JP3190776B2 - Online handwritten character recognition 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 on-line handwritten character recognition device, and more particularly, to a system for recognizing handwritten characters input from a tablet with a pen and displaying the recognition result on a CRT.
(Cathode Ray Tube: cathode ray tube), and an online handwritten character recognition device having a function of displaying on a liquid crystal or the like. For example, it is applied to a word processor, an electronic organizer, a personal computer, and the like.

[0002]

2. Description of the Related Art In an on-line handwritten character recognition apparatus which performs stroke recognition every time a stroke is input, feature extraction and stroke recognition of the input stroke are completed between the input of a stroke and the end of the input of the next stroke. It is desirable to do. However, when there are many basic stroke dictionaries, it takes a lot of time to recognize strokes, and there is a case where the accumulation of processing occurs.

[0003]

As described above, in the conventional on-line handwritten character recognition device, when the number of basic stroke dictionaries increases, the processing for each stroke is accumulated, and the number of processes from the end of character input to the display of the recognition result increases. It takes time. As one of the means to solve this, high-performance CP
U (central processing unit) or dedicated matching LS
I can be used.

[0004] The present invention has been made in view of such circumstances, and therefore, a preliminary selection of a basic stroke dictionary is performed by using information such as the geometric features of input strokes, the type of characters to be recognized, and the number of strokes. It is another object of the present invention to provide an online handwritten character recognition device capable of reducing the processing time required for stroke recognition.

[0005]

In order to achieve the above object, the present invention provides: (1) an online handwritten character recognized based on writing information (coordinate points, pen ON / OFF information) of an input character; An input unit for detecting the handwritten information; a stroke detecting unit for detecting a stroke (a curve in a section from a pen OFF → ON state to an ON → OFF state) based on the handwritten information; A basic stroke dictionary that describes a basic stroke and preliminary selection information added to the basic stroke,
The AND processing of the bit representation of the detected stroke information and the bit representation of the preliminary selection information is performed.
Performed, stroke order and basic stroke preliminary selection means for selecting a basic stroke of Matching object based on the processing result, the stroke recognition means for performing matching between the extracted features and basic stroke said selected beforehand character A character dictionary described in accordance with a stroke code, character matching means for matching the stroke recognition result with the character dictionary, and display means for displaying the recognition result. The preliminary selection information is stroke number information indicating in which stroke of the character the basic stroke is used, and (3) the preliminary selection information is character type information in which the basic stroke is used; Further, (4) the preliminary selection information is a starting point of a basic stroke.
-It is characterized by the position information of the end point .

[0006]

The online character recognition apparatus according to the present invention detects writing information of a character input by a user using input means such as a tablet, and detects a stroke from the writing information. A geometric feature required for recognition is extracted from the detected stroke, and a basic stroke dictionary is selected using various features. The input pattern created from the extracted features is matched with the basic stroke dictionary that has passed the preliminary selection, and the result of the stroke recognition is compared with a character dictionary in which characters are previously described in stroke order in stroke code. The recognition result is displayed by the display means. Based on such a configuration, by preliminary selection of the basic stroke dictionary, it is possible to prevent accumulation of processing for each stroke and to shorten the recognition time.

[0007]

Embodiments will be described below with reference to the drawings. FIG. 1 is a configuration diagram for explaining an embodiment of an online handwritten character recognition device according to the present invention. In the drawing, 1 is a tablet, 2 is a stroke detection unit, 3 is a feature extraction unit,
4 is a basic stroke preliminary selection unit, 5 is a stroke recognition unit, 6 is a basic stroke dictionary unit, 7 is a character dictionary matching unit,
Reference numeral 8 denotes a character dictionary unit, and 9 denotes a display unit.

An online handwritten character recognition apparatus according to the present invention is an online handwritten character recognition apparatus that recognizes based on handwritten information (coordinate points, pen ON / OFF information) input by a user. The input unit 1 such as a tablet detects writing information. The stroke detection unit 2 detects a stroke (pen is OFF → ON state to ON → O) based on the writing information.
FF state section) is detected. Feature extraction unit 3
Extracts geometric features (eg, line segment direction, length, etc.) required for recognition from the detected strokes. The stroke recognition unit 5 performs matching between the extracted feature and the basic stroke dictionary. The character dictionary collating unit 7 collates the stroke recognition result with a character dictionary in which characters are written in advance in stroke order with stroke codes. Display 9
Displays the recognition result. Preliminary selection information is added to the basic stroke dictionary 6, and the number of basic stroke dictionaries to be matched by preliminary selection is reduced.

[0009] As the preliminary selection information, a basic straw is provided.
The stroke number information indicating in which stroke of the character the character is used is added to the basic stroke dictionary 6, and the basic stroke dictionary 6 is reduced. Further, character type information using a basic stroke is added to the basic stroke dictionary 6 as preliminary selection information, and the basic stroke dictionary 6 is reduced. Further, position information of the start point and end point of the stroke is added to the basic stroke dictionary 6 as preliminary selection information, and the basic stroke dictionary 6 is reduced.

A case where the user inputs strokes in the order shown in Table 1 below as shown in FIG. 2 in order to input "a" will be described below.

[0011]

[Table 1]

FIG. 3 is a flowchart for explaining the operation of the online handwritten character recognition device according to the present invention. Hereinafter, the steps will be sequentially described. step1 : When the user inputs characters on the tablet 1 with a pen, the stroke detection unit 2
The writing information is sent to. The stroke detection unit 2 detects the pen-on writing information based on the writing information. next,
Until the writing state in the pen-off state is detected, the stroke information is held by the stroke detection unit 2 as coordinate data. Step 2: When the stroke detection unit 2 detects the writing information in the pen OFF state, the stroke detection unit 2 instructs the feature extraction unit 3 that a stroke has been input, and sends the coordinate data.

[0013] step3: feature extracting section 3 extracts a geometric features required for character recognition from coordinate data sent and sends the stroke preliminary selection section 4. According to the recognition method of the present invention, the direction sequence between coordinate points, the start point, the end point,
The minimum value, length, and the like are characteristics of the stroke. step 4: The basic stroke preliminary selection unit 4 checks preliminary input information added to each basic stroke of the basic stroke dictionary unit 6 describing the input stroke and the basic stroke as shown in FIG.

[0014] step5: stroke recognition unit 5 performs matching between the basic stroke dictionary satisfying the stroke and the preselection information input, and sends the recognition result for each stroke in the character collating unit 7. step 6 : The character collating unit 7 collates the result of the stroke recognition with the character dictionary unit 8 in which the characters are described in advance in the stroke order with basic strokes, and determines the characters satisfying the contents of the dictionary as the recognition results. In the case of “A”, the basic stroke code string is described as follows. Character dictionary A ... (01 or 02), (04 or 0
5), (06)

An actual embodiment will be described with reference to FIGS. As shown in FIG. 5, the basic stroke dictionary referred to in the preliminary selection and stroke recognition includes “basic stroke numbers”, “preliminary selection information”, and “stroke characteristics”. Various information is represented by bits in the storage location of the preliminary selection information.

First, a case where information on the number of strokes in which a stroke is used is used as preliminary selection information will be described.
Basic stroke "06" which is the third stroke of "A"
Is used as the first stroke of “no” and the second stroke of “me” other than “a”. That is, the basic stroke “06” may be matched with respect to the input of one to three strokes. Therefore, the bit corresponding to the number of pictures to be referred to is set to “1”, and is represented by 8 bits. The preliminary selection information of the basic stroke “06” is represented as shown in FIG. The basic stroke preliminary selection unit 4 performs an AND process on the “preliminary selection information” and the “number of strokes”, and determines whether or not the result of the processing is a matching target based on whether or not the result is 0. Table 2 shows the result of the AND process with the preliminary selection information of the basic stroke “06” for the first to four strokes of the input.

[0017]

[Table 2]

When the number of strokes becomes four or more, the result of the AND processing becomes 0
It can be seen that the basic stroke “06” is out of the matching target.

Next, a case where a character type is used as preliminary selection information will be described. The basic stroke "06", which is the third stroke of "A", may be matched only when "Hiragana" is included in the target character type. Therefore,
The bit corresponding to the character type to be used is set to '1' and is represented by 8 bits. The preliminary selection information of the basic stroke “06” is represented as shown in FIG.

The basic stroke preliminary selection section 4 performs an AND process on the "preliminary selection information" and the "character type to be recognized", and determines whether or not the result of the processing is a matching target based on whether or not the result is 0. (A) Kanji, Hiragana, Katakana, Alphabet, Number, Symbol, and (b)
The result of the AND process with the preliminary selection information of the basic stroke “06” for the case of alphabets, numbers and symbols is as follows:
Table 3 below.

[0021]

[Table 3]

In the case of (b), the result of the AND processing becomes 0, and it is understood that the basic stroke "06" is out of the matching target.

Next, a case where the position information of the starting point and the ending point of the stroke is used as the preliminary selection information will be described. As shown in FIG. 8, based on the center of gravity of the inputted stroke,
Consider two areas. In the basic stroke dictionary, an area bit that may be a start point / end point is set to “1”, and 8
Expressed in bits. The preliminary selection information of the basic stroke 01 is represented as shown in FIG.

In the basic stroke preliminary selection section 4, "preliminary selection information" and "position information of the start and end points of the input stroke"
Is determined by determining whether the result of the processing is equal to the “position information of the start point / end point of the input stroke”. When “A” as shown in FIG. 2 is input, the result of the AND process with the preliminary selection information of the basic stroke “01” is as shown in Table 4 below.

[0025]

[Table 4]

In the case other than one stroke, the result of the AND processing is different from the “position information of the input stroke start point / end point”, and it can be seen that the basic stroke “01” is out of the matching target. As described above, the online handwritten character recognition device of the present invention can add various types of preselection information to the basic stroke dictionary and reduce the number of basic stroke dictionaries that perform matching by preselection. We can block pool. Further, in the above embodiment, the method of performing the preliminary selection of the basic stroke dictionary by expressing the preliminary selection condition in 8 bits has been described, but the present invention is not limited to this.

[0027]

As is clear from the above description, the online character recognition device according to the present invention detects the writing information of the character input by the user using the input means such as a tablet, and detects the stroke from the writing information. Do. Extract the geometric features required for recognition from the detected stroke,
A basic stroke dictionary is selected using various features. The input pattern created from the extracted features is matched with the basic stroke dictionary that has passed the preliminary selection, and the result of the stroke recognition is compared with a character dictionary in which characters are previously described in stroke order in stroke code. The recognition result is displayed by the display means. Based on such a configuration, by preliminary selection of the basic stroke dictionary, it is possible to prevent accumulation of processing for each stroke and to shorten the recognition time.

[Brief description of the drawings]

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

FIG. 2 is a diagram illustrating a writing example of “A” in the present invention.

FIG. 3 is a flowchart for explaining a (name) operation according to the present invention.

FIG. 4 is a diagram showing a shape of a basic stroke in the present invention.

FIG. 5 is a diagram showing a structure of a basic stroke dictionary according to the present invention.

FIG. 6 is a diagram showing stroke number information of a basic stroke 06 in the present invention.

FIG. 7 is a diagram showing character type information of a basic stroke 06 in the present invention.

FIG. 8 is a diagram showing position information of a start point and an end point in the present invention.

FIG. 9 is a diagram showing position information of a basic stroke 01 according to the present invention.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Tablet, 2 ... Stroke detection part, 3 ... Feature extraction part, 4 ... Basic stroke preliminary selection part, 5 ... Stroke recognition part, 6 ... Basic stroke dictionary part, 7 ... Character dictionary collation part, 8 ... Character dictionary part, 9 Display part.

Claims (4)

(57) [Claims] 1. An on-line handwritten character recognition device for recognizing based on handwritten information of an input character, input means for detecting said handwritten information, stroke detecting means for detecting a stroke from said handwritten information, A feature extraction unit for extracting a geometric feature necessary for recognition from the stroke obtained, a basic stroke dictionary describing a basic stroke and preliminary selection information added thereto,
Bit information of the detected stroke information ,
AND processing with the bit representation of the preliminary selection information
Was carried out, and basic stroke preliminary selection means for selecting a basic stroke of Matching object based on the processing results,
A stroke recognition unit that matches the selected basic stroke with the extracted feature; a character dictionary in which characters are described in advance in a stroke code in a stroke code; and a comparison between the stroke recognition result and the character dictionary. An on-line handwritten character recognition device, comprising: a character collating means for performing the processing; and a display means for displaying a recognition result. 2. The online handwritten character recognition device according to claim 1, wherein the preliminary selection information is stroke number information indicating in which stroke of the character the basic stroke is used. 3. The pre-selection information is character type information used for a basic stroke.
Online handwritten character recognition device. 4. The method according to claim 1, wherein the preliminary selection information includes a basic stroke.
2. The position information of a start point and an end point.
Line handwritten character recognition device as claimed.