Method and device for identifying an electronic component

Abstract

The invention relates to a method for identifying an electronic component, in which at least one image, in particular a photograph, of the component is processed by means of image analysis. According to the invention, by means of the image analysis, components of the assembly are identified and a component type of at least individual of the components is determined. Suitably, different image analyses are carried out, and the results of the different image analyses are correlated to one another in order to identify components, determine component types and / or identify components. In one design of the invention, at least one of the image analyses is carried out by means of an artificial neural network which is trained to identify a component in the component and / or to determine which component type the respective component belongs to.

Description

Technical Field

[0001] The present invention relates to a method for identifying electronic components, wherein at least one image, particularly a photograph, of the component is processed by means of image analysis.

[0002] Furthermore, the present invention relates to a computer program product for implementing the method and an apparatus for identifying electronic components. Background Technology

[0003] Methods for identifying electronic components are well-known due to their existing applications. However, their application is severely limited by the low quality of the identification results. Summary of the Invention

[0004] The objective of this invention is to provide a method of the type described at the beginning, by means of which electronic components can be better identified.

[0005] The task is solved according to the present invention as follows: by means of image analysis, the components of the component are identified and the component type of at least some of the components is determined.

[0006] The previously mentioned measures are used to identify components arranged on carriers of electrical wires, such as circuit boards or films equipped with conductors. A second measure is used to determine the type of individual, multiple, or / and all components. This acquired understanding of the component's structure enables particularly accurate identification of the component.

[0007] In a particularly preferred design of the invention, different image analyses are performed, and the results of these different image analyses are correlated with each other in order to identify components, determine component types, and / or identify components. The accuracy of identification can be greatly improved by utilizing the different image analysis results and their correlation. Advantageously, incorrect image analysis results can be identified particularly easily, and correct image analysis results can be verified, if necessary, by stepwise comparison of the individual image analysis results.

[0008] In one embodiment of the invention, image analysis is performed or implemented at least once, preferably multiple times, and particularly preferably all image analyses using an artificial neural network. The artificial neural network is trained to identify components in a component and / or determine the type of a corresponding component. Suitably, the artificial neural network is trained to identify whether the corresponding component is an integrated switching circuit, a transistor, a resistor, a sensing component, a transformer, or another component type.

[0009] Advantageously, the function of a corresponding component in an assembly can be determined by knowledge of individual or multiple component types. By gaining knowledge of the corresponding function, the assembly, and especially its function, can be inferred.

[0010] Appropriately, determine the number of components in the assembly.

[0011] In another embodiment of the invention, the position of a corresponding component within the assembly is determined. This position determination may include a local arrangement of the components within the assembly, particularly on a conductor carrier. Suitably, this position determination includes the arrangement of the corresponding component relative to one or more other components arranged within the assembly. The arrangement may, for example, include the distance between the components.

[0012] In one design of the present invention, the proportions of the dimensions, especially the heights, of the components are determined.

[0013] Advantageously, the function of a component whose position is determined can be inferred by determining its positional and / or dimensional proportions. Furthermore, the recognition of the position of components relative to each other enables the identification of components.

[0014] In a particularly preferred design of the present invention, at least one feature of a corresponding component identified by the image analysis is compared with feature data of a component feature dataset in which the component features are assigned to a determined component type, and preferably, the identification of the component and / or the determination of the component type are performed by comparing one or more identified features with the feature data of the component feature dataset.

[0015] The characteristics of the corresponding components may include, for example, resistance, impedance, power, voltage, especially input voltage and / or output voltage, supply voltage, current intensity, number of windings, housing type, housing identification, manufacturer, manufacturer abbreviation, input type, interface, configuration, number of A / D converters, number of input terminals, pin size, number of pins, structure type, structure form, inductance, rated current, and diameter.

[0016] The feature dataset suitably exists as a database, which preferably includes multiple component features and / or assembly features. Suitably, the component feature dataset also includes the manufacturing cost and / or price of the components, information about the material composition of the components, and / or information about alternative and / or compatible components that can be used to replace or replace the corresponding components.

[0017] Suitably, at least one feature of a component identified by the image analysis is compared with feature data in a component feature dataset, in which component features and / or component features are assigned to the identified component, and preferably, the component is identified by comparing the identified features with the feature data in the component feature dataset. Suitably, the component feature dataset also includes the component's manufacturing cost and / or price, information about the component's material composition, and / or information about alternative and / or compatible components that can be used to replace or replace the corresponding component.

[0018] By comparing the identified characteristics of components and / or assemblies with component feature datasets and / or assemblies feature datasets, components can be explicitly configured using a relatively small number of features. As the number of identified features increases, the probability of correct component identification and / or assemblies labeling increases.

[0019] To identify components and / or to determine component types and / or assemblies, it is suitable to determine the color and / or other surface characteristics of the corresponding components or assemblies. Identification by color and / or surface characteristics is particularly well-suited because, in particular, the coloring of components and, if necessary, assemblies or parts thereof is predetermined, especially standardized. This determination is preferably carried out in consideration of color codes. Consideration of surface characteristics has proven particularly advantageous, primarily because some components in an assembly, especially components, possess characteristic surface characteristics. For example, they may have a defined metallic or alloy, plastic, glass, or ceramic metallic surface.

[0020] In one embodiment of the invention, the shape and / or size of the corresponding component is determined in order to identify the component and / or to determine the component type, and / or the shape and / or size of the component is determined in order to identify the assembly. The determination of shape and / or size has proven particularly effective for determining the component type because components typically used in assemblies not only have characteristic shapes but also characteristic dimensions, and these shapes and dimensions are often standardized. Thus, for example, coils, capacitors, diodes, resistors, or the like can be identified by image analysis using their shapes, and further, the corresponding component properties, such as inductance, capacitance, or resistance, can be determined by their corresponding dimensions. This determination is preferably made considering predetermined shapes and / or sizes, and more preferably considering standards regarding component shapes and / or sizes.

[0021] In a particularly preferred embodiment of the invention, the height of the components, especially with respect to the carrier, is determined.

[0022] In another design of the invention, in order to identify components and / or determine component types, the electrical contact closure of the corresponding components in the assembly is determined, wherein preferably the number and / or arrangement of electrical connectors, especially pins, on the components is determined. While the determination of electrical contact closure has also proven useful for identifying transistors, resistors, inductors, or transformers, it is particularly advantageous for identifying integrated switching circuits. Identifying these components is indeed particularly challenging because they are mostly housed within a housing and their components are often unidentifiable or only poorly identifiable in images. Conversely, the number and / or arrangement of electrical connectors, especially pins, allows for a fairly wide range of corresponding switching circuit types.

[0023] Suitably, in order to identify components and / or determine component types, markings arranged on and / or within the corresponding components are determined. This determination is preferably performed when text recognition and / or mark recognition are implemented, and if necessary, when analyzing the meaning of the text or other markings determined by text recognition. Suitably, the mark determination is used to determine trademarks, manufacturers, component codes or component codes, component types and / or component types, or other characteristics, by which components or components can be assigned to a determined matching dataset of component feature datasets and / or component feature datasets.

[0024] In one design of the present invention, an artificial neural network is specifically trained to recognize text and symbols commonly used to label components and parts.

[0025] In a further embodiment of the invention, at least one feature of one or more components and / or assemblies identified by the image analysis, preferably multiple identified features, is compared with feature data in a component feature dataset and / or a component feature dataset, wherein component features are assigned to a determined component in the component feature dataset. Preferably, the identification of components and / or assemblies is performed by comparing the identified features with the feature data in the component feature dataset and / or component feature dataset.

[0026] In one embodiment of the invention, the probability of determining the correctness of an image analysis result is considered for each defined feature or image analysis result. Suitablely, this probability is assigned to and preferably associated with the image analysis result. When associating image analysis results, it is preferable to weight the image analysis results using corresponding probabilities.

[0027] In a preferred embodiment of the invention, multiple images of the image analysis processing component are used. It is understood that the multiple images can image different aspects of the component, such as top views, rear views, and / or different side views of the component, particularly the equipped conductor carrier. Suitably, the multiple images image the component, particularly segments of the component, preferably segments of one or more components, from different viewing angles.

[0028] The image can be constructed using a single camera or, if necessary, multiple cameras simultaneously, such as a multi-camera array.

[0029] In a particularly preferred design of the invention, the images are preferably constructed using a light field imaging device, and more preferably using a light field camera. Advantageously, these images can be used to obtain 3D information about components and / or assemblies, which in particular enables the determination of the dimensions, especially the height, of the components and / or assemblies as described above. Furthermore, the analysis of multiple images of the same object from different viewing angles, as well as the analysis of corresponding multiple images of the same object, provides the possibility of verifying and merging the image analysis results for the same object. This increases the probability that the generated image analysis results are correct.

[0030] In one embodiment of the invention, at least some individual images of the image are generated using different filters, such as polarizing filters, ultraviolet blocking filters, color filters, infrared blocking filters, neutral density filters, edge filters, interference filters, Bayer filters, complementary color filters, and / or fluorescence filters. By using images generated with these filters, features of components and / or assemblies can be identified better or even for the first time in some cases. For example, text recognition can be significantly improved when using at least one polarizing filter, preferably different polarizing filters. Furthermore, the probability of finding the correct result can be significantly improved, especially when using neural networks.

[0031] In one design of the invention, the light field imaging device has at least one mirror, preferably multiple mirrors, and / or at least one prism, preferably multiple prisms, for generating an image on a receiving surface. The one or more mirrors and / or the one or more prisms are arranged in an optical unit such that multiple images of an object region can be generated on the receiving surface. Various, fundamentally different arrangements of the mirrors and / or prisms are possible for this purpose. It has proven particularly advantageous that the light field imaging device is configured to have multiple mirrors arranged such that different light beams entering in the incident opening pass through the light field imaging device without reflection according to the optical path, or are reflected (multiple times if necessary) on one or more mirror surfaces before leaving the light field imaging device again.

[0032] In a particularly preferred embodiment of the invention, the light field imaging device has a kaleidoscope. Suitably, such a kaleidoscope has at least one pair of planar mirrors, wherein the mirrors are arranged facing each other and spaced apart from each other. At least a portion, preferably all, of the light path extends through the space between the mirrors. Preferably, the mirrors are arranged parallel to each other. The kaleidoscope may have two or more pairs of mirrors. The pairs of mirrors can form a tube, which is polygonal, preferably rectangular, in cross-section. Alternatively, the kaleidoscope may be formed from a cylindrical glass rod that is polygonal in cross-section, the glass rod having silvered sides and end faces for light beam exit and incident. The glass rod preferably has an isosceles triangular, rectangular, especially square, regular pentagonal, hexagonal, heptagonal, or octagonal shape in cross-section.

[0033] Suitablely, the mirrors and / or prisms are configured such that different images are acquired from different viewing angles, presenting the object region. Thus, a light field camera can be formed using the optical unit.

[0034] In one design of the present invention, an imaging system is used for recording light field images, the imaging system having a plurality of imaging units arranged sequentially along the optical axis. A first imaging unit generates a true intermediate image of the object region in an intermediate image plane, a second imaging unit having the light field imaging device generates at least one virtual mirror image of the true intermediate image, the virtual mirror image being arranged offset from the true intermediate image in the intermediate image plane, and a third imaging unit generates a common image of the true intermediate image and the virtual mirror image as a true image on an image receiving surface to be arranged at an axial distance from the intermediate image plane.

[0035] Suitablely, image correction is performed on a common image that is the real image, the real intermediate image, and the at least one virtual mirror.

[0036] Preferably, the specific image correction is performed on the real intermediate image and the virtual image, and particularly preferably on each virtual image in the virtual image.

[0037] In a particularly preferred embodiment of the invention, the light field imaging device is arranged in an objective lens, the objective lens having at least one lens, the lens preferably forming the first imaging unit.

[0038] In a preferred embodiment of the invention, at least one light incident lens, preferably a plurality of light incident lenses forming a light incident lens system, is arranged in front of the optical unit along the optical path direction. Preferably, the light field imaging device forming the second imaging unit is suitably configured such that the light beam entering the light field imaging device is split according to its direction, so that these beams detect the object region from slightly different viewing angles, but still present the same object region.

[0039] The light incident lens is suitably configured such that it images the object region onto the input plane at the end of the optical unit facing the light incident lens.

[0040] In one embodiment of the invention, at least one light-emitting lens, preferably multiple light-emitting lenses forming a light-emitting lens system, is arranged after the optical unit when viewed along the optical path direction. The light-emitting lenses form the optical path, allowing the image to be generated onto the receiving surface in accordance with the invention. The one or more light-emitting lenses preferably form the third imaging unit.

[0041] Suitablely, the light-emitting lens system is configured such that the focal plane of the light-emitting lens system is the same as the input plane of the optical unit, especially the kaleidoscope.

[0042] It is understood that the successive arrangement of lenses and mirrors or prisms in the image can lead to potentially complex image errors that are difficult or impossible to correct using conventional image correction methods. Even a slight deviation from the intended position or shape of one of the lenses, mirrors, or prisms can result in the unwanted detection of different portions of the object region in the image or cause different imaging errors, such as distortion, between images.

[0043] In a particularly preferred embodiment of the invention, N×N images of the target area are formed on the receiving surface and generated side-by-side in an N×N grid, where N preferably represents an odd number. Suitably, the N×N images are formed such that these images, in particular, completely or at least almost completely occupy the readable portion of the receiving surface. These images are preferably arranged side-by-side vertically and horizontally in the grid.

[0044] Suitablely, the generation of the image enables the imaging of the object region from different viewing angles.

[0045] Particularly preferably, nine images of the object region are formed on the receiving surface, said images being generated in a 3×3 grid. Alternatively, for example, 25 images of the object region may be generated in a 5×5 grid, or 49 images may be generated in a 7×7 grid. It is understood that, in order to increase the number of achievable viewing angles, a larger number of images and a corresponding grid arrangement structure may also be provided.

[0046] The optical unit, particularly the objective lens, is described in WO2014 / 124982A1. This application incorporates the content of WO2014 / 124982A1 by reference. See specifically the text on page 7, paragraph 1 to page 11, paragraph 2, and... Figure 1 See Figure 3. A suitable light field imaging device is described there, and its working principle is explained.

[0047] Suitably, the objective lens has a housing in which the at least one lens and the light field imaging device are arranged. The objective lens preferably has a unit for mechanically fastening it to the camera housing, such as an objective lens thread or an objective lens mount. Furthermore, the objective lens may have a unit for electrical or electronic connection and / or data transmission with the camera housing.

[0048] Suitably, the receiving surface has at least one image acquisition sensor or is formed by at least one image acquisition sensor. In a preferred embodiment of the invention, the receiving surface is formed by a single image acquisition sensor. The image acquisition sensor is preferably a CCD sensor or a CMOS sensor.

[0049] Suitably, the imaging system has a unit for optical filtering, by means of which the real intermediate image and / or at least one virtual mirror image can be filtered separately from each other. Advantageously, this enables the filtering of at least individual images among those generated side-by-side in an N×N grid.

[0050] In a further embodiment of the invention, the artificial neural network is trained using multiple images of training components and / or training devices, wherein the images are preferably constructed using a light field imaging device. Suitably, a dataset, particularly a training dataset, is formed for a machine learning system used for digital image processing, particularly for image classification and / or object recognition and / or segmentation. Preferably, at least one set of digital images of the same acquired object, particularly at least one component and / or device, from different viewing angles is used to form the dataset.

[0051] In one design of the present invention, the artificial neural network is trained using a component feature dataset and / or a component feature dataset. Specifically, the artificial neural network is trained to associate the image with the component feature dataset and / or the component feature dataset, and particularly to assign identified component features and / or component features to the component feature dataset and / or component feature dataset.

[0052] In one embodiment of the invention, the artificial neural network is trained to identify components by means of identified components and component feature datasets and / or component feature datasets, and in particular to determine the function of the components.

[0053] Suitablely, the machine learning system is a deep learning system, such as a convolutional neural network. Convolutional neural networks are well known for their use in visual image analysis. They are primarily used in image and video recognition, image classification, and medical image analysis. Furthermore, the machine learning system can be a deep neural network system, a deep belief network system, or a recurrent neural network system.

[0054] In a particularly preferred embodiment of the invention, the machine learning system is configured for classification, segmentation, and / or object recognition.

[0055] Suitablely, multiple sets of digital images of the same object being detected, especially at least one component and / or part, are used, wherein at least two sets, preferably all sets, differ in the detection perspective.

[0056] Preferably, the dataset includes a set of real-world labeled images and / or a set of labeled images, which are provided for comparing and / or evaluating the performance of image processing, especially during training.

[0057] Alternatively, or additionally, unsupervised or semi-supervised learning models can also be used.

[0058] In a preferred embodiment of the invention, the method includes training a machine learning system using a training dataset for digital image processing, particularly classification, segmentation, and / or object recognition.

[0059] The present invention also relates to a trained artificial neural network that has been trained as explained above, and to the application of said artificial neural network for identifying components and / or parts.

[0060] The present invention also relates to a computer program product having instructions that, when executed by a computer, cause the computer to perform the methods described above.

[0061] The computer program product is suitably a computer program stored on a data carrier, preferably RAM, ROM, CD or the like, or a device, especially a personal computer, a device including an embedded processor, a computer embedded in a device, a smartphone, a device for constructing image recordings, especially a camera and / or camcorder, or a sequence of signals representing data suitable for transmission via a computer network, especially the Internet. The invention also relates to a data carrier signal that transmits the computer program product.

[0062] The present invention also relates to an apparatus for identifying electronic components, the apparatus having a unit for data processing, the unit having mechanisms for performing the methods described above. The apparatus suitably includes an imaging system and / or camera as described above, particularly a photographic and / or video camera. It is understood that the imaging system and / or camera are interconnected such that the constructed image is preferably automatically transmitted to the unit for data processing for image analysis and, if necessary, for identifying the component.

[0063] The data processing device appropriately utilizes the trained neural network.

[0064] This invention can be used, for example, in the following applications:

[0065] - Cost determination: The cost of a component can be determined by using component identification or component identification and associating it with a dataset of component features or component features that includes manufacturing costs and / or prices, as explained above.

[0066] - Quality control of electronic components, especially for monitoring equipment integrity, proper positioning of components, and the condition of components and / or electrical connectors, such as solder joints.

[0067] - Assembly of electronic components, especially checking whether the gripper used to assemble the components has accepted the set components. This is particularly relevant when components are supplied in bulk, as in these cases it is necessary to check whether the correct components have been retrieved, whether the components are intact, and where the gripper has retrieved the components, in order to assist the robot in the precise positioning of the components.

[0068] -Recycle:

[0069] a) Sorting components to be recycled using the markings according to the invention.

[0070] b) Separate the components and parts according to the components arranged on them.

[0071] c) Determining the value of the components to be recycled using the identifier according to the invention.

[0072] d) Identify components suitable for reuse as secondhand items.

[0073] - Identifying alternative components: If repairs are necessary, especially for older components, and not all components are available, it may be necessary to identify alternative components. This can be achieved by associating a database containing potential alternative components with component identifiers. Attached Figure Description

[0074] The invention is further explained below with the aid of embodiments and accompanying drawings relating to those embodiments. Wherein:

[0075] Figure 1 The device according to the invention is illustrated schematically; and

[0076] Figure 2 The electronic components are shown schematically. Detailed Implementation

[0077] Figure 1 An apparatus 1 according to the invention is shown, comprising a light field camera 2 for generating images from different viewing angles, a computer 3 connected thereto, and an imaging chamber 4, wherein a component 10, to be identified by means of apparatus 1, may be arranged in the imaging chamber. The camera 2 may be configured to generate images filtered using different filters.

[0078] Computer 3 is equipped with a computer program that can process images acquired using light field camera 2 by means of image analysis. The computer program includes at least one neural network trained to identify electronic components 12-20 and / or electronic components 10. A database 5 is provided, which the software can access, and the database contains component feature datasets and / or component feature datasets.

[0079] exist Figure 2 The electronic component 10 shown includes a circuit board 11, components namely resistors 12 and 13, coils 14 and 15, integrated circuits 16 and 17 with pins 18 and 19 for electrical contact connection, and capacitors 20 and 21 arranged on the circuit board and interconnected by electrical wires provided in the circuit board 11 (not shown here). For identification purposes, component 10 is arranged in the imaging chamber 4. This can be done, for example, by means of a conveyor belt, a robotic arm, or by hand, placing the component into the imaging chamber.

[0080] Images of component 10 are generated from different viewing angles, and if necessary, from different sides of component 10, using light field camera 2. For example, some of the images substantially show a top view, a rear view, and / or different side views of component 10 from multiple viewing angles. The images undergo one or more different image analyses, which are explained below and implemented using artificial neural networks trained for the respective image analyses.

[0081] A. Detection components.

[0082] B. Determine the spatial arrangement structure on circuit board 11 for each component.

[0083] C. For each component, determine whether it is marked, in particular text and / or markings. If markings are found, determine for each component whether the corresponding component type can be determined by means of the markings.

[0084] D. Determine the color and / or surface characteristics of each component.

[0085] E. Determine the shape and / or size of each component. In particular, 3D information about the corresponding component can be determined using depth information, which can be obtained by using images acquired by a light field camera from different viewing directions.

[0086] F. Determine the electrical contact connection of each component, wherein preferably the number and / or arrangement of electrical connectors, especially pins 18 and 19, on the components, particularly on the respective integrated circuits 16 and 17.

[0087] G. For each component, determine its spatial arrangement on circuit board 2 relative to other components.

[0088] In order to identify according to Figure 2 Component 10 detects components 12-20 on circuit board 11 for each image and determines the position of each component 12-20 on circuit board 11.

[0089] In this example, only integrated circuits 16 and 17 are marked, but it is understood that other components 12-15, 20, and 21 may also be marked. A mark containing textual components is placed on integrated circuit 16 and encoded. Integrated circuit 17 only has an identification code. The mark and encoding of the generated circuit 16 are identified and the associated information is stored. Furthermore, the identification code of integrated circuit 17 is identified and stored.

[0090] When determining the color of components 12-20, colors are identified on resistors 12 and 13, coils 14 and 15, and capacitors 20 and 21, with the colors corresponding to color codes and stored accordingly. When determining the surface characteristics of components 12-20, it is identified whether components 12-20 or their housings are made of metal, plastic, or ceramic, and this determined information is stored accordingly. The shape and size of the components are determined using each image in the images. To obtain 3D information, information acquired from different viewing angles is processed in a known manner, so that the height of components 12-20 can also be determined using depth information. The corresponding features identified are also stored. Furthermore, the electrical contact closure of components 12-20 is determined using the images. This can be achieved using electrical wires visible on the circuit board. For integrated circuits 16 and 17, the corresponding number and position of pins 18 and 19 are determined.

[0091] For each image analysis performed, feature information corresponding to the corresponding determined feature is stored, the probability and / or confidence value, especially the confidence interval, of the correctness of the corresponding determined feature is determined, and entries corresponding to the obtained feature information are stored in association with the probability or confidence value. One feature, or if necessary, multiple feature pieces of information, are used to determine which component type and component model are involved. This is achieved, if necessary, by using a neural network trained for this purpose through comparison with the component feature dataset.

[0092] Furthermore, the probability of correctly identifying the corresponding component type is determined by comparing the image analysis results of different images collected from different observation angles.

[0093] In another step, the identified features of components 12-17 are compared with the component feature dataset and thereby assigned to the component, especially component types with defined specifications.

[0094] In order to assign the identified components to the assembly, another neural network can be set up, which uses the identified components to determine the type of the assembly, especially the assembly equipped with the components.

[0095] The method according to the invention can be used to determine the value or cost corresponding to the identified components. For this purpose, the typical price or price range corresponding to the respective components 12-20 can be stored in a component feature dataset.

Claims

1. A method for identifying electronic components, wherein at least one image, particularly a photograph, of the component is processed by means of image analysis, characterized in that, The components of the component are identified by means of the image analysis, and the component type of at least some of the components is determined.

2. The method according to claim 1, characterized in that, Different image analyses are performed, and the results of the different image analyses are correlated with each other in order to identify the components, determine the type of the components, and / or identify the components.

3. The method according to claim 1 or 2, characterized in that, At least one image analysis is performed using an artificial neural network trained to identify components in the component and / or determine the type of the corresponding component, preferably determining whether the corresponding component is an integrated switching circuit, transistor, resistor, inductive component, transformer, or other component type, wherein the specifications of the corresponding component are preferably determined.

4. The method according to any one of claims 1 to 3, characterized in that, In order to identify the component and / or to determine the type of the component, the color and / or other surface characteristics of the corresponding component are determined, wherein the determination is preferably made in consideration of color codes.

5. The method according to any one of claims 1 to 4, characterized in that, In order to identify the component and / or to determine the type of the component, the shape and / or size of the corresponding component are determined, wherein the determination is preferably made in consideration of a predetermined shape and / or size, and preferably in consideration of standards regarding the shape and / or size of the component.

6. The method according to any one of claims 1 to 5, characterized in that, In order to identify the component and / or to determine the type of the component, the electrical contact of the corresponding component in the assembly is determined, wherein preferably the number and / or arrangement of the electrical connectors on the component, in particular the pins, is determined.

7. The method according to any one of claims 1 to 6, characterized in that, In order to identify the component and / or to determine the type of the component, the location of the corresponding component in the assembly is determined.

8. The method according to any one of claims 1 to 7, characterized in that, In order to identify the component and / or to determine the type of the component, a mark arranged on or / and in the corresponding component is determined, wherein the determination is preferably performed when text recognition and / or mark recognition is implemented, and if necessary, when analyzing the content of the text determined by means of the text recognition.

9. The method according to any one of claims 1 to 8, characterized in that, The features of at least one of the corresponding components identified by the image analysis are compared with the feature data of the component feature dataset, in which the component features are assigned to a determined component type, and preferably, the identification of the component and / or the determination of the component type are performed by comparing the identified features with the feature data of the component feature dataset.

10. The method according to any one of claims 1 to 9, characterized in that, The features of at least one of the components identified by the image analysis are compared with the feature data of the component feature dataset, in which the component features are assigned to the determined components, and preferably, the identification of the components is performed by comparing the identified features with the feature data of the component feature dataset.

11. The method according to any one of claims 1 to 10, characterized in that, For each image analysis result, determine the probability that the image analysis result is correct, assign the probability to the image analysis result, and weight the image analysis result with the corresponding probability when associating the image analysis results.

12. The method according to any one of claims 1 to 11, characterized in that, Multiple images of the component are processed using the image analysis method.

13. The method according to any one of claims 1 to 12, characterized in that, The multiple images image the component from different viewing angles, and the images are preferably constructed using a light field imaging device, and particularly preferably using a light field camera.

14. The method according to any one of claims 1 to 13, characterized in that, The multiple images image different sides of the component.

15. The method according to any one of claims 1 to 14, characterized in that, The artificial neural network is trained using multiple images of training components and / or training devices, wherein the images are preferably constructed using a light field imaging device, and particularly preferably using a light field camera.

16. A computer program product, comprising instructions that, when executed by a computer, cause the computer to perform the method according to any one of claims 1 to 14.

17. The computer program product according to claim 16, characterized in that, The computer program product is a computer program stored on a data carrier, preferably RAM, ROM, CD or the like, or a device, especially a personal computer, a device including an embedded processor, a computer embedded in a device, a smartphone, a device for constructing image recordings, especially a camera and / or camcorder, or a sequence of signals representing data suitable for transmission via a computer network, especially the Internet.

18. A data carrier signal, wherein the data carrier signal transmits the computer program product according to claim 16.

19. An apparatus for identifying electronic components, comprising a unit for data processing (3), said unit having a mechanism for performing the method according to any one of claims 1 to 15.

20. The apparatus according to claim 19, characterized in that, The device (1) includes a light field imaging system (2).

Images