KR102177550B1 - Method of automatically recognizing and classifying information of design in imaged PID drawings - Google Patents

Abstract

The main object of the present invention is to provide a method for recognizing and classifying design information by automatically digitizing an imaged P&ID drawing to digitize design information by counting design elements with high accuracy in a short time.
As a method of automatically recognizing and classifying design information in an imaged P&ID drawing to achieve the above object, a symbol area is extracted from the imaged P&ID drawing, and the origin and connection point of the corresponding symbol in the symbol area are identified. Automatically registering a symbol in a database after setting; Recognizing and extracting the pre-registered symbol from the imaged P&ID drawing, and recognizing and extracting the extracted symbol from the imaged P&ID drawing; Recognizing and extracting a line from the imaged P&ID drawing from which the symbol has been removed using a sliding window method that calculates a blob unit instead of a pixel unit; Calculating an aspect ratio from the imaged P&ID drawing from which the symbols have been removed to calculate an area where the text exists, and then recognizing and extracting the text from the corresponding area; Classifying the text detected in the drawing area among the extracted texts into respective attributes through a predefined attribute classification system; Associating the extracted symbol and line with the attribute of the extracted text based on the nearest distance, and associating the extracted symbol based on the device name recognized in the text when the extracted symbol is an equipment. do.

Description

{Method of automatically recognizing and classifying information of design in imaged PID drawings}

The present invention is a method of automatically recognizing and storing design information in an image P&ID. In detail, after recognizing a symbol, line, and text of an image P&ID drawing, the recognized design information is classified. And it relates to a method of storing in conjunction.

Typically, engineering companies, oil refining companies, chemical companies, etc. hold design drawings in CAD drawing formats such as AutoCAD and PDF or Hard Copy format. In a situation where AI and big data technologies are spreading due to the recent 4th industrial revolution, digitalization of the data held in the PDF or Hard Copy format is essential in order to apply the technology to the shipbuilding industry and the plant engineering industry as a whole.

Looking at the digitization process of the conventional design drawings by the contractor, first, when the P&ID (hereinafter referred to as "imaged P&ID") that is designed and created by the customer is an image file (hereinafter referred to as "imaged P&ID"), the design engineer is reminded The image filed P&ID drawing is created as a new P&ID drawing.

Here, the P&ID is an abbreviation of “Piping & Instrument Drawing”, which means a process flow diagram in which facilities, piping, and electrical instrumentation of a certain process are clearly expressed in a diagram format. As a design program for creating P&ID, Intergraph )'S SP P&ID (SmartPlant P&ID), Aveva's Aveva P&ID, and AutoDesk's AutoCAD Plant P&ID.

The material and quantity are described in the P&ID drawing created by the above construction company, and based on these drawings, the engineer collects the quantity data through a material calculation program run on a computer to create a bill of materials (BOM). Calculate the estimate through

However, the digitization process of the conventional design drawing described above was only used to check errors by comparing the digitized design drawing and the imaged drawing with the naked eye.This is the most important information and data necessary for automation of design, etc. There was a considerable problem. In other words, while comparing the imaged P&ID drawing with the new drawing created, it is necessary to check whether there is any missing or uncalculated item among the materials listed on the imaged P&ID drawing. There is a problem of the difficulty of checking the consistency.

In addition, since all material items including valves must be created while creating a new drawing, it takes a lot of unnecessary time to create a new drawing and fill in the quantity, and if the piping layout is changed, the calculation material is modified and updated. There is a problem that is very difficult to manage.

Due to this problem, most companies are not able to carefully check the results of calculations, and the accuracy and reliability of material calculations are remarkably low, resulting in delays in construction due to errors in the calculation of material quantities. It causes many problems such as

According to the above-described conventional method, the demand for technology to support digitization of existing drawing data in image format is increasing and the ripple effect of related industries is high, but the development of automatic drawing recognition technology is slow due to difficulty in generating profit in a short period of time.

The present invention was developed to solve such a conventional problem, and in the process of digitizing equipment and symbols and calculating estimates during the front-end engineering design (FEED) process of design, design information is automatically recognized from imaged P&ID drawings and Its main purpose is to extract and digitize it more accurately and quickly, and use this automatically recognized design information to effectively use it in creating P&ID design drawings.

The present invention for achieving the above object is a method for automatically recognizing and classifying design information in an imaged P&ID drawing, the present invention for automatically recognizing and classifying design information in an imaged P&ID drawing, Extracting a symbol area after removing lines and text from the imaged P&ID drawing, and automatically registering the symbol, the origin and connection points of the symbol in the corresponding symbol area in a database; Recognizing and extracting the pre-registered symbols from the imaged P&ID drawing in four directions, and removing the extracted symbols from the imaged P&ID drawing; Removing the trim line from the imaged P&ID drawing from which the symbol has been removed, and recognizing and extracting the line using a sliding window method; Calculating an aspect ratio from the imaged P&ID drawing from which the symbol has been removed to calculate an area where the text exists, and then recognizing and extracting the text from the corresponding area with OCR; Classifying the text detected in the drawing area among the extracted texts into respective attributes through a predefined attribute classification system; Link the attribute values of the extracted symbol and line with the classified attribute of the closest distance to the symbol and line in the text, and, in the case of a device among the extracted symbols, based on the device name recognized in the text It includes;

In addition, following the step of linking the extracted design information, a step of rearranging symbols in order of flow mark of lines to generate a topology may be further included.

In addition, the extracted design information and topology can be generated as an intermediate file in a compatible XML format.

In addition, a symbol in which an additional symbol is registered is first inspected, an equipment among symbols is first inspected, and a nozzle is found and extracted around a device area to increase a recognition rate.

In addition, in the step of recognizing the symbol, the recognized symbol may be compared with the feature points of the stored symbol, and recognized as a stored symbol only when a matching degree of the recognized symbol is higher than a set threshold.

In addition, in the step of recognizing and extracting the line, when a connection point and a line, and a line and a line of the extracted symbol are connected but the coordinates are different, the line may be extracted by correcting the coordinates in pixel units.

In addition, when the text is not recognized by the method of recognizing the text, a corresponding text image may be stored and text mapped from the stored image to train OCR to increase the recognition rate.

If drawings are automatically digitized through the above method, most tasks such as drawing generation, material calculation, basic design information such as equipment list, line list, instrument list calculation, etc. are automatically generated by counting design elements with high accuracy in a short time. It is possible, and it helps to improve work productivity by eliminating the simple and repetitive task of manually calculating design elements by advanced engineers.

In addition, if the drawing is automatically generated using the data generated by the above method, rather than the conventional method in which the drawing was drawn directly from the imaged P&ID, design product consistency is maintained and design quality can be improved. This solves the problems such as waste of time, omission of items, and errors of plant engineering companies that have been caused by the conventional visual inspection and drawing of drawings.

In addition, when verifying the accuracy after 3D modeling, it is possible to quickly and accurately verify the accuracy of the created drawing because it can be compared with the stored data.

1 is a diagram showing an embodiment of an imaged P&ID diagram of the present invention.
2 is a flowchart illustrating a classification method according to the present invention.
3 is an exemplary view showing a step of setting an origin and a connection point in symbol registration of the present invention.
4 is an exemplary view showing a step of recognizing a symbol among design information of the present invention.
5 is a diagram showing an embodiment after removing the symbols of the imaged P&ID diagram of the present invention.
6 is an exemplary view showing a step of recognizing a line in design information of the present invention.
7 is an exemplary view showing that a text area is extracted in the step of recognizing text among design information of the present invention.
8 is an exemplary view showing a step of associating an extracted symbol and line with design information of the present invention.
Figure 9 is an exemplary view showing the steps of rearranging in order of the flow mark of the present invention.
10 is a diagram showing a hierarchy structure of a database created as an intermediate file of the present invention.

Hereinafter, a method of automatically recognizing and classifying design information in an imaged P&ID drawing according to the present invention will be described in more detail with reference to the accompanying drawings.

However, the present invention is not limited to the embodiments disclosed below, but may be implemented in a variety of different forms, only these embodiments are intended to complete the disclosure of the present invention, and are common in the technical field to which the present invention pertains. It is provided to fully inform the knowledgeable person of the scope of the invention, and the invention is only defined by the scope of the claims. For reference, when it is determined that detailed descriptions of known functions or configurations related to the present invention may unnecessarily obscure the subject matter of the present invention, detailed descriptions thereof will be omitted.

1 is an embodiment of an imaged P&ID drawing, in which symbols, lines, and text are all included. The symbol is a diagram of materials excluding lines and texts in the drawing area, and consists of equipment, instrumentation, fittings, and operation page connection (OPC). 1 is composed of symbols such as a valve 100, an instrument 110, a device 120, a nozzle 130 attached to the device, and an OPC 140. The OPC is a symbol indicating another drawing connected to the drawing, and a direction indicating a direction of a process and a number of a connected P&ID drawing are written inside the OPC. The line is a straight line connected to symbols in FIG. 1 and is composed of a process line and a utility line. The process line is a piping line in which the main work of the plant is performed, and the utility line is a line for helping the operation of a process line such as an electric signal and a control line. The text is a part that describes the symbol and line, and includes text 150 for describing the device and a line number 160 for describing the line.

As described above, in the past, design engineers recreated each P&ID drawing in the form of an image file, usually a PDF file, into a new P&ID drawing. Accordingly, data between design products is inconsistent and FEED (Front-End Engineering Design) data is It took unnecessary time to set up early, and this caused a big problem, especially in conducting overseas projects.

In order to solve these problems, aggregate design elements through a short time and high accuracy in the initial FEED process, and utilize them for estimating work, and to improve work productivity and design quality by excluding simple repetitive tasks of advanced engineers, the present invention It provides the following automatic classification method.

2 is a flowchart of a method for automatically recognizing and classifying design information in an imaged P&ID drawing according to the present invention.

First, a symbol to be recognized is registered in the expected symbol area (S200). Here, the symbol predicted area is automatically extracted from the entire drawing through the contour algorithm. A symbol list is created according to the classification system defined in advance in the automatically extracted symbol area, and the symbol is registered in the database based on this.

The contour algorithm for extracting the symbol predicted area is to extract the symbol predicted area by connecting the edge borders of portions having the same color or color intensity. In the present invention, the symbol predicted area is extracted by distinguishing the margin of the drawing from the symbol area. Registering the symbols to be recognized by the user one by one in the drawing is an inefficient operation that takes a lot of time, and extracting the expected symbol area is to reduce the time required.

3 is an exemplary diagram illustrating setting connection information such as an origin and a connection point of a symbol in a symbol when registering a symbol in a database. The symbol in the exemplary diagram is a valve symbol, where the red dot 300 at the center is the origin and the blue dots 310 at both sides are the connection points. The reason for setting the connection information on the symbol is that the starting point can be set when a line is recognized through the coordinates of the connection point of the recognized symbol, and this connection information can be used to automatically create a future P&ID design drawing.

In the step of registering the symbol, when a symbol is formed as a set of a plurality of additional symbols, it may be characterized in that the symbol is registered as one symbol including the additional symbols forming the set. As a result, since the characteristics of the symbol become clear and the recognition rate increases in recognizing the symbol, which is a subsequent step, a small portion is not registered as one symbol, and additional symbols forming a set are registered as one symbol.

Next, a symbol is recognized and extracted from the imaged P&ID drawing based on the database in which the symbol is stored (S210). At this time, the recognized symbol is removed from the imaged P&ID drawing. Through this, it is possible to reduce the time it takes to recognize a subsequent symbol, to reduce a false recognition rate, and to prevent a case of erroneously recognizing a symbol as a line when recognizing a line that is a subsequent step.

As a method of recognizing symbols, a method of recognizing one symbol in a drawing by rotating the drawing by 0, 90, 180, 270 degrees can be used. If a symbol is recognized in four directions, it is possible to recognize even an unregistered symbol in a database in which the symbol differing only in the direction is registered, thereby increasing the recognition rate. In addition, since the recognized symbols are removed, it does not take much time even if the symbols are searched one by one.

In the step of recognizing the symbol, by comparing the recognized symbol with the feature points of the stored symbol, the symbol may be recognized as a stored symbol only when the matching degree of the recognized symbol is higher than a threshold value set by the user (320). Through this method, the user can arbitrarily set a threshold to adjust the recognition speed and recognition rate.

In the symbol recognition step, the symbol in which the additional symbol is registered is first checked, and the symbol is recognized and removed from the drawing. This is because, if a symbol including an additional symbol is recognized first, it is possible to increase the recognition rate by reducing duplicate recognition. Also, it is possible to prevent misrecognition by recognizing the basic symbol first.

In addition, in the step of recognizing the symbol, the image map of the symbol may be enlarged/reduced and compared according to the size of the symbol and the complexity of the drawing, and the symbol may be recognized and extracted.

In the step of recognizing and extracting the symbol, an equipment among the symbols is first checked, and a nozzle is found and extracted around the device area. In general, nozzles are located around the device, so they look for nozzles around the device area, and the symbol of the nozzle is small, so if it is recognized in the entire drawing area, there is a high probability of misrecognition. If extracted, the recognition rate of the nozzle is increased, and the recognition rate of all symbols can be improved.

4 is an exemplary diagram showing comparison of symbols one by one in a P&ID imaged with a png extension name as an initial step of recognizing a symbol. 13 PSD symbols, 28 DCS INSTRUMENTs, and 64 LOCAL MOUNT INSTRUMENTs were recognized and extracted. Currently, the Insertion Blind Open symbol 400 displayed on the right is being searched.

FIG. 5 shows an embodiment after removing the symbol of the imaged P&ID drawing, and it can be seen that the symbol is removed when compared with FIG. 2. As described above, this increases the recognition speed in recognizing a symbol and prevents the case of erroneously recognizing a symbol as a line when recognizing a line.

Next, a line connected to the connection point of the removed symbol in the imaged P&ID drawing is recognized and extracted using a sliding window method (S220). Before line recognition, a small object such as a trim line is removed from the imaged drawing.

6 is an exemplary view showing recognition of a line by the sliding window method. The sliding window method recognizes a line by calculating it in blob units rather than pixel units, and the time it takes for recognition is shorter than that in pixel units. Based on the recognized symbol 600, the sliding window 610 is moved up/down/left/right at the connection point of the symbol to recognize the line, and if the line is not found while moving left/right, the line is searched up/down at the end point. Even if more than a certain portion of the sliding window is occupied, it is recognized as a line, so it is possible to recognize a line even if the pixels are separated. Since the length of the sliding window can be arbitrarily adjusted by the user, the recognition accuracy and speed can be adjusted and used.

When extracting the coordinates of the line, the coordinates of the connection point of the symbol to which the line and the symbol are connected and the coordinates of the end point of the line may not exactly match due to the thickness of the symbol and the line on the image. In this case, lines and symbols separated by pixel units are finely adjusted with the coordinates of the connection point of the symbol. This is necessary to accurately connect lines and symbols when creating a new P&ID using the database of the present invention. The fine adjustment is applied in the same way between the line and the line, so that when the center of the horizontal line/vertical line is not connected due to its own thickness, it is processed to be connected.

Next, an aspect ratio is calculated from the imaged P&ID drawing from which the symbol has been removed to calculate an area in which the text exists, and then the text is recognized and extracted in the corresponding area (S230). As a method of recognizing text, an existing text recognition program such as OCR (Optical Character Reader) can be used. Since the OCR implementation principle is known, detailed descriptions are omitted. It is natural that known text recognition methods other than OCR can be used. Due to the nature of P&ID drawings in which lines, symbols and text are mixed, the text recognition rate is lower than that of general documents containing only text. Therefore, there is a need for a method of calculating a character aspect ratio in a drawing, extracting an area where text exists, and recognizing only the text in the corresponding area.

7 is an exemplary diagram illustrating a method of calculating the text area, and in the method of calculating the area, first, an outline is extracted to remove lines and instrument bubbles. After that, if the recognized part is out of the aspect ratio set by the Bounding Box, the part is removed, and if the recognized part is within the set aspect ratio, it is left as a text area. The entire text area is extracted by creating a contour bounding box of the entire text area in a manner that dilates the part recognized as the text area to a preset threshold value, and if the next recognized part is also determined to be a text area, is left. The reason for setting and extracting the text area is to recognize the corresponding text area as a design information unit and use it to classify and link attribute information extracted from the text that is a later stage.

When the area where the text exists is extracted by the above method, the text is recognized by applying OCR to the text in the image of the extracted area. However, since the recognition rate of the current highest level of OCR is not 100%, misrecognition or non-recognition occurs, so it is necessary to increase the recognition rate by training text. If the text is not recognized by the method of recognizing the text, the text image is first stored and characters are mapped in each image. As a method of mapping characters, a method of mapping a character most similar to the character of the image or a method in which a user individually designates a corresponding character may be used. After that, training data is created using the mapping data, and the generated training data is converted into a database and applied to text recognition.

Next, among the extracted texts, the text detected in the drawing area is classified into each attribute through a predefined attribute classification system (S240). The drawing area is a part consisting of a set of symbols and lines, not a description part of the drawing, and the reason for classifying the area is that attributes to be classified may be different. When text is detected in the Note, Revision Data, Title Block, or Description area other than the drawing area, it is determined as a recognition element by checking whether the detected text is included in the area after setting the area of each element to be recognized.

Text properties detected in the drawing area are divided into Line Number, Size, Tag Number, Instrumnet Type, Serial Number, Serial Number, and P&ID Name. The property can be arbitrarily specified according to the user. Line Number follows the business owner designation form, and is combined with the separators such as Size, Fluid, Serial Number, and Insulation. Size is composed of a combination of numbers and special characters (/,"), and the tag number consists of combinations of alphabets, numbers, and special characters (-,/,"), and Instrument Type is It is specified in the project.

7 is an embodiment of classifying attribute information by extracting a line number, and attribute information such as Fluid (700), Unit (710), Sequence (720), Material (730), Size (740), Insulation (750), etc. It is classified as

Next, attribute information that each symbol may have is defined in advance, and attributes suitable for the attribute type of the recognized symbol and line are found in the drawing and connected with the nearest attribute (S250). By linking the attributes, it is possible to grasp the devices and quantities required when calculating an estimate, model symbols and lines when creating a new P&ID in the future, and display text describing the symbols and lines. By defining attribute information in advance, it is possible to prevent an error in linking unnecessary or incorrect attributes to a symbol. The extracted symbol and line are associated with the attribute of the extracted text based on the nearest distance, and when the extracted symbol is an equipment, the extracted symbol and line are linked based on the device name recognized in the text. In the case of a device, the device name uses the Description in the Description field.

Symbol and line properties can be divided into Process/Utility line, Reducer, Equipment, Nozzle, Instrument, and OPC. The property can be arbitrarily specified according to the user. Process/Utility line uses Line Number among the attributes classified in text string, Main Size x Sub Size for Reducer, Tag Number for Equipment and Nozzle, Type, Serial Number for Instrument, and P&ID Number for OPC.

8 is an embodiment for explaining the step of associating the extracted symbol with the line. In 800, “E-234-009” in the Description area as a device among symbols is associated with a corresponding device as a device name. Is shown. 810 represents that "P-234-03303-CD3D-2"-N", "P-234-03305-CD3D-3"-N", which are closest to the line recognized as a line, are linked by Line Number, and 820 Is a symbol of Pressure Safety Valve that describes the connection of "PSV 2907" to the instrument type and serial number.

Through the present invention, a bill of materials (BOM) can be created based on design information linked to attribute information, or a design estimate required in the FEED process can be automatically calculated using equipment, instruments, etc. However, it can be used to automatically create a P&ID later by creating an intermediate file in XML format through topology creation by integrating objects and objects.

FIG. 9 is a schematic diagram of creating a topology by integrating objects and objects by connecting symbols connected to lines and lines through a recursive calling technique following the step of linking design information and rearranging the connected symbols in from-to order of lines. will be. First, the linkage relationship between recognized objects is defined as follows. Process/Utility Line may start at the device and end at the device, start at the line and end at the device, or start at the line and end at the line. In addition, the device has a nozzle dependent thereon, and the P&ID drawing has a linkage with other P&IDs.

The recursive algorithm means that an arbitrary function calls itself, and in the present invention, an arbitrary symbol or line calls another symbol or line connected by using the stored connection information, and the process continues. Repeatedly, the line and the symbol connected to the line are connected and rearranged in the from-to order.

To create a topology, first connect a line and each symbol connected to the line, and rearrange the connected symbols according to the flow mark of the line. If the connection is broken in the process of connecting the line and each symbol connected to the line, it is necessary to adjust the coordinates based on the center line to secure connectivity. The method of sorting in from-to order according to the flow mark is to find the line or object that connects From-to with the line connected to from or to in the line list as a starting point, and sort them in order. In FIG. 9, before alignment in the flow mark direction, the same symbols 2 and 3 were successively extracted during the process of recognizing the symbol, so that the symbols were arranged in no order to form a topology. They were arranged in order from left to right. If the topology is not generated in the from-to order, when creating a new P&ID based on the database created by the method of the present invention, a symbol must be modeled at random only with coordinate points, which results in a drawing different from the imaged P&ID drawing. Can be. Therefore, when creating a new P&ID drawing, it is necessary to create a topology to automatically create a design drawing that was done manually.

The process line and the utility line are not distinguished when extracting lines, but they can be distinguished during the topology generation process. In distinguishing lines, there may be parts that are originally incorrect or parts that are not recognized because it is most accurate to distinguish according to the objects connected to the process line and utility line in the process of connecting objects. When the connected object is an instrument, the line is classified as a utility line, and in other cases, it is classified as a process line.

In the process of connecting the device, the nozzle may be attached to the device or may be included in the shape, so a nozzle that overlaps the device shape is found and connected to the device. This is because the nozzles are generally small in shape near the device.

The process of connecting with other P&IDs can be included by using the P&ID Number displayed on the OPC. If the P&ID name displayed on the OPC and the actual file name are different, it is necessary to establish a relationship between the two. In addition, since there may be multiple P&IDs connected to one P&ID, the P&ID information connected to the P&ID is stored so that P&ID information on other pages can be easily obtained.

Next, it may be characterized by generating the extracted design information and topology as an intermediate file in a compatible XML format. 10 is an exemplary diagram of a hierarchy structure of a topology constructed through linkage between objects created as intermediate files. For Equipment, a device number is stored, and a nozzle stored in the order of "p1, p2, v1, v2" is linked to the stored device number. In this way, it shows that the symbols are connected in the order of "FOO41, Valve, Reducer, Foo41" in the line with the line number "8"-DMW-UW10029-AR5W-HC".

Claims (11)

As a method of automatically recognizing and classifying design information in imaged P&ID drawings,
Extracting a symbol area from the imaged P&ID drawing, setting the origin and connection point of a corresponding symbol in the symbol area, and automatically registering the symbol in a database;
Recognizing and extracting a symbol automatically registered in the database from the imaged P&ID drawing, and recognizing and extracting the extracted symbol from the imaged P&ID drawing;
Recognizing and extracting a line from the imaged P&ID drawing from which the symbol has been removed using a sliding window method that calculates a blob unit instead of a pixel unit;
Calculating an aspect ratio from the imaged P&ID drawing from which the symbols have been removed to calculate an area where the text exists, and then recognizing and extracting the text from the corresponding area;
Classifying the text detected in the drawing area among the extracted texts into respective attributes through a predefined attribute classification system;
Associating the extracted symbol and line with the attribute of the extracted text based on the nearest distance, and associating the extracted symbol based on the device name recognized in the text when the extracted symbol is an equipment. How to automatically recognize and classify design information in imaged P&ID drawings
The method according to claim 1,
In the registering of the symbol, when a symbol is formed as a set of a plurality of additional symbols, registering as one symbol including the additional symbols forming the set,
In the step of recognizing and extracting the symbol, the registered symbol including the additional symbol is first inspected and recognized. A method of automatically recognizing and classifying design information in an imaged P&ID drawing.
The method according to claim 1,
In the step of recognizing and extracting the symbol, design information is automatically extracted from the imaged P&ID drawing, characterized in that the symbol automatically registered in the database is rotated by 0, 90, 180, and 270 degrees to recognize and extract each symbol from the drawing. How to recognize and classify
The method according to claim 1,
In the step of recognizing and extracting the symbol, design information is automatically recognized and classified in an imaged P&ID drawing, characterized in that the device (equipment) among the symbols is first checked, and a nozzle is found and extracted around the device area. How to
The method according to claim 1,
The step of recognizing and extracting the symbol includes comparing the recognized symbol with the feature points of the symbol registered in the symbol registration step, and recognizing the symbol as a registered symbol only when the matching degree of the recognized symbol is higher than a set threshold. How to automatically recognize and classify design information in imaged P&ID drawings
The method according to claim 1,
In the step of recognizing and extracting the line, when the extracted symbol and the line are connected but the coordinates are different, or the line and the line are connected but the coordinates are different, the coordinates are corrected for each pixel to extract the line. How to automatically recognize and classify design information in imaged P&ID drawings
The method according to claim 1,
The step of recognizing and extracting the text comprises recognizing the text as OCR, and when the text is not correctly recognized, the imaged P&ID, characterized in that for training OCR by accurately mapping the text when the text is not correctly recognized. How to automatically recognize and classify design information in drawings
The method according to claim 1,
The linking of the extracted design information further comprises generating a topology by connecting a line and a symbol connected to the line through a recursive calling technique, and rearranging the connected symbols in from-to order of the line. How to automatically recognize and classify design information in imaged P&ID drawings

The method of claim 8,
A method for automatically recognizing and classifying design information in an imaged P&ID drawing, characterized in that generating the extracted design information and topology as a compatible XML format file

The method of claim 8,
In the step of generating the topology, if the object connected to the line is an instrument among symbols, it is classified as a utility line, and otherwise, as a process line, design information is automatically recognized and classified in an imaged P&ID drawing.
The method of claim 8,
The step of generating the topology comprises automatically recognizing and classifying design information in an imaged P&ID drawing, characterized in that the extracted text is connected to another P&ID by using a P&ID No appearing on an OPC (Operating page connection).

Images