JPS6084684A - Character recognizing system - Google Patents

Abstract

PURPOSE:To suppress classifying error rate of a basic stroke to low by a small processing quantity, and to obtain high recognition rate by limiting a basic stroke group being an object to be collated, by the number of strokes of an input character. CONSTITUTION:A stroke number counting part 5 cuts a character by a character spacing designation by touching some area on a tablet 1 with a pen point of an input pen 2, and a pen touch off-state exceeding some prescribed time, etc. As a result, the number of strokes of an input character, and an order of each stroke are derived from on and off of a pen touch of the input pen 2, and sent to a stroke recognizing part 6 and a character recognizing part 9. In a pre- processing part 4, stroke information of every stroke is sent to the stroke recognizing part 6. In the stroke recognizing part 6, the stroke information obtained from the per-processing part 4 is collated with a stroke pattern of a basic stroke in accordance with the number of strokes of the input stroke by a stroke number counting part 5.

Description

DETAILED DESCRIPTION OF THE INVENTION (Technical Field) The present invention relates to a character recognition method that performs character recognition based on information regarding strokes that constitute a character.

(Prior Art) Conventionally, a system has been adopted in which input strokes are classified into basic strokes prepared in advance based on information regarding the strokes that constitute a character, and a manually created character is recognized from a set of basic strokes. However, in B ice,
Many characters are used, including kanji, hiragana, katakana, alphabets, and numbers, and there are also many basic strokes for classifying the strokes. However, there are many strokes that make up hiragana that are not used for kanji, katakana, alphabets, numbers, etc., and there are also strokes that make up alphabets and numbers that are not used for other characters. . If there are many types of characters to be recognized in this way, the number of basic strokes that must be prepared will also inevitably increase, so errors in basic stroke classification that occur when dividing input strokes into basic strokes result in a low recognition p. Ta.

(Point 11) The present invention aims to keep the classification error rate low for No. 1 (basic stroke by limiting the main stroke group and use a small amount of processing acid) and loaf to be matched based on the number of strokes of the input character, and to achieve a high recognition rate of 111. The purpose is to provide a character recognition method that can.

(Example) Hereinafter, an example of the present invention will be described in detail with reference to the drawings.

FIG. 1 is a block diagram showing the configuration of a character recognition device which is an embodiment of the present invention. In the figure, 3 is a character information input device for inputting character information to be recognized, which is composed of a tablet 1 and an input pen 2. By using the input pen 2 to draw character information to be recognized on the tablet 1, Input is made. 4 is a character information input device 3
5 is a pre-processing unit that performs noise removal, normalization, etc. on the input character information; 5 is the coordinate information of the input pen 2 on the tablet 1, and whether the human pen 2 is touching the tablet 1 or not is turned on or off A stroke number counting section 6 calculates the number of strokes of the input character from the information; 6 is a stroke recognition section that recognizes the input stroke of the input character based on information from the preprocessing section 4 and information from the stroke number counting section 5; 7 is for stroke recognition; 8 is a character information processing unit that processes the length of each input stroke, positional relationship, etc., and 9 is the result obtained from the stroke recognition unit 6. and a character recognition unit that recognizes input characters based on character information obtained from the character information processing unit; lO is a character dictionary unit in which a plurality of types of character patterns are stored;
Reference numeral 1 denotes an output unit that outputs the result recognized by the character recognition unit 9.

FIG. 2 is an example of basic strokes stored in the basic stroke dictionary section 7 of FIG. 1, and displays basic strokes used only for characters with a stroke 1d number and representative strokes 1 to 6. The arrow of the representative stroke represents the direction of movement of the input pen 2. The strokes with "0" in the third column are basic strokes used only for hiragana, katakana, alphabets, and numbers, and are used only for characters with a stroke count of 1 or more and 6 or less.

Next, the present embodiment will be described with reference to FIGS. 1 and 2.

When the operator writes a character using the input pen 2 on the tablet 1-L-, the coordinate information of the input pen 2 on the tablet 1 and whether the tip of the input pen 2 is touching the tablet 1 at regular intervals is determined. Information on whether or not there is
and is sent to the stroke number counting section 5. The stroke counting unit 5 specifies character separation by touching a certain area on the tablet 2 with the tip of the manual pen 2, cuts out characters by touching off the pen after a certain period of time, and inputs the input pen 2. The number of strokes of the input character and
The number of each stroke is determined and sent to the stroke recognition section 6 and character recognition section 9. The preprocessing unit 4 performs processing such as noise removal and smoothing on the input character information, and then cuts out strokes based on the on/off information of the bend touch of the input pen 2, and extracts stroke information for each stroke. It is sent to the stroke recognition section 6. In addition, the length of each stroke of the input character, the start point and end point of the stroke, the coordinates of the point of change in the moving direction of the input pen 2, the intersection of each stroke, etc.
The character information processing section 8 sends "No."

The stroke recognition unit 6 compares the stroke information obtained from the preprocessing unit 4 with the stroke pattern of basic strokes according to the number of input strokes obtained from the stroke count counting unit 5, and classifies it as one of the basic strokes. .

Now, the kanji "ji" is tablet) 1 to "/".

r-J, rlJ, NJ'+ rlJ, r-J
"Tsukasa, r-", r\",, r/", r-J.

It is assumed that the characters r-J and rlJ are input as 13-stroke characters. Stroke r' from 1st stroke to 6th stroke.

When r-J, N'', N'', r-+'', , r-J is input, it is not known whether the number of strokes of this input character is 6 or more F or 7 or less, so each stroke is divided into all Kyogi strokes. It matches the pattern according to a predetermined algorithm and classifies it into basic strokes. However, when the 7th character "-" is input, it is known that the number of strokes of this character is 7 or more, so the subsequent strokes are basic strokes excluding the basic strokes that only appear in characters with 1st to 6th strokes. The pattern is matched according to a predetermined algorithm and classified into basic strokes. If strokes from 1st to 6th stroke are classified as basic strokes that appear only in characters from 1st to 6th stroke, then The stroke pattern is compared with a predetermined algorithm, and processing is performed, such as classifying the stroke as a basic stroke or rejecting it as an unrecognizable stroke and performing predetermined processing. As described above, when all the strokes of one character are processed by the stroke recognition unit 6, the character recognition unit 9 processes the stroke recognition unit 6 to the 1d number/h-1 character information processing unit 8 for each character stroke.
Stroke position information, stroke intersection information, stroke length character information, and the number of strokes of the input character are obtained from the stroke number counting section 5, and the results are compared with the character pattern registered in the character dictionary section 10 and the recognition result is output from the output section 11. do.

In the M embodiment, the basic stroke group to be compared was changed depending on whether the number of strokes of one character was 6 or less or 7 or less, or the stroke ID number 7.
10.1B, 34, 35° 36 strokes, loaf, strokes that appear only in the second car, strokes in Id Nan ~ 16, 17, etc. There are strokes that appear only in a certain number of strokes, so the number of strokes in a character. In addition, the group of basic strokes to be compared may be limited depending on which car the input stroke belongs to.

(Effects) As is clear from the explanation of 1. According to the present invention, the basic stroke length to be verified is limited by the number of strokes of a human character, and a high stroke recognition rate can be obtained with a small amount of processing. It is possible to increase the recognition rate.

[Brief explanation of drawings]

FIG. 1 is a diagram showing the configuration of a character recognition device according to an embodiment of the present invention, and FIG. 2 shows a basic stroke dictionary stored in the basic stroke dictionary section shown in FIG. In this figure, 3 is a character information input device, 5 is a stroke number of 51, 6 is a stroke recognition section, 7 is a basic stroke dictionary section, 9 is a character recognition section, and 10 is a character dictionary section. Junjin Canon Co., Ltd.

Claims (1)

[Claims] A character recognition method, characterized in that, in a character recognition device that performs character recognition based on information regarding strokes constituting a character, a group of basic strokes to be compared is changed according to the number of strokes of an input character.