investigating Materials By The Use Of Optical Means Adapted For Particular Applications (AREA)
- Testing Of Optical Devices Or Fibers (AREA)
Abstract
Description
Claims
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| DE102018114005.3A DE102018114005A1 (de) | 2018-06-12 | 2018-06-12 | Materialprüfung von optischen Prüflingen |
| PCT/EP2019/064759 WO2019238518A2 (de) | 2018-06-12 | 2019-06-06 | Materialprüfung von optischen prüflingen |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| EP3807838A2 true EP3807838A2 (de) | 2021-04-21 |
Family
ID=66821241
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| EP19729717.9A Pending EP3807838A2 (de) | 2018-06-12 | 2019-06-06 | Materialprüfung von optischen prüflingen |
Country Status (5)
| Country | Link |
|---|---|
| US (1) | US11790510B2 (de) |
| EP (1) | EP3807838A2 (de) |
| CN (1) | CN112243519B (de) |
| DE (1) | DE102018114005A1 (de) |
| WO (1) | WO2019238518A2 (de) |
Families Citing this family (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP4066001B1 (de) | 2019-11-30 | 2026-03-04 | Intrinsic Innovation LLC | Systeme und verfahren zur transparenten objektsegmentierung unter verwendung von polarisationshinweisen |
| US11734814B2 (en) * | 2019-12-12 | 2023-08-22 | Ricoh Company, Ltd. | Enhanced print defect detection |
| FR3108176B1 (fr) * | 2020-03-10 | 2022-09-02 | Data Pixel | Dispositif de détection de défauts de surface sur au moins une face terminale d’au moins une fibre optique. |
| DE102020112496A1 (de) * | 2020-05-08 | 2021-11-11 | Carl Zeiss Smt Gmbh | Verfahren und Vorrichtung zur Inspektion der Oberfläche eines optischen Elements |
| US11900581B2 (en) | 2020-09-22 | 2024-02-13 | Future Dial, Inc. | Cosmetic inspection system |
| US11836912B2 (en) * | 2020-09-22 | 2023-12-05 | Future Dial, Inc. | Grading cosmetic appearance of a test object based on multi-region determination of cosmetic defects |
| CN114648480B (zh) * | 2020-12-17 | 2025-09-05 | 杭州海康威视数字技术股份有限公司 | 表面缺陷检测方法、装置及系统 |
| DE102021110149A1 (de) | 2021-04-21 | 2022-10-27 | Genesys Elektronik Gmbh | Prüfvorrichtung zum Kontrollieren von Bauteiloberflächen und Verfahren dafür |
| DE102021210572B4 (de) | 2021-09-23 | 2024-11-28 | Zf Friedrichshafen Ag | Vorrichtung und Verfahren zum Prüfen eines Prüflings |
| FR3128527B1 (fr) * | 2021-10-21 | 2023-11-17 | Fogale Nanotech | Procédé de caractérisation fonctionnelle d’objectifs optiques |
| CN114061909B (zh) * | 2021-10-29 | 2024-07-12 | 歌尔光学科技有限公司 | 用于测试光机模组的连接装置和光机模组的测试系统 |
| EP4449087B1 (de) * | 2021-12-16 | 2026-04-15 | Alcon Inc. | Verfahren und system zur inspektion einer ophthalmischen linse in einem automatisierten linsenherstellungsverfahren |
| EP4343302A1 (de) * | 2022-09-26 | 2024-03-27 | Essilor International | Computerimplementiertes verfahren zur bestimmung einer lokalisierung mindestens eines defekts in einer geprüften ophthalmischen linse |
| TWI833471B (zh) * | 2022-11-29 | 2024-02-21 | 孟申機械工廠股份有限公司 | 瑕疵檢測方法及其裝置 |
| DE102023204481A1 (de) | 2023-05-12 | 2024-07-25 | Carl Zeiss Smt Gmbh | Verfahren zur automatisierten Defektcharakterisierung von optischen Elementen |
| US11971482B1 (en) * | 2023-08-18 | 2024-04-30 | Mloptic Corp. | Six-dimensional optical position sensor |
| DE102023128512A1 (de) * | 2023-10-18 | 2025-04-24 | Bayerische Motoren Werke Aktiengesellschaft | Verfahren zur Herstellung eines Kraftfahrzeugs |
| US20250156522A1 (en) * | 2023-11-14 | 2025-05-15 | Via Science, Inc. | Certifying camera images |
Family Cites Families (29)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5500732A (en) * | 1994-06-10 | 1996-03-19 | Johnson & Johnson Vision Products, Inc. | Lens inspection system and method |
| DE19739250C2 (de) * | 1996-09-13 | 2003-01-02 | Fraunhofer Ges Forschung | Optische Erfassung von Freiformflächen |
| DE19914115A1 (de) * | 1998-04-20 | 1999-11-04 | Gfai Ges Zur Foerderung Angewa | Verfahren und System zur Fehleranalyse bei polykristallinen Wafern, Solarzellen und Solarmodulen, insbesondere zur Bestimmung der prozeß- und strukturbedingten mechanischen Spannungen |
| EP1148333A1 (de) * | 2000-02-05 | 2001-10-24 | YXLON International X-Ray GmbH | Verfahren zur automatischen Gussfehlererkennung in einem Prüfling |
| US20020106134A1 (en) * | 2000-09-22 | 2002-08-08 | Dundon Thomas A. | Multiple-orientation image defect detection and correction |
| DE102008060621B3 (de) * | 2008-12-05 | 2010-08-12 | Carl Zeiss Ag | Optische Anordnung zum berührungslosen Messen oder Prüfen einer Körperoberfläche |
| DE102011079382B4 (de) * | 2011-07-19 | 2020-11-12 | Carl Zeiss Smt Gmbh | Verfahren und Vorrichtung zum Analysieren und zum Beseitigen eines Defekts einer EUV Maske |
| DE102013003900A1 (de) | 2012-03-28 | 2013-10-02 | Carl Zeiss Microscopy Gmbh | Lichtmikroskop und Verfahren zur Bildaufnahme mit einem Lichtmikroskop |
| JP6164212B2 (ja) * | 2012-05-16 | 2017-07-19 | ソニー株式会社 | 撮像光学系、撮像装置 |
| JP6408259B2 (ja) * | 2014-06-09 | 2018-10-17 | 株式会社キーエンス | 画像検査装置、画像検査方法、画像検査プログラム及びコンピュータで読み取り可能な記録媒体並びに記録した機器 |
| DE102014112242A1 (de) | 2014-08-26 | 2016-03-03 | Carl Zeiss Ag | Phasenkontrast-Bildgebung |
| US10504218B2 (en) | 2015-04-21 | 2019-12-10 | United Technologies Corporation | Method and system for automated inspection utilizing a multi-modal database |
| DE102015208084A1 (de) | 2015-04-30 | 2016-11-03 | Carl Zeiss Microscopy Gmbh | Verfahren zum Generieren eines Kontrastbildes einer Objektbeschaffenheit und diesbezügliche Vorrichtungen |
| DE102015208087A1 (de) | 2015-04-30 | 2016-11-03 | Carl Zeiss Microscopy Gmbh | Verfahren zum Generieren eines reflektionsreduzierten Kontrastbildes und diesbezügliche Vorrichtungen |
| DE102015009728B4 (de) | 2015-07-31 | 2018-10-25 | Airbus Defence and Space GmbH | Probabilistische Fehlererkennung in verlegten Faserbändchen |
| US20170069075A1 (en) * | 2015-09-04 | 2017-03-09 | Canon Kabushiki Kaisha | Classifier generation apparatus, defective/non-defective determination method, and program |
| US10176588B2 (en) * | 2015-09-14 | 2019-01-08 | Sightline Innovation Inc. | System and method for specular surface inspection |
| US10436720B2 (en) * | 2015-09-18 | 2019-10-08 | KLA-Tenfor Corp. | Adaptive automatic defect classification |
| DE102015218917B4 (de) | 2015-09-30 | 2020-06-25 | Carl Zeiss Smt Gmbh | Verfahren zur Ermittlung einer Position eines Strukturelements auf einer Maske und Mikroskop zur Durchführung des Verfahrens |
| CN108463838A (zh) * | 2015-11-16 | 2018-08-28 | 物化股份有限公司 | 增材制造过程中的差错检测 |
| DE102015122712B4 (de) | 2015-12-23 | 2023-05-04 | Carl Zeiss Microscopy Gmbh | Vorrichtung und Verfahren zur Bildaufnahme |
| FR3047451B1 (fr) * | 2016-02-09 | 2019-03-22 | Sncf Reseau | Procede, dispositif et systeme de detection de defaut(s) d’un pantographe d’un vehicule en mouvement sur une voie ferree |
| US11210777B2 (en) * | 2016-04-28 | 2021-12-28 | Blancco Technology Group IP Oy | System and method for detection of mobile device fault conditions |
| WO2017200524A1 (en) | 2016-05-16 | 2017-11-23 | United Technologies Corporation | Deep convolutional neural networks for crack detection from image data |
| US11580398B2 (en) * | 2016-10-14 | 2023-02-14 | KLA-Tenor Corp. | Diagnostic systems and methods for deep learning models configured for semiconductor applications |
| US11047806B2 (en) * | 2016-11-30 | 2021-06-29 | Kla-Tencor Corporation | Defect discovery and recipe optimization for inspection of three-dimensional semiconductor structures |
| DE102016226206A1 (de) | 2016-12-23 | 2018-06-28 | Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V. | System und Verfahren zum Erfassen von Messbildern eines Messobjekts |
| DE102017108873A1 (de) | 2017-04-26 | 2018-10-31 | Carl Zeiss Microscopy Gmbh | Phasenkontrast-Bildgebung mit Übertragungsfunktion |
| US20190318469A1 (en) * | 2018-04-17 | 2019-10-17 | Coherent AI LLC | Defect detection using coherent light illumination and artificial neural network analysis of speckle patterns |
-
2018
- 2018-06-12 DE DE102018114005.3A patent/DE102018114005A1/de not_active Withdrawn
-
2019
- 2019-06-06 CN CN201980039102.5A patent/CN112243519B/zh active Active
- 2019-06-06 WO PCT/EP2019/064759 patent/WO2019238518A2/de not_active Ceased
- 2019-06-06 EP EP19729717.9A patent/EP3807838A2/de active Pending
- 2019-06-06 US US17/251,415 patent/US11790510B2/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| WO2019238518A2 (de) | 2019-12-19 |
| US20210279858A1 (en) | 2021-09-09 |
| WO2019238518A3 (de) | 2020-03-12 |
| US11790510B2 (en) | 2023-10-17 |
| DE102018114005A1 (de) | 2019-12-12 |
| CN112243519A (zh) | 2021-01-19 |
| CN112243519B (zh) | 2024-09-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2019238518A2 (de) | Materialprüfung von optischen prüflingen | |
| EP1488377B1 (de) | Verfahren zur selbstüberwachung eines mikroskopsystems, mikroskopsystem und software zur selbstüberwachung | |
| DE102019114012A1 (de) | Mikroskopieverfahren, Mikroskop und Computerprogramm mit Verifikationsalgorithmus für Bildverarbeitungsergebnisse | |
| EP3186672A2 (de) | Phasenkontrast-bildgebung | |
| EP3608831A1 (de) | Verfahren zur bereitstellung eines auswertungsmittels für wenigstens ein optisches anwendungssystem einer mikroskopbasierten anwendungstechnologie | |
| EP3195250B1 (de) | Verfahren zum erzeugen eines ergebnisbilds und optische vorrichtung | |
| EP3602163B1 (de) | Strukturierte beleuchtung mit optimierter beleuchtungsgeometrie | |
| DE102014206309A1 (de) | System und Verfahren zum Erhalten von Bildern mit Versatz zur Verwendung für verbesserte Kantenauflösung | |
| DE102021205541A1 (de) | Verfahren zur Bestimmung einer Abbildungsqualität eines optischen Systems bei Beleuchtung mit Beleuchtungslicht innerhalb einer zu vermessenden Eintrittspupille | |
| DE102021100444A1 (de) | Mikroskopiesystem und verfahren zum bewerten von bildverarbeitungsergebnissen | |
| EP3071928B1 (de) | Bildaufnahmesimulation in einem koordinatenmessgerät | |
| DE102022121543A1 (de) | Mikroskopiesystem und Verfahren zur Qualitätsüberprüfung eines maschinell gelernten Bildverarbeitungsmodells | |
| DE102019208114B4 (de) | Vorrichtung und Verfahren zur 3D Vermessung von Objektkoordinaten | |
| DE102015210619A1 (de) | Dreidimensional-Formmessapparatur, Messdaten-Verarbeitungseinheit, Messdaten-Verarbeitungsverfahren und Computerprogramm | |
| DE102021101439A1 (de) | Mikroskopiesystem und verfahren zur rotationsüberprüfung einer mikroskopkamera | |
| WO2021008927A1 (de) | Verfahren zur untersuchung von substanzen, computerprogrammprodukt zur verwendung in diesem verfahren, sowie mobiles elektronisches gerät | |
| EP3988989B1 (de) | Verfahren und mikroskop mit einer einrichtung zum erfassen von verlagerungen einer probe gegenüber einem objektiv | |
| WO2019007698A1 (de) | Digitale bestimmung der fokusposition | |
| DE10250503A1 (de) | Mikroskopsystem und Verfahren zur Detektion und Kompensation von Veränderungen eines aufgenommenen Bildinhalts | |
| EP4477995B1 (de) | Verfahren und vorrichtung zur berührungslosen vermessung von objekten | |
| DE102024131391A1 (de) | Verfahren zum Trainieren eines Bildverarbeitungssystems mit einem Maschinenlernmodell zum Ausführen einer virtuellen Mehrwinkel-Rekonstruktion von mit einem Lichtblattmikroskop aufgenommenen Bild-Stapeln | |
| DE102022130975A1 (de) | Lokalisierung von verschmutzungen in mikroskopiesystemen | |
| DE102024111857A1 (de) | Verfahren und Vorrichtung zum Bestimmen einer aktuellen Kalibrierung und/oder Überprüfen einer initialen Kalibrierung eines Kamerasystems eines Mikroskops | |
| EP4231233A1 (de) | Computerimplementiertes verfahren für die schädigungserkennung in einem körper, trainiertes neuronales netz für die schädigungserkennung sowie prüfsystem | |
| DE102022128078A1 (de) | Verfahren und vorrichtung zur mikroskopie |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: UNKNOWN |
|
| STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: THE INTERNATIONAL PUBLICATION HAS BEEN MADE |
|
| PUAI | Public reference made under article 153(3) epc to a published international application that has entered the european phase |
Free format text: ORIGINAL CODE: 0009012 |
|
| STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: REQUEST FOR EXAMINATION WAS MADE |
|
| 17P | Request for examination filed |
Effective date: 20201209 |
|
| AK | Designated contracting states |
Kind code of ref document: A2 Designated state(s): AL AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HR HU IE IS IT LI LT LU LV MC MK MT NL NO PL PT RO RS SE SI SK SM TR |
|
| AX | Request for extension of the european patent |
Extension state: BA ME |
|
| DAV | Request for validation of the european patent (deleted) | ||
| DAX | Request for extension of the european patent (deleted) | ||
| STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: EXAMINATION IS IN PROGRESS |
|
| 17Q | First examination report despatched |
Effective date: 20240510 |