JPS6222186A - Drawing reader - Google Patents

Abstract

PURPOSE:To reduce a load on a character recognition part by making the part perform the recognition process of a character only when the type of the character to be recognized is covered by recognition applied character type. CONSTITUTION:A drawing image read and inputted through a drawing input part 1 is stored at a frame memory 2, applying a drawing process on it. A diagram processor 4 detects the linked area of a graphic component from the inputted drawing image and performs a decision process, and a table processor 5 detects the row of characters written on and detects the degree of complexity of the character type. Regarding an input image, the character type of which is decided as that of KANA (Japanese syllabary)/KANJI (Chinese character) character, the coordinate value and the image pattern of the image area stored as an input character image, and the input image, the character type of which is decided as that of alpha-numeric character, is given to a character recognition part 6, applying a character recognition process on it. Information processed at each area is consolidated and is outputted.

Description

DETAILED DESCRIPTION OF THE INVENTION [Technical Field of the Invention] The present invention relates to a drawing reading device for processing drawings in which various kinds of character information are written.

[Technical background of the invention and its problems]

In recent years, various drawing reading devices have been developed that read and input handwritten drawings into a computer system, transcribe them, and file them as documents.

A wide variety of drawings are processed by this type of apparatus, including atomic couplant drawings, civil engineering and architectural drawings, mechanical drawings, electrical block drawings, and even body maps.

However, these drawings are generally composed of original drawing information consisting of figures and patterns expressed by combinations of lines, connection lines, etc., and character information consisting of character strings appended to the drawing information. .

This character information consists of, for example, (1) a character string representing management information such as an element name or line name, and (2) a character string representing comment-like document information.

However, when reading and inputting such drawings, it is necessary not only to accurately read the drawing information and process it accurately, but also to accurately read and recognize the character strings attached to the drawing information. is necessary. Therefore, a drawing reading device is constructed by incorporating a character recognition device into various drawing processing devices that have been proposed in the past, and for example, it is possible to integrally process the above-mentioned drawing information and character information of drawings read and input by OCR. is considered.

By the way, character strings written in drawings can be broadly classified into (1) character strings consisting of alphanumeric characters and (2) character strings including kana and kanji. However, although it is relatively easy to recognize the above alphanumeric characters using current character recognition technology,
Accurately recognizing kana and kanji requires a fairly sophisticated recognition processing device. For this reason, there is a problem in the feasibility of incorporating a character recognition device capable of recognizing characters including kana and kanji into a drawing processing device.

Furthermore, character strings in drawings are often unspecified in their writing position, writing direction, character arrangement direction, etc. For this reason, it has been extremely difficult to accurately recognize each character in a character string whose writing format is not fixed.

[Purpose of the invention]

The present invention has been made in consideration of these circumstances, and its purpose is to easily process character strings written in drawings and to effectively utilize them in the cleanup process of drawings. An object of the present invention is to provide a highly practical drawing reading device.

[Summary of the invention]

The present invention detects a graphic area in a drawing image read and input via OCR, performs graphic processing, detects a character writing area in the drawing image, and processes the characters written in the character writing area. The character type is identified by checking the complexity, for example, from the number of strokes and the number of feature points, and the character type is set as the character type to be recognized by the character recognition unit.For example, only when alphanumeric characters are written in the character writing area. The recognition process is performed for each character individually.

If the character type of the character string written in the character writing area is not the recognition target character type, the input image information of the character writing area is output as is, or the frame information indicating the character writing area is output. This is what we did.

〔Effect of the invention〕

Thus, according to the present invention, only character strings of character types with a small number of character strokes, such as alphanumeric characters, can be effectively recognized, and complex character strings with a large number of character strokes, such as kanji, can be recognized in the read input image. It prompts other processing by outputting the character as it is, or by outputting the frame information of the character writing area, which greatly reduces the burden on the character recognition unit and makes effective use of this character recognition unit. It becomes possible to perform processing on input drawings.

In addition, the character writing area is detected, and the type of character written in each character writing area is effectively determined based on the complexity of the character to determine whether or not to perform character recognition processing. Effects such as easy control of processing procedures can be achieved.

[Embodiments of the invention]

Hereinafter, one embodiment of the present invention will be described with reference to the drawings.

FIG. 1 is a schematic configuration diagram of an embodiment of the apparatus, and l is a drawing input section consisting of a facsimile, a drum scanner, a hand scanner, etc. For example, as shown in FIG. 2, a drawing image consisting of graphic information and various character information read and inputted through the drawing input section 1 is stored in the frame memory 2 and subjected to drawing processing.

The address controller 8 manages the two-dimensional address of the drawing image stored in the frame memory 2, and in response to this address management, the drawing image is read out to the line graphic processing processor 4.

Basically, the line figure processing processor 4 detects connected regions of figure components from the input drawing image, as shown in FIG. It is determined whether the area is drawing information or character information (step b). On this occasion,
When the graphical connected component is character information, a process such as detecting a circumscribing frame for the character string and identifying the inside of the frame as a character writing area is executed.

The table processing processor 5 executes drawing processing on the drawing information in the input drawing image according to the processing result of the processor 4 (step C in FIG. 3). This drawing processing involves, for example, vectorizing each line segment of the connected region to correctly recognize the figure, normalize the figure, and convert it into a figure code.
It consists of processing to determine the position of the figure in the entire drawing.

Further, this table processing processor 5 receives information on the character writing area from the processor 4 and detects the character string written therein (step d in FIG. 3). After each character is thinned (step e in FIG. 3), the complexity of the character is detected (step f in FIG. 3).

The complexity of the character is determined, for example, by determining the number of strokes of the character, determining the number of feature points such as branch points, and determining the density of the feature points with respect to the size of the character.

The character type is determined by identifying such information on the complexity of the character (step g in FIG. 3).

This character type determination is performed by determining whether the character is an alphanumeric character or a kana/kanji character, since the character recognition unit B uses alphanumeric characters as the character type to be recognized.

Then, for the input image of the character writing area whose character type is determined to be kana-kanji, its coordinate values and image pattern are stored as the input character image (Fig. 3, step h).
. Further, the input image of the character writing area whose character type is determined to be alphanumeric is provided to the character recognition section 6 and subjected to the character recognition processing (FIG. 3, step i).

The information of each area processed in this way, that is, the graphical information of the drawing area, the recognition result of alphanumeric characters written in the character writing area, and the input image of kana and kanji in the character writing area, are integrated and output, At this point, the cleanup process of the input drawing image is completed (step j in FIG. 3). Then, the above-mentioned kana-kanji character string is appropriately encoded and transcribed by an interactive information input means such as kana-kanji conversion processing using a keyboard input.

A more detailed explanation of the character string (character writing area) detection process and the character type determination process is as follows.

In other words, in this device, the characters in the input drawing are processed in character string units.
It is detected as the character writing area and processed. Therefore, it becomes possible to process the character very easily. Incidentally, it is very difficult to accurately detect characters one by one from an input drawing in which the direction of writing, the position of writing, and the direction of arrangement of characters are unspecified.

For example, if the character string ``Hachi'' exists, it is extremely difficult to determine whether it consists of 4 consecutive characters ``Ki'' or 2 consecutive characters ``Hayashi''. . However, as described above, in this device, the character writing area is detected using the character string as a unit, so the detection is easy and reliable.

Furthermore, although it is very difficult to identify the character type of each individual character, if you focus on the most complex character in the string and determine the character type, you can determine whether the string contains kanji or not. can be identified relatively easily.

That is, many kanji are composed of intersections of a plurality of vertical lines, horizontal lines, and diagonal lines. Therefore, the number of branch points representing feature points of a line figure within a rectangle based on the character string width is measured, and the maximum value is compared with the general number of branch points for alphanumeric characters. Specifically, the measured number of branching points is compared with a predetermined threshold value to determine whether the character string is alphanumeric. In this way, it is possible to relatively easily determine whether the character type of the character string is alphanumeric or kanji.

Therefore, according to the present apparatus, which controls whether or not to perform character recognition processing on each character writing area as a processing unit, as described above, it is possible to significantly reduce the processing load on character information in a drawing.

Note that each character in the character writing area can be detected, for example, as a connected area of a plurality of black pixels. Therefore, for example, as shown in FIG.
40"41' to "490, etc., and calculate the coordinate values (Xa, Ya, Xb, Y
b) Obtained as le. Then, the information of each of these areas is
In addition to creating a table as shown in the JS diagram, the connection relationship between each area is determined based on whether the distance between the areas is less than a predetermined threshold, and the areas in that connection relationship are connected with a pointer. For example, it becomes possible to extract and detect characters one by one from this table information.

In this way, recognition processing may be performed for each character extracted from the character string.

Note that the present invention is not limited to the embodiments described above. For example, the character type may be determined based on the fact that kanji characters have a greater number of strokes than alphanumeric characters. Alternatively, the black pixel density constituting the character pattern may be measured, and if a locally high pixel density is detected, this may be determined to be a Chinese character.

In addition, in the embodiment, the input image of the character writing area where the character type is determined to be kanji is output as is, but only the circumscribing frame is output, and a kanji character string is input interactively into this frame to edit the drawing. You may also do this. In short, the present invention can be implemented with various modifications without departing from the gist thereof.

[Brief explanation of the drawing]

The figures show one embodiment of the present invention, in which Fig. 1 is a schematic configuration diagram of the embodiment device, Fig. 2 is a diagram showing an example of an input drawing, and Fig. 3 is a diagram showing an example of an input drawing.
4 is a diagram showing an example of a processing procedure for an input drawing image, FIG. 4 is a diagram showing an example of a character pattern, and FIG. 5 is a diagram showing an example of the configuration of a connected area table for character detection. ■... Drawing input section, 2... Frame memory, 3...
- Address controller, 4... Line figure processing processor, 5... Table processing processor, B... Character recognition section. Applicant's agent Patent attorney Takehiko Suzue Figure 1 Figure 2

Claims (4)

[Claims] (1) A means for detecting a graphical area in a drawing image that has been read and input and performing graphical processing, and a character recognition unit that detects a character writing area in the drawing image and recognizes and processes the characters written in the character writing area. The control means of the character recognition unit checks the complexity of the character written in the character writing area detected from the drawing image to identify the character type, and the control means of the character recognition unit identifies the character type. 1. A drawing reading device that performs recognition processing for each character written in a character writing area only when the character type is a recognition target character type. (2) The drawing according to claim 1, wherein the complexity of a character is comprised of information on the number of strokes and the number of feature points of a character, and the character type is determined by comparing the complexity with a predetermined threshold. reader. (3) The drawing reading device according to claim 1, wherein the character recognition unit uses alphanumeric characters as recognition target character types. (4) The drawing reading according to claim 1, wherein the information on the character writing area excluded from the character recognition processing target is processed as input image information or as frame information indicating the character writing area. Device.