JPS61114388A - Character input device - Google Patents

Abstract

PURPOSE:To improve the recognizing accuracy of the similar characters of a specific writer by making use of a fact that the characters read wrong or disused and the groups of recognized candidate characters are substantially invariable in case the writer is specified. CONSTITUTION:When plural recognized candidate characters are available, a deciding part 4 retrieves a similar character identification table 5. The table 5 stores the correspondence between the similar character groups and output characters for each writer. The part 4 checks whether or not the table 5 stores a similar character group which is coincident with the recognized candidate character group sent from a recognizing part 2. If said coincidence is obtained, the output character stored in response to the similar character group coincident with said recognized candidate character group is delivered as the result of recognition. While the result of recognition is disused when no similar character group coincident with the recognized candidate character group exists.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a single-character hexagonal device, and more particularly to a character input device with high g-sensitivity accuracy used for a specific small number of scribes.

[Conventional technology]

Conventional single-character input devices generally use recognition dictionaries designed with an unspecified number of scribes in mind, and recognize and input characters written or printed on a recording medium such as a form. widely known.

[Problem that the invention seeks to solve]

In the conventional character input device as described above, when a certain number of scribes use the above-mentioned recognition dictionary, the characters may be deformed due to the scribe's unique writing habits.

In particular, many misreadings and rejections may occur. In order to solve this problem, a method has been proposed in which a unique recognition dictionary is designed for each person, just like a scribe.

This takes a lot of time to design the recognition dictionary, and
There were problems such as huge development costs.

This invention was made to solve this problem, and if it is limited to specific scribes, the handwriting will remain unchanged and the shape of the letters will remain unchanged unless the conditions at the time of writing (scribbling speed, writing instrument, etc.) change significantly. Since the variation is small, a kind of regularity occurs in the input characters and recognition results of this type of character input device. In other words, the characters that are misread or rejected and their
! It is an object of the present invention to provide a character input device that can improve the recognition accuracy of similar characters of a specific scribe by taking advantage of the fact that a group of candidate characters remains almost unchanged when the number of scribes is limited.

[Means for solving problems]

The character input device according to the present invention has a similar character identification table in which similar character groups and output characters are stored in correspondence like a scribe, and outputs a recognition result by referring to this similar character identification table. be.

[Effect]

In the character input device of the present invention, the scanning unit scans characters written by a specific scribe, the recognition unit compares the input character pattern with a recognition dictionary, and the recognition unit outputs a plurality of recognition candidate character groups. When a character type is included, the determination unit searches whether the recognition candidate character group is stored as a similar character in the similar character identification table for scribes, and if it is stored, the recognition candidate character group is By outputting the output characters stored in correspondence, it is possible to improve a particular scribe's recognition of similar characters.

〔Example〕

FIG. 1 is a block diagram showing the configuration of a character input device that is an embodiment of the present invention. In fig.

1 is a scanning unit that scans and photoelectrically converts characters written or printed on a recording medium such as a form, and 3 is a recognition dictionary.

2 is a recognition unit that compares the input character pattern sent from the scanning unit 1 with the recognition dictionary 3 and outputs a group of recognition candidate characters obtained by narrowing down the characters to be recognized; 4 is a similar character identification table 5 and recognition candidate characters; A determination unit 5 which compares groups and outputs output characters is a similar character identification table in which similar character groups and output characters are stored in association with each other for each scribe.

In the character input device configured as described above, an input character pattern scanned and input by the scanning unit 1 is compared with the GM dictionary 3 and recognized by the recognition unit 2. If you keep it, set the predetermined threshold values α, β, and the recognition target character C11ll! ! The similarity between the target character Ci and the input character pattern is calculated as 5i (i-1 to n,
n is the number of characters to be recognized), Cm&X the character with the maximum similarity
If the similarity of the r character CmaX is smhx, the recognition unit 2
sends all the characters Ci to be recognized whose similarity degree Si is greater than α and whose similarity is greater than smam mini-si to the determination unit 4 as recognition candidate characters. When there is only one recognition candidate character, the determining unit 4 selects the recognition candidate character as s! ! ! output as a recognition result. Also, if there are no recognition candidate characters, the recognition result is output as rejected.If there are multiple recognition candidate characters, the recognition result is output as rejected.
The determination unit 4 searches the similar literature identification table 5. The similar character identification table 5 stores similar character groups and output characters in correspondence for each scribe. The determining unit 4 searches the similar character group of the similar character identification table 50 of the scribe to see if there is a character that matches the recognition candidate character group sent from the recognizing unit 2. Therefore, if there is a similar character group that matches the recognition candidate character group, the output character stored in association with it is output as the recognition result. Furthermore, if there is no similar character group that matches the recognition candidate character group, the recognition result is output as rejected.

FIG. 2 shows an example of a similar character identification table used in the character input device of FIG. 1. As shown in FIG. Output characters 7 are associated and registered for each light reporter. For example, suppose that recognition unit 2 recognizes the character ``ri'' (katakana) written by a scribe named Tabe Mitsubishi, and the recognition candidate characters are ``ri'' (katakana) and ``7'' (number). Then, the above similar character identification table 5a for scribes
When searching for , there is a similar character group 6 consisting of "ri" (katakana) and "7" (numeral), and "ri" (katakana) is registered as the output character 7, so the determination unit 4 uses it as a recognition result. "
Outputs ``(katakana).

As explained above, in the character input device of the present invention, for example, when the character "ri" (katakana) of scribe A is recognized, one character "ri" (katakana) and "7" (number) are input. Even in cases where the similarity to the character pattern is extremely close and the recognition result is often rejected, when the character "7" (number) of scribe A is recognized, the character "ri" (katakana) of the input character pattern is recognized. Since it is empirically known that the similarity to K: is low and is almost always correctly recognized as r7J (number), when the characters I & reporter are recognized, the recognition candidate character K ” (katakana) and “7” (number) are included, and the similarity of the two input character patterns is extremely close.

If the recognition result is ``ri'' (katakana), most of the input characters that would otherwise have been rejected will be correctly recognized, making it possible to improve the recognition accuracy of similar characters.

In the above embodiment, the case where there are two recognition candidate characters has been described, but the present invention is not limited to this, and may be used when there are three or more recognition candidate characters.

In the above embodiment, even if the number of characters in the recognition candidate character group sent from the recognition unit 2 does not match the number of characters in the similar character group 6 stored in the similar character identification table 5, if it is a subset thereof, then It is also possible to output the output character 7 registered in .

〔Effect of the invention〕

The present invention provides the above-described one-character input device with a similar character identification table in which similar character groups and output characters are stored in correspondence, like a scribe.

Since the recognition results are output with reference to this similar character identification table, the present invention has an excellent effect in that the recognition accuracy of similar characters can be significantly improved compared to the conventional apparatus of this type.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing the configuration of a character input device according to this embodiment. The second factor is a diagram showing an example of a similar character identification table used in the character input device of FIG. be. In the figure, 1... scanning section, 2... recognition section, 3...
- Recognition dictionary, 4... Determination unit, 5... Similar character identification table. 6...Similar character group, 7...Output character.

Claims (1)

[Claims] A character input device that recognizes and inputs characters written or printed on a recording medium such as a form includes a scanning unit that scans and photoelectrically converts characters, a recognition dictionary, an input character pattern sent from the scanning unit, and the input character pattern. A recognition unit that outputs a group of recognition candidate characters obtained by comparing recognition dictionaries and narrowing down the characters to be recognized; a similar character identification table that stores similar character groups and output characters in correspondence for each scribe; It includes a determining unit that compares the recognition candidate character group with a similar character identification table and outputs an output character, the scanning unit scans the characters written by a specific scribe, and the recognition unit compares the input character pattern with the recognition candidate character group. When the recognition candidate character group output from the recognition unit includes a plurality of character types after checking the recognition dictionary, the determination unit determines that the recognition candidate character group is included in the similar character identification table for the scribe as a similar character. A character input device characterized in that it searches whether or not the character is stored, and if the character is stored, outputs the stored output character in correspondence with the recognition candidate character group.