JPH07129717A - Online character segmenting device - Google Patents

Abstract

PURPOSE:To provide an online character segmenting device reduced in character segmentation error of separate characters, characters with voiced sound symbols, and characters with semi-voiced sound symbols. CONSTITUTION:The online character segmenting device is provided with a voiced sound symbol discriminating means which discriminates whether two continuous strokes inputted by writing are a voiced sound symbol or not and discriminates two strokes as one segment of the voiced sound symbol in the case of the voiced sound symbol, a voiced sound symbol/semi-voiced sound symbol coupling means 5 which couples a voiced sound symbol or a semi-voiced sound symbol to the just preceding fundamental segment, and a KANA (Japanese syllabary) separate character coupling means 6 which couples two continuous fundamental segments in the case of two fundamental segments constituting a KANA separate character, and characters with voiced sound symbols, characters with semi-voiced sound symbols, KANA separate characters, and characters easy to separate at the time of writing are segmented.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an online character slicing device for slicing online each character data from written character string data, and is particularly suitable for an online character recognition device.

[0002]

2. Description of the Related Art In an online character recognition device, data of each character is cut online from a character string data written by an input person using a coordinate input device such as a tablet or a stylus pen. Naturally, this cutting accuracy also affects the character recognition rate, and it is required to be performed accurately. As a method of cutting out the data of each character from the data of a written character string, conventionally, a method of displaying a character frame (mesh) and allowing the user to write in it has been generally used. JP-A-55-83
970 publication "). In addition to such a cutout method, for example, character recognition processing that cuts out characters based on the validity of the combination by using character recognition that is performed for each element such as bias and structure that composes a character. A method utilizing "(Japanese Patent Application Laid-Open No. 61-29982") has already been proposed.

[0003]

However, in the above method, since the character width information is used for clipping, when the character width of the written character is out of the predetermined range, the character width is erroneously cut out. There was a problem that it did. In addition, a lot of hiragana and katakana are included in general Japanese sentences. Separation characters such as “ha” and “ho”, dakuten characters such as “ga” and “gi”, and “pa” and “pi”. "
In some cases, the semi-voiced characters such as "" could not be cut out correctly.

The present invention solves the above-mentioned problems of the prior art, and provides an online character segmentation device with less character segmentation errors for dakuten characters, semi-dakuten characters, kana separated characters, and characters that are easily separated during writing. The purpose is to

[0005]

In order to solve the above problems, the present invention provides a coordinate input device for outputting position information of an input character when writing a character as coordinate data in an online character cutting device. , Based on the coordinate data output from the coordinate input device, stroke coding means for sequentially extracting and coding strokes as the input character is written, and whether two consecutive strokes are "dakuten" In the case of “dakuten”, two strokes are treated as a dakuten and one segment is judged as one segment, and the overlap of each stroke and the segment output by the dakuten judging means is detected and overlapped. When the basic segment is composed of only the dakuten or the semi-dakuten, and the basic segment dividing means that outputs an item as one segment In addition, there is provided a dakuten / semi-dakuten connecting means for connecting the immediately preceding basic segment and a kana separated character connecting means for connecting two basic segments when two consecutive basic segments are kana separated characters. .

[0006]

According to the present invention, the online character cutting device is constructed as described above, and each means operates as follows.
The coordinate input means successively outputs position information of characters written by the input person as coordinate data. The stroke coding means is
A stroke is extracted from the coordinate data and coded. The encoding is performed by classifying strokes according to the signature of each segment, the angle of the segment, and the rotation angle between the segments, for example, based on the characteristic points of the input data string for one stroke. The characteristic points are extracted by, for example, a method of obtaining a sign in the x and y directions between the data of the input data string and using a change point of the state of the sign as a characteristic point. The sign here is
It indicates whether the difference between the coordinate values in the X direction and the Y direction between the input data is positive, negative, or zero. The dakuten judging means judges that the two strokes are dakuten and outputs the two strokes as one segment when the two consecutive strokes are both “small strokes on the upper side”. The basic segment dividing means projects the segments output by each stroke and the dakuten judging means on the X-axis, and outputs the overlapping strokes and segments as one basic segment to the dakuten / semi-dakuten combining means.
The dakuten / semi-dakuten combining means connects with the immediately preceding basic segment when the basic segment is composed of only the dakuten and the semi-dakuten. The condition for the semi-voiced point is that the number of strokes of the basic segment is 1, and the stroke is small and round. Whether the stroke is round or not can be easily determined by the stroke code. The kana separated character combining means determines whether two consecutive basic segments satisfy the condition for being a kana separated character. Kana separation characters are separation characters for hiragana and katakana such as "ke", "ni", "ha", "ho", and "ha". The condition for being a kana separated character is described by the number of strokes, stroke code, and the like. Two basic segments are combined when the condition for being a kana separation character is satisfied. With the above configuration, it is possible to cut out dakuten characters, semi-dakuten characters, and kana separated characters that frequently appear in Japanese sentences and are easy to separate during writing.

[0007]

Embodiments of the present invention will now be described in detail with reference to the drawings. FIG. 1 is a diagram showing a configuration of an online character cutting device according to an embodiment of the present invention. As shown in the figure, this online character cutting device includes a coordinate input device 1, a stroke coding means 2, a dakuten judging means 3, a basic segment dividing means 4, a dakuten / semi-dakuten combining means 5, and a kana separated character combining means 6. , Basic segment output control means 7,
It is composed of the character cutting means 8. In the figure, 9 indicates an output terminal to an online character recognition device (not shown). Figure 2
2 is a flow chart showing the outline of the operation of the apparatus of FIG. 1, in which the stroke coding means 2 carries out step 101, the dakuten judging means 3 carries out steps 102 to 104, the basic segment dividing means 4 carries out step 105, and the dakuten / semi-dakuten. The combining means 5 executes steps 106 to 108, the kana separated character combining means 6 executes steps 109 to 111, and the basic segment output control means 7 executes step 112. FIG. 3 is a diagram showing an example of a condition for two consecutive basic segments being a kana separated character. The operation of the apparatus of this embodiment will be described below for each means.

First, the coordinate input device 1 is for inputting characters, and is realized by, for example, a tablet, a mouse, a light pen, a touch panel, or the like. When a character is input, the coordinate input device 1 extracts a data string {(xi, yi), i = 1,2, ..., nj} j and sends it to the stroke encoding means 2. The stroke coding means 2 extracts strokes from the data string and assigns each stroke to several kinds of codes prepared in advance according to the shape thereof based on the characteristic point information for each stroke. The characteristic points are extracted by, for example, using the change points of the signs in the x and y directions of the input data string as the characteristic points. The coding is performed, for example, by classifying the X and Y signs of each segment, the angle of the segment, and the rotation angle between the segments. This coded data is used by the dakuten judging means 3, the dakuten / semi-dakuten combining means 5, and the kana separated character combining process 6.

When the two consecutive strokes are both "small strokes on the upper side", the dakuten judging means 3 combines the two strokes into one segment. FIG. 4 is a diagram for explaining the dakuten determination. The condition of "a small stroke in the upper part" is that the horizontal width of the stroke is Xh and the vertical width of the stroke is
Yh, stroke height is Height, and line height is Lineheig
If ht, α, β are constants of 1 or less, Xh <α × Lineheight (1) Yh <α × Lineheight (2) Height> β × Lineheight (3) All conditions of the expressions (1) to (3) This is the case when it is satisfied.
α and β are set appropriately depending on the writer and the writing conditions.

The basic segment dividing means 4 projects the segments output from each stroke and the dakuten judging means 3 on the X-axis, and makes the overlapping strokes and segments one basic segment to the dakuten / semi-dakuten combining means 5. Output. FIG. 5 shows basic segments divided by the basic segment dividing means 4.

Next, a method of performing segment division will be described in detail. First, the maximum value and the minimum value in the X direction are obtained for each segment output by the stroke and cloud point determination means 3. The respective values are XMAXi, XMINi (i indicates the order in which they are written; i = 1 to SMAX: SMAX is the number of strokes that are not combined by the dakuten judging means 3 and the number of segments output by the dakuten judging means 3. Is the sum). First, it is assumed that the segments output by the stroke and dull point determination means 3 are each one segment. When processing the kth segment, the minimum value in the X direction of that segment (XMINk)
Does not fall below the maximum value (XMAXi; i <k) in the X direction of the previous segment (which reflects the processing results up to that point) (XMINk <= XMAXi). i is changed in order from 1, and the one that satisfies the condition first is adopted. if,
When the condition is satisfied, the i-th segment to the k-th segment are regarded as one segment. The above operation is k 1
It is possible to divide the basic segment by changing from SMAX to SMAX and processing sequentially. The basic segment generated by the division is output to the dakuten / semi-dakuten combining means 5.

The dakuten / semi-dakuten combining means 5 joins the immediately preceding basic segment when the basic segment is composed of only the dakuten and the semi-dakuten. The condition that the basic segment is an independent dakuten is that there are only two strokes in the basic segment and that the dakuten judgment processing 3 determines that the stroke is a dakuten. The condition that the basic segment is an independent semi-voiced point is that the number of strokes in the basic segment is 1
, The stroke code of the included stroke must be a code indicating a circle.

The kana separated character combining means 6 determines whether or not two consecutive basic segments satisfy the condition for being a kana separated character. The conditions for being a kana separator are:
For example, as shown in FIG. 3, it is described by the stroke number of each basic segment, the stroke code of the included stroke, and the like. However, the stroke number and stroke code of the basic stroke do not include dakuten and semi-dakuten. As shown in FIG. 6, when "ji" is one basic segment, the number of strokes is "1" and the stroke code is "shi". When all the conditions for the two basic segments to be kana separation characters are satisfied, the two basic segments are combined.

When the basic segment is determined as one character (voiced character, semi-voiced character, kana separated character), the basic segment output control means 7 skips the cutting means and outputs the basic segment to the output terminal 9. Output. When the basic segments are unprocessed, the continuous unprocessed basic segments are collected into one and output to the character cutting means 8.

The character cutting-out means 8 performs character cutting-out by combining the basic segment groups sent from the basic segment output control means 7. For character cutout,
For example, two basic segments are combined only when the width when two consecutive basic segments are combined is equal to or less than a certain threshold value, and when the condition is not satisfied, one basic segment is determined as one character. . The above operation is performed for all input basic segments. Further, in the present invention, the above method is used, but the character cutting means 6 can use any existing character cutting. For example, it is possible to use various means such as a method of detecting the pen-up time to perform clipping and a method of using character recognition.

[0016]

As described above in detail, according to the present invention, by providing the dakuten judging means, the dakuten / semi-dakuten combining means, and the kana separated character combining means as described above, Characters that appear frequently and are easy to separate during writing,
It is possible to cut out dakuten characters, semi-dakuten characters, and kana separated characters.

[Brief description of drawings]

FIG. 1 is a diagram showing a configuration of an example.

FIG. 2 is a flowchart showing the operation of the embodiment.

FIG. 3 is a diagram showing conditions for being a kana separated character in a kana separated character combining means.

FIG. 4 is a diagram for explaining a dakuten determination performed by a dakuten determination means.

FIG. 5 is a diagram showing basic segments divided by a basic segment dividing unit.

FIG. 6 is a diagram showing how to determine a stroke number and a stroke code in a kana separated character combination process.

[Explanation of sign]

1 ... Coordinate input device, 2 ... Stroke coding means,
3 ... dakuten determination means, 4 ... basic segment dividing means, 5
... voiced sound / semi-voiced sound combining means, 6 ... kana separated character combining means, 7 ... basic segment output control means, 8 ... character cutting means.

Claims (3)

[Claims] 1. A coordinate input device that outputs position information of an input character when writing a character as coordinate data, and based on the coordinate data output from the coordinate input device,
Stroke coding means for sequentially extracting and coding strokes as input characters are written, and determining whether two consecutive strokes are "voiced points". If "voiced points", the two strokes are voiced points. And a basic segment dividing means for detecting the overlap between each stroke and the segments output by the dakuten determination means, and outputting the overlapping ones as one segment, When the basic segment is composed of only dakuten or semi-dakuten, the dakuten that is combined with the immediately preceding basic segment
An online character segmentation device comprising a semi-voiced point combining means and a kana separated character combining means for connecting two basic segments when two consecutive basic segments are kana separated characters. 2. The cloud point determination means sets the conditions for the cloud point determination as horizontal width of stroke, vertical width of stroke,
The online character cutting device according to claim 1, wherein the stroke is determined from the height of the stroke and the height of one line. 3. The kana separated character combining means determines the condition for being a kana separated character from the stroke code and stroke number of the stroke excluding the dakuten and semi-dakuten in the basic segment. Item 1. The online character cutting device according to item 1.