JP2905334B2 - Online handwritten character recognition dictionary creation method and online handwritten character recognition dictionary creation device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an online handwritten character recognition dictionary creation method and an online handwritten character recognition dictionary creation device used for registering new characters in a dictionary of an online handwritten character recognition device.

[0002]

2. Description of the Related Art Online dictionaries for handwritten character recognition are based on data of the same character written by many authors, and have redundancy that allows variations in the character shape due to the writer or chance. . However, when the user adds a character to be recognized, since the character data of a large number of people cannot be collected, the conventional online handwritten character recognition device uses the stroke position shape feature from the character input once by the user. It is configured to extract and register it as dictionary data as it is.

[0003]

However, in the above-described conventional configuration, the variation of the character shape due to the writer or accident cannot be taken into the dictionary data, so that the additionally registered characters cannot be recognized correctly with only a slight change in the character shape. There was a problem. SUMMARY OF THE INVENTION The present invention has been made in view of the above circumstances, and a dictionary creation method for online handwritten character recognition and online handwriting that can create dictionary data with redundancy allowing glyph variation from glyph data input by handwriting only once. An object of the present invention is to provide a dictionary creation device for character recognition.

[0004]

According to a first aspect of the present invention, a user inputs a part dictionary by processing and storing stroke position and shape characteristics for each radical or symbol constituting a character. A feature extraction step of obtaining a position and shape feature of a stroke constituting a handwritten character; and a stroke position and shape feature of the handwritten character obtained in the feature extraction step are described in the component dictionary for each component such as a radical or a symbol constituting the character. A component recognition step of comparing with the component, and determining a component constituting the handwritten character; and a dictionary data combining step of combining dictionary data of the handwritten character from the dictionary data of each component constituting the handwritten character recognized in the component recognition step. Is performed.

[0005] A second aspect of the present invention provides a handwritten character input unit for a user to input a handwritten character, and a feature extraction unit for obtaining a position and shape characteristic of a stroke constituting the handwritten character input to the handwritten character input unit. And a component dictionary that processes and stores stroke position / shape features for each radical or symbol that constitutes the character, and a stroke position / shape feature of the handwritten character obtained by the feature extraction unit. A part recognition unit that compares the parts dictionary for each part such as a neck and a symbol to determine a part constituting the handwritten character, and a handwritten character from the dictionary data of each part constituting the handwritten character recognized by the part recognition unit. And a dictionary data synthesizing unit for synthesizing the dictionary data.

According to a third aspect of the present invention, the position of a stroke constituting a handwritten character input by a user is obtained by using a character recognition dictionary which processes and stores stroke position and shape characteristics of a handwritten character to be recognized. A feature extraction step for obtaining a shape feature, and a stroke position shape feature of the handwritten character obtained in the feature extraction step is compared with the character recognition dictionary for each part such as a radical or a symbol constituting the character, and the handwritten character is compared with the character recognition dictionary. Executing a component recognition step of determining a component to be configured and a dictionary data synthesizing step of synthesizing dictionary data of a handwritten character from dictionary data of each component constituting the handwritten character recognized in the component recognition step. I have.

According to a fourth aspect of the present invention, there is provided a handwritten character input unit for a user to input a handwritten character, and a feature extracting unit for obtaining a position and shape characteristic of a stroke constituting the handwritten character input to the handwritten character input unit. And a character recognition dictionary that processes and stores stroke position and shape features of handwritten characters to be recognized, and the stroke position and shape characteristics of handwritten characters obtained by the feature extraction unit are used as radicals and For each part such as symbols, collate with the character recognition dictionary,
A component recognizing unit that determines a component constituting the handwritten character, and a dictionary data combining unit that combines dictionary data of the handwritten character from dictionary data of each component constituting the handwritten character recognized by the component recognizing unit. It is characterized by.

[0008]

According to the first aspect of the present invention, in a feature extraction step, a user inputs in a feature extraction step using a component dictionary that processes and stores stroke position and shape features for each radical or symbol constituting a character. In the component recognition step, the stroke position and shape characteristics of the handwritten characters obtained in the feature extraction step are obtained by obtaining the position and shape characteristics of the strokes that constitute the handwritten characters in the component dictionary for each of the components such as radicals and symbols that constitute the characters. Collation, determine the parts that make up the handwritten characters,
In the dictionary data synthesizing step, dictionary data of the handwritten character is synthesized from the dictionary data of each component constituting the handwritten character recognized in the component recognition step.

In the invention according to claim 2, the handwritten character input section is for a user to input handwritten characters. The feature extraction unit obtains a position and shape feature of a stroke constituting a handwritten character input to the handwritten character input unit.
The component dictionary processes and stores stroke position and shape features for each radical or symbol constituting a character. The component recognition unit
The stroke position / shape characteristics of the handwritten character obtained by the feature extraction unit are collated with the component dictionary for each component such as a radical or a symbol that constitutes the character, and the component that constitutes the handwritten character is determined. The dictionary data combining unit combines the dictionary data of the handwritten characters from the dictionary data of each component constituting the handwritten characters recognized by the component recognition unit.

In a third aspect of the present invention, in the feature extraction step, the stroke position and shape features of the handwritten characters to be recognized are processed and stored in a character recognition dictionary.
The position and shape characteristics of the strokes constituting the handwritten character input by the user are obtained, and in the component recognition step, the stroke position and shape characteristics of the handwritten character obtained in the characteristic extraction step are converted to the radicals and symbols constituting the character. Each component is compared with the character recognition dictionary to determine the components that make up the handwritten character.In the dictionary data synthesis step, the dictionary data of the handwritten character is extracted from the dictionary data of each component that makes up the handwritten character recognized in the component recognition step. Combine.

[0011] In the invention of claim 4, the handwritten character input section is for a user to input handwritten characters. The feature extraction unit obtains a position and shape feature of a stroke constituting a handwritten character input to the handwritten character input unit.
The character recognition dictionary processes and stores stroke position and shape features of handwritten characters to be recognized. The component recognizing unit compares the stroke position and shape characteristics of the handwritten character obtained by the feature extracting unit with a character recognition dictionary for each component such as a radical or a symbol that constitutes the character, and determines a component that constitutes the handwritten character. . The dictionary data combining unit combines the dictionary data of the handwritten characters from the dictionary data of each component constituting the handwritten characters recognized by the component recognition unit.

[0012]

Embodiments of the present invention will be described below in detail with reference to the drawings. (Embodiment 1) FIG. 1 is a block diagram of an online handwritten character recognition dictionary creation device according to a first embodiment of the present invention. The online handwritten character recognition dictionary creation device includes a handwritten character input unit 1 and a feature extraction unit 2. A component dictionary 3, a component recognition unit 4, a dictionary data synthesizing unit 5, a dictionary data registration unit 6,
And a character recognition dictionary 7. Handwritten character input unit 1
Is composed of a tablet for detecting a coordinate position and an input pen, and is for a user to input handwritten characters. The feature extracting unit 2 obtains a position and shape feature of a stroke constituting a handwritten character input by the user using the handwritten character input unit 1. The component dictionary 3 stores stroke position and shape characteristics of radicals and symbols constituting characters. The component recognizing unit 4 checks the stroke position / shape characteristics of the handwritten character obtained by the feature extracting unit 2 with the component dictionary 3 for each component such as a radical or a symbol constituting the character, and determines a component constituting the handwritten character. decide. The dictionary data synthesizing unit 5 synthesizes dictionary data of handwritten characters from dictionary data of each component constituting the handwritten characters recognized by the component recognition unit 4. The dictionary data registration unit 6 additionally registers the dictionary data of the input characters synthesized by the dictionary data synthesis unit 5 in the character recognition dictionary 7.

FIG. 2 is an explanatory diagram of dictionary data stored in the component dictionary 3, where A and B are "A then B", (A | B)
Is "A or B", and [A] is a notation meaning "may or may not have A". Thus, the parts dictionary 3
Has redundancy in order to allow for variations in character shapes. Next, the operation will be described. The character handwritten by the user using the handwritten character input unit 1 is supplied to the feature extracting unit 2 as a coordinate sequence of the locus of the pen as shown in FIG. Thereby, the feature extracting unit 2 approximates the supplied coordinate sequence with line segments in eight directions as shown in FIG. 4, obtains a direction code sequence as shown in FIG. This direction code sequence is referred to as a feature F. Thereby, the component recognition unit 4
Are the features F: f1, f2,..., Fn from the feature extraction unit 2.
And the dictionary data D stored in the parts dictionary 3.
t: dt1, dt2,..., and dtm are collated, and if the following condition is satisfied, it is determined that the input character has dictionary data Dt as a component.

(1) Among the features F, f1, f2,.
i matches one or more parts. (2) fj, fj + 1,..., Fk satisfying i <j coincide with Dt. (3) There is no Dt ′ that matches fj, fj + 1, ‥, fl with i <j, k <l.

That is, in the case of the character shown in FIG.
b matches “pan”, but strokes a,
b, c, d, e, f, g, h, i are "gomben (words)"
"Pan" is not determined to be a component. If it is determined that "Gomben (word)" is a component, the collation work with the component dictionary 3 is performed again with the stroke j as the first stroke of the component. Then, the dictionary data of each recognized component is supplied to the dictionary data synthesizing unit 5. As a result, the dictionary data synthesizing unit 5 synthesizes the dictionary data of the input characters from the dictionary data of the individual components from the component recognizing unit 4 as shown in FIG. 7, and supplies the dictionary data to the dictionary data registration unit 6. I do. Thereby, the dictionary data registration unit 6 additionally registers the dictionary data from the dictionary data synthesis unit 5 in the character recognition dictionary 7.

As described above, the handwritten character input unit 1 for the user to input handwritten characters, and the feature extraction unit 2 for obtaining the position and shape characteristics of the strokes constituting the handwritten characters input to the handwritten character input unit 1 A part dictionary 3 that processes and stores stroke position / shape characteristics of each radical or symbol that constitutes a character; A component recognizing unit 4 for collating each component such as a neck and a symbol with the component dictionary 3 to determine a component constituting a handwritten character;
And a dictionary data synthesizing unit 5 for synthesizing the dictionary data of the handwritten characters from the dictionary data of each component constituting the handwritten characters recognized by the method. Generated dictionary data. (Embodiment 2) FIG. 8 is a configuration diagram of an online handwritten character recognition dictionary creation device according to a second embodiment of the present invention. The online handwritten character recognition dictionary creation device includes a handwritten character input unit 11 and a feature extraction unit 12. And the character recognition dictionary 13,
A component recognition unit 14, a dictionary data synthesis unit 15, and a dictionary data registration unit 16 are provided. Handwritten character input unit 11
Is composed of a tablet for detecting a coordinate position and an input pen, and is for a user to input handwritten characters. The feature extraction unit 12 allows the user to input the handwritten character input unit 11
Is used to obtain the position and shape characteristics of the strokes constituting the input handwritten character. The character recognition dictionary 13 stores stroke position shape characteristics of handwritten characters. The component recognizing unit 14 compares the stroke position and shape characteristics of the handwritten character obtained by the feature extracting unit 12 with the character recognition dictionary 13 for each component such as a radical or a symbol that constitutes the character. To determine. The dictionary data synthesis unit 15
The dictionary data of the handwritten character is synthesized from the dictionary data of each component constituting the handwritten character recognized by the component recognition unit 14. The dictionary data registration unit 16 is a dictionary data synthesis unit 15
Is additionally registered in the character recognition dictionary 13.

Next, the operation will be described. The character handwritten by the user using the handwritten character input unit 11 is supplied to the feature extracting unit 12 as a coordinate sequence of the locus of the pen. Thus, the feature extracting unit 12 approximates the supplied coordinate sequence by line segments in eight directions, obtains a direction code sequence, and supplies the direction code sequence to the component recognizing unit 14. This direction code sequence is referred to as a feature F. As a result, the component recognizing unit 14 receives the features F: f1, f from the feature extracting unit 12.
2, ‥, fn and a part of the dictionary data Dt: dt1, dt2, ‥, dtm of the character recognition dictionary 13,
When the following conditions are satisfied, the input character is dictionary data Dt.
It is determined that some of the components dtp, dtp + 1, ‥, dtq are included in the component parts.

(1) Among the features F, f1, f2,.
i matches one or more parts. (2) fj, fj + 1,..., Fk satisfying i <j coincide with a part dtp, dtp + 1,. (3) Dt with qp being the largest among those satisfying the above (2).

(4) Among those satisfying the above (3), the one having the minimum stroke number m of the dictionary data. That is, in the case of the character "Wa" in FIG.
Since c, d, and e match the ABCDE of the “role” dictionary, it is determined that one part is obtained by the above five strokes. Therefore,
The components of the input character "Wa" are part of the "ro", "people",
Becomes a "woman". Then, a part of the dictionary data recognized as the component is supplied to the dictionary data synthesizing unit 15. Thereby, the dictionary data synthesizing unit 15 synthesizes the dictionary data of the input characters from a part of the dictionary data from the component recognition unit 14 and supplies the dictionary data to the dictionary data registration unit 16. As a result, the dictionary data registration unit 16
Is additionally registered in the character recognition dictionary 13.

As described above, the component dictionary 3 in the first embodiment
, The character dictionary 13 is used, so that the component dictionary 3 unique to external character registration becomes unnecessary.

[0021]

As described above, according to the present invention, the position and shape characteristics of the strokes constituting the handwritten character input by the user are obtained, and the stroke position and shape characteristics are converted to the radicals and symbols constituting the characters. For each component, such as the component dictionary, the components that make up the handwritten character are determined, and the dictionary data of the handwritten character is synthesized from the dictionary data of the component. Dictionary data with redundancy that allows

Also, instead of a component dictionary that processes and stores stroke position / shape characteristics for each radical or symbol that constitutes a character, stroke position / shape characteristics of a handwritten character to be recognized are processed and stored. If a character recognition dictionary is used, a component dictionary unique to external character registration can be omitted.

[Brief description of the drawings]

FIG. 1 is a configuration diagram of an online handwritten character recognition dictionary creation device according to a first embodiment of the present invention.

FIG. 2 is an explanatory diagram of data stored in a component dictionary.

FIG. 3 is an explanatory diagram of a feature extraction method.

FIG. 4 is an explanatory diagram of a feature extraction method.

FIG. 5 is an explanatory diagram of a feature extraction method.

FIG. 6 is an explanatory diagram of a component recognition method.

FIG. 7 is an explanatory diagram of a component recognition method.

FIG. 8 is a configuration diagram of an online handwritten character recognition dictionary creation device according to a second embodiment of the present invention.

FIG. 9 is an explanatory diagram of a component recognition method.

[Explanation of symbols]

 REFERENCE SIGNS LIST 1 handwritten character input unit 2 feature extraction unit 3 component dictionary 4 component recognition unit 5 dictionary data synthesis unit 11 handwritten character input unit 12 feature extraction unit 13 character recognition dictionary 14 component recognition unit 15 dictionary data synthesis unit

Claims (4)

(57) [Claims] 1. Using a part dictionary that processes and stores stroke position and shape characteristics for each radical or symbol that constitutes a character, the position and shape characteristics of strokes that constitute a handwritten character input by a user are determined. The feature extraction step to be determined, and the stroke position shape feature of the handwritten character obtained in the feature extraction step is compared with the component dictionary for each component such as a radical or a symbol that constitutes the character, and the component that constitutes the handwritten character is checked. An online handwritten character, comprising: executing a component recognition step of determining; and a dictionary data synthesizing step of synthesizing dictionary data of the handwritten character from dictionary data of each component constituting the handwritten character recognized in the component recognition step. How to create a dictionary for recognition. 2. A handwritten character input unit for a user to input a handwritten character, a feature extracting unit for obtaining a position and shape feature of a stroke constituting the handwritten character input to the handwritten character input unit, A part dictionary that processes and stores stroke position and shape features for each radical or symbol, and the stroke position and shape characteristics of handwritten characters obtained by the feature extraction unit, such as radicals and symbols that constitute characters. A component recognizing unit that determines a component constituting a handwritten character by comparing the component dictionary with the component dictionary for each component; and synthesizes dictionary data of the handwritten character from the dictionary data of each component configuring the handwritten character recognized by the component recognizing unit. A dictionary creation device for online handwritten character recognition, comprising: 3. A feature for obtaining a position and shape feature of a stroke constituting a handwritten character input by a user by using a character recognition dictionary in which a stroke position and shape feature of a handwritten character to be recognized is processed and stored. Extracting the stroke position and shape characteristics of the handwritten character obtained in the extracting step with the character recognition dictionary for each part such as a radical or a symbol constituting the character, and determining a part constituting the handwritten character And a dictionary data synthesizing step of synthesizing dictionary data of handwritten characters from dictionary data of each component constituting the handwritten characters recognized in the component recognition step. Dictionary creation method. 4. A handwritten character input unit for a user to input a handwritten character, a feature extraction unit for obtaining a position and shape characteristic of a stroke constituting the handwritten character input to the handwritten character input unit, A character recognition dictionary that processes and stores the stroke position and shape characteristics of a certain handwritten character, and the stroke position and shape characteristics of the handwritten character obtained by the feature extraction unit are converted into parts such as radicals and symbols constituting the character. A component recognition unit that determines a component that constitutes a handwritten character by comparing with the character recognition dictionary; and synthesizes dictionary data of the handwritten character from the dictionary data of each component that constitutes the handwritten character recognized by the component recognition unit. An online dictionary for recognizing handwritten characters, comprising: a dictionary data synthesizing unit.