Method for identifying at least one component without a marker and for tracing components in a production process

Abstract

Method for the identification of at least one component without a physical identifier, in which the optical surface structure of at least one part of the component is captured as an image using an image processing system and the unique, characteristic, optical (intrinsic) surface features of the component are determined, and a digital signature is generated as a unique identifier of the latter and this is stored in an electronic data processing system as a reference for the identifier of at least one part of the component, and a number of the unique, characteristic, optical (intrinsic) surface features of the component are extracted and determined from the image data of the image of at least one part of the component via corresponding feature vectors.The system is characterized by the fact that, from the extracted number of intrinsic surface features of the image of at least one part of the component, the signature forming the unique identifier of the component is created using adapted predefined software algorithms and stored as a reference for the unique identifier in the data processing system, and that the captured image of the at least one part of the component is at least partially masked centrally in the form of rectangles, squares, circles, annuli and / or ellipses for the identification areas to be considered, wherein the surface areas for the identification of the individual component types are configured independently of each other by limiting the surface areas for individualization through the masking of the unprocessed areas.

Description

Technical area

[0001] Method for the identification of at least one component without the use of a mark, by means of an image processing system capturing the optical surface structure of at least one part of the component as an image and determining the unique, characteristic, optical (intrinsic) surface features of the component, and generating a digital signature as a unique identification of the latter and storing this digital signature as a reference for the identification of at least one part of the component in an electronic data processing system, a number of the unique, characteristic, optical (intrinsic) surface features of the component are extracted and determined from the image data of the image of at least one part of the component using corresponding feature vectors,from the extracted number of the intrinsic surface features of the image of at least one part of the component, the signature of the latter forming the unique identification of at least one part of the component is created by means of adapted predefined software algorithms and is stored as a reference of the unique identification in the data processing system,

[0002] The invention also relates to a method for tracking (tracing) components in a production process with one or more process steps, wherein the tagless identification method according to one of claims 1 and 2 is used and process parameters and / or production data for the component are assigned to the stored tag of the latter and stored. State of the art

[0003] DE 10 2011 110 478 A1 discloses a method for recognizing a first three-dimensional object. In a data acquisition step, characteristic data of the first three-dimensional object are acquired, compiled, processed, and stored. In an identification step, data from a second three-dimensional object to be identified are acquired, processed, and stored. The stored data of the first three-dimensional object are compared with the stored data of the second three-dimensional object. The following steps are performed: - in the data acquisition step, a one-dimensional surface profile of the first three-dimensional object is acquired by scanning as the first input signal, - in the data acquisition step, the first input signal is converted into first transformed data by a linear frequency domain transformation with several different scalings, - in the data acquisition step, the first transformed data or data generated therefrom are converted into a shift- and scale-invariant representation, thereby generating first shift- and scale-invariant data, - in the identification step, a one-dimensional surface profile of the second three-dimensional object is acquired by scanning as a second input signal, - in the identification step, the second input signal is converted into second transformed data by the same linear frequency domain transformation with several different scalings, - in the identification step, the second transformed data or data generated therefrom are converted into the same shift- and scale-invariant representation, thereby generating second shift- and scale-invariant data, - the first shift and scale invariant data are compared with the second shift and scale invariant data, - if the deviations between the compared data do not exceed a predetermined level, an output signal is emitted which signals an identity between the first three-dimensional object and the second three-dimensional object and otherwise an output signal is emitted which signals no identity between the first three-dimensional object and the second three-dimensional object.

[0004] DE 10 2009 016 289 A1 discloses a method and a device for the mark-free identification of objects, in particular photovoltaic (PV) wafers, as well as for tracking and tracing objects in a production process. For the mark-free identification of an object, unique, characteristic optical features of the optical surface structure of the object are captured as an image by a camera, and the determined features are stored as the object's mark in an electronic data processing device. Likewise, all production data and process parameters for all systems through which the object passes are stored in the electronic data processing system.To track and / or trace an object in a production process with one or more process steps, the previously described method for the identification of the object without a mark is applied, and the process parameters and / or process data for the object in the respective process step are assigned and stored as process step data to the stored mark of the object. If unique, characteristic, optical features of the optical surface structure of the object change during a process step, the changed unique, characteristic, optical features form the current mark of the object after the process step. By repeating the identification of the substrate without a mark based on its unique, characteristic, optical features, a check is carried out after it has passed through a system to determine whether the tracking within a system during a process step is faulty.The item can be stored temporarily between individual process steps.

[0005] The device known from DE 10 2009 016 289 A1 for the identification of an object without a mark, optionally for tracking (or tracing) the object in the manner described above, comprises a camera for capturing and identifying the object on the basis of the unique, characteristic, optical features of the optical surface structure of the object before or in a processing system, an electronic data processing device for storing the unique, characteristic, optical features of the optical surface structure of the object in digital form and for assigning a mark to the object, and a device for tracking and managing the object within the processing system.

[0006] DE 10 2012 205 347 A1 also discloses a method for authenticating and identifying an object with the following steps: Acquiring an initial acquisition image of a selected inspection area of ​​an object surface of the object to generate initial acquisition image data representing characteristic properties of the object surface in the inspection area; Generating third-party image data representing a third-party image different from the initial capture image; Determining first comparison data by a first comparison operation in which the first acquisition image data is compared with the third image data using predetermined comparison methods; Capturing a test image of a test area of ​​an object surface of a test object to be tested in order to generate test image data that represent characteristic properties of the object surface of the test object to be tested; Determining second comparison data by a second comparison operation in which the test image data are compared with the third image data using the predetermined comparison method; Authenticating the test object by comparing the first comparison data with the second comparison data to determine authenticity properties, as well as by evaluating the authenticity properties.

[0007] WO 2012 / 019 614 A1 also discloses a method for identifying a structural design of a product with multiple components as known, wherein the following method steps are carried out: Attaching a contactless readable data carrier to each of the several components and Storing a first information regarding the identity of the respective component on each data carrier and Storing a second piece of information on each data carrier, indicating which of the multiple components the respective component is installed on. Storing the first and second pieces of information involves a contactless transfer of data to the data carrier. The first and / or second pieces of information can each consist of a triple subject-predicate-object.

[0008] In the methods for the identification of a component without a license plate, as described above, a relatively large area of ​​the component is inspected, which can be performed at any position within the camera's field of view. This means that twists, minor positional deviations, and rotations of the component are irrelevant during the identification without a license plate. The component can also be identified without a license plate across partial areas.

[0009] The use of intrinsic surface properties inherent in the object, such as clearly distinguishable features of forged structures or cast surfaces resulting from manufacturing, as well as surface structures characterized by machining traces such as turning marks or grinding marks, form the basis for the identification of components without the need for markings. Scratches, stains, and / or discolorations can also provide distinguishable surface inhomogeneity that can be used for marking-free identification.

[0010] In the automotive industry, components of the same type are machined in a process cycle and assembled with other components in subsequent process steps to form an assembly. The present invention therefore relates in particular to a method and a device for component tracking without a license plate in the automotive industry, especially since previous component tracking methods there are based on additional identification or marking of the individual components. In automotive production, this is usually done by marking the components with a laser (data matrix code; alphanumeric labeling), which requires an additional work step with appropriate component handling and a visual inspection of readability.

[0011] The current method for the mark-free identification of components and for the traceability of components in a production process is considered to be disadvantageous in that, for example, in automotive manufacturing, the identification of millions to billions of components, e.g., gears to be installed, generates an extremely large number and size of signature descriptions or image data for a subsequent similarity search of the component. This data must be stored and handled, transferred, or retrieved in a logistics system. If the data is stored in a central database, appropriate access to the database must be provided for the identification of the components, which is costly and time-consuming. Objective, task and solution

[0012] The present invention therefore aims to provide for a mark-free component identification and component traceability, particularly in the automotive industry, which reduces the amount of data to be handled and transmitted and can ensure higher evaluation speeds.

[0013] The object of the present invention is therefore to provide a method according to the type mentioned at the outset, with which the image data acquisition of the surface structure of the various component types to be identified can be reduced and evaluation speeds can be increased.

[0014] This object is achieved by a method for the identification of at least one component without the use of a license plate according to the type mentioned at the outset, in which the captured image of at least one part of the component is at least partially masked centrally in the form of rectangles, squares, circles, circular rings and / or ellipses for identification areas to be viewed, wherein the surface areas for the identification of the individual component types without the use of a license plate are each configured independently of one another by delimiting the surface areas for individualization by masking the unprocessed surfaces.

[0015] Preferably, the independent configurations of the individual component types are each calculated according to a given image generation algorithm.

[0016] All registrations and identifications occurring in the process chain are based on the masked areas. Masking has the advantage that the entire component no longer needs to be examined, which results in faster evaluation speed for component identification.

[0017] The object of the invention is also achieved by a method for tracing a component in a production process with one or more process steps before its processing or assembly by a) applying the markless identification method according to one of claims 1 and 2, b) Performing a similarity search of the currently generated digital signature among the references stored in the reference database, c) Determining the identification of the component using the maximum detection probability, and d) Assigning and storing process parameters and / or process data to the identifier of the stored component, whereby the corresponding component history is created.

[0018] Preferably, a threshold value is experimentally defined for the recognition probability such that if the recognition probability is below the threshold value when determining the identification of the component, the corresponding components are considered not to be recognized, whereas if the recognition probability is above the threshold value, the database entry with the highest value is output as the identified component.

[0019] The registration and identification algorithms can be checked using different exposure times of the component.

[0020] The present invention also relates to a method for tracking and recognizing components in a production process with several process steps between different production lines by: a) applying the markless identification method according to one of claims 1 and 2 and registering the components after their mechanical processing at the end of the corresponding production line, b) Intermediate storage (buffering) of the registered, processed components in a so-called supermarket c) feeding at least one component temporarily stored in the supermarket to an assembly line of the production process after prior application of the identification method according to one of claims 1 and 2, and d) data-based assignment of the tagless identification of each component fed into the assembly line to an RFID tag of the respective component carrier, e) Reading the RFID tag of the respective component carrier in order to identify the corresponding component via the latter in the assembly line (corresponding production line).

[0021] The independent configurations of the individual component types can be calculated using a predefined image generation algorithm based on generating parameters. The coordinates of the configurations are defined for each component type in a configuration file.

[0022] Masking eliminates interference from outside the areas being viewed during image analysis. These masks should be sufficiently large and positioned as centrally as possible to allow for localization of the areas during subsequent identification. This is particularly important for rotationally symmetrical components, as these must be identified in various rotational positions.

[0023] The prerequisite for tracking and recognizing components in a production process with multiple process steps between different production lines in the manner described above is that the respective component is always imaged from the same side in which it was registered. Existing irregularities on the surface of the components must undergo no or only minimal changes in the period between their registration and their identification without a mark, i.e., contamination, additional processing, and / or corrosion should be avoided.

[0024] Ideally, unprocessed areas of the component's surface are used for registration and mark-free identification. If this is not possible, processed areas of the component's surface with usable processing traces can also be used. According to the method according to the invention, if unique, characteristic optical features change during a process step, the changed unique optical features form the current identification of the component after the process step. Brief description of the drawings

[0025] The present invention will now be explained with reference to the drawings, in which: Fig. 1 a basic flow diagram of the procedure for the identification of at least one component without the use of a mark and the tracing of components in a production process, Fig. 2 a schematic representation of the structure of an exemplary system for the identification of at least one component without a mark and for tracing components in a production process outside the scope of the patent claims, Fig. 3 a schematic representation of a device of the exemplary system according to Fig. 2 for registering and identifying components, Fig. 4 a basic representation of the exemplary system according to Fig. 2, wherein the device according to Fig. 3 is shown in use for both registration and identification, Fig. 5 a representation of a feature extraction of a component using the example of a gear, and Fig. 6 a schematic representation of an embodiment of the method according to the invention, wherein recognition of the components takes place between two production lines. Description of preferred embodiments of the invention

[0026] Fig. Figure 1 illustrates the principle of component identification for component traceability using a system consisting of lighting, a camera, and image processing software. According to block a, component registration is initially performed by using an image processing system to capture the optical surface structure of the component and determine the unique, characteristic, optical (intrinsic) surface features. A digital signature (block b) is generated as a unique identification of the component. The digital signature is generated centrally by image processing. The digital signature of the registered component is stored in a database (block c) of a central computer (block d) as a reference to the latter's identification. The relevant process data (block e) are also stored in the database (block c).

[0027] Fig. Figure 2 schematically shows the structure of an exemplary system for the identification of at least one component without the use of a mark and for tracing components in a production process outside the scope of protection of the patent claims. In a station A for registering the components, each component of the current production line is fully or partially optically recorded using an image processing system. The identification areas to be viewed, such as unprocessed areas of the component surface, are masked (block f). The individual surface features are then determined from the masked image data using corresponding feature vectors. A digital signature is then determined for each component from the features using mathematical algorithms (block g). The signature is stored as a reference (block h) in a database D.

[0028] The identification of the component for its traceability is carried out as Fig. 2, at one or more stations B1 ... B n in the processing and / or assembly line of the production process. According to station A, in each station B 1-nEach component is optically recorded in full or in part (e.g., if obscured) using an image capture system (block i). The component must be within the camera's field of view and may only be partially obscured. Twists or rotations relative to the reference position are permitted. According to station A, the identification areas are at least partially masked (block i). The individual surface features of each component are determined from the masked identification areas, and a corresponding digital signature is generated for each component (block j). A similarity search is then carried out for the currently generated signature among the references in database D (block k). The signature with the highest recognition probability corresponds to the identified component (block 1).

[0029] A fixed threshold prevents components with a lower detection probability from leading to false identifications. If the component is detected, the corresponding manufacturing parameters can be assigned to the reference of the respective component in database D (Block 1).

[0030] The exemplary system described can be used to trace a wide variety of component types.

[0031] As the block k in Fig. 2, configurations of the test settings can also be created and taught in as new components in both station A and station B for the purpose of optically recording their surface structure and at least partially masking the corresponding identification areas to be observed.

[0032] Out of Fig. 3 shows the operation of a device for registering and identifying components of the exemplary system according to Fig. 2. Here, the components are successively guided through a light barrier L and are imaged one after the other by a camera with a lens, to which a corresponding illumination with a homogeneous light field is assigned. The camera is coupled to an image processing system consisting of a computer and a screen. From the imaged optical surface structure of the respective component, the unique, characteristic, optical (intrinsic) surface features of the component are determined by the image processing system and converted into a digital signature of the component as a unique identification of the latter. This digital signature is stored by the image processing system as a reference in the database D ( Fig. 2) is entered for storage

[0033] Fig. 4, which shows a basic representation of the exemplary system according to Fig. 2 illustrates the use of the device according to Fig. 3 both for registering and for identifying the component in the exemplary system according to Fig. 2.

[0034] Out of Fig. Figure 5 illustrates feature extraction using masking according to the inventive method. The captured image of a gear is shown on the left, the gear with masked irrelevant image areas in the center, and the stored masked image of the gear with the cropped area used for registration and identification is shown on the right. The unprocessed surfaces of the component used for evaluation are masked and cropped, which minimizes disruptive influences, for example, during component recognition and increases the evaluation speed. Furthermore, the amount of data stored in the database is reduced.

[0035] The masking of irrelevant image areas of the gear is carried out by first creating the original image with an auxiliary circle for determining the hole position of the gear, then generating the profile line of the auxiliary circle for determining the hole edge position in a pre-processed image, then carrying out the circle search based on the hole edge position using redundancy in case of error detection, and finally saving the masked image of the gear in the batch run.

[0036] Fig. Figure 6 shows a schematic representation of an embodiment of the method according to the invention, wherein a recognition of the components takes place between two production lines. In the upper part of the Fig. 4 shows a mechanical processing line of a production process, along which successive component types are processed and at the end of the mechanical processing line, each one is assembled by means of a device according to Fig. 3 for component registration and then temporarily stored in a so-called supermarket.

[0037] In the lower part of the Fig. Figure 6 shows an assembly line with successive stations: pre-assembly, final assembly, and testing of the production process, to which component types requested from the supermarket are to be supplied. For recognition, the component types requested from the supermarket are first identified at the beginning of the assembly line using the device shown in Fig. 3 subjected to the described component identification and, if recognized, then forwarded to pre-assembly.

Claims

[1] A method for the identification of at least one component without the use of a mark, in which the optical surface structure of at least one part of the component is captured as an image by means of an image processing system and the unique, characteristic, optical (intrinsic) surface features of the component are determined, and a digital signature is generated as a unique identification of the latter and this is stored in an electronic data processing system as a reference for the identification of at least one part of the component, a number of the unique, characteristic, optical (intrinsic) surface features of the component are extracted and determined from the image data of the image of at least one part of the component using corresponding feature vectors,from the extracted number of the intrinsic surface features of the image of at least one part of the component, the signature of the latter forming the unique identification of the component is created by means of adapted predefined software algorithms and is stored as a reference of the unique identification in the data processing system, , characterized by that the captured image of at least one part of the component is at least partially masked centrally in the form of rectangles, squares, circles, circular rings and / or ellipses for identification areas to be viewed, wherein the surface areas for the identification of the individual component types without the need for a mark are each configured independently of one another by delimiting the surface areas for individualization by masking the unprocessed surfaces. [2] Method according to claim 1, characterized bythat the independent configurations of the individual component types are each calculated according to a given image generation algorithm on the basis of generating parameters. [3] Method for tracing a component in a production process with one or more process steps before its processing or assembly by a) applying the markless identification method according to one of claims 1 and 2, b) Performing a similarity search of the currently generated digital signature among the references stored in the reference database c) Determining the identification of the component using the maximum detection probability, and d) Assigning and storing process parameters and / or process data to the identifier of the stored component, whereby a corresponding component history is created. [4] Method according to claim 3, characterized bythat a threshold value is experimentally defined for the probability of recognition in such a way that if the probability of recognition is below the threshold value when determining the identification of the component, the corresponding components are considered not to be recognized, whereas if the probability of recognition is above the threshold value, the database entry with the highest value is output as the identified component. [5] Method for tracking and recognizing components in a production process with multiple process steps between different production lines by: a) applying the markless identification method according to one of claims 1 and 2 and registering the components after their mechanical processing at the end of the corresponding production line, b) Intermediate storage (buffering) of the registered, processed components in a so-called supermarket, c) feeding at least one component temporarily stored in the supermarket to an assembly line of the production process after prior application of the identification method according to one of claims 1 and 2, and d) data-based assignment of the tagless identification of each component fed into the assembly line to an RFID tag of the respective component carrier, e) Reading the RFID tag of the respective component carrier in order to identify the corresponding component via the latter in the assembly line (corresponding production line). [6] Method according to one of claims 1-5, wherein, if unique, characteristic, optical features change during a process step, the changed unique, optical features form the current characteristic of the component after the process step.

Images