JP2006227824A - Drawing recognition method and device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To perform recognition processing of CAD data from electronic data on a drawing wherein a character and a figure are mixed, in a short time with high accuracy. <P>SOLUTION: This drawing recognition device has: a raster symbol pattern recognition means 5 performing recognition processing of a symbol with raster drawing data as input; a diagramming symbol pattern recognition means 7 performing symbol recognition with editing enabling format drawing data as input from candidates of the symbol recognized by the raster symbol pattern recognition means 5; a symbol recognition means 9 imparting ID information as the CAD data to the symbol recognized by the diagramming symbol pattern recognition means 7; a character area estimation means 10 area-dividing the drawing in reference to a recognized symbol position, and grouping a character string reflecting the area division; a character string recognition means 11 recognizing the character string of an estimated character area as attribute information of the symbol; and an integration recognition means 12 integrating the recognized symbol position and attribute information, and the ID information, and storing it in a CAD database. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a method and apparatus for recognizing necessary information as engineering data from the representation of drawing information including mixed characters, symbols, and line segments as electronic data, and more particularly to character strings / symbols represented by drawings and mutual information. The present invention relates to a method and apparatus for automatically recognizing relationships and recognizing them as electronic data that can be used for design data.

  Conventionally, as a technique of reading drawing information in which characters, symbols, and line segments are mixed as electronic data that can be used as design data, the character, symbol, In some cases, the relative positional relationship between line segments and a registered drawing rule are compared, and the recognition result is corrected using the knowledge to improve the accuracy.

  In Non-Patent Document 1, recognition accuracy is improved by performing recognition processing again on the recognized character, symbol, line segment, and a relative positional relationship registered in advance between them. The way it is shown.

JP 2001-92967 A (page 4-6, FIGS. 3 and 6) Kasahara et al., “Development of drawing recognition system”, 2003 IEEJ Symposium, 2003. Image Processing-Text Processing Standard Textbook, Foundation for Image Information Education. Graphic file format handbook, by David C. Kay / John R. Levine, translated by MbCD, ASCII Publishing. “DXF Handbook”, Ohmsha. Programming Microsoft Windows, Nikkei BP Soft Press Seiichi Nakagawa, “Pattern Information Processing”, p153, Maruzen. Seiichi Nakagawa, “Pattern Information Processing”, pp 239-245, Maruzen. Mike Heath, et.al “Comparison of Edge Detectors”, Computer Vision and Image Understanding Vol.69, No.1, pp.38-54, January, 1998 Otsu et al., “Pattern Recognition”, pp62-64, Asakura Shoten J. O’Rouke “Computational Geometry”, Cambridge University Press, 1988.

  When extracting character strings and symbols from the image data of drawings as in the method described in Patent Document 1, especially when there is a limit to the recognition accuracy of character strings and data that does not allow errors such as engineering data, the final confirmation It takes a lot of work. The effort required for confirmation is considered to be smaller as the recognition error is smaller, and the final quality is considered to be at least higher than the automatic recognition accuracy.

On the other hand, although it has not been given meaning as design information in recent years, only a line and a character string have been widely distributed in a data format that can be edited separately (Non-Patent Document 2, Non-Patent Document 3, Non-patent). If this is used, it is considered that character string information at a specific position can be obtained with 100% accuracy. However, the data described in these data formats have the following problems because they are only apparent data that does not consider meaning.
(1) Characters may not be grouped. (For example, when the device ID is “RA-124”, if “RA” and “124” are broken due to space limitations, they may be treated as different character strings in the editable format. is there.
(2) In general, symbols are also described separately as a collection of lines. In general, a very large amount of calculation is required for recognizing a disjoint line as a symbol by following the connection relation from the positional relation of the end points. In an extreme case, even a single line segment may be divided at an arbitrary position, and may be handled as continuous data of a short line segment inside.

  An object of the present invention is to obtain desired engineering data with high accuracy and high speed from drawing data in a vector format composed of character information and line information.

  In order to achieve the above object, the present invention makes it possible to edit image data of a drawing including characters and symbols, that is, raster drawing data, and the drawing in a format in which characters and figures can be individually edited. By using together with the format drawing data, rough symbols are recognized with the raster drawing data, and among the candidates, the recognition target is further narrowed down using the editable format drawing data.

  Specifically, the present invention that achieves the above-described object includes an image storage device that stores raster drawing data of drawings including characters and symbols, and editable format drawing data that can individually edit characters and graphics. Editable format drawing data storage device for storing the drawing expressed in the above, a CAD database for storing recognition results as CAD data, and a raster pattern storage device for storing a raster pattern in which symbols are expressed in a raster drawing data format A plotting pattern in which the symbol is described in an editable format format, a plotting pattern storage device that stores the relative positional relationship between the symbol and the attribute character, and ID information of the symbol type on the CAD database; A symbol library storage for storing information for associating the raster pattern and the diagrammatic pattern A raster symbol pattern recognizing means for collating a raster pattern stored in the raster pattern storage device with raster drawing data stored in the image storage device and narrowing a range of symbol positions in the drawing; The plotting pattern stored in the plotting pattern storage device is compared with the editable format drawing data stored in the editable format drawing data storage device, and the editable format format is used from the narrowed range. A diagrammatic symbol pattern recognizing unit for recognizing the described symbol, and ID information as symbol position CAD data is given to the recognized diagrammed pattern by referring to information in the symbol library storage device. Symbol recognition means and a character area for estimating a character area of raster drawing data stored in the image storage device The character string recognition means for recognizing the character string of the estimated character area, the character string position information and the content of the character string, and the symbol position recognized by the diagrammatic symbol pattern recognition means And a unified recognition means for recognizing the position of the symbol and the attribute character string related to the symbol and storing them in the CAD database, and a drawing recognition apparatus for recognizing characters and symbols in the drawing and their mutual relationship.

  According to the above configuration, when a symbol is recognized from electronic data of a drawing including characters and symbols, the symbol to be recognized is represented by raster drawing data from image data such as bitmap data of the drawing, that is, raster drawing data. Narrow down the candidate (position) of the symbol with reference to the raster pattern, and use the diagram pattern that describes the symbol in the editable format for the narrowed down candidate (position). Recognize symbols from the rendered editable format drawing data.

  Therefore, using a diagram pattern in which symbols are described in an editable format, the amount of calculation is reduced compared to the case where symbols are recognized from the entire editable format drawing data in which the drawing is expressed in an editable format. The time required for recognition is reduced.

  The character area estimation means is preferably configured to divide a character string from raster drawing data using position information of recognized symbols. As described above, the character string recognition accuracy is improved by dividing the character string from the raster drawing data using the position information of the recognized symbol.

  Further, it is desirable to provide a drawing position correspondence data means for setting a correspondence relationship between the raster drawing data and the editable format drawing data.

  Further, the editable format drawing data is input, and the character information of the data is extracted together with the position information, and the editable format drawing data is deleted from the editable format drawing data, It is good also as a drawing recognition apparatus containing the non-character part image data generation means converted into raster drawing data. By providing such means, even when only editable format drawing data is available, raster drawing data in which characters are deleted from editable format drawing data is generated, and the raster drawing data and editable Drawing recognition by the drawing recognition apparatus using the format drawing data can be performed.

  The above-described problem is also stored in the raster drawing data of the drawing including characters and symbols, and the procedure of storing the drawing expressed in the editable format drawing data capable of editing the character and the graphic individually. The raster drawing data and the raster pattern in which the symbol is expressed in the raster drawing data format in advance are referred to, and the range of the position of the symbol on the drawing is narrowed, and the diagram is a diagram pattern in which the symbol is described in the editable format format. Referring to the drawing expressed in the editable format drawing data capable of individually editing the character pattern and the figure, and the information for associating the raster pattern with the diagrammed pattern, on the narrowed drawing A procedure for recognizing a symbol described in an editable format from a range of positions, and a diagram of the recognized symbol The ID information as symbol position CAD data by referring to the ID information of the symbol type on a preset CAD database, and the recognized symbol position, A procedure for estimating the character area of the stored raster drawing data, a procedure for recognizing the character string of the estimated character area, position information of the character string and the contents of the character string, pre-stored symbols and attribute characters And a procedure for recognizing a symbol position and an attribute character string related to the symbol and storing them in a CAD database using the recognized symbol position, and the character, symbol, and symbol in the drawing. This is also achieved by a drawing recognition method for recognizing the mutual relationship.

  This method also reduces the time required for recognizing the drawing, as described above.

  According to the present invention, drawing information in which characters, symbols, and line segments are mixed can be automatically recognized with higher accuracy and in a shorter time than when using only editable format drawing data.

Embodiments of the present invention will be described below with reference to the drawings.
(First embodiment)
FIG. 1 is a block diagram showing the overall configuration of a drawing recognition apparatus according to the first embodiment of the present invention. The drawing recognition apparatus shown in the figure is roughly divided into a computer, an image data input device 1, a data output device 16, and an interface connected to the computer.

  The interface includes a display screen, a keyboard, a mouse, and devices for controlling them, and includes drawing position correspondence data setting means 18, recognition result correction means 14, and recognition result display means 13 as software.

  The image data input device 1 is a device in which a scanner having a function of reading a paper drawing as a raster drawing data format, processing software, and a file input device for reading editable format drawing data are combined.

  The data output device 16 is a device that outputs data held in the computer to a storage medium such as a CD-ROM.

  The computer includes a RAM, a CPU, and a storage device. The RAM temporarily holds the processing results. The image storage device 2 holds raster drawing data input from the image data input device 1, and editable format drawing data input from the image data input device 1. It comprises an editable format drawing data storage device 3 to be held and a drawing position correspondence data storage device 17 to hold data inputted from the drawing position correspondence data setting means 18.

  The CPU executes processing of each software of the raster symbol pattern recognizing means 5, the diagrammatic symbol pattern recognizing means 7, the symbol recognizing means 9, the character area estimating means 10, the character string recognizing means 11, and the integrated recognizing means 12.

  The storage device retains data permanently, and corresponds to a hard disk and its control device, a raster pattern storage device 6, a diagrammed pattern storage device 8, a symbol library storage device 15, and the recognition result correction. Means 14 and a CAD database 4 in which the processing results of the integrated recognition means 12 are stored. The raster pattern storage device 6 stores a raster pattern as a reference for symbol recognition in advance.

  When a paper drawing is input to the scanner of the image data input device 1, raster drawing data is output from the scanner and stored in the image storage device 2. When the editable format drawing data is input to the file input device of the image data input device 1, the input data is stored in the editable format drawing data storage device 3.

  The types of raster drawing data include BMP (bitmap) format (Non-Patent Document 2), tiff format (Non-Patent Document 3), and the like. The editable format is distinguished from characters and graphics in the DXF (Autodesk) format (non-patent document 4), Windows (registered trademark) (non-patent document 5) metafile (Microsoft) format, etc. A format that can be registered together with the coordinates. Further, there is a one-to-one correspondence between the specific position coordinates (X, Y) in the raster drawing format and the coordinates (X ′, Y ′) in the corresponding editable format at the same point. Yes, the corresponding point coordinates can be calculated mutually.

  The setting of the two correspondence relationships is executed by the drawing position correspondence data setting means 18 incorporated in the interface. FIG. 15 shows a display example on the interface. In this example, a raster drawing (for example, BMP format) is displayed on the left side, an editable format drawing is displayed on the right side, and corresponding points are sequentially set with a pointing device, thereby setting corresponding points in the drawing. In the example of FIG. 15, the corresponding points 1 are set so that the points 1501 and 1502 correspond to the corresponding points 3, and the points 1503 and 1504 correspond to each other. How many set points should be associated depends on the method used, but when using the most common method, affine transformation (Non-Patent Document 10), if a correspondence relationship of at least 4 points is set, Correspondence can be defined across the two drawings. In FIG. 15, the corresponding points 2 and 4 are not shown. Data set by the interface of FIG. 15 is stored in the drawing position correspondence data storage device 17.

  FIG. 16 shows an example of drawing position correspondence data stored in the drawing position correspondence data storage device 17. A set of corresponding points in the raster drawing (image data drawing) and the editable format drawing is stored in a table format. Even if there is no interface for performing the display as shown in FIG. 15, the user can create the data shown in FIG. 16 and import it into the drawing position correspondence data storage device 17 to realize the same function.

  FIG. 2 shows the DXF format (Non-Patent Document 4) as one example of the editable format drawing data, and the lines are displayed as a set of end points connecting the lines with POLYLINE tags. On the other hand, the character string is distinguished from the line by the TEXT tag and is described together with the rectangular coordinates of the character region and the direction of the character.

  FIG. 3 shows an example of a raster diagram of a single-line diagram. Such raster diagram data is, for example, a value of 0 or 1 for each coordinate position in the BMP monochrome format. Is retained.

  On the other hand, it is necessary to extract and register the meaning represented by the drawing as CAD data. In the example of FIG. 3, four devices are connected to the bus, and it is necessary to recognize what each device is from the shape of the figure hanging from the bus and convert it to CAD data. The device number and sequence number of each device are described in the lower table of the figure, and these data need to be converted into CAD data together with the device type.

  FIG. 4 shows a schematic processing procedure for creating CAD data by using the apparatus having the configuration shown in FIG. 1 and using two types of formats: raster drawing data and editable format drawing data.

  First, in step 200, raster image data and editable format drawing data are read by the image data input device 1 and stored in the image storage device 2 and editable format drawing data storage device 3, respectively.

  In step 201, the raster symbol pattern recognition means 5 performs pattern matching between the reference raster pattern stored in the raster pattern storage device 6 in advance and the raster drawing data stored in the image storage device 2. The symbols in are recognized. As a method therefor, a template matching method (Non-patent Document 6) that directly shapes the shape, a combination of partial features (Non-Patent Document 7), a Fourier transform, a wavelet transform, or the like for these processes. A known method such as a method using image processing (Non-patent Document 8) can be used. These templates and partial features necessary for processing are registered in the raster pattern storage device 6 in advance. FIG. 13 shows an example of a raster pattern of valve symbols. This example is an example of a symbol template of the BMP file format expressed by 0 and 1, where 1 indicates a line. In the process of the procedure 201, since it is premised that further detailed collation is performed in the process of the next procedure 202, even if the patterns do not exactly match, the symbol candidate area (position) is extracted.

  The process of step 202 is executed by the diagrammatic symbol pattern recognition means 7. For the symbol candidate area (position) narrowed down in the process of step 201, the chart symbol pattern (diagram pattern) registered in the chart pattern storage device 8 in advance is used as a reference. The final symbol recognition is executed by comparing with the editable format drawing data stored in the editable format drawing data storage device 3. FIG. 5 shows an example of a plotting pattern that is stored in the plotting pattern storage device 8 in advance and serves as a reference for recognition. The diagram pattern is described by the position coordinates of a line existing at a position relative to the reference coordinate position of the symbol pattern. For this reason, a large amount of calculation is required when collating the pattern of such a diagram in the entire drawing without narrowing down the symbol position candidates as in the present invention.

  The left side of FIG. 5 is a raster display so that the pattern object can be seen visually, and the location of reference numeral 501 is the reference point. The right side shows a part of the corresponding diagram pattern. The raster pattern 505 corresponds to the raster pattern 503, the raster pattern 504 corresponds to the raster pattern 507, and the raster pattern 507 corresponds to the raster pattern 504. The diagram pattern 506 corresponds to 502. In the diagram pattern, it is represented by relative position coordinates from the reference point, and in this example, the y coordinate is downward. For example, the portion 505 is a polyline consisting of 4 points, and the coordinates of 4 points. (−200.0,50.0), (−200.0,70.0), (−150.0,70.0), and (−150.0,50.0). Such a notation is common. Actually, there may be a case where the position of the pattern narrowed down by the raster drawing data is slightly deviated, so a line is searched with a margin around this area. Also, the grouping as polylines in the editable format may not always match the pattern, so the search is a line for each line connecting each point defined in the polyline It is carried out by checking whether or not.

  Procedure 203 is executed by the symbol recognition means 9. In step 203, for the pattern recognized in step 202, using the symbol library defined in the symbol library storage device 15, the symbol type ID for use in CAD, and the number and type of attributes to be assigned Is given. These are used in the integrated recognition of procedure 206. FIG. 14 shows an example of a symbol library defined in the symbol library storage device 15. In this example, a symbol library corresponding to the diagrammed pattern A in FIG. 5 is registered, and a symbol ID in CAD and accompanying attribute data are set.

  Next, the procedure 204 is executed by the character area estimation means 10. In step 204, attribute character string areas necessary for CAD data conversion are recognized as a group. Some character strings in the drawings are not necessarily required for CAD data conversion. In addition, for example, the character string data such as the valve attribute is line-breaked in the middle due to space or the like, as in the sequence numbers in the lower table of FIG. 3, such as “QA-” and “231”. Even if there is a line break, it is necessary to recognize “QA-231” as one attribute character string.

  In such a case, as a method for improving the recognition accuracy by using both the raster drawing data and the editable format drawing data, (1) the relative positional relationship between the position of the symbol and the attribute character described in Patent Document 1 is used. (2) If you straddle lines other than characters and symbols, use the fact that it is unlikely to be an attribute string of the same group, and (3) Other than symbols There are three known means: a method of grouping character string attributes by distinguishing remaining lines and character strings. Here, as an example, a detailed processing procedure for grouping character string regions in step 204 will be described using a method using these three in combination.

  FIG. 6 shows the flow of the process in step 204. Roughly speaking, the remaining straight lines and character lines other than symbols are represented based on the lengths of the line segments when the outline is approximated to the line segments. Separate the drawing using the rectangular area of the symbol, the non-character line in the figure, and the rectangular area of the symbol, and increase the threshold for grouping the character strings across the areas to Group.

  First, in step 601, the symbol area already recognized in step 203 is deleted from the raster drawing. You can delete the area recognized as a symbol from the raster drawing, or if the raster drawing is generated from editable format drawing data, the polyline corresponding to the symbol of the editable format drawing data This can be achieved by rasterizing the data only.

  Next, in step 602, the outline of the drawing area left in step 601 is generated. This can be easily realized by Canny's algorithm (Non-Patent Document 8). The line segment is approximated by the approximation in which the contour line is set. This can be realized by approximation such that a deviation from an approximate line segment in a certain pixel is equal to or less than a threshold (Non-Patent Document 9). Usually, the attribute character string to be CAD data is less than a certain size range, so if one line approximated by a line segment is larger than a certain threshold value, it cannot be considered as an attribute character line of CAD data. It is considered as a non-character line.

  In step 603, it is determined whether one line approximated to the line segment is larger than a threshold value, and only a long line segment equal to or greater than the threshold value is extracted. The threshold value is set in the entire drawing or defined in the attribute position data of the symbol pattern data as 511, 512, and 513 in FIG. 511, 512, and 513 in FIG. 5 are data that define the presence of attribute character strings corresponding to the positions of 508, 509, and 510, respectively. The position of the attribute character string is polyline data, and the attribute name is an attribute tag. For example, in the example of 511, it is defined as “device name”, and the size of the character string is defined as a character size tag, and in the example of 511, the size of a rectangle of one character is defined as 20 pixels. Has been. For example, a line segment having a length twice or more than the size of the character string defined in this way is not considered a part of the character string, and is therefore regarded as a line other than a character.

  In step 604, the drawing is divided into regions as shown in FIG. 7 using the lines recognized in this way and the upper, lower, left and right lines of the rectangular area of the symbol.

  In step 605, the neighborhoods of the contour line composed of short straight lines are grouped in consideration of both the region division in FIG. 7 and the attribute positions for the respective symbols shown in FIG. FIG. 8 shows only the character string and the boundary line by deleting the lines other than the symbol area and the character from the original FIG. In this example, grouping does not compete across the boundary of the attribute character string in FIG. 5 and the boundary. As a result, the character strings are grouped as shown by the dotted frame in FIG. When grouping, by setting a threshold value or the like, the character string region candidates in the same attribute range defined in FIG. 5 are divided into the same group and dividing lines for dividing the region in FIG. Crossing groups are grouped to avoid as much as possible. If these two requests cause a conflict, the position of the line that sets the region (if the line or symbol that sets the region is far from the character string to be grouped) The dividing line is considered to be less effective), the type of attribute string (depending on the attribute, the attribute pattern in FIG. Grouping is performed with comprehensive consideration of name restrictions (alphabet-4-digit pattern, etc.).

  When the character string area is estimated and grouped in step 204, the process proceeds to step 205. In step 205, the character string area estimated in step 204 by the character string recognition means 11 and the editable format drawing data stored in the editable format drawing data storage device 3 are shown in the example of FIG. A recognition character string is generated by comparing the rectangular coordinates of the TEXT defined as described above and extracting characters in the estimated area from the editable format. Alternatively, it is possible to activate character recognition software and extract a character string for the estimated region.

  The generated character string is given as an attribute character string to each symbol by the integrated recognition means 12 as conceptually shown in FIG. 10 (step 206). Which character string is assigned to which symbol is mainly limited by the positional relationship between the symbol and the character, the character type, and the character content. The data (CAD data) finally stored in the CAD database 4 is a set of symbol type and position and recognized attribute data as shown in FIG.

  In step 207, the generated CAD data is displayed on the display screen by the recognition result display means 13, and is confirmed by the user. If necessary, the user corrects the displayed data by the recognition result correcting means 14 via the keyboard. As the recognition result display means 13, for example, CAD software such as Auto CAD, which can display DXF data, or CAD software specific to piping design and map data creation can be used. The recognition result correction means 14 can use the same CAD software used for the display, but for example, use an interface screen for inputting correction of attribute data in a table format as shown in FIG. Is also possible.

  The confirmed / corrected CAD data is registered in the CAD database 4 in step 208. The registered CAD data can be taken out from the data output device 16 and used as necessary.

According to the present embodiment, when a symbol is recognized, first, a region in which the symbol exists in the raster drawing data by pattern matching the reference raster pattern and the raster drawing data stored in the image storage device 2. And then editable format drawing data using the symbolized symbol pattern registered in advance in the diagrammed pattern storage device 8 for the narrowed candidate region (position) of the symbol The final symbol recognition is executed by comparing with the editable format drawing data stored in the storage device 3. Therefore, the target area when collating with the editable format drawing data stored in the editable format drawing data storage device 3 using the diagrammatic symbol pattern registered in the plotting pattern storage device 8 as a reference However, the time required for drawing recognition, that is, conversion of necessary information into CAD data is shortened.
(Second Embodiment)
In the present embodiment, only the editable format drawing data is input, and the editable format drawing data is converted into raster drawing data such as BMP, and the same effects as in the first embodiment are obtained. can get. Such conversion from editable format drawing data to raster drawing data such as BMP is generally possible.

FIG. 11 shows a functional block diagram according to the second embodiment of the present invention. The configuration of the present embodiment is different from the configuration of the first embodiment in the following points (i) to (c), and the other configurations are the same as those of the first embodiment. The reference numerals are attached and the description is omitted.
I. The image data input device 1101 includes non-character part image data generation means 1122 and character string data extraction means 1121. The RAM comprises a non-character part image data storage device 1102, a non-character editable format drawing data storage device 1103 A, and a character limited editable format drawing data storage device 1103 B. c. FIG. 12 is a schematic flowchart showing a processing procedure according to the present embodiment. First, in step 1200, editable format drawing data is input in the image data input device 1101. Next, in step 1201, the character string data extraction unit 1121 analyzes the data, extracts only the character string data, and stores the character string and the character string in the character limited edit enable format drawing data storage device 1103B. Data consisting of areas (position information) is stored. For example, in the case of the DXF format, such extraction can be easily executed because the TEXT tag is attached to the character string as shown in FIG.

  On the other hand, the non-character portion image data generating means 1122 generates editable format drawing data in which only the portion with the tag of the character string is removed, and raster drawing data from which the character string is removed based on the data. Is generated and stored in the non-character part image data storage device 1102 (procedure 1202). The character string is removed in advance in the editable format drawing data because the character is extracted in advance as electronic data in the character-limited editable format drawing data storage device 1103. This is because if the column remains, it can be confused with the symbol. The editable format drawing data from which only the portion with the tag of the character string is removed is stored in the editable format drawing data storage device 1103A other than characters.

  Next, in step 1203, the raster symbol pattern recognition means 1105 performs the recognition processing of the raster symbol pattern using the raster pattern stored in the raster pattern storage device 1106 to narrow down the symbol position. Symbol positions are extracted by the pattern recognition means 1107 with reference to the diagrammed pattern storage device 1108. In step 1204, the character area estimation unit 1110 uses the extracted symbol position and the relative positional relationship between the symbol and the attribute character stored in the diagram pattern storage device 1108 in the format of FIG. The character area is estimated. This is very different from the first embodiment. In the first embodiment, the remaining portion that is neither a symbol nor a line is subjected to contour processing, and is used as an initial candidate for character region grouping. In the embodiment, the character limited edit enabled format drawing data stored in the character limited edit enabled format drawing data storage device 1103 is used as an initial candidate for grouping. As in the first embodiment, the threshold value for grouping is set based on the recognition result of the symbols as shown in FIG. 8 so that the boundaries of the divided area lines are not easily exceeded. .

  The processing after the procedure 1205 is as described in FIG. In the procedure shown in FIG. 4, the character string areas are grouped in step 204 and then recognized in step 205. In this embodiment, the editable format drawing data uses the TEXT shown in FIG. Like the line, each character is already known, so there is no need to recognize the character again.

  In the first and second embodiments, both the raster drawing data and the editable format drawing data are used for symbol recognition, but either one can be realized.

  The present embodiment is an example of a method of generating raster drawing data from editable format drawing data when there is no raster drawing data, and recognizing the position of the symbol pattern using the generated raster drawing data. As in the first embodiment, the target area when collating with the editable format drawing data is limited using the diagrammatic symbol pattern as a reference, and the amount of calculation can be reduced. Therefore, the time required for converting necessary information into CAD data is shortened.

1 is a block diagram showing an overall configuration of a drawing recognition apparatus according to a first embodiment of the present invention. It is a figure which shows the example of the editable format drawing data based on the 1st Embodiment of this invention. It is a top view which shows the example of drawing used as the object of CAD data generation. It is a flowchart which shows the example of the process sequence of drawing recognition which concerns on the 1st Embodiment of this invention. It is a figure which shows the example of the diagram pattern based on the 1st Embodiment of this invention. FIG. 5 is a flowchart showing an example of a procedure for character area grouping in the flowchart shown in FIG. 4. FIG. It is a figure which shows the example of progress of the area | region division process of drawing for character area grouping which concerns on the 1st Embodiment of this invention. It is a figure which shows the example of progress of the area | region division process of drawing for character area grouping which concerns on the 1st Embodiment of this invention. It is a figure which shows the example of the result of the character area grouping process which concerns on the 1st Embodiment of this invention. It is a figure which shows the example of the conversion result to CAD data with an attribute which concerns on the 1st Embodiment of this invention. The block diagram which shows the function structure of the drawing data recognition apparatus which concerns on the 2nd Embodiment of this invention. The flowchart which shows the implementation procedure of drawing data recognition which concerns on the 2nd Embodiment of this invention. It is a figure which shows the example of the raster pattern which concerns on the 1st Embodiment of this invention. It is a figure which shows the example of the symbol library which concerns on the 1st Embodiment of this invention. It is a top view which shows the example of the input screen of the drawing position corresponding | compatible data setting means based on the 1st Embodiment of this invention. It is a figure which shows the example of the drawing position corresponding | compatible data which concern on the 1st Embodiment of this invention. It is a top view which shows the example of the input screen of the recognition result correction means which concerns on the 1st Embodiment of this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Image data input device 2 Image storage device 3 Editable format drawing data storage device 4 CAD database 5 Raster symbol pattern recognition means 6 Raster pattern storage device 7 Diagram symbol pattern recognition device 8 Diagram pattern storage device 9 Symbol recognition Means 10 Character region estimation means 11 Character string recognition means 12 Integrated recognition means 13 Recognition result display means 14 Recognition result correction means 15 Symbol library storage device 16 Data output device 18 Drawing position correspondence data setting means

Claims (5)

  An image storage device for storing raster drawing data of drawings including characters and symbols, and an editable format drawing for storing the drawings expressed by editable format drawing data capable of individually editing characters and graphics A data storage device, a CAD database in which recognition results are stored as CAD data, a raster pattern storage device for storing a raster pattern in which a symbol is expressed in a raster drawing data format, and a line describing the symbol in an editable format format A diagrammatic pattern storage device for storing the diagrammatic pattern and the relative positional relationship between the symbol and the attribute character; symbol type ID information on the CAD database; and information for associating the raster pattern and the diagrammatic pattern. The symbol library storage device to be stored and the raster pattern storage device Raster symbol pattern recognition means for collating the raster pattern thus obtained with raster drawing data stored in the image storage device to narrow down the range of symbol positions in the drawing, and lines stored in the diagrammed pattern storage device A diagram that compares a plotting pattern with editable format drawing data stored in the editable format drawing data storage device and recognizes a symbol described in the editable format format from the narrowed range Symbol pattern recognizing means, symbol recognizing means for giving ID information as symbol position CAD data with reference to information in the symbol library storage device to the recognized diagrammed pattern, and the image storage device Character area estimation means for estimating the character area of the stored raster drawing data, and a character string of the estimated character area The character string recognition means, the character string position information and the contents of the character string, and the symbol position recognized by the diagrammatic symbol pattern recognition means to recognize the symbol position and the attribute character string related to the symbol. A drawing recognition apparatus configured to include integrated recognition means for storing in a CAD database and recognizing characters and symbols in the drawing and their mutual relationship.   2. The drawing recognition apparatus according to claim 1, wherein the character area estimation means is configured to divide a character string from a raster drawing using positional information of recognized symbols. Recognition device.   3. The drawing recognition apparatus according to claim 1, further comprising drawing position correspondence data means for setting a correspondence relation between positions of raster drawing data and editable format drawing data.   4. The drawing recognition apparatus according to claim 1, wherein editable format drawing data is input, character information of the data is extracted together with position information, and deleted from the editable format drawing data. A drawing recognition apparatus comprising: non-character portion image data generation means for converting the editable format drawing data from which characters have been erased into raster drawing data. Raster drawing data of drawings including characters and symbols, and a procedure for storing the drawings expressed in editable format drawing data capable of individually editing characters and graphics, and the stored raster drawing data and A procedure for narrowing a range of positions of the symbol on the drawing with reference to a raster pattern representing the symbol in the raster drawing data format, a diagram pattern describing the symbol in an editable format format, the raster pattern and the line Editable from the range of positions on the narrowed-down drawing with reference to the drawing expressed in the editable format drawing data in which the information for associating the drawing pattern and the character and the figure can be individually edited A procedure for recognizing a symbol described in a standardized format and a diagrammatic pattern of the recognized symbol A procedure for assigning ID information as symbol position CAD data with reference to ID information of a symbol type on a set CAD database, and the stored raster drawing with reference to the recognized symbol position A procedure for estimating the character area of the data, a procedure for recognizing the character string of the estimated character area, the position information of the character string and the content of the character string, the relative positional relationship between the symbol and the attribute character stored in advance, and Using the recognized symbol position, and recognizing the character position in the drawing and the mutual relationship between the symbol position and the attribute character string related to the symbol and storing the attribute character string in the CAD database. Drawing recognition method.

Images