KR20230141582A - Method of extracting drawing information from 2-dimensional drawing image in graph structure and method of comparing one drawing image with another drawing image - Google Patents

Abstract

A method for extracting drawing information in a graph structure from a two-dimensional drawing image is disclosed. The method is performed by a controller and includes recognizing symbol elements from a two-dimensional drawing image; Recognizing line elements from a two-dimensional drawing image; Recognizing text elements from a two-dimensional drawing image; Comparing the symbol element, line element, and text element and setting them as nodes and edges; And it may include extracting the nodes and edges into a graph structure.
A method of comparing a drawing image extracted by the above method with another drawing image is further disclosed.

Description

A method of extracting drawing information from a two-dimensional drawing image in a graph structure and a method of comparing one drawing image with another drawing image {METHOD OF EXTRACTING DRAWING INFORMATION FROM 2-DIMENSIONAL DRAWING IMAGE IN GRAPH STRUCTURE AND METHOD OF COMPARING ONE DRAWING IMAGE WITH ANOTHER DRAWING IMAGE}

The present invention relates to a method of extracting drawing information from a two-dimensional drawing image in a graph structure and a method of comparing a two-dimensional drawing image and a three-dimensional drawing image. More specifically, the present invention relates to extracting drawing information from a two-dimensional drawing image regardless of its shape. It relates to a method of extracting drawing information into a graph structure and a method of comparing one drawing image with another drawing image by extracting drawing information into a graph structure unrelated to its shape.

In order to convert a hard copy drawing image (drawing image drawn on paper, etc.) to an electronic file format (dwg file, etc.), it is necessary to recognize the drawing image and extract drawing information from the drawing image. Previously, the focus was on advancing recognition technology for individual drawing elements to recognize and digitize drawing images. In other words, the focus was on how well drawing elements such as symbols, text, and lines are recognized, and what methods are used to recognize drawing elements well.

In order to confirm whether the drawing information is accurately recognized from the drawing image and to confirm part of the drawing image with the corresponding drawing information for a specific purpose, the recognized drawing information is expressed in a state dependent on the shape.

However, because the same drawing/system information may exist in various forms, it may not be easy to analyze and compare form-dependent drawings. For example, in order to manufacture a ship, a two-dimensional drawing image must be converted into a three-dimensional drawing image, but the conversion of the drawing image cannot be performed automatically and is performed manually by people. In the process of manually converting a drawing image, the drawing image may be converted incorrectly due to human error. Additionally, even if the conversion of the drawing image is performed automatically, there may be errors in the conversion process. Therefore, it is necessary to compare the three-dimensional drawing image with the corresponding two-dimensional drawing image to check whether both drawing images are identical. In order to compare both drawing images, drawing information must be recognized from both drawing images. Drawing information recognized from a 3D drawing image is 3D, and drawing information recognized from a 2D drawing image is 2D, so both drawing images It is not easy to compare with each other.

The matters described in this background art section have been prepared to enhance understanding of the background of the invention, and may include matters that are not prior art already known to those skilled in the art in the field to which this technology belongs.

An embodiment of the present invention seeks to provide a method for extracting drawing information from a two-dimensional drawing image into a graph structure unrelated to shape.

Another embodiment of the present invention seeks to provide a method of comparing drawing information extracted from one drawing image in a shape-independent graph structure with drawing information extracted in a shape-independent graph structure from another drawing image. .

A method of extracting drawing information in a graph structure from a two-dimensional drawing image according to an embodiment of the present invention is performed by a controller and includes the steps of recognizing symbol elements from the two-dimensional drawing image; Recognizing line elements from a two-dimensional drawing image; Recognizing text elements from a two-dimensional drawing image; Comparing the symbol element, line element, and text element and setting them as nodes and edges; And a step of extracting the nodes and edges into a graph structure, wherein recognition of symbol elements includes recognizing symbol category information, symbol label information, and symbol coordinate information, and recognition of line elements includes line coordinate information and line size. It includes recognizing information and line direction information, and the recognition of text elements includes recognizing text coordinate information and text information, and the step of comparing the symbol elements, line elements, and text elements to set them as nodes and edges It may include setting nodes and edges based on symbol category information and symbol coordinate information of the symbol element, and line coordinate information, line size information, and line direction information of the line element.

Nodes set by comparing the symbol element, line element, and text element include the starting point and end point of the line element, the point where the line element meets the line element, the point where the line element meets the symbol element, the point where the size changes in the line element, and It can be the center point of a symbol element.

Based on symbol category information, nodes related to specific symbol elements or nodes whose expression is unnecessary may be eliminated from the set nodes.

Line elements between set nodes and virtual line elements between a node that is the center point of a symbol element and a point where the symbol element and the line element meet can be set as edges.

The step of extracting the nodes and edges into a graph structure includes assigning node coordinate information and edge direction information to the nodes and edges using symbol coordinate information, line coordinate information, and line direction information; assigning node type information to the node using symbol category information and symbol label information of symbol elements related to the node; It may also include the step of assigning edge type information and edge fitting information to the edge using symbol category information and symbol label information of the symbol element related to the edge, and line size information of the line element.

The step of extracting the nodes and edges into a graph structure compares the symbol coordinate information of the symbol element, the line coordinate information of the line element, and the text coordinate information of the text element, relates the symbol element to the corresponding text element, and compares the symbol coordinate information of the symbol element to the corresponding text element. relating text elements; extracting nodes and edges related to the symbol element and line element; Additionally, the step of assigning node type information and node label information to the extracted node using text information of the corresponding text element and providing edge type information and edge fitting information to the extracted edge may be further included.

The step of extracting the nodes and edges into a graph structure may further include listing the extracted nodes with or without node information and listing edges connecting the nodes with or without edge information. there is.

The method may further include the step of displaying nodes and edges extracted in a graph structure in a graph form.

A method of comparing a first drawing image and a second drawing image according to another embodiment of the present invention is performed by a controller and includes extracting first drawing information from the first drawing image into a first graph structure; extracting second drawing information from the second drawing image into a second graph structure; Comparing node label information of target nodes of the first graph structure and target corresponding nodes of the second graph structure or receiving identical node pairs and detecting corresponding identical node pairs in the first and second graph structures; Detecting target corresponding nodes in a second graph structure that correspond to target nodes in the first graph structure; Comparing an edge between any two neighboring target nodes with a corresponding edge between two neighboring target corresponding nodes in a second graph structure; And if it is determined that all target nodes of the first graph structure are identical to all target corresponding nodes of the second graph structure and all edges of the first graph structure are identical to all corresponding edges of the second graph structure, the first drawing image and the It includes the step of notifying that the two drawing images are the same, and at least one of the first and second drawing images may be a two-dimensional drawing image, and the two-dimensional drawing image is extracted into a graph structure according to the method according to the embodiment of the present invention. It can be.

In one example, the first drawing image is a two-dimensional drawing image, the second drawing image is a three-dimensional drawing image, the three-dimensional drawing image includes a plurality of blocks each having block information, and from the second drawing image Extracting the second drawing information into a second graph structure includes collecting block information of a plurality of blocks included in the second drawing image; Setting nodes and edges by comparing block information; It may also include the steps of giving node information to a set node and giving edge information to a set edge using block information.

The step of detecting target corresponding nodes corresponding to target nodes of the first graph structure in the second graph structure may be performed based on the similarity between node type vectors according to node distances.

Calculation of similarity between node type vectors according to node distance includes classifying all nodes in the first drawing information according to the node distance from the target node; generating a node type vector according to node distance based on node type information of all nodes in the first drawing information; Classifying all nodes in the second drawing information according to the node distance from the temporary target corresponding node; generating a corresponding node type vector according to node distance based on node type information of all nodes in the second drawing information; increasing the node distance and calculating a vector distance between the node type vector and the corresponding node type vector according to the same node distance; Calculating similarity according to node distance based on a weight according to the vector distance and the node distance; And a step of detecting the target corresponding node based on the similarity according to all node distances to all temporary target corresponding nodes, where the node distance to a random node is defined as the minimum number of edges between a random node and the corresponding node. , the node type vector according to the node distance can be defined as the distribution of node types that exist at the corresponding node distance.

The node distance is a value between 0 and the maximum node distance, and the node type vector according to the node distance of 0 may be the node type vector of the target node or the node type vector of the temporary target node.

The weight according to the node distance may be preset to increase as the node distance decreases.

The weight according to the node distance may be preset to become smaller as more nodes are detected as the same node pair at the corresponding node distance.

The step of detecting the target corresponding node includes calculating the final similarity for any temporary target corresponding node by adding the similarities according to all node distances; And, it may include detecting the temporary target corresponding node with the minimum final similarity as the final target corresponding node.

At least one target node of the first graph structure is not identical to at least one target corresponding node of the second graph structure, or at least one edge of the first graph structure is not identical to at least one corresponding edge of the second graph structure. If it is determined that the first drawing image and the second drawing image are not the same, the method may further include notifying that the first drawing image is not the same.

The method may further include listing non-identical node pairs and/or non-identical edge pairs.

The method may further include displaying non-identical node pairs and/or non-identical edge pairs in one of the first and second graph structures.

The method may further include displaying symbol elements, line elements, and/or text elements corresponding to non-identical node pairs and/or non-identical edge pairs on at least one of the first and second drawing images.

According to the present invention, the extracted drawing information can be utilized for various purposes by extracting drawing information from a two-dimensional drawing image into a graph structure unrelated to its shape. For example, a two-dimensional drawing image or a three-dimensional drawing image can be compared with a corresponding two-dimensional drawing image or a three-dimensional drawing image to check changes or to check the identity of both drawing images.

By displaying different parts in both drawing images on at least one of the two drawing images, the user can check whether the drawing information has been accurately extracted and can quickly and easily recognize changed or added parts.

In addition, effects that can be obtained or expected due to embodiments of the present invention will be disclosed directly or implicitly in the detailed description of the embodiments of the present invention. That is, various effects expected according to embodiments of the present invention will be disclosed in the detailed description to be described later.

Embodiments of the present specification may be better understood by referring to the following description in conjunction with the accompanying drawings, where like reference numerals refer to identical or functionally similar elements.
1 shows one example of a system that can implement a method according to an embodiment of the invention.
Figure 2 shows one example of a database used in the system of Figure 1.
Figure 3 is a flowchart of a method for extracting drawing information in a graph structure from a two-dimensional drawing image according to an embodiment of the present invention.
Figure 4 shows one example of a two-dimensional drawing image.
Figure 5 shows symbol recognition from the drawing image of Figure 4.
Figure 6 is a table listing information on symbols recognized from the drawing image in Figure 5.
Figure 7 shows line recognition from the drawing image of Figure 4.
FIG. 8 is a table listing information on lines recognized from the drawing image in FIG. 7.
FIG. 9 shows text recognition from the drawing image of FIG. 4.
FIG. 10 is a table listing text information recognized from the drawing image in FIG. 9.
FIG. 11 is an example showing drawing information extracted in a graph structure from the two-dimensional drawing image of FIG. 4.
Figure 12 is a flowchart of a method for comparing one drawing image with another drawing image according to another embodiment of the present invention.
Figure 13 shows an example of a two-dimensional drawing image and drawing information extracted from the example in a graph structure.
FIG. 14 shows a 3D drawing image created from the 2D drawing image of FIG. 13.
FIG. 15 shows drawing information extracted in a graph structure from the 3D drawing image of FIG. 14.
FIG. 16 is a diagram showing three arbitrary nodes (A node, B node, and C node) from the drawing information extracted in the graph structure of FIG. 13.
FIG. 17 is a diagram showing three corresponding nodes (A' node, B' node, and C' node) corresponding to the three nodes shown in FIG. 16 in the drawing information extracted with the graph structure of FIG. 15.
The drawings referenced above are not necessarily drawn to scale and should be understood as presenting rather simplified representations of various preferred features illustrating the basic principles of the present disclosure. The specific design features of the present disclosure, including, for example, specific dimensions, orientation, location, and shape, will be determined in part by the particular intended application and usage environment.

The terminology used herein is for the purpose of describing specific embodiments only and is not intended to limit the disclosure. As used herein, singular forms are intended to also include plural forms, unless the context clearly indicates otherwise. The terms “comprise” and/or “comprising”, when used herein, specify the presence of stated features, integers, steps, operations, elements and/or components, but It will also be understood that this does not exclude the presence or addition of one or more of the features, integers, steps, operations, elements, components and/or groups thereof. As used herein, the term “and/or” includes any one or all combinations of the associated listed items.

Additionally, it is understood that one or more of the methods or aspects thereof below may be implemented by at least one or more controllers. The term “controller” may refer to a hardware device that includes memory and a processor. The memory is configured to store program instructions, and the processor is specifically programmed to execute the program instructions to perform one or more processes described in more detail below. A controller may control the operation of units, modules, components, devices, or the like, as described herein. It is also understood that the methods below can be performed by an apparatus that includes a controller along with one or more other components, as will be appreciated by those skilled in the art.

Additionally, the controller of the present disclosure may be implemented as a non-transitory computer-readable recording medium containing executable program instructions executed by a processor. Examples of computer-readable recording media include ROM, RAM, compact disk (CD) ROM, magnetic tapes, floppy disks, flash drives, smart cards, and optical data storage devices. It is not limited to this. The computer-readable recording medium may also be distributed throughout a computer network so that program instructions can be stored and executed in a distributed manner, for example, on a telematics server or a Controller Area Network (CAN).

According to the present disclosure, drawing information is extracted from a two-dimensional drawing image in a shape-independent graph structure. Specifically, symbol elements, line elements, and text elements are recognized from two-dimensional drawing images, nodes and edges are set by comparing the recognized symbol elements, line elements, and text elements, and the set nodes and edges are converted into a graph structure. Extract. Since the drawing information is extracted in a graph structure independent of its shape, the drawing information can be utilized for various purposes. For example, a two-dimensional drawing image or a three-dimensional drawing image can be compared with a corresponding two-dimensional drawing image or a three-dimensional drawing image to check changes or to check the identity of both drawing images. Additionally, by displaying different parts in both drawing images on at least one of the drawing images, the user can check whether the drawing information has been accurately extracted and can quickly and easily recognize changed or added parts.

Hereinafter, embodiments of the present disclosure will be described in detail with reference to the attached drawings.

Figure 1 shows one example of a system that can implement a method according to an embodiment of the invention, and Figure 2 shows one example of a database used in the system of Figure 1.

1 and 2, a system capable of executing methods according to embodiments of the present invention includes an input interface 10, a controller 20, and an output interface 30. The input interface 10, the controller 20, and the output interface 30 are connected to each other wirelessly or wired to transmit or receive data.

The input interface 10 may be configured to receive a two-dimensional drawing image or a three-dimensional drawing image from the user. For example, when trying to extract drawing information from a two-dimensional drawing image drawn on paper, etc., a user can capture the two-dimensional drawing image using a photographing device. In this case, the input interface 10 may be a photographing device. Additionally, when drawing information is to be extracted from a digitized drawing image, the digitized drawing image can be displayed or loaded on a display connected to the controller 20. In this case, the input interface 10 may be a display. The input interface 10 may include, but is not limited to, an imaging device, display, keyboard, mouse, touch screen, microphone, joystick, jog shuttle, stylus, etc.

The controller 20 is configured to receive a drawing image input through the input interface 10 and recognize at least one drawing element from the drawing image. If the drawing image is a two-dimensional drawing image, the drawing elements include symbol elements, line elements, and text elements. Recognizing drawing elements from the drawing image means recognizing symbol elements, recognizing line elements, and recognizing text elements from the drawing image. It means to do. Additionally, recognizing a symbol element includes recognizing symbol category information, symbol label information, and symbol coordinate information, and recognizing a line element includes line coordinate information, line size information (information about the diameter of the line element), and It includes recognizing line direction information, and recognizing text elements includes recognizing text coordinate information and text information. To this end, the controller 20 may include an image recognition algorithm capable of recognizing drawing elements within an image and a text recognition algorithm capable of recognizing text among drawing elements. An image recognition algorithm may recognize the shape, coordinates, and/or size of drawing elements in an image, and a text recognition algorithm may recognize characters constituting text among drawing elements.

The controller 20 is configured to extract drawing information in a graph structure using recognized drawing elements. Here, drawing information includes nodes and edges, and extracting drawing information in a graph structure means processing drawing elements to set nodes and edges, giving node information corresponding to each node, and creating an edge corresponding to each edge. It means providing information and arranging (arranging or displaying) nodes and edges in a graph structure together with or separately from that information. Here, the graph structure means expressing the drawing image with corresponding nodes and edges instead of symbol elements, line elements, and text elements. That is, symbol elements, line elements, and text elements are extracted into nodes and edges, node information is given to nodes, and edge information is given to edges. Node information given to a node includes node type information, node coordinate information, and node label information, an edge is a line connecting neighboring nodes, and edge information given to an edge includes edge type information, edge fitting information, and edge direction. Contains information.

In order to recognize drawing elements from drawing images and extract drawing information in a graph structure, the controller 20 includes a database 40 or is connected to the database 40 wirelessly or wired to send or receive data. Figure 2 shows one example of database 40, but database 40 is not necessarily limited to the database 40 illustrated in Figure 2. The database 40 may be constructed through input and/or learning.

As shown in FIG. 2 , the database 40 may include a shape database 42, a symbol database 44, a node attribute database 46, and an edge attribute database 48.

The shape database 42 contains data regarding the shape of drawing elements. Based on the data stored in the shape database 42, the controller 20 distinguishes whether the drawing element in the drawing image is a symbol element, a line element, or a text element, and distinguishes between the differentiated symbol elements (symbol coordinate information) and line elements. (line coordinate information and line direction information), and/or text elements (text coordinate information) can be recognized.

Symbol database 44 contains data regarding symbol categories and/or symbol labels that correspond to the shapes of symbol elements. For example, in FIG. 5, the shape of symbol number 4 indicates that the symbol category is valve and the symbol label is a valve, and the shape of symbol number 1 indicates that the symbol category is equipment and the symbol label is a tank. Data is provided in the symbol database 44. It is stored in The controller 20 may recognize symbol category information and/or symbol label information of symbol elements based on data stored in the symbol database 44.

The node attribute database 46 includes node type information and/or node label information corresponding to data about the shape of drawing elements, symbol category information, symbol label information, and/or text information. Node types included in the node type information include, but are not limited to, “EQUIP”, “TEE”, “3VV”, “NORMAL”, “CONNECT”, “REDU”, and “INTERFACE”. "EQUIP" is a node type that represents equipment, "TEE" is a node type that represents a point where three or more pipes meet, "3VV" is a node type that represents a three way valve, and "CONNECT" is a node type that represents a device. It is a node type that represents the point where pipes are connected, "REDU" is a node type that represents the point where the diameter of the pipe changes, "INTERFACE" is a node type that represents the connection point between drawings, and "NORMAL" is described previously. This is a node type given to nodes that are not included in one node type. Additionally, the node label information includes the device name.

The edge attribute database 48 includes edge type information, edge fitting information, and/or edge direction corresponding to data about the shape of drawing elements, symbol category information, symbol label information, and/or line direction information. Edge types included in edge type information include, but are not limited to, “PIPE” and “EQUIP”. "PIPE" is an edge type that represents a general piping, and "EQUIP" is an edge type that represents an edge connecting a node with a node type of "CONNECT" and a node with a node type of "EQUIP". Edge fitting information includes information about special parts such as valves or orifices included in the pipe. Edge direction information includes information about the start node and end node of the edge.

If the drawing image is a digitized drawing image, particularly a 3D digital drawing image, the controller 20 may collect drawing information included in the 3D drawing image and extract it into a graph structure. In general, equipment or piping included in a digitized drawing image is formed by gathering multiple blocks, and each block has block information, similar to nodes and edges, and the block information includes block coordinate information, block size information, and block information. It includes type information, block label information, block fitting information, and/or block direction information. The controller 20 collects block information of a plurality of blocks included in the digitized drawing image, sets nodes and edges by comparing the block information, provides node information for the set nodes, and provides edge information for the set edges. can be given. For example, if the block coordinate information and block direction information of the five blocks are consecutive and the block size information, block type information, block label information, and block fitting information of the five blocks are the same, the controller 20 A block can be set as one edge, the block direction information of five blocks can be processed into edge direction information, and the block type information and block fitting information of five blocks can be saved as edge type information and edge fitting information, respectively. Here, a method for extracting drawing information from a digitized drawing image in a graph structure is illustrated. However, the present invention is not limited to the exemplified methods.

Additionally, the controller 20 is configured to compare one drawing image with another drawing image using drawing information extracted in a graph structure. For example, in order to check whether a two-dimensional drawing image and a three-dimensional drawing image created from the two-dimensional drawing image are the same, that is, whether the three-dimensional drawing image is accurately produced, the controller 20 determines whether the two-dimensional drawing image is produced from the two-dimensional drawing image. Drawing information is extracted into one graph structure, drawing information is extracted from a 3D drawing image into another graph structure, and nodes and edges in one graph structure are converted to corresponding nodes and edges in another graph structure. You can compare to determine whether both drawings are the same.

For this purpose, the controller 20 may be implemented with one or more processors operated by a set program, and the memory of the controller 20 contains a two-dimensional drawing image according to an embodiment of the present invention through the one or more processors. Program commands programmed to perform each step of the method of extracting drawing information in a graph structure and comparing one drawing image with another drawing image are stored.

The output interface 30 may display the extracted drawing information in a graph structure as shown in FIGS. 11, 13, and 15. In addition, the output interface 30 displays drawing images to be compared, and displays different parts in both drawing images on at least one of the drawing images, so that the user can check whether the drawing information has been accurately extracted and any changed or added parts. can be recognized quickly and easily. For example, by comparing a 2D drawing image with a 3D drawing image created from the 2D drawing image, it can be confirmed whether the 3D drawing image was created accurately.

Hereinafter, with reference to FIGS. 3 to 11, a method for extracting drawing information in a graph structure from a two-dimensional drawing image according to an embodiment of the present invention will be described in more detail.

FIG. 3 is a flowchart of a method for extracting drawing information in a graph structure from a two-dimensional drawing image according to an embodiment of the present invention, FIG. 4 shows an example of a two-dimensional drawing image, and FIG. 5 is a drawing of FIG. 4. It shows recognition of a symbol from an image, FIG. 6 is a table listing information on symbols recognized from the drawing image in FIG. 5, FIG. 7 shows recognition of a line from the drawing image of FIG. 4, and FIG. 8 shows 7 is a table listing information on lines recognized from the drawing image, FIG. 9 shows text recognized from the drawing image in FIG. 4, and FIG. 10 is a table listing text information recognized from the drawing image in FIG. 9. FIG. 11 is an example showing drawing information extracted in a graph structure from the two-dimensional drawing image of FIG. 4.

As shown in FIG. 3, the method of extracting drawing information in a graph structure from a two-dimensional drawing image according to an embodiment of the present invention begins with a user inputting a two-dimensional drawing image through the input interface 10.

When a two-dimensional drawing image from which drawing information must be extracted in a graph structure is input through the input interface 10, the controller 20 recognizes at least one drawing element from the two-dimensional drawing image. That is, the controller 20 recognizes symbol elements from the two-dimensional drawing image (S100), recognizes line elements from the two-dimensional drawing image (S110), and recognizes text elements from the two-dimensional drawing image (S120).

Below, the process of recognizing drawing elements from the two-dimensional drawing image shown in FIG. 4 will be described in more detail.

The controller 20 recognizes the shape, coordinates, and/or size of drawing elements from the two-dimensional drawing image shown in FIG. 4 through an image recognition algorithm, and recognizes the drawing elements based on the shape, coordinates, and/or size. Classifies drawing elements into symbol elements, line elements, and text elements. When the drawing elements are classified into symbol elements, line elements, and text elements, the controller 20 recognizes the symbol elements, line elements, and text elements from the two-dimensional drawing image (S100, S110, S120).

As shown in FIG. 5, the controller 20 recognizes drawing elements classified as symbol elements from the two-dimensional drawing image (S100). There are five symbol elements in FIG. 5, and the results of recognition by the controller 20 for each symbol element are shown in FIG. 6. For example, the symbol category of symbol 1 is equipment (EQ), the symbol label is tank, the coordinates of one corner are (200, 200), and the coordinates of one corner are (200, 200), and the other corner (on the diagonal of one corner) is Recognize that the coordinates of the located corner are (400, 400). Additionally, the recognized information is stored as symbol category information, symbol label information, and symbol coordinate information.

Additionally, as shown in FIG. 7, the controller 20 recognizes drawing elements classified as line elements from the two-dimensional drawing image (S110). Here, a line element means a straight line extending in one direction. There are six line elements in FIG. 7, and the results recognized by the controller 20 for each line element are shown in FIG. 8. For example, it is recognized that the starting point coordinates of line 3 are (300, 50) and the ending point coordinates are (850, 50). Additionally, the recognized information is stored as line coordinate information, line size information, and line direction information. Although line size information is omitted in FIG. 8, line size information may be included to recognize the point where two pipes of different diameters arranged in a straight line meet as a node.

Additionally, as shown in FIG. 9, the controller 20 recognizes drawing elements classified as text elements from the two-dimensional drawing image (S120). Here, there are three text elements in FIG. 9, and the results recognized by the controller 20 for each text element are shown in FIG. 10. For example, the coordinates of one corner of text number 2 are (710, 370), the coordinates of another corner (a corner located diagonally of one corner) are (770, 385), and the coordinates of the text composing the text are (770, 385). Recognizes that the character is "Pump". Additionally, recognized information is stored as text coordinate information and text information.

When all drawing elements have been recognized from the two-dimensional drawing image, the controller 20 sets nodes and edges by comparing the recognized symbol elements, line elements, and/or text elements (S130). That is, the controller 20 sets nodes and edges based on the symbol category information and symbol coordinate information of the symbol element, and the line coordinate information, line size information, and line direction information of the line element. More specifically, the starting and ending points of line elements, the points where line elements meet, the points where line elements and symbol elements meet, the points where the size (e.g., diameter) changes in the line elements, and the center point of the symbol elements are Can be set as a node. Additionally, based on symbol category information, nodes related to some (specific) symbol elements are eliminated from the node. For example, if a symbol element located on a line element whose symbol category information is an orifice or valve, nodes related to the symbol element (the point where the symbol element and the line element meet and the center point of the symbol element) are eliminated from the node. In addition, the line elements between the finally set nodes and the virtual line elements between the node that is the center point of the symbol element and the point where the symbol element and the line element meet are set as edges.

When the nodes and edges are set by comparing the recognized drawing elements, the controller 20 extracts the nodes and edges into a graph structure (S140). More specifically, the controller 20 provides node coordinate information and edge direction information to nodes and edges using symbol coordinate information, line coordinate information, and line direction information. In addition, using the symbol category information and symbol label information of the symbol element related to the node, the controller 20 gives node type information to the node, symbol category information and symbol label information of the symbol element related to the edge, and line element. Using the line size information, the controller 20 provides edge type information and edge fitting information to the corresponding edge. Additionally, the controller 20 compares the symbol coordinate information of the symbol element, the line coordinate information of the line element, and the text coordinate information of the text element to associate the symbol element with the corresponding text element and associate the line element with the corresponding text element. . The controller 20 extracts nodes and edges related to the symbol element and line element, assigns node type information and node label information to the extracted node using text information of the corresponding text element, and assigns an edge type to the extracted edge. Provides information and edge fitting information. Afterwards, nodes to which node information is given and edges to which edge information is given are listed in a graph structure. That is, the controller 20 lists the extracted nodes with or without node information, and lists the edges connecting the nodes with or without edge information.

Optionally, the controller 20 may display nodes and edges extracted from the graph structure in a graph form (S150). That is, the controller 20 can display nodes and edges in a graph form on the display. In FIG. 11, nodes and edges extracted in a graph structure from the two-dimensional drawing image of FIG. 4 are displayed in a graph form. The controller 20 places the extracted nodes using the node coordinate information of the extracted nodes, and connects the nodes with edges using the edge coordinate information and edge direction information of the extracted edges. In FIG. 11, node information is described in nodes and edge information is described in edges, but node information and edge information corresponding to nodes and edges displayed in a graph form do not necessarily need to be described.

Hereinafter, with reference to FIGS. 12 to 17, a method of comparing one drawing image with another drawing image according to another embodiment of the present invention will be described in detail.

Figure 12 is a flowchart of a method for comparing one drawing image with another drawing image according to another embodiment of the present invention.

As shown in FIG. 12, a method of comparing one drawing image with another drawing image according to another embodiment of the present invention involves the user selecting two drawing images that need to be compared through the input interface 10. It begins. In one example, a user may select two drawing images that need to be compared by inputting the first and second drawing images through the input interface 10. In another example, the user can select two drawing images that need to be compared by inputting a first drawing image and loading a second drawing image stored in the memory of the controller 20. Here, the first drawing image may be a reference drawing image that serves as a standard for comparison (i.e., accurately created).

Below, the first drawing image is a two-dimensional drawing image, the second drawing image is a three-dimensional drawing image, and the second drawing image is a drawing image produced from the first drawing image. However, the present invention is an example. It is not limited to the examples provided. In other words, the present invention is intended to compare drawing information extracted from a drawing image in the form of a graph, and although the drawing information includes node coordinate information and edge coordinate information, drawing information to which node coordinate information and edge coordinate information correspond is compared. Since it is not used to do so, the present invention can be applied to both two-dimensional drawing images and three-dimensional drawing images.

When the first and second drawing images requiring comparison are input, the controller 20 extracts the first drawing information from the first drawing image into a first graph structure (S200) and extracts the second drawing information from the second drawing image. 2Extract as a graph structure (S210).

FIG. 13 shows an example of a two-dimensional drawing image and drawing information extracted from the example in a graph structure, FIG. 14 shows a three-dimensional drawing image produced from the two-dimensional drawing image of FIG. 13, and FIG. 15 shows drawing information extracted in a graph structure from the 3D drawing image of FIG. 14.

As shown in FIG. 13, if the first drawing image is a two-dimensional drawing image, as described above, a method of extracting drawing information in a graph structure from the two-dimensional drawing image according to the embodiment of the present invention shown in FIG. 3. The first drawing information can be extracted into the first graph structure using . The first drawing information includes n target nodes and target edges. The first drawing information extracted from the first graph structure is shown in FIG. 13.

As shown in FIG. 14, if the second drawing image is a 3D digitized drawing image, the controller 20 collects the second drawing information included in the 3D digitized drawing image and extracts it into a second graph structure. For example, the controller 20 collects block information of a plurality of blocks included in a digitized drawing image, sets nodes and edges by comparing the block information, provides node information for the set nodes, and sets the edges. For this, edge information is given, and through this, the second drawing information is extracted into the second graph structure. The second drawing information includes target-corresponding nodes and target-corresponding edges.

When the first drawing information is extracted in the first graph structure and the second drawing information is extracted in the second graph structure, the controller 20 compares the node label information of the target nodes and target corresponding nodes or selects the same node pairs from the user. It receives input and detects corresponding identical node pairs in the first and second graph structures (S220). In one example, the controller 20 determines whether the target node and the target corresponding node are the same by comparing whether the device name recorded in the node label information of the target node and the device name recorded in the node label information of the target corresponding node are the same. You can. In another example, the controller 20 may receive input of the same node pair from the user.

After detecting or simultaneously detecting the corresponding identical node pair in the first and second graph structures, the controller 20 detects the target corresponding node corresponding to the n target nodes (S230 to S260).

More specifically, the controller 20 substitutes 1 into i (S230) and detects the first target corresponding node corresponding to the first target node in the first graph structure from the second graph structure (S240). Afterwards, the controller 20 checks whether i is equal to n (S250), and if i is not n, substitutes (i+1) into i (S260). Afterwards, the controller 20 repeats S240, S250, and S260 until i becomes equal to n.

Here, the method of detecting the i'th target corresponding node corresponding to the i'th target node is based on the similarity between node type vectors according to the node distance. Hereinafter, with reference to FIGS. 16 and 17, a method for calculating similarity between node type vectors according to node distance will be described in more detail.

FIG. 16 is a diagram showing three arbitrary nodes (A node, B node, and C node) from the drawing information extracted with the graph structure of FIG. 13, and FIG. 17 is a diagram showing the drawing information extracted with the graph structure of FIG. 15. This is a diagram showing three corresponding nodes (A' node, B' node, and C' node) corresponding to the three nodes shown in Fig. 16. Here, the process of determining whether the A node, B node, and C node of FIG. 16 are the same node pair as the A' node, B' node, and C' node of FIG. 17, respectively, will be described in detail.

First, all nodes in the first drawing information are classified according to the node distance from node A (target node). Here, the node distance from the target node is defined as the minimum number of edges between the target node and the corresponding node. For example, there are 2 nodes at node distance 1 from node A, including node C, and there are 4 nodes at node distance 2 from node A, including node B. In this way, if all nodes in the first drawing information are classified according to the node distance from node A, the controller 20 generates a node type vector (Vx) according to the node distance based on the node type information. Here, x is the node distance and is a natural number between 0 and p, p is the maximum node distance based on the target node, and V0 is the node type vector of the target node. Additionally, the node type vector is a 1×m vector, and m is the number of node types included in the node type information. As described previously, if the node types included in the node type information are "EQUIP", "TEE", "3VV", "NORMAL", "CONNECT", "REDU", and "INTERFACE", m is 7. The node type vector (Vx) according to the node distance represents the distribution of node types that exist at that node distance. For example, if the number of nodes with node type "EQUIP" is 2, the number of nodes with node type "3VV" is 1, and no nodes with other node types exist at node distance 1, then node distance 1 The node type vector (V1) according to is (2 0 1 0 0 0 0). Here, the arrangement of the node types can be arbitrarily preset by a person skilled in the art, but the arrangement of the node types used when calculating the similarity between node type vectors must be the same.

When a node type vector (V0, V1, ..., Vp) according to the node distance from node distance 0 to node distance p is generated, all nodes in the second drawing information are divided into node distances with the A' node (temporary target corresponding node). Classify according to the node distance, and generate a corresponding node type vector (Vx') according to the node distance (step of generating a corresponding node type vector according to the node distance to one temporary target corresponding node).

The controller 20 calculates the vector distance (Sx) between the node type vector (Vx) and the corresponding node type vector (Vx') according to the same node distance from 0 to p, and adds to the vector distance (Dx) The similarity (Sx) according to the node distance is calculated by multiplying the weight (Wx) according to the node distance. That is, Sx = Dx * Wx. Here, the weight (Wx) according to the node distance is set to increase as the node distance decreases. That is, W0 ≥ W1 ≥ W2 ≥ … ≥ Wp. Additionally, the weight (Wx) according to the node distance is set to become smaller as more nodes are detected as the same node pair at the node distance x. The weight (Wx) according to the node distance can be set in advance through various methods, such as multiple analysis and neural network models.

The controller 20 calculates the final similarity (S) for one temporary target corresponding node by adding the similarity according to the node distance from 0 to p. That is, S = S1 + S2 + … +Sp.

The controller 20 generates a corresponding node type vector according to the node distance to the other temporary target corresponding node and calculates the final similarity (S) to the other temporary target node. The controller 20 calculates the final similarity (S) for all temporary target corresponding nodes, and detects the temporary target corresponding node with the minimum final similarity (S) as the final target corresponding node. In this way, it is determined that the A node, B node, and C node of FIG. 16 correspond to the A' node, B' node, and C' node of FIG. 17, respectively. That is, the A node of the first graph structure is identical to the A' node of the second graph structure, the B node of the first graph structure is identical to the B' node of the second graph structure, and the C node of the first graph structure is It is judged to be the same as the C' node of the second graph structure.

When all target corresponding nodes of the second graph structure corresponding to all target nodes of the first graph structure are detected, the controller 20 selects the edge between any two neighboring target nodes in the first graph structure to the second graph structure. Compare with the corresponding edge between two neighboring corresponding target nodes (S270). For example, the controller 20 compares the edge information of the edge between node A and node C in FIG. 16 with the edge information between node A' and node C' in FIG. 17 and determines that the edge and the corresponding edge are the same. Determine awareness.

The controller 20 repeatedly performs S270 on all edges of the first graph structure and compares all edges of the first graph structure with all corresponding edges of the second graph structure (S280).

If, as a result of S280, it is determined that all target nodes of the first graph structure are identical to all target corresponding nodes of the second graph structure and all edges of the first graph structure are identical to all corresponding edges of the second graph structure (in S290, ' Yes'), the controller 20 can inform the user through the output interface 30 that the first drawing image and the second drawing image are the same. If it is determined that some target nodes of the first graph structure are not identical to some target corresponding nodes of the second graph structure or some edges of the first graph structure are not identical to some edges of the second graph structure ('No in S290 '), the controller 20 may inform the user through the output interface 30 that the first drawing image and the second drawing image are not the same. In addition, the controller 20 optionally lists non-identical node pairs and/or non-identical edge pairs through the output interface 30, or lists non-identical node pairs and/or non-identical edge pairs in the first, 2Drawing elements (symbol elements, line elements, and/or text elements) displayed in at least one of the graph structures, or corresponding to non-identical node pairs and/or non-identical edge pairs, among the first and second drawing images. You can mark at least one. The user can check whether the first and second drawing information has been accurately extracted into the first and second graph structures from the first and second drawing images using the nodes, edges, and/or drawing elements displayed on the output interface 30, or , you can compare the second drawing images to check for changes, or check whether the first and second drawing images are the same.

Although preferred embodiments of the present invention have been described above, the present invention is not limited to the above embodiments, and can be easily modified by a person skilled in the art from the embodiments of the present invention to provide equivalent equivalents. Includes all changes within the scope deemed acceptable.

Claims (20)

Recognizing, by a controller, a symbol element from a two-dimensional drawing image;
Recognizing, by a controller, line elements from a two-dimensional drawing image;
Recognizing, by a controller, a text element from a two-dimensional drawing image;
Comparing the symbol element, line element, and text element and setting them as nodes and edges, by a controller; and
Extracting, by a controller, the nodes and edges into a graph structure;
Includes,
Recognition of symbol elements includes recognizing symbol category information, symbol label information, and symbol coordinate information,
Recognition of line elements includes recognizing line coordinate information, line size information, and line direction information,
Recognition of text elements includes recognizing text coordinate information and text information,
The step of comparing the symbol element, line element and text element and setting them as nodes and edges is based on the symbol category information and symbol coordinate information of the symbol element and the line coordinate information, line size information and line direction information of the line element. A method of extracting drawing information in a graph structure from a two-dimensional drawing image including the step of setting edges. According to paragraph 1,
Nodes set by comparing the symbol element, line element, and text element include the starting point and end point of the line element, the point where the line element meets the line element, the point where the line element meets the symbol element, the point where the size changes in the line element, and A method of extracting drawing information in a graph structure from a two-dimensional drawing image that is the center point of a symbol element. According to paragraph 2,
A method of extracting drawing information in a graph structure from a two-dimensional drawing image in which nodes related to specific symbol elements are excluded from set nodes based on symbol category information. According to paragraph 3,
The line elements between set nodes and the virtual line elements between the node, which is the center point of the symbol element, and the point where the symbol element and the line element meet are set as edges. Drawing information is extracted in a graph structure from a two-dimensional drawing image. method. According to paragraph 1,
The step of extracting the nodes and edges into a graph structure is
Giving node coordinate information and edge direction information to nodes and edges using symbol coordinate information, line coordinate information, and line direction information;
assigning node type information to the node using symbol category information and symbol label information of symbol elements related to the node; and
assigning edge type information and edge fitting information to the edge using symbol category information and symbol label information of the symbol element related to the edge, and line size information of the line element;
A method of extracting drawing information in a graph structure from a two-dimensional drawing image including. According to clause 5,
The step of extracting the nodes and edges into a graph structure is
Comparing the symbol coordinate information of the symbol element, the line coordinate information of the line element, and the text coordinate information of the text element to relate the symbol element to the corresponding text element and to relate the line element to the corresponding text element;
extracting nodes and edges related to the symbol element and line element; and
assigning node type information and node label information to the extracted node using text information of the corresponding text element and assigning edge type information and edge fitting information to the extracted edge;
A method of extracting drawing information in a graph structure from a two-dimensional drawing image further comprising: According to clause 6,
The step of extracting the nodes and edges into a graph structure further includes listing the extracted nodes with or without node information, and listing edges connecting the nodes with or without edge information. 2 A method of extracting drawing information from a dimensional drawing image into a graph structure. According to paragraph 1,
A method of extracting drawing information in a graph structure from a two-dimensional drawing image, further comprising displaying nodes and edges extracted in a graph structure in a graph form by a controller. Extracting, by a controller, first drawing information from a first drawing image into a first graph structure;
Extracting, by a controller, second drawing information from the second drawing image into a second graph structure;
Comparing node label information of target nodes of the first graph structure and target corresponding nodes of the second graph structure, by the controller, or receiving identical node pairs and detecting corresponding identical node pairs in the first and second graph structures. ;
Detecting, by the controller, target corresponding nodes corresponding to target nodes in the first graph structure in the second graph structure;
Comparing, by the controller, an edge between any two neighboring target nodes with a corresponding edge between two neighboring target corresponding nodes in a second graph structure; and
If it is determined by the controller that all target nodes of the first graph structure are identical to all target corresponding nodes of the second graph structure and all edges of the first graph structure are identical to all corresponding edges of the second graph structure, Notifying that the image and the second drawing image are the same;
Includes,
At least one of the first and second drawing images is a two-dimensional drawing image,
A method of comparing a first drawing image and a second drawing image in which the two-dimensional drawing image is extracted in a graph structure according to the method according to any one of claims 1 to 8. According to clause 9,
The first drawing image is a two-dimensional drawing image, the second drawing image is a three-dimensional drawing image,
A 3D drawing image includes a plurality of blocks each having block information,
The step of extracting the second drawing information from the second drawing image into a second graph structure is
Collecting block information of a plurality of blocks included in the second drawing image;
Setting nodes and edges by comparing block information; and
Giving node information to a node set using block information and giving edge information to a set edge;
A method of comparing a first drawing image and a second drawing image including. According to clause 9,
The step of detecting target corresponding nodes corresponding to target nodes of the first graph structure in the second graph structure is performed by selecting the first drawing image and the second drawing image based on the similarity between the node type vectors according to the node distance. How to compare. According to clause 11,
The calculation of similarity between node type vectors based on node distance is
Classifying all nodes in the first drawing information according to the node distance from the target node;
generating a node type vector according to node distance based on node type information of all nodes in the first drawing information;
Classifying all nodes in the second drawing information according to the node distance from the temporary target corresponding node;
generating a corresponding node type vector according to node distance based on node type information of all nodes in the second drawing information;
increasing the node distance and calculating a vector distance between the node type vector and the corresponding node type vector according to the same node distance;
Calculating similarity according to node distance based on a weight according to the vector distance and the node distance; and
detecting a target corresponding node based on similarity according to all node distances to all temporary target corresponding nodes;
Includes,
The node distance from any node is defined as the minimum number of edges between any node and that node,
A method of comparing the first drawing image and the second drawing image, where the node type vector according to the node distance is defined as the distribution of node types existing at the corresponding node distance. According to clause 12,
The node distance is a value between 0 and the maximum node distance,
A method of comparing the first drawing image and the second drawing image, where the node type vector according to the node distance of 0 is the node type vector of the target node or the node type vector of the temporary target node. According to clause 12,
A method of comparing a first drawing image and a second drawing image in which the weight according to the node distance is preset to increase as the node distance decreases. According to clause 12,
A method of comparing a first drawing image and a second drawing image in which the weight according to the node distance is preset to be smaller as more nodes are detected as the same node pair in the corresponding node distance. According to clause 12,
The step of detecting the target corresponding node is
Calculating the final similarity for any temporary target corresponding node by adding the similarities according to all node distances; and
detecting a temporary target corresponding node with the minimum final similarity as a final target corresponding node;
A method of comparing a first drawing image and a second drawing image including. According to clause 9,
By the controller, at least one target node of the first graph structure is not identical to at least one target corresponding node of the second graph structure or at least one edge of the first graph structure is equal to at least one corresponding edge of the second graph structure. If it is determined that the first drawing image and the second drawing image are not the same, the method of comparing the first drawing image and the second drawing image further includes the step of notifying that the first drawing image and the second drawing image are not the same. According to clause 17,
A method for comparing a first drawing image and a second drawing image, further comprising listing non-identical node pairs and/or non-identical edge pairs. According to clause 17,
A method for comparing a first drawing image and a second drawing image, further comprising displaying non-identical node pairs and/or non-identical edge pairs in one of the first and second graph structures. According to clause 17,
The first drawing image and the second drawing image further comprising the step of displaying symbol elements, line elements, and/or text elements corresponding to non-identical node pairs and/or non-identical edge pairs on at least one of the first and second drawing images. How to compare two drawing images.

Images