JPS62231389A - Character recognizing device - Google Patents

Abstract

PURPOSE:To obtain a character code on which a bit of en information is added, and to simplify the operation of an operator, by providing an en identifying part which outputs the bit of en information representing that an input character is an en character. CONSTITUTION:A character measuring part 3 scans an input form 1, and converts it to an image pattern, then stores it on an image memory 4. A character segmenting part 5 segments an image pattern 6 of MXN corresponding to one character frame, from the memory 4, and segments a character pattern 7 by a circumscription detecting process. Next, it is detected whether the pattern 7 exists only at the left half or the right half of the pattern 6 by an en identifying part 8, based on X1 and X2 of an X-axis, and the en information, or a bit of en information is outputted. A code output part 14 converts an inputted character number to the character code using a code memory 15, and adds the en information, or the en information obtained at the identifying part 8, then stores it on a candidate memory 16, and also, outputs it to a control part 17.

Description

DETAILED DESCRIPTION OF THE INVENTION (Field of Industrial Application) The present invention relates to a character recognition device that uses, as an input sheet, a sheet on which a plurality of character frames for writing characters corresponding to one character have been recorded in advance. Concerning character processing.

(Example) As shown in the prior art Japanese Patent Application No. 195316/1983, in an offline character recognition device, a sheet of paper with a plurality of character frames for writing characters corresponding to one character recorded in advance is usually used. is used as an input form, and the recognition unit first extracts the feature pattern of the input character written in the character box, and then compares this pattern with the standard feature pattern in the dictionary to determine the character code of the candidate character with a high degree of similarity. can get.

(c) Problems to be solved by the invention In the conventional technology, only the character code of the candidate character was obtained as a recognition result, so when creating a document by connecting a character recognition device to a word processor, half-width When trying to input a character, it cannot be achieved by simply writing the character on an input form; after the recognition result is obtained, the operator must use keys to add half-width attributes to the character code of the candidate character. There wasn't.

(d) Means to solve the problem The present invention detects that input characters are written only within a specific range on the left or right side of a character frame, and indicates that the input characters are half-width characters. This is provided with a half-width recognition section that outputs half-width information.

(E) Function According to the present invention, a character code to which half-width information is added can be obtained by simply writing characters within a specific range, such as the left half or the right half, of the character frame of the input form.

(F) Embodiment FIG. 1 is a block diagram showing an embodiment of the present invention, and FIG. 2 is a diagram showing a specific example of input paper.
A rectangular character frame (2) of a predetermined size is printed in a dropout color for each character for writing characters.

The character observation unit (3) scans the input paper (1) and reads the COD
Using a photoelectric exchange element such as an image sensor, at least one line of characters is converted into an image pattern having one of two values, and the pattern is stored in an image memory (4). As shown in FIG.
), cut out an image pattern (6) of MXN dots surrounded by solid lines corresponding to one character frame (2), and then use circumscribing frame detection processing to extract the actual character pattern (7) surrounded by wavy lines. At this time, in the image pattern of MXN dots, the tentacles extend toward the center from each side, and when the character part is first encountered, the circumscribed frame of that side is detected. Therefore, when cutting out the character, each circumscribed frame is detected. frame coordinates xr
.

x2 and yI+72 are detected.

Therefore, in the present invention, the half-width recognition unit (8) detects whether the character pattern (7) exists only in the left half or the right half of the image pattern (6) based on the work coordinates χ1 and work. are doing. In other words, if ≦M/2, the right half is blank and the character pattern exists only in the left half, and if ≧M/2, the left half is blank and the character pattern is in the right half. Therefore, if the character pattern exists only in the right or left half, output half-width information, and if the character pattern exists in both the left and right halves, output full-width information. ing.

By the way, the extracted character pattern (7) is recognized by the recognition unit (
The normalization unit (10) in 9) normalizes the size and position, and the feature extraction unit (11) extracts the feature pattern. It is compared with the stored standard characteristic pattern common to full-width and half-width characters, and the character number of the candidate character with the highest degree of similarity is output.

The code output unit (14) uses the code memory (15>) to convert the incorrectly input character number into a character code, and adds half-width or full-width information obtained by the half-width identification unit (8) to this character code. Then, it is stored in the candidate memo IJ (16) and sent to the control unit (17).

(18) is a control part (17), a text (19),
It is a document creation device such as a word processor that is equipped with a character generator (20) and a display section (21), and information from a code output section (14) is stored in text (19).

If the information stored in the text (19) is full-width information and a character code, the control section (17) reads out the full-width dot pattern corresponding to the character code from the character generator (20) and directly displays it on the display section ( 21) and displays the candidate character in full-width. In the case of half-width information and a character code, the control unit (17) reads out the full-width dot pattern corresponding to the character code and converts it into a half-width dot pattern. The candidate characters are then sent to the display unit (21) to display half-width candidate characters.

When assigning a character code different from a full-width character to a half-width character, a half-width character code is created using the character number and half-width information in the code output section (14), and the created code is stored in the candidate memory (16). In this case, if half-width dot patterns corresponding to half-width character codes are stored in advance in the character generator (20), half-width display can be achieved by directly reading out the half-width dot patterns.

By the way, half-width characters generally do not exist in all characters; for example, half-width characters are not used in kanji and the like.

Therefore, the code output unit (14) determines whether the character number as a recognition result is a character that can be half-width.
If the half-width information is ignored for characters that cannot be used, even more accurate half-width attributes can be added. Specifically, you can set up a table that registers characters that can be half-width and check the recognition results against this table, or you can use the fact that character numbers are assigned within a predetermined range for each character type. , it is sufficient to detect whether the character number as a recognition result is within a predetermined range.

All character frames (2) on the input form are marked with (22) (2) in Figure 2.
As shown in 3), if you print wavy lines or dots indicating a specific range in dropout color, it will be easier to write in half-width characters.

(g) Effects of the Invention According to the present invention, when entering characters in the character frame of an input form, a character code with a half-width attribute added can be obtained by simply writing within a specific range on the left or right. Operation is simplified and extremely convenient.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing an embodiment of the present invention, FIG. 2 is an explanatory diagram showing a specific example of an input sheet, and FIG. 3 is an explanatory diagram for explaining cutting out of a character pattern. (1)...Input paper, (8)...Half-width identification section,
(9)... Recognition unit, (16)... Candidate memory,
(19)...Text, (21)...Display section. 71 Figure 2 Figure 3 ゝ---＼---ノ'

Claims (1)

[Claims] (1) In a character recognition device that uses a sheet of paper on which multiple character frames for character entry corresponding to one character are recorded in advance as an input sheet, the input character is written only within a specific range to the left or right of the character frame. 1. A character recognition device comprising: a half-width identification unit that detects that the input character is a half-width character and outputs half-width information indicating that the input character is a half-width character.