JPS6292087A - Recognizing device for on-line handwritten character - Google Patents

Abstract

PURPOSE:To recognize quickly handwritten characters by reading out only the characters corresponding to a selection information stored previously in response to a basic stroke when the degree of resemblance is calculated among the handwritten strokes of input handwritten characters. CONSTITUTION:A quantization processing means 2 calculates the handwritten strokes of the handwritten characters supplied from a handwritten character input means 1. A basic stroke selecting means 3 reads out only the basic stroke that contains at least a piece of information on the stroke order, the number of changing points of (x) and (y) directions and the character types from a basic stroke dictionary 4 against the handwritten strokes. This read basic stroke is informed to a resemblance degree calculating means 5 for the calculation of the degree of resemblance of the handwritten strokes. This calculated degree of resemblance is delivered in the form of the recognized character information and the minimum degree of resemblance is recognized as the handwritten characters. In such a way, the handwritten characters can be quickly recognized.

Description

[Detailed Description of the Invention] [Summary] In an online handwritten character recognition device that recognizes manually handwritten characters, information on the stroke order, the number of change points in the X direction and the y direction, and the character type is provided for each basic stroke. A basic stroke dictionary storing selection information including selection information obtained from one handwritten stroke and a basic stroke selection means for selecting and reading out only basic strokes having selection information corresponding to the selection information obtained from one handwritten stroke, and calculating the calculated similarity. Among them, the one corresponding to the smallest value is recognized as the input handwritten character.

[Industrial application field]

The present invention adopts a configuration in which, when calculating the similarity of handwritten strokes of 1 handwritten input characters, only those corresponding to selection information (character type, etc.) registered in advance in association with basic strokes are read out. By doing so, you can quickly
The present invention relates to an online handwritten character recognition device configured to recognize handwritten characters by calculating Q1 degrees.

[Problems to be solved by conventional technology and invention] Conventional 2
The device that recognizes handwritten characters online is 1.
The handwritten strokes of the handwritten characters were compared with all the basic strokes to be compared, and the degree of similarity was calculated to recognize the handwritten characters. For this reason, in order to convert handwritten characters online, comparison with all basic strokes is required, making it difficult to quickly recognize them.

[Means for solving problems]

In order to solve the above-mentioned problems, the present invention provides selection information f that is registered in advance in association with basic strokes when calculating the similarity of handwritten strokes of characters input by hand.
By adopting a configuration that reads only those that correspond to il (character type, etc.).

Quickly calculate similarity and change handwritten letters to L? l! I try to understand.

FIG. 1 shows the basic configuration of the present invention. In the figure, l represents handwritten character input means, 2 represents quantization processing means, 3 represents basic stroke selection means, 4 represents basic stroke dictionary, and 5 represents similarity calculation means.

In FIG. 1, five handwritten character input means 1 are devices for inputting nine characters by hand, such as a tablet.

The quantization processing means 2 calculates handwritten strokes of characters input by the handwritten character input means 1. This handwritten stroke is characterized by direction and stroke (length).

Basics) Q,'7 The selection means 3 selects only the basic strokes corresponding to the character type information of the handwritten characters from the basic stroke dictionary 4 for the handwritten strokes of the handwritten characters notified from the quantization processing means 2. It is selected and read out.

In the case of the present invention, the basic stroke dictionary 4 includes one basic stroke and, in the case of the present invention, a stroke order 1 in which the basic stroke is used, the number of change points in the X direction and the y direction, and information on the character type in correspondence with the basic stroke. It is designed to be stored.

This information may be informed by a notification from the host or entered in some other way.

The similarity calculation means 5 compares the handwritten strokes quantized by the quantization processing means 2 with the basic strokes selectively read out from the basic stroke dictionary 4 by the 5 basic stroke selection means 3 and calculates 1 similarity. It calculates degrees.

Information regarding the degree of similarity calculated by this degree of similarity calculating means 5 (12 characters information) is outputted 5 based on this information (combination of basic strokes).

The characters are determined.

[Effect]

When the configuration shown in FIG. 1 is adopted and a handwritten character is input from the handwritten character input means 1, the quantization processing means 2 calculates the handwritten stroke of the handwritten character. With respect to the handwritten stroke notified to the basic stroke selection means 3, at least one piece of information exists regarding the stroke order, the number of change points in the X direction and the y direction, and the character type corresponding to the handwritten input character. Only basic strokes are read from the basic stroke dictionary 4. This read basic stroke is notified to the similarity calculation means 5, and the similarity for one handwritten stroke is calculated.

This calculated degree of similarity is output as recognized character information, and a character corresponding to a combination of basic strokes having a small degree of similarity is searched for and one handwritten character is recognized.

As described above, only basic strokes for which at least one selection information such as character type exists corresponding to one handwritten stroke are selected, and the similarity of one handwritten stroke is calculated. One handwritten character is recognized by calculating the combination of basic strokes for which the calculated degree of similarity is the smallest. Therefore, it is not necessary to calculate the degree of similarity for all basic strokes, and it is possible to quickly calculate the degree of similarity and recognize handwritten characters.

〔Example〕

FIG. 2 shows the configuration of one embodiment of the present invention. In the figure.

1-1 is a tablet, 6 is a character coordinate reading means, 7 is a normalization processing means, 8 is a selection information specifying means, 8-1 is stroke order information, 8-2 is X-direction change point number information, 8-3 is Information on the number of change points in the X direction, 8-4 is mode information, and 9 is basic stroke extraction means.

10 represents a character selection means, and 11 represents a character dictionary. still.

In the figure, numerals 2 to 5 are equivalent to those shown in FIG.

In FIG. 2, the coordinates (x, y) of the trajectory of a character manually handwritten from the tablet 1-1 using a pen are read by the character coordinate reading means 6.

The read handwritten characters are reduced or enlarged by the normalization processing means 7 and normalized to a predetermined size. This normalized handwritten character notified to the quantization processing means 2 is extracted as one handwritten stroke (expressed by direction and length).

The selection information specifying means 8, which has received the notification of the extracted handwritten strokes, notifies the basic stroke selection means 3 of the extracted handwritten strokes, and also selects the host for the notified handwritten strokes. Notified stroke order information 8-1°X direction change point number information 8-2.
Information on the number of change points in the y direction! [! 8-3. and mode information 8-4 are also notified. Upon receiving this notification, the basic stroke selection means 3 selects the stroke ranking information 4-1, which corresponds to the notified stroke ranking information 8-1, etc. X direction change point number information 4-2. Information on the number of change points in the y direction 1 [14
-3, and mode information base -4, only the basic strokes that have at least one earth are selected in the basic stroke dictionary 4.
, and notify the similarity calculation means 5. The similarity calculating means 5, which has received the notification of this basic stroke, calculates the similarity between the two handwritten strokes. The basic stroke extracting means 9, which has been notified of the calculated similarity, extracts the combination of basic strokes with the smallest similarity value and notifies the character selection means 10 of the combination. Upon receiving this notification, the character selection means 10 searches the one-character dictionary 11 for a character corresponding to the combination of basic strokes for which one notification has been received, and outputs the searched character as a recognized character.

As explained above, according to the present invention, the extracted stroke ranking information 8-1. X direction change point number information 8
-2. X direction change point number information 8-3 and mode information I
Only the basic strokes that correspond to at least one of 48-4 are read out from the basic stroke dictionary 4 and 1 (12 degrees is calculated, so unrelated basic strokes are read out and the similarity is calculated. Because there are no markings, handwritten characters can be recognized quickly.

FIG. 3 shows an example of basic stroke selection information. This basic stroke selection information is registered in the basic stroke dictionary N4 in association with the basic stroke.

The "number" shown at the top of FIG. 3 is an address to search for a pattern (basic stroke).

The stroke order shown in the upper part of the figure shows the number of handwritten strokes in which the pattern specified by the number appears. For example, the number "32" appears only second.

The number of change points in the X direction shown in the upper part of the figure represents the number of change points in the X direction.

The number of change points in the X direction shown in the upper part of the figure represents the number of change points in the X direction.

The mode shown in the upper part of the figure represents the type of character written by hand. In the diagram, “Ao is alpha head.

“N” is a number, “K” is katakana, “J” is hiragana 1
°C" is a kanji character, and "S" is a special symbol. For example, when creating a slip for business use using only numbers, the second symbol indicates that only handwritten characters using the numbers are used. By notifying the mode information 8-4 in the figure, the basic stroke dictionary 4 can be changed from the inside to the mode information 8-4.
By reading out only the pattern (basic stroke) of the number specified by the number "N", it is possible to calculate the number "A" or higher. Therefore, since there is no need to read basic strokes unrelated to numbers and calculate the degree of similarity, it is possible to quickly calculate the degree of similarity and recognize handwritten characters.

〔Effect of the invention〕

As explained above, according to the present invention, when calculating the similarity of handwritten strokes of characters input by hand, selection information f[
! Since the system is configured to read out only the basic strokes whose (character type, etc.) corresponds to the above-mentioned manually drawn strokes, it is possible to quickly calculate the degree of similarity and recognize handwritten characters.

[Brief explanation of drawings]

FIG. 1 shows the basic configuration of the present invention, FIG. 2 shows the configuration of one embodiment of the invention, and FIG. 3 shows an example of basic stroke selection information. In the figure, 1 is a handwritten character input means, 2 is a quantization processing means, and 3
4 is a basic stroke selection means, 4 is a basic stroke dictionary,
5 represents a similarity calculation means.

Claims (1)

[Claims] An online handwritten character recognition device that recognizes characters input by hand, comprising: a quantization processing means for calculating handwritten strokes of characters input by hand; and a stroke order and x direction in association with each basic stroke. A basic stroke dictionary that stores selection information including the number of change points in the and y directions, and character type information, and selection information obtained for handwritten strokes of characters input by hand, such as stroke order, x direction, and y
basic stroke selection means for selecting and reading out only basic strokes that correspond to at least one of information on the number of change points in direction and character type; and basic strokes selected by this basic stroke selection means and hand A similarity calculation means for calculating the similarity with calligraphy strokes, and a similarity calculation means calculated by this similarity calculation means,
An online handwritten character recognition device characterized in that it is configured to recognize the smallest value as an input handwritten character.