JPH0714007A - Method and device for online handwritten character recognition - Google Patents

Abstract

PURPOSE:To correctly recognize a handwritten character even in the case of the erroneous oder of making strokes in writing a Chinese character or input like stroke addition. CONSTITUTION:Stroke data of a handwritten character is inputted by an input part 1, and the inputted stroke data string is transposed based on position information by a stroke arranging part 2, and matching with a recognition dictionary is performed based on received stroke data in a recognition part 3 to recognize the character. Preferably, next stroke data is taken out if the character cannot be recognized with only stroke data taken out before at the time of character recognition, thereby eliminating a need of especially segmenting the character.

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 method and apparatus.

[0002]

2. Description of the Related Art In a conventional online handwritten character recognition apparatus, an input person sequentially inputs stroke data (point sequence data of input coordinates starting with pen down and ending with pen up) from an input unit composed of a tablet and a pen. Then, the recognition unit receives the stroke data from the input unit, matches information such as stroke coordinates and stroke order with the recognition dictionary,
The input stroke sequence is recognized as a character. A technique for recognizing characters from this stroke data by matching with a recognition dictionary is, for example, Japanese Patent Publication No. 5-2.
Since this is a well-known technique as disclosed in Japanese Patent No. 2949, detailed description thereof will be omitted.

A conventional online handwritten character recognition apparatus will be described below with reference to the drawings. First, 1
A conventional online handwritten character recognition device of the type in which characters are input character by character will be described with reference to FIGS. 6 and 7. In FIG. 6, reference numeral 21 is an input section composed of a pen and a tablet, and 22 is a recognition section for input handwritten characters. FIG. 7 shows an example of stroke data of a character input. s1 to s5 indicate strokes in the order of input.

Next, the operation will be described. Input person is input section 2
Use the pen to write characters on the tablet of 1. As the pen moves over the tablet, its coordinates are detected by the tablet. The written characters are delivered to the recognition unit 22 in stroke units. When the input person finishes writing one character, the recognition unit 22 recognizes based on the input stroke data for one character. In the recognition unit 22,
Based on the received stroke data for one character, matching is performed with the recognition dictionary to recognize the character as a character.

To determine the end of writing one character, a method of explicitly instructing the input person to start recognition and notifying that the input of characters has been completed, and providing a plurality of input boxes to enter another input box There are known a method of deciding that the input of the corresponding character is completed when the character is input, and a method of deciding that the input of the character is completed when there is no input for a predetermined time. However, depending on the input person, the input may be made in the wrong stroke order, so in the conventional online handwritten character recognition, a plurality of dictionaries are prepared to deal with this. For example, in FIG.
When writing the character "right" of s1, s1 → s2 → s3 → s4
→ It is correct to write in s5 stroke order, but s2 → s1 → s3 →
In many cases, the stroke order is written from s4 to s5. In such a case, these two stroke order patterns are prepared in a dictionary to support this.

Next, a conventional on-line handwritten character recognition device of the type in which a plurality of characters are input at one time and the characters are cut out and then recognized will be described with reference to FIGS. 8 and 9. In FIG. 8, 31 is an input unit, 32 is a cutout unit,
33 is a recognition unit. FIG. 9 shows the input characters and the cutout state. Next, the operation will be described. The input person writes a character on the tablet of the input unit 31 using a pen. As the pen moves over the tablet, its coordinates are detected by the tablet. The input person can write multiple characters at once. After inputting the characters for a plurality of characters, the input person gives an instruction to cut out and execute recognition.

The cutout unit 32 cuts out the stroke data for several characters received from the input unit 31 in character units. For example, as shown in FIG. 9, “patent application”
The case where is input will be described. Enterer is "special"
Enter characters by handwriting in the order of "Hu", "Exit", "Wish". All "special" strokes are within the range indicated by a1, and all "allow" strokes are within the range indicated by a2. As a result, the cutout unit 32 cuts out the character between the last stroke of “special” and the first stroke of “allow”. The cut-out strokes are passed to the recognition unit 33 on a character-by-character basis, and the recognition unit 33 performs matching with a recognition dictionary based on the received series of stroke data and recognizes them as characters.

[0008]

However, in the above-described conventional online handwritten character recognition device, input is made such that the stroke order of the characters is incorrect, or a point or stroke that has been forgotten after writing the character is added. I could not recognize it correctly. Further, in the case of dealing with stroke order variation by having a plurality of contents of the recognition dictionary, there is a problem that the dictionary is increased in order to deal with many stroke order variations.

Further, when a plurality of characters are input at one time, the stroke order variation as described above spreads over a plurality of characters, so that there is a problem that the characters cannot be cut out properly. The present invention, in view of the above conventional problems,
It is an object of the present invention to provide an online handwritten character recognition method and an apparatus thereof that can be correctly recognized even when an input such as an incorrect stroke order or a stroke is added.

[0010]

In order to achieve the above object, an online handwritten character recognition method according to the invention of claim 1 recognizes characters by extracting input stroke data in an order corresponding to a stroke position. It is characterized by According to the invention of claim 2, in the invention of claim 1, the next stroke data is extracted when the character cannot be recognized only by the previously extracted stroke data in character recognition.

The online handwritten character recognition apparatus according to the third aspect of the present invention includes a handwritten character input unit, a stroke alignment unit for extracting the input stroke data in the order corresponding to the position of the stroke, and the extracted stroke alignment unit. And a recognition unit for recognizing characters from the stroke data. According to the invention of claim 4, in the invention of claim 3, means for requesting the next stroke data to the stroke aligning portion when the recognition portion cannot recognize as a character only by the stroke data extracted previously. It is characterized by having.

[0012]

According to the invention of claim 1, since the stroke data is extracted in the order corresponding to the position and the character is recognized,
It is possible to recognize handwritten characters without being affected by the stroke order variation regardless of the stroke error of the input person or the correction of the character by adding strokes, and it is not necessary to add a recognition dictionary for dealing with the stroke order variation.

According to the second aspect of the invention, even when a plurality of characters are input at one time, the character recognition is performed by the stroke data sequentially extracted according to the stroke position, and the character recognition cannot be performed. By requesting the next stroke, the character will be cut out when it is recognized as a character, and it is not necessary to prepare a special character cutting method or a character cutting device and to recognize it by dividing it into character units in advance. It is possible to prevent erroneous recognition due to redundancy in the number of strokes, such as erroneous recognition when a plurality of characters are input in proximity to each other in an attempt to recognize the plurality of characters as one character.

Further, according to the inventions of claims 3 and 4, the same operations as those of the inventions of claim 1 and claim 2 are achieved.

[0015]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to FIGS. In FIG. 1 showing the configuration of an online handwritten character recognition apparatus according to an embodiment of the present invention, 1 is an input unit, 2 is a stroke alignment unit, and 3 is a recognition unit.
FIG. 2 shows an example of stroke data of characters input.

Next, the operation will be described in detail. The input unit 1 inputs stroke data of handwritten characters. There are a method of inputting stroke data directly handwritten using a pen and a tablet, a method of inputting stroke data which is handwritten in advance and filed, and the like. The input stroke data string for one character is passed to the stroke alignment unit 2.

The stroke arranging unit 2 rearranges the received stroke data sequence based on its position information.
The position information means, for example, coordinates of end points of each stroke, coordinates of a circumscribed rectangle of the stroke, or distance of the stroke from specific coordinates. In this case, the abscissa of the end points of each stroke, that is, the one with the smaller X coordinate is used, and the rearrangement is performed by arranging the X coordinate with the smallest X coordinate, that is, the leftmost one.

For example, the case where the character "must" as shown in FIG. 2 is passed to the stroke alignment unit 2 in the order of s1 → s2 → s3 → s4 → s5 will be described. The stroke aligning unit 2 checks the X coordinate of the end point of each received stroke data. Of the two end points, the end point having the smaller X coordinate is used as the position information of the stroke. When the strokes are rearranged from the smallest X coordinate, s2 → s4 → s3 → s1 → s5. As a result, the stroke alignment unit 2 stores the stroke data in s.
The sequence is passed to the recognition unit 3 in the order of 2 → s4 → s3 → s1 → s5.

The recognition unit 3 performs matching with a recognition dictionary based on the received stroke data and recognizes this as a character. The order of the strokes received, that is, the stroke order, does not depend on the input person, but the stroke alignment unit 2
It is necessary to prepare a recognition dictionary corresponding to the position information because it is rearranged according to the position information, but it is not necessary to prepare a plurality of dictionaries in order to cope with stroke order variation and the like.

Next, another embodiment of the present invention will be described with reference to FIGS.
This will be described with reference to FIG. In FIG. 3, which shows the configuration of the online handwritten character recognition apparatus in this embodiment, 11
Is an input unit, 12 is a stroke alignment unit, and 13 is a recognition unit. FIG. 4 shows a flowchart of the operation of the recognition unit 13. 5 shows an example of stroke data of a character string input.

Next, the operation will be described in detail. The input unit 11 inputs stroke data of handwritten characters. There are a method of inputting stroke data directly handwritten using a pen and a tablet, a method of inputting stroke data which is handwritten in advance and filed, and the like. The input stroke data sequence is passed to the stroke alignment unit 12.

The stroke arranging unit 12 rearranges the received stroke data sequence based on its position information. The position information is, for example, the coordinates of the end points of each stroke,
Alternatively, it refers to the coordinates of the circumscribed rectangle of the stroke, the distance of the stroke from specific coordinates, and the like. In this case, the abscissa of the end points of each stroke, that is, the one with the smaller X coordinate is used, and the rearrangement is performed by arranging the X coordinate with the smallest X coordinate, that is, the leftmost one. The rearranged stroke data is recognized by the stroke recognition unit 1
Passed to 3.

In the recognition unit 13, the stroke alignment unit 12
The stroke data is received from the stroke data one by one (step # 1), and the character is recognized from the stroke data (step # 2). For example, at the time point when the stroke s2 is received in FIG. 5, the recognition unit 13 cannot recognize even if the recognition is performed because there is no valid candidate character. At this time, the recognition unit 13 requests the next stroke data s3 from the stroke alignment unit 12 (returns from step # 3 to step # 1). In response to this, the stroke alignment unit 12 passes the next stroke data s3 to the recognition unit 13. The recognition unit 13 recognizes the character "mu" from the stroke data s1, s2, s3. At this point, the stroke data s4
Indicates that the stroke is not a stroke forming this character "mu", and the stroke data s4 is recognized as a stroke forming another character. That is, the character "mu" is cut out and recognized. Therefore, even if "mu" and "shi" are input in proximity to each other, it is possible to prevent erroneous recognition due to redundancy of the number of strokes such as erroneous recognition by making them one character.

The present invention is not limited to the above embodiments, and various modifications can be made based on the spirit of the present invention, and these modifications are not excluded from the scope of the present invention.

[0025]

As described above, according to the first aspect of the invention, the stroke data is taken out in the order corresponding to the position and the character is recognized, so that the correction of the character by the stroke error of the input person or the addition of the stroke can be performed. Nevertheless, handwriting recognition is possible without being affected by these stroke order variations, and it is not necessary to add a recognition dictionary for dealing with stroke order variations.

Further, according to the second aspect of the invention, even when a plurality of characters are input at one time, the character recognition is performed by the stroke data sequentially extracted according to the stroke position, and the character recognition cannot be performed. By requesting the next stroke, the character will be cut out when it is recognized as a character, and it is not necessary to prepare a special character cutting method or a character cutting device and to recognize it by dividing it into character units in advance. It is possible to prevent erroneous recognition due to redundancy in the number of strokes, such as erroneous recognition when a plurality of characters are input in proximity to each other in an attempt to recognize the plurality of characters as one character.

Further, according to the inventions of claims 3 and 4, the same effects as those of the inventions of claims 1 and 2 can be obtained.

[Brief description of drawings]

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

FIG. 2 is an explanatory diagram of stroke data of a character input in the same embodiment.

FIG. 3 is a block diagram showing a configuration of an online handwritten character recognition device according to another embodiment of the present invention.

FIG. 4 is an operation flowchart of a recognition unit in the embodiment.

FIG. 5 is an explanatory diagram of stroke data of a character string input in the same example.

FIG. 6 is a block diagram showing a configuration of a conventional online handwritten character recognition device.

FIG. 7 is an explanatory diagram of stroke data of a character input in the conventional example.

FIG. 8 is a block diagram showing the configuration of an online handwritten character recognition device in another conventional example.

FIG. 9 is an explanatory diagram showing an input character string and a cutout state in the conventional example.

[Explanation of symbols]

 1 Input Unit 2 Stroke Alignment Unit 3 Recognition Unit 11 Input Unit 12 Stroke Alignment Unit 13 Recognition Unit

Claims (4)

[Claims] 1. An online handwritten character recognition method characterized in that input stroke data is extracted in the order corresponding to the position of the stroke and character recognition is performed. 2. The on-line handwritten character recognition method according to claim 1, wherein the next stroke data is extracted when the character cannot be recognized only by the previously extracted stroke data in character recognition. 3. A handwriting character input section, a stroke arranging section for extracting input stroke data in an order corresponding to a stroke position, and a recognition section for recognizing characters from the extracted stroke data. An on-line handwritten character recognition device. 4. The recognizing unit is provided with means for requesting the next stroke data to the stroke arranging unit when the previously extracted stroke data alone cannot be recognized as a character. Item 3. The online handwritten character recognition device according to Item 3.