Method for identifying a quantity of modules
Patent Information
- Authority / Receiving Office
- EP · EP
- Patent Type
- Applications
- Current Assignee / Owner
- HOFLE ROGER
- Filing Date
- 2024-08-12
- Publication Date
- 2026-04-29
Smart Images
Figure EP2024072698_27022025_PF_FP_ABST
Abstract
Description
[0001] Method for identifying a set of building blocks
[0002] The present invention relates to a method for identifying a set of building blocks, each having a shape and a color.
[0003] A relevant application for such a method is toy kits, especially LEGO. It often happens that the building blocks originally assigned to a kit are found unsorted and mixed in a pile with building blocks from many other kits, and users try in vain to find individual building blocks in the pile in order to assemble a specific object from a kit. A method of this type offers users considerable assistance and can provide the basis for corresponding problems, such as comparing the identified building blocks with a kit parts list or identifying missing building blocks.
[0004] In the current state of the art, the process for recognizing a set of building blocks is generally initiated by scanning the set of building blocks. For example, the set can be photographed using a smartphone or tablet, so that the scan or photo forms the basis for the object recognition process for identifying the shapes and colors. The process for recognizing the shape is possible using various known AI methods or models, in particular machine learning such as deep neural networks, whereby shape recognition already functions relatively reliably.
[0005] However, the problem remains that color recognition has so far been unsatisfactorily resolved. The scans often exhibit highly variable and / or poor quality, primarily due to unfavorable lighting conditions. As a result, the colors of the building blocks are often distorted, cannot be distinguished from one another, and cannot be correctly assigned. This challenge is particularly acute with gray, yellow, blue, or red tones.
[0006] To solve this problem, prior art methods often employ color calibration methods using color targets with reference color values, such as the Munsell target or the IT8 target. Unfortunately, this method requires that a color target be scanned. For example, US 2014329598 A1 proposes arranging the components on a background with color targets, where the background could be a printed cardboard. Similarly, the method of CN 108986173 A requires a single-color calibration board as a background.
[0007] The requirement to provide a color target, a background or a calibration board (in short: color calibration element) in addition to the set of building blocks is extremely impractical and disadvantageous for users.
[0008] The object of the present invention is therefore to provide a method for identifying objects that is improved over the prior art and, in particular, does not require a color calibration element. This object is achieved by a method for identifying a set of building blocks, each with a shape and a color, wherein the following steps are carried out:
[0009] Scanning the set of building blocks and obtaining a set of recognized building blocks, each with an associated color shade, selecting at least one reference block from the set of recognized building blocks by identifying a known shape of a known building block from a set of known building blocks, assigning a known color to the at least one reference block and preferably carrying out a color calibration, in particular by means of a color calibration matrix, using the at least one reference block, assigning known colors to the rest of the set of recognized building blocks based on a known, preferably relative, frequency distribution of the known colors of the known building blocks and / or the results of the color calibration.
[0010] The identification of the building blocks is therefore based on the known frequency distribution of all known colors of all known building blocks and / or on the color calibration.
[0011] In short, according to a central aspect of the invention, color calibration is performed by recognizing the shape of at least one reference brick whose color is known with certainty or with a sufficiently high probability. It should be noted that the recognized bricks are not yet identified initially, but are merely obtained by scanning. During the process, the recognized bricks are assigned to known bricks and thus become identified bricks at the end of the process.
[0012] The great advantage of the invention is that the assignment of the known color of at least one reference brick and further to the rest of the set of recognized bricks takes place directly via the recognized bricks. This means that the color calibration takes place via the bricks themselves and not via an additional color calibration element. Thus, only the set of bricks to be identified needs to be scanned; scanning an additional color calibration element is not necessary.
[0013] Compared to prior art object recognition methods, the method according to the invention represents a significantly simpler option with less effort. Despite the simplicity of the method, reliable color recognition and assignment to the building blocks is at least as reliable.
[0014] Another advantage in this context is that no exact color comparison between a scanned color and a color of a color calibration element is required; by using the known frequency distribution, a very high hit rate in color recognition, in particular of at least one reference stone, is already guaranteed.
[0015] This results in a simplified assignment of a known color to at least one reference brick. Furthermore, a simplified assignment of known colors to the rest of the set of recognized bricks follows based on a known frequency distribution of the known colors of the known bricks and / or the color calibration.
[0016] Depending on the lighting conditions and / or the number of reference bricks and / or the distinguishability of the colors of the detected building blocks, it may be possible to dispense with the frequency distribution and / or color calibration. The worse the lighting conditions and / or the smaller the number of reference bricks and / or the lower the color distinguishability, the more advantageous it becomes to consider the frequency distribution in combination with color calibration.
[0017] The scanning of the set of building blocks is generally carried out using a camera, which creates a digital image of the recognized building blocks and makes it available for further processing. The recognized building blocks are not yet identified; they are only just recognized.
[0018] Preferably, the camera is integrated into a smartphone, tablet or similar, so that the detected building blocks can be further processed using an app.
[0019] The recognized building blocks each have a color shade assigned to the respective building block. The color shades are not yet identified or assigned to a known color; they have only just been recognized.
[0020] The color shades and / or known colors are generally characterized as RGB values. In principle, any known building block that is a building block in the set of recognized building blocks can be selected as a reference block. This means that only building blocks that are actually contained in the set of recognized building blocks can actually be reference blocks.
[0021] Particularly preferably, several, preferably at least four, reference stones are selected. Less preferably, exactly one reference stone is selected.
[0022] The color calibration is carried out using at least one reference stone, wherein preferably at least two, particularly preferably at least four, reference stones are selected.
[0023] It is particularly preferred to have at least four reference stones in four different colour shades.
[0024] Particularly in the case of very large quantities or heaps of building blocks and / or heterogeneous lighting conditions, the heaps can preferably be divided into different areas or sub-areas, wherein preferably in each sub-area at least one reference brick is selected and / or used for color calibration.
[0025] Preferably, two or more reference stones are present in each sub-area; particularly preferably, there are at least four reference stones in different color shades. In this case, color calibration is performed separately for each sub-area.
[0026] Alternatively, the piles can also be manually divided into sub-areas so that the exposure situation is as homogeneous as possible for the entire pile of building blocks.
[0027] Preferably, at least one reference brick is selected whose known shape, in comparison to all known shapes of the known building blocks, has a complex surface contour, i.e. is in particular asymmetrical and / or is formed by a plurality of mathematically describable surfaces merging into one another.
[0028] Furthermore, at least one reference stone is preferably selected whose color shade and / or anticipated known color is a known color with a high probability, i.e. with a probability of more than 50% and / or more than 70% and / or more than 90%.
[0029] The recognized building blocks can also be printed with at least one additional color, whereby the additional color can differ from the color shade assigned to the recognized building block. However, the assigned color shade is preferably used exclusively to carry out the method.
[0030] The well-known yellow man's head (design number 3626) is particularly preferred as a reference brick for a set of LEGO bricks.
[0031] Other possible reference bricks for a set of LEGO bricks include, for example, green trees or plants and / or red roof tiles, since, like the yellow man's head, they are relatively likely to be in the aforementioned colors. Other possible reference bricks are LEGO brooms, stone picks, pickaxes, swords, and / or tires. Other possibilities also exist with regard to well-known road sets (intersection, T-junction, curve) or railway tracks. In principle, however, any known LEGO brick can be used as a reference brick.
[0032] Preferably, there is a list of preferred reference bricks, wherein the list is a defined subset of all known bricks and / or wherein the preferred known bricks are particularly easy to identify.
[0033] Assigning the known color to at least one reference stone is to be understood in particular as a calibration of the reference stone with respect to its color. Advantageously, no additional color calibration element is required for this color calibration.
[0034] The assignment of known colors to the remainder of the set of recognized building blocks, i.e. all recognized building blocks of the set of recognized building blocks less the at least one reference block, is to be understood as a recalculation of the colors taking into account the known frequency distribution of the known colors and / or the color calibration. In other words, each assigned color shade of each recognized building block is assigned to a known color; this is done by there being a tendency for the color shade to match a certain known color due to the known frequency distribution and / or the color calibration.
[0035] In a preferred embodiment, the assignment of known colors to the rest of the set of recognized building blocks is carried out exclusively by taking into account the known frequency distribution of the known colors and the color calibration.
[0036] In another embodiment, the assignment of known colors to the rest of the set of recognized building blocks is carried out exclusively by taking into account the known frequency distribution of the known colors.
[0037] In a further embodiment, the assignment of known colors to the rest of the set of recognized building blocks is carried out exclusively by taking color calibration into account.
[0038] The tendency to match can be considered as actual matching, so that a final assignment to a known color and / or colors is made.
[0039] The tendency towards agreement can also be viewed as a vague agreement or estimate, whereby a first assignment of all color shades of all recognized building blocks to known colors is made, and a current frequency distribution of the recognized building blocks is created, which is compared with the known frequency distribution and, based on this comparison, a correction of the assignment is made. This means that at least a second assignment of all color shades of all recognized building blocks to known colors can be made, which is based on the first or previous assignment.
[0040] The quality of the initial estimate can be improved, in particular, by applying the color calibration result to the color shades of all recognized bricks using at least one reference brick. However, the current frequency distribution of the recognized bricks can be easily determined for informational purposes and does not necessarily have to be used for a corrected assignment of the recognized bricks to known bricks.
[0041] For example, current frequency distributions can be stored at least temporarily and / or used to update the known frequency distribution.
[0042] With evenly distributed, recognized building blocks under ideal lighting conditions, e.g., under laboratory conditions, the color shade would match a known color relatively precisely. In this case, the recorded RGB values of the assigned color shade would correspond to the RGB values of one of the known colors of the known building blocks, so that, theoretically, the known frequency distribution and / or the color calibration results would not need to be used to assign the recognized building blocks to known building blocks.
[0043] Further preferred embodiments and developments of the invention are defined in the dependent claims.
[0044] It is preferably provided that the known frequency distribution of the known colors is updated, preferably continuously and / or regularly.
[0045] The known frequency distribution can be updated through user input, by taking into account current frequency distributions of the sets of recognized building blocks and / or using publicly available data on frequency distributions of known and / or new building blocks.
[0046] It can be provided that each known color of the known building blocks is and / or will be assigned a multitude of color shades.
[0047] The multitude of color shades of the respective known color can, for example, be data which is used in combination with an image recognition process to assign the color shade to a known color.
[0048] It is particularly preferred that the frequency distribution of the known colors, known in particular for the at least one reference brick, be two-dimensional, and that the known colors correlate with known shapes of the known building blocks. In other words, the colors and shapes of the building blocks are known and, in particular for the at least one reference brick, are related to one another through the known frequency distribution.
[0049] It is preferably provided that the selection of the at least one reference brick includes a search for a known building brick with a known color and a known shape.
[0050] Particularly preferably, the selection of at least one reference stone begins with the search for a known building block.
[0051] It is preferably provided that the known color assigned to the known shape using the frequency distribution is present with a probability of over 50%. With LEGO bricks, in particular, the man-heads, trees, plants, roof tiles, and railroad tracks are present with a probability of over 50% as yellow man-heads, green trees, green plants, red roof tiles, and gray railroad tracks.
[0052] It is particularly preferably provided that the assignment of the known colors to the rest of the set of recognized building blocks follows directly or not directly after a color calibration of the known color of the at least one reference brick, wherein the known color serves as a calibration basis, in particular for a color calibration algorithm.
[0053] In principle, various algorithms can be used for color calibration, with the choice of algorithm preferably depending on the number of different color shades or known colors to be assigned. For example, if only one or two color shades or known colors to be assigned are present, the well-known 3x3 diagonal model can be used. If, for example, four or more different color shades or known colors to be assigned are present, the well-known color calibration algorithm by S. can be used.
[0054] Wolf (S. Wolf; "Color Correction Matrix for Digital Still and Video Imaging Systems"; NTIA Technical Memorandum TM-04-406; 2003) can be used. In general, a higher number of reference stones has a positive effect on the quality of color calibration.
[0055] It is particularly preferred that, preferably by means of the two-dimensional frequency distribution and / or the results of the color calibration, the known shapes of the known building blocks are assigned to the remainder of the set of recognized building blocks. It is preferably provided that the assignment of the known shapes takes place in conjunction with an image recognition method, in particular one based on a K1 method. Such image recognition methods are, in and of themselves, state of the art.
[0056] It can be provided that the known shapes of the known building blocks are each assigned and / or will be assigned a plurality of perspective representations of the respective known shapes.
[0057] The multitude of perspective representations of a respective known shape can, for example, be data which is used in combination with an image recognition method in order to assign the recognized shape to a known shape.
[0058] It can be provided that absolute dimensions, such as length, width and height, and / or relative dimensions are and / or are assigned to the known shapes, wherein the dimensions can be used to assign a recognized shape to a known shape.
[0059] It can be provided that at least one grouping of color-identical and / or shape-identical and / or color- and shape-identical recognized building blocks takes place into at least one group, preferably wherein at least one count of the recognized building blocks takes place within a group. In this way, the recognized building blocks can be sorted and an overview of the set of recognized building blocks, in particular of an unsorted pile of building blocks, can be created. In this case, the recognized building blocks are fully identified building blocks, i.e. both their shape and color are identified.
[0060] It is preferably provided that a list is created from the recognized or identified building blocks, wherein the recognized building blocks are each identified by a known color and known shape and have a number.
[0061] In this case, it may be provided that the list is compared with at least one assembly instruction and / or parts list of a kit containing building blocks and is checked for identical color, shape, and number of building blocks. This can be used, for example, to determine the extent to which a selected kit is available and / or which building blocks are missing to complete the selected kit.
[0062] It can further be provided that the list of the at least one parts list is assigned to at least one kit if the list is a complete subset of the at least one parts list. This makes it possible to show which kits can be fully assembled using the identified components.
[0063] It is also conceivable that identified missing components from a parts list of a known kit are consolidated and, for example, issued to users.
[0064] It is also possible for missing modules to be reprocured automatically. Users can also reprocure the identified missing modules themselves. It is also conceivable that functionally equivalent replacement modules will be suggested for the missing modules.
[0065] Since the process makes it possible to clearly distinguish between known colors and known shapes, even assemblies that have already been assembled and / or appear to be optically homogeneous can be segmented in such a way that known building blocks within this assembly can be identified.
[0066] A preferred variant of the method according to the invention is described below using an example:
[0067] - first, a pile of building blocks is scanned so that scanned or recognized building blocks are obtained;
[0068] - four reference stones are selected from the scanned or recognized building blocks and / or found among the scanned or recognized building blocks: e.g. a yellow head, a red roof tile, a gray stone pick and a green tree;
[0069] - the respective RGB values of the above-mentioned reference bricks are determined in the scan and compared with the respective RGB values of the known original bricks;
[0070] - the required color calibration matrix is determined from the difference between the RGB value pairs (original building blocks vs. scan of the reference blocks);
[0071] - Color calibration using the required color calibration matrix improves the match between the respective RGB values of the scanned reference bricks and the RGB values of the original bricks, although an exact match will only be achieved in very rare cases; - the color calibration results are applied to the RGB values of the remaining recognized bricks and converted;
[0072] - the converted RGB values are compared with the known frequency distribution so that all recognized building blocks are assigned known colors and are thus identified;
[0073] - optionally, a current frequency distribution of the identified building blocks is determined and compared with the known frequency distribution; if there is an approximate match, the process is terminated; if there are major differences between the current and the known frequency distribution, further calculations and / or recalibration can be carried out.
[0074] Further advantages and details of advantageous variants and
[0075] Further developments of the invention are apparent from the figures and the associated descriptions. They show:
[0076] Fig. 1-3 a schematic explanation of an application of the method according to the invention,
[0077] Fig. 4 is a schematic representation of the first
[0078] Process step of the process according to the invention,
[0079] Fig. 5a-5c a set of recognized building blocks under different lighting conditions,
[0080] Fig. 6a-6c a schematic representation of the second process step of the process according to the invention,
[0081] Fig. 7-8 schematic representations of the third
[0082] Process step of the process according to the invention,
[0083] Fig. 9 shows a known frequency distribution, Fig. 10 shows a schematic representation of the third and fourth method steps of the method according to the invention,
[0084] Fig. 11 is a schematic representation of a
[0085] Further development of the method according to the invention,
[0086] Fig. 12 shows a schematic explanation of a related application of the method according to the invention,
[0087] Fig. 13-16 schematic representations of further developments of the method according to the invention,
[0088] Fig. 17 a recognized assembly,
[0089] Fig. 18 is a further schematic representation of the fourth method step of the method according to the invention and
[0090] Fig. 19a-19c schematic representations of a
[0091] Further development of the method according to the invention.
[0092] Figs. 1-3 show a schematic explanation of an application of the method according to the invention: Construction kits 6, in particular for toys such as LEGO, are usually delivered with assembly instructions 8 for assembling one or more objects 14. Initial assembly is supported by dividing the construction kits 6 into specific subsets of the construction kits 12 for assembling partial objects 13 and preparing them in packaging (see Fig. 1). Complete objects 14 are then assembled from the partial objects 13 (see Fig. 2).
[0093] The construction sets 6 are designed so that they can be taken apart and reassembled at a later time. Since many users mix their construction sets 6 together, thereby creating a confusing set of building blocks 1, there are several difficulties when reassembling one or more objects 14, see Fig. 3. For example, when there are several construction sets 6 of a similar type (e.g. LEGO City), it is particularly difficult to assign the set of building blocks 1 to specific objects 14. However, the problem also exists with very complex individual construction sets 6 or when known building blocks 5 are lost.
[0094] Fig. 4 shows a schematic representation of the first method step of the method according to the invention, wherein the set of building blocks 1 is scanned and a set of recognized building blocks 2 is obtained, wherein each recognized building block 2 is assigned a color shade 3.
[0095] Figs. 5a-5c each show the same set of recognized building blocks 2 with color shades 3 assigned to recognized building blocks 2 under different lighting conditions. This schematic representation is intended to illustrate that the problem lies in the fact that the quality of the scans can be very different or poor, which is why it is difficult to differentiate and assign the colors with sufficient accuracy. In the case of Figs. 5a-5c, the situation is further complicated by the fact that the recognized building blocks 2 are located in a pile in which the building blocks overlap one another.
[0096] Figs. 6a-6c each show a schematic representation of the second method step of the method according to the invention under different exposure ratios. In this second method step, at least one reference brick 4 is selected from the set of recognized building blocks 2, wherein the selection is made by identifying a known shape of a known building block 5 from a set of known building blocks 5. In the present exemplary case, which is also retained in the following figures and figure descriptions, the cylindrical recognized building block 2 is selected as the reference brick 4.
[0097] This recognized building block 2 and / or reference block 4 can have different color shades 3 depending on the lighting situation (cf. Fig. 5a-5c). At the same time, the recognized building block 2 and / or reference block 4 can have one of the known colors from a list of known colors or a color list for a known building block 5, cf. Fig. 7.
[0098] Preferably, a color list is available for each known building block 5, with which the recognized color of the recognized building block 2 and / or the reference block 4 can be compared.
[0099] The reference brick 4 , which is also a recognized building brick 2 , can be, for example, a LEGO head of a LEGO man, cf. Fig. 8 .
[0100] Fig. 9 shows a known, preferably relative, frequency distribution 7, which represents the frequency of a known building block 5 and thus also of the recognized building block 2 and / or reference block 4 in the known colors.
[0101] Preferably, such a frequency distribution 7 is known for each known building block 2.
[0102] The frequency distribution 7 can be compared with the color shades 3 of the recognized building blocks 2 and / or the reference brick 4. Fig. 10 shows a schematic representation comprising the third and fourth method steps of the method according to the invention. In the third step, the known color is first assigned to the reference brick 4. In the fourth step, further known colors are assigned to the remainder of the set of recognized building blocks 2 based on the known, preferably relative, frequency distribution 7 of the known colors of the known building blocks 5 and / or the results of the color calibration.
[0103] This means that from the initially recognized building blocks 2, identified building blocks 10 are created through a successful comparison with known building blocks 5.
[0104] The assignment of the known colors to the rest of the set of recognized building blocks 2 can be carried out directly or indirectly on the basis of a calibration of the known color of the at least one reference block 4, wherein the known color serves as a calibration basis, in particular for a color calibration algorithm.
[0105] In a particularly preferred variant of the method, the known frequency distribution 7 of the known colors is a two-dimensional frequency distribution 9 and the known colors correlate with known shapes of the known building blocks 5, in particular wherein the known shapes of the known building blocks 5 can be assigned to the rest of the set of recognized building blocks 2.
[0106] Fig. 11 shows a schematic representation of a further development of the method according to the invention, wherein the selection of the reference brick 4 includes a search for a known building brick 5 with a known color and a known shape, preferably wherein the known color assigned to the known shape by means of the frequency distribution 7 is present with a probability of more than 50%.
[0107] This search for an anticipated reference brick 4 refers to the set of bricks 1 , but is essentially carried out on the basis of the recognized bricks 2 .
[0108] Fig. 12 shows a schematically illustrated explanation of a related application of the method according to the invention: In the case of very large quantities of building blocks 1, for example of many construction sets 6 with many building instructions 8 and in particular in the form of an unsorted pile, a large number of recognized building blocks 2 are obtained, wherein the problem of "the needle in the haystack" merely generates a second, analogous problem of "a second needle in a second haystack". Compare in this regard to the large number of highlighted recognized building blocks 2 in Fig. 12, which in turn is confusing.
[0109] To enable users to grasp the corresponding recognized building blocks 2 as quickly and efficiently as possible, the limitations of human perception must be taken into account and information must be provided in such a way that users can use it appropriately. Schematically shown in Fig. 13, the assembly instructions 8, in conjunction with an underlying parts list 11, are preferably segmented into individual subsets of the kit 12.
[0110] The data obtained, in particular the segmentation obtained, can be processed in a database 16. Then, by means of the method according to the invention, all recognized building blocks 2 of the set of building blocks 1, in particular those relevant for the assembly of a specific object 14, are recognized (represented in Fig. 13 by the fact that both dashed and dotted building blocks in the set of building blocks 1 are highlighted). In contrast, however, only the recognized building blocks 2 of a single subset of the kit 12 are shown on a smartphone display (in this case, only the relevant recognized building blocks 2 of subset A of the kit 12 are shown on the smartphone display).
[0111] The subsets of the kit 12 can be reasonably limited so that they form a self-contained unit, e.g., a wall, a roof, or similar, according to the assembly instructions 8. If necessary, further subsets or partial objects 13 can be formed.
[0112] Fig. 14 schematically shows a further development of the method according to the invention, which aims to further improve information acquisition for users. For this purpose, found or relevant recognized building blocks 2 can be successively removed from the original overall overview or set of recognized building blocks 2 after scanning the set of building blocks 1 and after they have become identified building blocks 10.
[0113] In Fig. 14, all identified building blocks 10 of a sought-after subset of the kit 12 are highlighted on the display of the smartphone shown on the left. After an identified building block 10 has been found or spatially located as part of the subset of the kit 12 during scanning (smartphone in the middle), the found relevant recognized building block 2 is no longer displayed in the overall overview of the set of recognized building blocks 2 (smartphone on the right-hand side of Fig. 14). This allows the user's perception to be continuously focused and the findability of the remaining recognized building blocks 2 to be improved.
[0114] According to the schematic representation of Fig. 15, it is also conceivable to indicate to users at any time, for example by means of pie charts, what proportion of found or detected identified building blocks 10 is in the subset of a kit 12 and / or of the entire kit 6.
[0115] The detected, identified building blocks 10 can be separated, already separated, or still be in the unsorted pile or in the set of building blocks 1, wherein the set of building blocks 1 can consist of various assemblies 15 and / or various kits 6 and / or parts thereof. Users can thus estimate the probability that a specific object 14 of a kit 6 can be completed.
[0116] According to the schematic representation of Fig. 16, a comparison of the identified building blocks 10 with the parts list of a kit 11 for the assembly of at least one object 14 can be carried out, wherein the data can be processed in a database 16 and / or a check can be carried out for building blocks that are identical in color and shape and identical in number.
[0117] Preferably, the comparison is performed using a list created from the recognized building blocks 2, wherein the recognized building blocks 2 are each identified by a known color and known shape and have a number. For example, it can be determined which and / or how many known building blocks 5 might need to be purchased as part of a replacement purchase. The purchase can be automated and / or manual.
[0118] Fig. 17 shows an assembly 15 consisting of a set of recognized building blocks 2 with associated different color shades 3. Here, the recognized building blocks 2 are in an assembled, form-fitting state.
[0119] Fig. 18 shows a further schematic representation of the fourth method step of the method according to the invention, wherein known colors are assigned to the recognized building blocks 2 of an assembly 15 on the basis of the color calibration by means of the reference block 4.
[0120] Figs. 19a-19c show schematic representations of a further development of the method according to the invention, wherein, by means of known dimensional ratios or relative dimensions, e.g., length:width:height, assemblies 15 can be segmented into their individual recognized building blocks 2. Thus, assemblies 15 that appear visually homogeneous can be segmented to such an extent that the presence of known building blocks 5 in such an assembly 15 can be deduced.
[0121] Fig. 19a is intended to illustrate that the boundaries between the detected building blocks 2 are not easily recognizable, for example due to poor scan quality. Fig. 19b is intended to illustrate that the boundaries can be identified using the known dimensional ratios and / or color calibration. Fig. 19c is intended to illustrate the situation when the boundaries of the detected
[0122] Building blocks 2 represent . Key to the reference numbers:
[0123] (set of) building blocks
[0124] (amount) of detected
[0125] Building blocks
[0126] Assigned
[0127] Color shading
[0128] Reference stone
[0129] Known building block(s)
[0130] Kit / Kits
[0131] Acquaintance
[0132] Frequency distribution
[0133] Assembly instructions
[0134] Two-dimensional
[0135] Frequency distribution
[0136] Identified building blocks
[0137] Parts list of at least one kit
[0138] Subset of at least one kit
[0139] Subobj ect ( s )
[0140] Ob j ect
[0141] Assembly (s)
[0142] database
Claims
Patent claims 1. A method for identifying a set of building blocks (1) each having a shape and a color, wherein the following steps are carried out: Scanning the set of building blocks (1) and obtaining a set of recognized building blocks (2), each with an associated color shade (3), selecting at least one reference block (4) from the set of recognized building blocks (2) by identifying a known shape of a known building block (5), assigning a known color to the at least one reference block (4) and preferably carrying out a color calibration using the at least one reference block (4), assigning known colors to the rest of the set of recognized building blocks (2) based on a known, preferably relative, frequency distribution (7) of the known colors of the known building blocks (5) and / or the results of the color calibration.
2. Method according to claim 1, wherein the known frequency distribution (7) of the known colors is updated, preferably continuously and / or regularly.
3. Method according to one of the preceding claims, wherein each known color of the known building blocks (5) is and / or is assigned a plurality of color shades.
4. Method according to one of the preceding claims, wherein the known frequency distribution (7) of the known colors is a two-dimensional frequency distribution (9) and the known colors correlate with known shapes of the known building blocks (5).
5. Method according to one of the preceding claims, wherein selecting the reference brick (4) includes a search for a known building brick (5) with a known color and a known shape.
6. Method according to one of the preceding claims, wherein the known color assigned to the known shape by means of the frequency distribution (7) is present with a probability of more than 50%.
7. Method according to one of the preceding claims, wherein the assignment of the known colors to the rest of the set of recognized building blocks (2) follows a color calibration of the known color of the reference brick (4), wherein the known color serves as a calibration basis, in particular for a color calibration algorithm.
8. Method according to one of the preceding claims, wherein, preferably by means of a two-dimensional frequency distribution (9) according to claim 4, the known forms of the known building blocks (5) are assigned to the rest of the set of recognized building blocks (2).
9. Method according to the preceding claim, wherein the assignment of the known shapes is carried out in conjunction with an image recognition method.
10. Method according to one of the preceding two claims, wherein the known shapes of the known building blocks (5) are each assigned and / or are assigned a plurality of perspective representations of the respective known shape.
11. Method according to one of the preceding claims, wherein absolute and / or relative dimensions are and / or are assigned to the known shapes.
12. Method according to one of the preceding claims, wherein at least one grouping of color-identical and / or form-identical and / or color- and form-identical recognized building blocks (2) takes place into at least one group, preferably wherein at least one counting of the recognized building blocks (2) takes place within a group.
13. Method according to one of claims 8 to 12, wherein a list is created from the recognized building blocks (2), wherein the recognized building blocks (2) are each identified by a known color and known shape and have a number.
14. Method according to the preceding claim, wherein the list is compared with at least one parts list of at least one kit (11) comprising building blocks and checked for color, shape and number identical building blocks.
15. Method according to the preceding claim, wherein the list is assigned to the at least one parts list of the at least one kit (11) if the list is a complete subset of the at least one parts list of the at least one kit (11).