JPS60136884A - Optical character reader - Google Patents

Abstract

PURPOSE:To register a necessary external character as necessary by constituting so that an external character can be registered at the time point when the external character appeared. CONSTITUTION:A Chinese character OCR is constituted of a character recoginizing part 1 and an external character registering part 2. When a character which cannot be read exists, it is rejected, and an input pattern and a candidate character are displayed on a display. An operator watches the display, and selects the candidate character by using a keyboard. When the operator watches the display, and recognizes that it is an external character, he depresses an external character input key. When its external character is registered already, a similar character set is displayed, and it is selected by the keyboard. On the other hand, in case its external character is not registered, it is changed to an external character registering mode by depressing an external character registering key. When the device becomes the external character registering mode, an external character registering control program inputs a Chinese character code of the external character by using the keyboard and the display. The candidate character is derived in advance by the character recognizing part, therefore, said external character code is registered additionally to the similar character set corresponding to the respective candidate characters. When the external character registration is ended, the device is reset to its original read mode.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Application of the Invention] The present invention relates to an optical character reading device, and more particularly to a method for registering external characters in a kanji reading device.

[Background of the invention]

The problem with optical character reading devices (hereinafter abbreviated as Kanji 00) that read kanji is that there are many types of kanji, so kanji OCR
, is the inability to read all kanji. Kanji OC
The readable characters of R are defined as internal characters, and the unreadable kanji are defined as external characters. Kanji O is how to input foreign characters efficiently.
This is one guideline for making C-box easier to use.

Methods for inputting non-standard characters include inputting a kanji code from the keyboard, and additionally registering a standard button (mask) to make non-standard characters a new recognition target (converting them into internal characters).

The former method has the drawback of requiring operators to look up a code table and input the kanji code each time a non-standard character appears, which places a burden on the operator. For the latter, a standard pattern learning function must be provided for Kanji 0C, and additionally registered masks may have a negative effect on masks in other categories, in which case it may be necessary to modify the recognition logic. . For this reason, additional learning has not yet been put into practical use.

On the other hand, as recently reported in the auxiliary project performance report of the Electronic Meter A method has emerged that displays characters as characters and inputs them by selecting keys.

This method is a good method that does not put a burden on the operator, as it allows the user to enter a foreign character in the same way as keying in a rejected character.

[Purpose of the invention]

The purpose of the present invention is to create a similar character set for external characters in Kanji OC.
The objective is to provide a method for registering with R.

[Summary of the invention]

There are two ways to register external characters.

The first method is when a rejection occurs during Kanji reading p. If the operator looks at the pattern displayed on the screen and determines that the pattern has been rejected because it is a non-standard character, and that non-standard character registration is necessary, the operator switches to the non-standard character registration mode by operating a key.

Since internal character candidates close to the external character are known from the recognition results, similar character set r external character codes for each candidate category are registered. The external character code is entered by key-in.

The second method is to register non-standard characters using a non-standard character registration sheet. Write the external characters on the sheet along with the kanji code,
Set Kanji 0C 几 to external character registration mode and read it. When writing external characters, if you include multiple characters instead of one character at a time, you will have greater tolerance for modified characters. Recognizes the characters written on the registration sheet and registers the external characters in the similar character set. The kanji codes for external characters are written on the form, so there is no need to key in them.

[Embodiments of the invention]

An embodiment of the present invention will be described below with reference to FIG. 1st
In the figure, the kanji 0CII is composed of a character recognition section 1 and a non-standard character registration section 2. The character recognition unit reads the characters written on the form. If there are unreadable characters, they will be rejected and an input button and candidate characters will be displayed on the display. The operator looks at the display and selects a candidate character using the keyboard.

When the operator looks at the display and sees that it is a custom character, he presses the custom character input key. If the external character is registered, a set of similar characters will be displayed next to the candidate character, and you can select it using the keyboard.

If the external character is not registered, press the external character registration key to change to external character registration mode.

The external character registration section consists of a control program for registering external characters and a memory for registering similar character sets.

When the user enters the external character registration mode, the external character registration control program causes the user to input the kanji code of the external character using the keyboard and display. Since candidate characters close to the external characters are stored in the character recognition section, the external character code is additionally registered in the similar character set corresponding to each candidate character.

When the external character registration is completed, the original reading mode is restored.

FIG. 2 shows another embodiment of the invention. The example in Figure 2 is
This is an example in which the mode is not changed to the external character registration mode only when rejected, but the mode is changed to the original character registration mode.

Kanji 00 几 can be switched between the ``Music'' character reading mode and the external character registration mode using the format program or an external switch. In the character reading mode, the external character registration program is not activated and normal character reading is performed. In the external character registration mode, the external character registration program is always activated, not only when rejecting. Use a custom character registration sheet to register custom characters. The non-standard character registration sheet has a format as shown in Figure 3, for example.The form has a character frame for entering the kanji code of the non-standard character and a character frame for entering the non-standard character. The reason why multiple character frames for writing external characters are provided is to absorb character deformation.

When a form is read, the recognition program first reads the kanji code, then reads the external characters one by one, and selects a candidate category for internal characters that are close to the external character. When the eight candidate categories are complete, a similar character set corresponding to each of the eight candidate categories is determined. Register the external character code.

In this embodiment, the reason why a plurality of external characters are written instead of one character at a time is because even if the characters are written by the same person, they do not necessarily form the same stamp and are easily deformed. The more kanji you write, the more deformed characters will appear, and if you register a set of similar clicks corresponding to each deformation, when you actually want to input a non-standard character, you will be able to input a non-standard non-standard character even if the click is different from the one you registered. It will be less likely that the person will be deemed not to have registered.

〔Effect of the invention〕

According to the present invention, a non-standard character can be registered at the time when the non-standard character appears, so necessary non-standard characters can be registered as needed.

If you know the external characters in advance, you can not only eliminate the influence of deformed characters due to the user's habits by filling out the external character registration entry sheet and registering them, but also by entering the kanji code at the same time. You can save the trouble of entering the kanji code.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing one embodiment of the invention, FIG. 2 is a block diagram showing another embodiment of the invention, and FIG. 3 is a format diagram of a non-standard character registration sheet. 1st eye 2nd mouth

Claims (1)

[Claims] l. In an optical character reading device that optically scans and reads unknown characters on a form, there is a character recognition device that recognizes the unknown character, and a character pattern rejected by the recognition device as a candidate similar to the unknown character. A device that is equipped with a display device that displays categories and a keyboard for selecting candidate categories displayed on the display device, and the unknown character is an external character that is not recognized by the character g recognition device. , by switching the mode, the similar character set of characters similar to the foreign character is registered, the similar character set is displayed instead of the candidate category when inputting the foreign character, and the character can be selected using the keyboard. An optical character reading device that enables input.