JPS63316189A - Optical character recognition device - Google Patents

Abstract

PURPOSE:To attain efficient and accurate correction by displaying a character pattern indicated by a character code converted by a character recognition section together with an image pattern extracted from a character picture not recognized so as to allow manual correction. CONSTITUTION:A picture of an OCR original 1 to be read is inputted to a picture input section 2 and stored once in an input picture storage section 3. A character recognition section 4 converts the stored picture into a character code recognized normally and a reject code not recognized and stores the result to a code data storage section 5. A character pattern generating section 6 generates a character pattern indicated by the character code being the result of conversion by the normal recognition and a character picture extracting means 12 extracts a character picture not recognized by the recognition section 4 to generate the image pattern. The character pattern recognized normally and the image pattern not recognized are displayed on a display section 8. Thus, the operator uses an input device 11 to operate a correction processing means 10 to attain efficient and accuracy correction.

Description

DETAILED DESCRIPTION OF THE INVENTION Field of the Invention The present invention relates to an OCR device (optical character recognition device) that recognizes handwritten characters written in a predetermined entry field and converts them into character codes.

BACKGROUND OF THE INVENTION Conventional OCR devices of this type recognize handwritten character images located at predetermined locations in an image input by an image reading device such as a facsimile, character by character, and convert them into character codes. At this time, if some of the handwritten character images are not recognized because they are broken or quirky, the unrecognized character images are temporarily converted to a jijikuto code and output.

From this reject code, an appropriate symbol pattern indicating that it is blank or unrecognizable is generated and output and displayed together with the character pattern generated from the regular character code. Character images that are not recognized are converted into regular character codes through manual correction work.

Problems to be Solved by the Invention However, with this configuration, the correction work of manually converting unrecognized character images into regular character codes requires both the input image and the output display screen to be corrected each time one character is corrected. Since the user has to refer to the information alternately, there are problems in that work efficiency is poor and correction errors are likely to occur due to misunderstandings.

The above problem arises for the following reasons. That is, it cannot be determined whether or not a character image has been properly recognized without looking at the output display screen, and it is necessary to look at the input image in order to examine the character content of a character image that has not been recognized.

This invention was made in view of the above-mentioned problems, and it is possible to eliminate the trouble of having to alternately refer to both the input image and the output display screen each time the operator corrects a single character.
To provide an optical character recognition device capable of efficiently and accurately performing correction work of manually converting unrecognized character images into regular character codes while viewing only an output displayed screen.

Means for Solving the Problems In order to solve the above-mentioned problems, the present invention provides a character image recognition process for character images at predetermined locations in an image input from an image reading device such as a facsimile machine and converting them into character codes. a recognition unit; a character pattern generation unit that generates a character pattern specified by the character code converted by the character recognition unit; and a character pattern generation unit that extracts character images not recognized by the character recognition unit from the input image and generates the character pattern. a display section for displaying the character pattern generated by the character pattern generation section together with the image pattern generated by the character image extraction means; and a correction processing means for converting the pattern part into a regular character code based on a manual input operation.

Effect of the Invention With the above-described configuration, the present invention uses regular character patterns generated based on recognized character images as well as images and images generated based on unrecognized character images.
Since the pattern is also output and displayed, the character content of the unrecognized character image can be directly viewed and examined from only the output and displayed screen. This allows the operator to manually correct unrecognized characters while looking only at the output display screen, without having to go through the trouble of alternately referring to both the input image and the output display screen each time a worker corrects a character. Efficiently perform correction work to convert to character code.
And you can definitely do it.

Embodiment FIG. 1 shows a schematic configuration of an OCR apparatus according to an embodiment of the present invention.

The OCR device shown in the figure first includes an image input section 2, an input image storage section 3, a character recognition section (OCR) 4, a code/data storage section 5, a character pattern generation section 6, a display image storage section 7,
It has a display section (display) 8, a printer 9, a correction processing means 10, an input device 11 such as a keyboard, and a character image extraction means 12.

The image input unit 2 inputs an image of the OCR original 1 read by a facsimile or the like.

The OCR manuscript 1 has a header mark IA indicating that it is an OCR manuscript, as shown in FIG. 2 as an example of its format.
Timing mark IB indicating the line.

and a character entry field IC for each character.

The image 2 input from the image input section 2 is temporarily stored in the input image storage section 3.

The character recognition section 4 extracts handwritten character images one by one from predetermined locations of the input image stored in the input image storage section 3 and subjects them to recognition processing. Recognized characters are converted into character codes one by one. At this time, if the original handwritten characters are broken or quirky and cannot be recognized as characters, only the unrecognized character images are temporarily converted into a kind of dummy reject handwriting code. do. This reject code is outputted and stored in the code data storage unit 5 in the order of conversion together with the normally recognized character code.

The character pattern generating section 6 generates a character pattern indicated by a character code that is properly recognized and converted from among the codes stored in the code/data storage section 5. Although not shown, this character pattern is generated by quoting from a character font set prepared in advance. The generated character pattern is stored in the display image storage section 7 in the form of a bitmap image.

The display section 8 is constructed using a CRT display device or the like, and displays the image data stored in the display image storage section 7 in a bitmap format. At this time, the character pattern generated by the character pattern generating section 5 is displayed at a position corresponding to the entry field for the original handwritten character.

The printer 9 prints out a hard fist copy of the character code string stored in the code/data storage section 5 or the image stored in the display storage section 7.

The modification processing means 10 has a kind of editing function for modifying the contents stored in the code/data storage section 5 based on an operation input from an input device 11 such as a keyboard.

The code corrected by this correction processing means 10
The stored contents of the data storage section 5 are also output to the outside as output data of the OCR device.

Then, the character image extraction means 12 performs 1. The above character recognition section 5
The character images that are not recognized in the above are extracted from the input image, and an image pattern that replaces the character pattern is generated from the extracted character images. This character image extraction means 2 is
The character recognition unit 4 operates when the reject code is temporarily converted, and extracts the handwritten character image that caused the reject code. The extracted character image is converted into an image pattern having the same dot configuration as the regular character pattern (for example, 24×24 dots) through processing such as reduction or enlargement.

The image pattern thus generated is stored in the display image storage section 7 in the form of a bitmap image, in the same way as the regular character pattern generated by the character pattern generation section 6.

Accordingly, the display section 8 displays the following information as shown in FIG.
Image patterns 14 that directly represent images of unrecognized handwritten characters mixed with the normally generated character patterns 13 are displayed at positions corresponding to the original entry fields. In this case, preferably only the image pattern 14 is displayed in a manner different from the regular character pattern by color coding, inversion of black and white, or by enclosing a frame.

As described above, the image pattern of the unrecognized character image is output and displayed on the display unit 8 together with the character pattern based on the normally recognized character image. Thereby, it is possible to directly view and examine the character content of the unrecognized character image only from the screen output and displayed on the display unit 8. This allows the operator to manually correct unrecognized characters while looking only at the output display screen, without having to go through the trouble of alternately referring to both the input image and the output display screen each time a worker corrects a character. The correction work of converting to the character code can be performed efficiently and without error.

Effects of the Invention As is clear from the above explanation, the present invention generates an image pattern from an unrecognized character image,
By outputting and displaying the regular character pattern generated based on the recognized character image, the operator does not have to alternately refer to both the input image and the output display screen each time he/she corrects a single character. This has the effect of allowing you to manually convert unrecognized characters into regular character codes efficiently and without error while only looking at the output screen.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing a schematic configuration of an OCR device (optical character recognition device) according to an embodiment of the present invention, and FIG. 2 is an OCR document in which characters to be recognized by the device shown in FIG. 1 are written. FIG. 3 is a plan view showing a part of the apparatus shown in FIG. 1, and FIG. 3 is a diagram showing an example of an output display screen in the apparatus shown in FIG. 1...OCR manuscript, IC...character entry field, 2...
Image input section, 3... Input image storage section, 4... Character recognition section, 5... Code/data storage section, 6... Character pattern generation section, 7... Display image storage section, 8 . . . Display unit, 9 . . . Printer, 10 . . . Correction processing means, 11
. . . Input device such as a keyboard, 12 . . . Character image extraction means. Name of agent: Patent attorney Toshio Nakao and 1 other person No. 1
Figure 2 Figure 3

Claims (1)

[Claims] a character recognition unit that recognizes and processes a character image at a predetermined location of an input image and converts it into a character code; a character pattern generation unit that generates a character pattern indicated by the character code converted by the character recognition unit; a character image extraction means for extracting a character image not recognized by the character recognition unit from the input image and generating an image pattern to replace the character pattern; The image generated by the extraction means
An optical character recognition device comprising: a display section that displays the image pattern together with the pattern; and a correction processing means that converts the image pattern portion into a regular character code based on a manual input operation.