JPS63158669A - Drawing recognition device - Google Patents

Abstract

PURPOSE:To efficiently extract a character or a graphic with a high by extracting the quantity of feature constituting the character or the graphic from the picture data of a drawing and sorting and shaping the quantity of the feature according to the format of an area to which the quantity of the feature belongs. CONSTITUTION:The drawing to be read is divided into the plural formats and the specific graphic is enters the respective formats. The quantity of feature extraction part 1 extracts the quantity of the feature constituting the character or the graphic from the picture data of the drawing. Specific format information promised in the respective formats and the position information of the area are stored in a format information memory 4. An area extraction part 2 corrects a dislocation or an inclination at the time of reading the drawing with respect to the quantity of the feature extracted from the drawing to extract the format area to which the respective quantities of the feature belong. Then, the quantity of the feature is sorted and shaped as the specific graphic corresponding to the format according to the output of the memory 4 by a character and graphic processing part 3. In such a way, the character or the graphic can be efficiently extracted with the high recognition ratio.

Description

[Detailed Description of the Invention] Industrial Application Field The present invention relates to a device that recognizes characters and figures by reading drawings etc. with a scanner etc. and processing them with a computer system. The present invention relates to a drawing recognition device that performs recognition processing for characters and figures according to a format area in which the characters and figures overlap.

2. Description of the Related Art Conventionally, as a method for recognizing characters and figures from drawings in which characters and figures coexist, there is a method relating to recognition of circuit drawings, for example.

An example of the above-mentioned conventional drawing recognition device will be described below with reference to the drawings. FIG. 5 shows a block diagram of a conventional drawing recognition device.

In FIG. 5, 61 is a feature extraction unit that extracts feature quantities from image data, 62 is a feature classification unit that extracts character areas, etc. from the feature quantities, and 53 is a formatting unit for the feature quantities to prevent cuts, blurring, etc. 54 is a character 1 that recognizes characters and figures from the classified and formatted feature amounts;
This is a figure recognition section. The processing flow of the drawing recognition apparatus configured as described above will be described below. A feature extraction unit 61 extracts line segment areas such as horizontal lines and vertical lines, areas other than the line segment areas, etc. from the image data as feature quantities constituting characters and figures. Next, the feature amount classification section 52 extracts character areas and the like from the feature amounts.

When inputting a drawing using a scanner or the like, line breakage or blurring may occur. , performs shaping to connect the horizontal lines, vertical lines, etc. that have been separated by the line break or the like. Based on the formatted features and the image data of the character area, the character and figure recognition unit 54 recognizes characters and figures (for example, Information Processing Society of Japan 32nd Gold Country Conference Proceedings 4N-1p, p. , 13sei).

Furthermore, a method for classifying and recognizing characters and figures from a formatted drawing containing a mixture of characters and figures using the format information of the drawing is unprecedented.

Problems to be Solved by the Invention In formatted drawings such as timing charts, formats are determined for each area, such as areas where only characters are written, areas where specific characters and symbols are mixed, etc. There is. For such a drawing, format information cannot be used with the above configuration, so when formatting a line break as shown in FIG. 6, incorrect formatting may be performed beyond the area, or Alternatively, processing that also targets graphics may be performed in an area where only characters are normally written, resulting in a decrease in processing efficiency, or a problem may arise in which graphics are erroneously extracted in the area, resulting in a decrease in recognition rate.

In view of the above-mentioned problems, the present invention utilizes the format information of the drawing in a formatted drawing and performs processing suitable for each format area, thereby efficiently and efficiently editing characters and figures in the drawing. The present invention provides a drawing recognition device that performs extraction based on recognition rate.

Means for Solving the Problems In order to solve the above problems, the drawing recognition device of the present invention includes:
A feature extraction unit extracts features such as horizontal lines and vertical lines from image data of an input drawing, and after correcting each feature for tilt and positional deviation at the time of drawing input, the feature is extracted from the image data of the drawing. A region extraction unit that accurately extracts which 7-ormanoto region is written in, and a graphic processing that performs feature amount classification, shaping, and character 3 shape recognition processing according to the format in each formant region. and a format information memory for storing format information, etc. of each format area.

Effect of the present invention With the above-described configuration, the feature values are extracted from a formatted drawing, and by accurately extracting the format area to which each of the feature values belongs, features suitable for the format are extracted within each format area. Classification of quantity,
Performs formatting processing to efficiently recognize characters and shapes.
This is done with a high recognition rate.

Embodiment Hereinafter, a drawing recognition device for formatted drawings according to an embodiment of the present invention will be described with reference to the drawings.

FIG. 1 shows a block diagram of a drawing recognition device according to an embodiment of the present invention. In Fig. 1, 1 is a feature extraction unit that inputs image data of a drawing and extracts feature quantities in the drawing, and 2 is a feature quantity extraction unit that corrects deviations and inclinations at the time of drawing input for the nff feature quantity, and 6f7 an area extraction unit that extracts which format area in the drawing each feature belongs to; 3;
classifies and formats the feature amounts according to the format,
A character 3 graphic processing section recognizes character 1 graphics, and 4 is a format information memory that stores position information and formant information of each 7-omanoto area in the drawing. FIG. 2 shows an example of a formatted drawing in an embodiment of the present invention. In FIG. 2, the thick solid-line frames are markers for detecting shifts and inclinations of the input drawing, and the broken-line frames indicate the boundaries of each format area (7-ohmanoto frame) in which characters and figures are written. Further, symbols a to g in the figure indicate examples of format types of each area.

The operation will be explained below using FIGS. 1 to 4.

First, based on the image data of the drawing, the feature amount extraction unit 1 extracts feature amounts such as horizontal lines, vertical lines, areas other than line segments (for example, Information Processing Society of Japan Box 32 Gold National Conference Proceedings 4N-
1p p 1385).

Next, each of the feature amounts is subjected to the processing shown in FIG. 3 in the region extracting section 2, and seven ohmast regions to which each feature amount belongs are extracted.

First, in step 31, horizontal lines and vertical lines that are candidates for the marker shown by the thick solid line in FIG. , compare the position information of the marker in the input drawing with the position information of the marker registered in the format information memory 4,
Detects deviations, tilts, and expansion/contraction when inputting drawings. The deviation, inclination, and expansion m at the time of inputting the drawing can be obtained by calculating the coefficients of affine transformation, for example. In step 34, [)
If it is confirmed that the rI offset, inclination, and expansion/contraction are within tolerance values, coordinate transformation is performed on each feature in step 36 to correct the position information. As a coordinate transformation,
For example, there is a method of performing an inverse transformation of the affine transformation obtained in step 33 on the central coordinates of each feature. Note that the coordinate transformation may use coordinate values other than the center position of each feature. Next, in step 36, the positions of the respective feature quantities whose positions have been corrected are compared with the positions of the boundaries of the respective format areas stored in the format information memory 4, and the format areas to which the respective feature quantities belong are extracted. In step 37, if the information on the input drawing is incorrect in step 32 or step 34, a transfer process is performed, such as stopping the process.

Next, the character 6 figure processing section 3 performs the processing shown in FIG. 4 to recognize the characters and figures in the drawing.

By referring to the format information memory 4 in the unsorting step 41, the feature amounts in each area are efficiently classified according to the specific format information specified for each format. For example, if it is an area where only characters are written, all the features in the area are classified as character candidates,
For an area where a mixture of characters and figures is written, the feature amounts in the area are classified into character candidates, figure candidates, etc. by examining their shapes and connection relationships. Step 42
Then, the features other than characters in each region are modified to remove line breakage, blurring, etc. In step 42, erroneous shaping as shown in FIG. 6 can be avoided by paying attention to the feature amounts in each area and performing shaping according to the format of the area.

In other words, in the case of the example shown in FIG.
The set of 2 and 72 is the target of formatting, and m of h, and v2 is not the target of formatting.

In step 43, character recognition is performed using the image data of the feature amounts that were deemed to be characters in step 41 (for example, Information Processing Society of Japan, Box 32, Gold Country Conference Proceedings 4N-4p,
p. 1391). In step 44, a character string is generated by referring to the character recognition result and the format information memory 4.

Effects of the Invention As described above, the present invention extracts feature amounts from image data of formatted drawings in which characters and figures are drawn, and extracts features according to the format of the region to which a small amount of text and figures belong. Since it performs quantity classification and formatting processing,
Even in a short recognition process, it is possible to prevent erroneous extraction of characters and figures and improve the recognition rate of drawings. Since positional corrections are made for shifts and inclinations when inputting drawings, the format area to which characters and figures belong can be efficiently and accurately extracted.

[Brief explanation of the drawing]

Fig. 1 is a block diagram showing the overall configuration of a drawing recognition device in an embodiment of the present invention, Fig. 2 is an explanatory diagram of the format targeted by the present invention, and Fig. 3 is an area extraction diagram of the drawing recognition device. FIG. 4 is a flowchart of the processing of the character 1 graphic processing section of the same apparatus. 1...Feature quantity extraction unit, 2...Region extraction unit, 3...Character/figure processing unit, 4......
Format information memory. Name of agent: Patent attorney Toshio Nakao and one other name
1 Figure 2 α~? Format 41 Figure 3 Figure 4

Claims (2)

[Claims] (1) A drawing to be read is divided into multiple formats, and a specific figure is entered in each format, and a feature quantity extraction unit extracts the feature quantities constituting characters and figures from the image data of the drawing. a format information memory that stores specific format information and area position information promised in each of the formats, an area extractor that detects in which format area in the drawing the feature amount is written, and the format It is characterized by comprising means for classifying and formatting the feature amount as a specific figure corresponding to the format based on the output of the information memory, and a character and figure processing unit that recognizes the output figure of the classification and formatting means. drawing recognition device. (2) The area extracting unit corrects the deviation and inclination of the feature quantities extracted from the drawing when reading the drawing, and extracts the format area to which each feature quantity belongs. The drawing recognition device according to item 1.