JPH0239388A - Character recognizing device - Google Patents

Abstract

PURPOSE:To recognize a character-string which is connected arbitrarily by inputting a character, and also, a specific symbol annexed thereto, detecting a position and a size of the character and the character-string from its digital image and storing them. CONSTITUTION:A character position detecting means 3 inputs a digital image consisting of a white picture element and a black picture element, which is stored in an image storage means 2, detects a specific symbol in the digital image, and outputs a position of a character. A character size detecting means 5 derives a size of the character from the specific symbol which is detected by the character position detecting means 3 and output it. A character size storage means 6 stores the size of the character which the character size detecting means 5 has outputted, and a character-string detecting means 7 inputs the digital image which is stored in the image storage means 2, and outputs information for showing a fact to be a character-string from the specific symbol in the digital image.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a character recognition device that recognizes characters from a read digital image of a drawing.

[Conventional technology]

Figure 6 shows, for example, the Information Processing Society of Japan's 31st Gold Country Convention in Showa 60.
It is an overall configuration diagram showing a character recognition device that can be found in Proceedings of the Second Half of the Year 6N-, z1 "One Method for Extracting Character Strings from Drawings".

In the figure (11 is a drawing input means for reading the drawing t- and outputting a digital image of Hli, (2) is an image storage means for storing the digital image output by the drawing input means 10, (l is an image storage means ( 2) character extraction means for inputting the digital image stored in the process and outputting the position of each character; character information storage means for storing the position of the next character outputted by the 41μ character extraction means US;翰 is a character string extraction means for inputting character position information stored in the character information storage means (4) and outputting the characters constituting the character string as character string information; (9) a character string information storage means (9) for storing character string information, inputting the digital image stored in the image storage means (21) and character position information stored in the character information storage means (4); character image cutting means for outputting a character image; aa is character image storage means for storing the character image outputted by the character image cutting means (9); α11 is character image storage means for recognizing the character image stored in the character image storage means σ1; character recognition means that outputs a character code; az is a character storage means that stores the character code output by the character recognition means αυ; fi3 is a character string information storage means (character string information stored by 81 and character storage means α2); (141 is a correction result storage means that stores the character code corrected by the recognition result correction means 11).

Explain the action to the arrow.

First, in order to obtain more reliable results, the following conditions are usually imposed on the input character string to be gratified.

(a) Character strings are written in the positive direction of the X and y axes.

(b) Character fluctuation is less than the width of the preceding and following characters.

(c) Character spacing is below a certain value.

(d) The character string interval is greater than or equal to a certain value.

(e) When a character string is surrounded by a rectangular area, it does not intersect with the rectangular areas of other character strings.

When the drawings with the objects to be read arranged under these certain conditions are inputted by the image input means (11),
The image is stored in the image storage means (2) as a binary digital image consisting of white pixels and black pixels. In the character extraction means α9,
When this digital image is input and the mutual positional relationship of the graphical elements constituting two characters satisfies the conditions (b) and (c) above, it is determined that the graphical elements constitute one character, and these The position of the circumscribed rectangle surrounding the graphic element is output to the character information storage means (4) as the position of the character. The character string extraction means αG inputs the character position i4 information from the character information storage means (4), and determines the mutual positional relationship of the circumscribed rectangles surrounding two characters as shown in (a) above.

When conditions such as (d) and (e) are satisfied, it is determined that the character strings are one set, and a character string information storage means stores a pointer to the character extracted by the character extraction means aS for every two character strings as character string information. Output to (8). Character image cutting means (9)
1. Input the digital image from the image information storage means (2) and the character position information from the character information storage means (4), and output the image inside the circumscribed rectangle that surrounds the character to the character image storage means αG. do. Character recognition means αυ is 2 character image storage means? A character image is inputted from the 1CI, and character codes of ff2j are prioritized and outputted to the 2-character storage means 62 as candidates for character recognition for each character. In recognition result correction processing fee 3.

The character string information storage means (8) and the character storage means (U2 input the character codes of the characters constituting the character string, and if they are not correct as one character string, the one character code is corrected according to the priority order, and the correction result storage means σ Output to gradient.

Note that the correctness of a character string is determined by comparison with a word dictionary.

[Problem to be solved by the invention]

Conventional character recognition devices are configured as described above, so if a character is extracted incorrectly.

Correct character recognition results cannot be obtained. Further, there are problems that need to be solved, such as when a character string is extracted incorrectly, correct recognition result correction processing is not performed.

In addition, in order to improve the recognition rate of two characters and character strings, restrictions have been added to the input characters and character strings that can be read, making it possible to recognize character strings of any size and any concatenation in any position. There wasn't.

This invention was made to solve the above-mentioned problems, and eliminates errors in extracting two characters and character strings, and consists of characters of arbitrary sizes in arbitrary positions. The purpose of the present invention is to obtain a device that can recognize arbitrarily concatenated character strings. .

[Means for Solving the Problems] The character recognition device according to the present invention inputs two characters and an accompanying specific symbol, and detects and stores the position, size, and character string of the characters from this digital image. It is something.

[Effect]

The character recognition device according to the present invention uses the coordinate position of a digital image of a specific symbol input together with two characters as the character position, determines the size of the character from the shape of the digital image of this specific symbol, and determines the size of the character from the shape of the digital image of this specific symbol. Detect strings from those that indicate concatenation of characters. With this specific symbol, the position and size of the two characters can be freely changed, and the direction of the character string can also be freely changed.

[Embodiments of the invention]

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

FIG. 1 is an overall configuration diagram showing one embodiment of the present invention, in which ([) indicates a screen input means, (2) an image storage means,
(4) is character information storage means, (8) is character string information storage 1.
1 means, (9) is a character image cutting means, 011B is a character image storage means, αυ is a character recognition means, 113 is a character storage means, 03 is a recognition result correction means, α is a correction result storage means, and these are This is similar to the conventional one shown in FIG. (3) is a character position detection means that inputs a digital image consisting of white pixels and black pixels stored in the image storage means (2), detects a specific symbol in the digital image, and outputs the position of one character; 5) is a character size detecting means for determining and outputting the character size from the specific symbol detected by the character position detecting means (3).

(6) is a character size storage means for storing the size of the characters outputted by the character size ejection means (5), and (7) is for inputting the digital image stored by the image storage means (2), and inputting the digital image stored in the image storage means (2). This is a character string detection means that outputs information indicating that it is a character string based on a specific symbol.

Next, the character recognition device shown in FIG. 1, which is an embodiment of the present invention, will be explained using FIGS. 2 to 5.

Figure 2 is a drawing depicting the characters to be read, Figure 3
The figure shows specific symbols corresponding to FIG. 2. In the example in Figure 3, a square whose side length is the size of two characters is used as a symbol to represent the position and size of two characters, and a line connecting these squares is used as a symbol to represent the connection of two character strings. There is.

When drawings such as those shown in FIGS. 2 and 3 are inputted by the image input means il+, they are stored in the C0 image storage stage (2) as binary digital images consisting of white pixels and black pixels.

FIG. 4 shows an enlarged portion of the digital image corresponding to FIG.

The character position detection means (3) first detects the digital image of FIG. 3 stored in the two-image storage means (2).

White pixels that are continuous in four directions: top, bottom, left, and right.

In other words, find a white pixel area in which 4 connected white pixels are grouped together, and when the vertical and horizontal sizes fall within a certain size range, the position of 2 characters and the size of the character correspond to a specific symbol. Detected as a white pixel area.

Next, find the circumscribing rectangle of the detected white pixel area.

The position is outputted to the character information storage means (4) as a character position. In FIG. 5, the white pixel area detected from FIG. 4 is shown in a hatched pattern. Further, the position information of the character to be output is the coordinate value of the X mark shown in FIG. Note that the order in which the character position detection means (3) detects the white pixel area is used as a pointer for two-character position information, and is also used as a reference pointer for the character or white pixel area in subsequent steps.

The character size detection means (5) uses the one character information storage means (4).
), the size of one character is calculated and output to the character size storage means (6).

Here we will explain how to calculate the font size.

The character position information is the lower left and upper right coordinate values of the circumscribed rectangle in the white pixel area detected by the two character position detection means (3). Since this circumscribed rectangle almost matches the specific symbol, if the lengths of the two sides of the circumscribed rectangle are calculated, this can be used as the character size. In this example, since the specific symbol is a square, the size of the character can be determined only by calculating one side.

The character string detection means (7) first selects the top, bottom, left, right, and 45
A black pixel area is obtained in which black pixels that are continuous in eight directions in the ° direction, that is, eight connected black pixels are grouped into one set. Next, based on the obtained black pixel area, the string information storage means (8
). Here, what is output as the connection order is the reference pointer of the white pixel area mentioned above, and the same pointer is also used to refer to the character position information written in the character information storage means (4).

Next, the two-character image cutting means (9) inputs the digital image stored in the two-image storage means (2) and the character position information from the two-character information storage means (4), and extracts the image within the white pixel area. It is output to the character image storage means IIα.

The character recognition means αD inputs the character size from the two character size storage means (6) and the character image from the one character image storage means αG, and prioritizes n character codes as candidates for character recognition for one character. Rank them.

Output to character storage means 03.

In the recognition result correction means αJ, 2 character string information storage means (8)
The character codes of the characters constituting the character string are input from the character storage means fi3, and if they are not correct as one character string, the character codes are corrected according to the priority order, and the correction result storage means α4
Output to. Note that the correctness as a nine-character string is determined by comparison with a word dictionary.

In the above embodiment, a square is used as the specific symbol, but any symbol may be used as long as the position and size of the two characters can be specified, such as a rectangle or an s-shape.

Further, the line connecting the squares shown in the above embodiments may be any symbol as long as the character string can be determined, and may be an arrow mark pointing to the next character in two character strings.

In addition, 2. In the above example, the case where one character is placed in the positive vertical and horizontal directions is shown, but even if two characters are placed rotated, the coordinates are determined by a specific symbol, so even if the characters are tilted, they can be recognized. .

Further, in the above embodiment, the case where the size of characters within the same character string is the same is illustrated, but even if there are characters of different sizes within the same character string, they can be recognized because the size can be determined by a specific symbol.

Note that the effects of the present invention will not be lost even if the specific symbol is input at the same time as the object to be read or input separately.
Furthermore, it is not limited to only characters that can be recognized; any code pattern that has been encoded and registered may be used.

〔Effect of the invention〕

As described above, according to the present invention, the position, size, and character string of a character can be determined by a specific symbol, so even if two characters are located at any position, at any size, and in any direction, they can be accurately determined. It has a discernible effect. Furthermore, since it does not recognize anything that is not accompanied by a specific symbol, even if the input image contains figures other than characters, they are ignored and do not cause any problems.

[Brief explanation of the drawing]

FIG. 1 is an overall configuration diagram showing an embodiment of the present invention. Figures 2 to 5 are diagrams to help explain the character recognition device shown in Figure 1. Figure 2 is a diagram depicting the characters to be read, and Figures 8g3 and 8g3 are specific symbols corresponding to Figure 2. FIG. 4 is an explanatory diagram showing an enlarged part of the digital image corresponding to FIG. 3. Figure 5 is an explanatory diagram showing the white pixel area and character position information obtained from Figure 4, and Figure 6 is an overall configuration diagram showing a conventional character recognition device. (2) is an image storage means, (3) is a character position detection means, (4) is a character information storage means, (5) is a character size detection means, +61H character size storage means, and (7) is a character string detection means. means, (81 is a character string information storage means, (9) is a character image cutting means, +
1CI is character image storage means, fLll is character recognition means, (13 is character storage means, t13 is recognition result correction means, (141 is correction result storage means, (!9 is character extraction means, net character string extraction means Note that the same reference numerals in each figure indicate the same or equivalent parts.

Claims (1)

[Scope of Claims] A character recognition device that recognizes characters from a digital image of a read drawing, comprising (a) means for detecting and storing the position and size of a character from a digital image of a specific symbol attached to a character; (b) a character; comprises a means for detecting and storing a character string from a digital image of a specific symbol attached to the image; the character is recognized using the position and size of the character stored by the means (a) above, and the character string is recognized by the means (b) above. A character recognition device characterized by recognizing a reading target from a stored character string.