WO2024185071A1 - Component identification device, component identification method, and component identification program - Google Patents

Abstract

An input unit (105) inputs an image containing a first component. A feature amount extraction unit (107) extracts, for the first component, a feature amount in disposition of an element forming a component from the image. A similarity calculation unit (108) reads, from a database storing a plurality of pieces of identification information, which is information for identifying components and a plurality of feature amounts in association with each other for second components, a feature amount related to the second component and compares the read feature amount to the feature amount extracted by the feature amount extraction unit to calculate a similarity. The output section (110) uses the similarity to identify a similar component, which is a component similar to the first component, from among the second components, and then reads and outputs the identification information related to the similar component from the database.

Description

PART IDENTIFICATION DEVICE, PART IDENTIFICATION METHOD, AND PART IDENTIFICATION PROGRAM

This disclosure relates to a part identification device, a part identification method, and a part identification program.

　There is a known design support system that makes product design more efficient by using design data for parts that were previously developed. For example, Patent Document 1 discloses a design support system that has a database that stores the correspondence between part functions and the shapes that realize those functions, making it easy to search for and extract shape data that corresponds to required functions.

JP 2008-21159 A

When designing a product to replace equipment with a new one, it is sometimes desirable to design it with a shape similar to that of the existing equipment. For example, when replacing an existing control panel implemented with hardware in operation at a factory, etc., with a control panel implemented with a graphic screen simulated by software, it is advantageous for the layout of the elements that make up a part such as a meter (for example, scales and numerical display) to be similar to that of the original part, so that users can start using it without feeling uncomfortable after the replacement. However, while the above-mentioned conventional technology makes it possible to extract shape data for parts corresponding to functions, there is a problem in that the user needs to select appropriate shape data to achieve a layout similar to that of the original part, which increases the design load.

This disclosure has been made to solve the above-mentioned problems, and aims to show parts that have a similar arrangement of elements to the original parts.

The part identification device according to the present disclosure includes an input unit to which an image including a first part is input, a feature extraction unit that extracts, for the first part, feature amounts in the arrangement of elements that constitute the part from the image, a similarity calculation unit that reads out feature amounts for the second part from a database that stores a plurality of identification information, which is information for identifying parts, and feature amounts in association with each other for second parts, and calculates a similarity by comparing the feature amounts with the feature amounts extracted by the feature extraction unit, and an output unit that uses the similarity to identify similar parts that are similar to the first part from among the second parts, and reads out the identification information for the similar parts from the database and outputs it.

The part identification device according to the present disclosure extracts features of the arrangement of elements that make up a part from an image of a first part, reads the features from a database that stores multiple features related to a second part, calculates a similarity by comparing the extracted features, and identifies and outputs similar parts from among the second parts using the similarity. This makes it possible to show parts that are similar to the first part in terms of the arrangement of elements.

FIG. 2 is a hardware configuration diagram of the part identification device according to the first embodiment. FIG. 2 is a functional configuration diagram of the part identifying device according to the first embodiment. FIG. 2 is a diagram showing an original part and its elements according to the first embodiment. FIG. 2 is a diagram showing the data structure of a feature DB according to the first embodiment. 4 is a flowchart showing the operation of the part identifying device according to the first embodiment. FIG. 13 is a diagram showing the data structure of a feature DB according to the second embodiment. FIG. 13 is a diagram showing a linear meter according to a second embodiment and its elements. FIG. 11 is a functional configuration diagram of a part identifying device according to a second embodiment.

Embodiment 1.
Hereinafter, the first embodiment will be described in detail with reference to the drawings. In the first embodiment, an example will be shown in which a part identification device identifies parts (hereinafter, referred to as similar parts) similar to parts (hereinafter, referred to as original parts) constituting a control panel before replacement, and outputs the results. In other words, a user can grasp the similar parts by viewing the information output by the part identification device, and can select which parts will actually replace the parts constituting the control panel before replacement.

The original part corresponds to the first part in the claims.

FIG. 1 is a hardware configuration diagram of a part identification device according to the first embodiment. The part identification device 100 includes a processor 101, a memory 102, an input/output IF 103 (interface), and a bus 104. The processor 101, the memory 102, and the input/output IF 103 are connected to each other via the bus 104 so as to be able to communicate with each other.

The processor 101 may be, for example, a CPU (Central Processing Unit), a DSP (Digital Signal Processor), a GPU (Graphical Processing Unit), or an FPGA (Field-Programmable Gated Array).

The memory 102 is, for example, a static random access memory (SRAM), a dynamic random access memory (DRAM), or a read-only memory (ROM). If the memory capacity of the memory 102 alone is insufficient, the part identification device 100 may include an auxiliary storage device (not shown) as necessary. The auxiliary storage device is, for example, a hard disk drive (HDD) or a solid state drive (SSD).

The input/output IF 103 is an interface for inputting and outputting information from outside the part identification device 100. For example, a mouse, a keyboard, a display, an external memory, and a network cable are connected to the input/output IF 103.

FIG. 2 is a functional configuration diagram of part identification device 100 according to the first embodiment. Part identification device 100 includes input unit 105, type identification unit 106, feature extraction unit 107, similarity calculation unit 108, output unit 110, and feature DB 109 (Data Base). Part identification device 100 accepts input of an image of the control panel before replacement via input unit 105, and outputs information indicating parts whose element arrangement is similar to that of the original parts constituting the control panel before replacement from output unit 110.

The input/output IF 103 is implemented as an input unit 105 and an output unit 110. The processor 101 is implemented as a type identification unit 106, a feature extraction unit 107, and a similarity calculation unit 108. The memory 102 is implemented as a feature DB 109.

The input unit 105 accepts input of an image of the control panel before replacement. Note that the image input to the input unit 105 may be an image of the original component alone, rather than an image of the control panel.

The type identification unit 106 recognizes the original part shown in the image received by the input unit 105, and identifies the elements that make up the original part and the type of the original part.

FIG. 3 is a diagram showing an original part and its elements according to the first embodiment. Original part 111 is composed of needle 111a, scale 111b, numeric display 111c, center circle 111d, and display board 111e. Here, type identification unit 106 is preset to identify the type of a part as an analog meter when the part is composed of a needle, scale, numeric display, center circle, and display board. In other words, type identification unit 106 recognizes that original part 111 shown in the image is composed of five elements, namely needle 111a, scale 111b, numeric display 111c, center circle 111d, and display board 111e, and identifies the type of original part 111 as an analog meter.

Returning to the explanation of FIG. 2, the feature extraction unit 107 extracts features related to the arrangement of elements for the original part 111 recognized by the type identification unit 106.

The similarity calculation unit 108 calculates the similarity by comparing the features between the original part 111 and the candidate part. Here, the candidate part is a part that is selected as a part that constitutes the control panel after replacement, and a plurality of candidate parts are prepared in advance. The similarity calculation unit 108 uses the feature DB 109 when making the comparison.

FIG. 4 is a diagram showing the data structure of the feature DB 109 according to the first embodiment. The feature DB 109 stores, for three candidate parts whose type is an analog meter (part IDs 112 are A, B, and C, respectively), part IDs 112 and feature values 113 in association with each other. The part IDs 112 correspond to identification information that is information for a user or device to identify each part. The identification information may also be the name of the part or a thumbnail image of the part. The feature DB 109 is not limited to the example of the first embodiment, and may store identification information and feature values 113 in association with each other.

For ease of understanding, the feature DB 109 in FIG. 4 is shown storing only three candidate parts whose type is analog meter, but in reality, the feature DB 109 also stores candidate parts whose type is other than analog meter, with part IDs and feature values associated with each other. In the example of FIG. 4, the feature values 113 indicate feature values for three points: the position of the center circle, the distance between the scale and the numerical display, and whether or not the scale is outside the numerical display.

For example, the feature 113 of a candidate part whose part ID 112 is A indicates that the position where the central circle is located is coordinate (20, 15). Here, coordinates indicate a position based on a predetermined origin. In the first embodiment, the origin is the top left pixel of the image of the cut-out part, and coordinates (20, 15) indicate a point that is moved 20 pixels to the right and 15 pixels down from the origin. In this way, the feature 113 may be a coordinate value that indicates where within the part an element is located. The position of the origin is arbitrary, and may be, for example, the center of the part. The unit of the coordinates is also arbitrary, and may be, for example, a relative value based on the dimensions of the part (e.g., vertical or horizontal width).

Furthermore, the feature amount 113 of the candidate part whose part ID 112 is A indicates that the distance between the scale and the numerical display is 5. In this way, the feature amount 113 may be the relative distance between elements.

Furthermore, the feature amount 113 of the candidate part whose part ID 112 is A indicates that whether or not the scale is outside the numerical display is "True." In this way, the feature amount 113 may indicate the relative positional relationship between elements. Furthermore, the feature amount 113 may be a multiple-choice value instead of a numerical value.

In FIG. 4, the feature 113 of the candidate part whose part ID 112 is B indicates that the position of the center circle is (20, 23), the distance between the scale and the numerical display is 7, and whether or not the scale is outside the numerical display is "False." Additionally, the feature 113 of the candidate part whose part ID 112 is C indicates that the position of the center circle is (20, 17), the distance between the scale and the numerical display is 2, and whether or not the scale is outside the numerical display is "False."

In the present disclosure, it is assumed that the part identification device 100 stores the feature amount DB 109 in advance. The feature amount DB 109 may be created by a user, or may be generated by artificial intelligence such as machine learning.

The similarity calculation unit 108 compares the features of the original part 111 and the candidate parts to calculate the similarity. The candidate parts that the similarity calculation unit 108 compares are of the same type as the original part 111. Since the type of the original part 111 is an analog meter, three candidate parts with part IDs 112 of A, B, and C are compared. The similarity calculation unit 108 finds the difference between the feature values of the original part 111 and the feature values 113 of the candidate parts, and calculates a higher similarity the smaller the difference is.

For example, the feature amount of the original part 111 indicates that the position of the center circle is (20, 18), the distance between the scale and the numerical display is 2, and whether or not the scale is outside the numerical display is "False". Also, consider an example in which a comparison is made with a candidate part with a part ID of A in FIG. 4. The similarity calculation unit 108 calculates that the difference in the position of the center circle is (20, 18) - (20, 15) = (0, 3). Here, the difference in the position of the center circle is a two-dimensional value, and since it cannot be added to other one-dimensional differences as it is, it may be converted to one dimension by calculating the norm as necessary. For example, |0| + |3| = 3 is used as the one-dimensional difference in the position of the center circle. Furthermore, the difference in the distance between the scale and the numerical display is calculated to be 2 - 5 = -3. Furthermore, regarding whether or not the scale is outside the numerical display, if the values match, the difference is calculated to be 0, and otherwise the difference is calculated to be 1. In other words, in this example, the difference is calculated to be 1. The similarity calculation unit calculates the absolute value of each difference and adds them together to determine the overall difference. In other words, the overall difference is |3|+|-3|+|1|=3+3+1=7.

In the above example, the differences between multiple feature amounts are simply added up, but the differences between each feature amount may be multiplied by a predetermined coefficient before being added up. This allows you to set the degree to which each of the multiple feature amounts is weighted when calculating the similarity.

Returning to the explanation of FIG. 2, the output unit 110 uses the similarity calculated by the similarity calculation unit 108 to identify similar parts, which are parts similar to the original part 111, from among the candidate parts. Specifically, the output unit 110 determines that a candidate part whose similarity exceeds a predetermined threshold value is a similar part. The output unit 110 reads out identification information related to the similar parts from the feature DB 109 and outputs it.

An example of outputting identification information related to similar parts using similarity is described below. For example, the output unit 110 may output identification information related only to similar parts as a search result. This ensures that the parts output as a search result are similar parts, and the user can easily select similar parts. The output unit 110 may also output a list of similar parts associated with similarity as a search result. This allows the user to understand the similarity of the similar parts and easily recognize similar parts with high similarity. The output unit 110 may also display the search result so that parts with high similarity stand out. For example, the output unit 110 may list similar parts in descending order of similarity. Alternatively, the output unit 110 may output similar parts together with candidate parts that are not similar parts, while displaying only similar parts in a way that makes them stand out (such as making the text bold or red, making the background color fluorescent, or surrounding them with a red line). This makes candidate parts with higher similarity more noticeable to the user, and the user can easily recognize candidate parts with higher similarity.

FIG. 5 is a flowchart showing the operation of the part identification device 100 according to the first embodiment.

The input unit 105 accepts input of an image of the control panel before replacement (step S101), and proceeds to step S102.

The type identification unit 106 recognizes the original part 111 shown in the image received by the input unit 105, identifies the elements that make up the original part 111 and the type of the original part 111 (step S102), and proceeds to step S103.

The feature extraction unit 107 extracts features related to the arrangement of elements for the original part 111 (step S103), and the process proceeds to step S104.

The similarity calculation unit 108 selects one candidate part of the same type as the original part 111 from among the candidate parts stored in the feature DB 109 (step S104), and proceeds to step S105.

The similarity calculation unit 108 calculates the similarity between the original part 111 and the candidate part selected in step S104 by comparing the features (step S105), and proceeds to step S106. The similarity calculation unit 108 obtains the features 113 of the candidate part by reading them from the feature DB 109.

The similarity calculation unit 108 determines whether the calculation of similarities has been completed for all candidate parts of the same type among the candidate parts stored in the feature DB 109 (step S106), and if it is determined that the calculation is completed, proceeds to step S107, otherwise proceeds to step S104.

The output unit 110 uses the similarity calculated by the similarity calculation unit 108 to identify similar parts from among the candidate parts, reads out and outputs identification information related to the similar parts from the feature DB 109 (step S107), and ends the flowchart in FIG. 5.

As described above, the part identification device 100 according to the first embodiment extracts, for a first part, feature amounts in the arrangement of elements constituting a part from an image, reads out feature amounts 113 for the second part from a database that stores a plurality of identification information, which is information for identifying parts, and feature amounts 113 in association with each other for a second part, calculates a similarity by comparing it with the extracted feature amounts, identifies similar parts that are similar to the first part from among the second parts using the similarity, and reads out and outputs the identification information for the similar parts from the database. In this way, the part identification device 100 can indicate parts that are similar to the first part in terms of element arrangement, and the user can select parts that are highly similar to the first part in terms of element arrangement.

The first part is the original part 111 that constitutes the control panel before replacement, and the second part is a candidate part that is a part that is a candidate for selection as a part that constitutes the control panel after replacement realized on the graphic screen. This allows the part identification device 100 to indicate parts that are similar to the original part 111 that constitutes the control panel before replacement, and to assist in replacement with the control panel realized on the graphic screen.

Furthermore, the feature amount 113 related to the candidate part and the extracted feature amount indicate the position of the elements within the part, or the distance between the elements, or the positional relationship between the elements. This allows the part identification device 100 to indicate, as a similar part, a part that is similar to the original part 111 in terms of the position of the elements within the part, or the distance between the elements, or the positional relationship between the elements.

In the first embodiment, an example has been shown in which the user views the information output by the output unit 110, but the information output by the output unit 110 according to the present disclosure can also be input information to another device. The other device is, for example, a device that automatically replaces a control panel by automatically selecting similar parts. As a simple example, the other device can automatically replace a control panel by selecting the part with the highest similarity from among the similar parts indicated in the information output by the part identification device 100.

Embodiment 2.
In the second embodiment, a method of replacing parts to give a sense of unity to the entire control panel will be described.

6 is a diagram showing the data structure of the feature DB 109 according to the second embodiment. The difference from the first embodiment is that a group ID 114 is added. The group ID 114 is a value for grouping candidate parts with similar element arrangements. A candidate part with a component ID of A is associated with group 1 as the group ID 114, a candidate part with a component ID of B is associated with group 2 as the group ID 114, and a candidate part with a component ID of C is associated with group 3 as the group ID 114. Although only three candidate analog meter parts are shown in FIG. 6, in reality, group IDs 114 are also assigned to candidate parts of other types (for example, linear meters, which will be described later), and analog meters and linear meters that are similar to each other in terms of element arrangement are assigned the same group ID 114. In other words, if there are two original parts, an analog meter and a linear meter, in the control panel before replacement, candidate parts associated with the same group ID 114 are preferentially output as candidate parts to replace the two, thereby making it possible to replace the control panel with a sense of unity as a whole.

FIG. 7 is a diagram showing a linear meter and its elements according to the second embodiment. The linear meter 116 is composed of a meter 116a, a meter 116b, a scale 116c, and a numerical display 116d. Here, one of the features of the linear meter 116 is the distance between the scale 116c and the numerical display 116d. This feature can be considered to be the same as the feature 113 relating to the distance between the scale 111b and the numerical display 111c in the analog meter. Therefore, the feature of the distance between the scale and the numerical display is compared with each other, and the same group ID 114 is associated with an analog meter and a linear meter 116 whose difference is sufficiently small (for example, less than a predetermined threshold value). In this way, even if the parts are of different types, there may be some common elements and comparable features. In such a case, the features are compared with each other, and if the difference is sufficiently small, the same group ID 114 is associated.

FIG. 8 is a functional configuration diagram of the part identification device 100 according to the second embodiment. The difference from the first embodiment is that a group judgment unit 115 is added. Consider the replacement of a certain control panel. The group judgment unit 115 collects the group IDs 114 of the identified similar parts in the replacement of the control panel. The group judgment unit 115 performs the above-mentioned process for each of the multiple original parts that make up the control panel before replacement. The group judgment unit 115 counts the distribution of the collected group IDs 114 and judges the group ID that is suitable for the control panel before replacement. For example, the group judgment unit 115 may count the group IDs 114 of the similar parts with the highest similarity for each original part, and judge that the group ID 114 that is the most common as a result of the count is the most suitable for the control panel before replacement. This allows a majority vote to be taken on an original part basis, and the group IDs 114 that are considered suitable can be narrowed down to one. Note that the group determination unit 115 according to the present disclosure is not limited to the above operation, and may extract multiple group IDs 114 suitable for the control panel before replacement. This allows multiple groups to be presented to the user from the part identification device 100, allowing the user to have a wider range of parts to choose from. In this manner, the group determination unit 115 determines the group ID 114 suitable for the control panel before replacement, and passes the group ID 114 to the output unit 110.

The output unit 110 outputs information indicating candidate parts corresponding to group ID 114 suitable for the control panel before replacement. Specifically, the output unit 110 may output candidate parts corresponding to group ID 114 suitable for the control panel before replacement in addition to outputting similar parts as described in the first embodiment. Alternatively, the output unit 110 may output only those of the identified similar parts that correspond to group ID 114 suitable for the control panel before replacement.

As described above, the part identification device 100 according to the second embodiment obtains the group ID 114 for grouping the candidate parts stored in the database in association with the feature quantities of the candidate parts, determines the group of the candidate parts suitable for the control panel before replacement using the obtained group ID 114, and reads and outputs identification information from the database for the candidate parts belonging to the group suitable for the control panel before replacement. This makes it possible to output candidate parts that are similar not to a single part but to the entire control panel, enabling replacement with a sense of unity to the entire control panel.

100 Part identification device, 101 Processor, 102 Memory, 103 Input/output IF, 104 Bus, 105 Input section, 106 Type identification section, 107 Feature extraction section, 108 Similarity calculation section, 109 Feature DB, 110 Output section, 111 Original part, 111a Needle, 111b Scale, 111c Numeric display, 111d Center circle, 111e Display panel, 112 Part ID, 113 Feature, 114 Group ID, 115 Group judgment section, 116 Linear meter, 116a Meter, 116b Meter, 116c Scale, 116d Numeric display

Claims (6)

an input unit to which an image including a first part is input;
a feature extraction unit that extracts, from the image of the first part, a feature in an arrangement of elements that constitute the part;
a similarity calculation unit that reads out a feature amount related to a second part from a database that stores a plurality of identification information, which is information for identifying a part, and feature amounts associated with the second part, and calculates a similarity by comparing the feature amount with the feature amount extracted by the feature amount extraction unit;
an output unit that identifies a similar part that is a part similar to the first part from among the second parts using the similarity, and reads out the identification information related to the similar part from the database and outputs the identification information;
A part identification device comprising: The first component is an original component constituting a control panel before replacement,
2. The part identification device according to claim 1, wherein the second part is a candidate part that is a part that is a candidate for selection as a part that constitutes a post-replacement control panel realized on a graphic screen. 3. The part identification device according to claim 2, wherein the feature amount related to the candidate part and the feature amount extracted by the feature amount extracting unit indicate the position of an element in the part, the distance between elements, or the positional relationship between elements. a group determination unit that obtains a group ID for grouping the candidate parts, the group ID being stored in the database in association with the feature quantities of the candidate parts, for the similar parts, and determines a group of the candidate parts suitable for the control panel before replacement using the obtained group ID;
3. The part identification device according to claim 2, wherein the output unit reads out from the database the identification information for the candidate parts belonging to the group suitable for the control panel before replacement, and outputs the identification information. An image including a first part is input;
extracting from the image a feature amount of an arrangement of elements constituting a part with respect to the first part;
reading out the feature amount for the second part from a database that stores a plurality of pieces of identification information, which is information for identifying a part, and feature amounts in association with each other for the second part, and calculating a similarity by comparing the feature amount with the feature amount extracted from the image;
a similar part that is a part similar to the first part is specified from among the second parts using the similarity, and the identification information relating to the similar part is read out from the database and output. an input process in which an image including a first part is input;
a feature extraction process for extracting, from the image of the first part, a feature in an arrangement of elements constituting a part;
a similarity calculation process of reading out a feature amount related to a second part from a database that stores a plurality of identification information, which is information for identifying a part, and feature amounts in association with each other for the second part, and calculating a similarity by comparing the feature amount with the feature amount extracted by the feature amount extraction process;
an output process of identifying similar parts, which are parts similar to the first part, from among the second parts using the similarity, and reading out the identification information related to the similar parts from the database and outputting the identification information;
A part identification program for executing the above.

Images