JP2002109471A - Device for processing input character - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a character recognition device which can accurately recognize sloppy handwriting character. SOLUTION: A type of stroke which is input in a tablet for handwriting input character is judged (S3). When it is a straight stroke (S4: YES), an intersection between the strokes is detected (S5). When the strokes are intersected (S6: YES), an angle of the intersection or inside product is computed, compared to the prescribed threshold, and judged whether it is overlap stoke or not (S8). When it is an overlap stroke (S8: YES), overlapped stokes are divided (S9), and offline character recognition is processed (S10).

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a character recognition device that recognizes input handwritten characters and converts them into character codes.

[0002]

2. Description of the Related Art Conventionally, methods for recognizing handwritten characters include:
One line (hereinafter, referred to as “stroke”), which is a constituent line of an input character, is used as a recognition unit, and online recognition that performs recognition using time-series information in the input order of strokes, and an input character Off-line recognition is known in which recognition is performed using information in the entire form. Online recognition is
Although the recognition process is performed when a stroke is input, the handwriting character recognition process is fast.However, the drawback is that it is vulnerable to variations in the stroke order and strokes, such as incorrect stroke order, continuation characters, and overwriting. is there. On the other hand, the off-line recognition has a feature that since the recognition is performed using information on the shape of the entire input character, there is little erroneous recognition due to an error in the stroke order or overwriting.

[0003]

However, off-line recognition has a drawback that it is brittle with respect to the overall balance of input characters and the displacement of stroke positions. In particular, in the case where strokes that originally do not intersect with each other are written even when the strokes are misaligned, the offline method has a problem that recognition is difficult.

SUMMARY OF THE INVENTION The present invention has been made to solve the above-described problem. In performing handwritten character recognition by an off-line method, even if strokes that originally do not intersect and are written by overlapping, the strokes are not changed. It is an object of the present invention to provide a character recognition device capable of detecting and separating and correctly recognizing input characters including overlapping strokes.

[0005]

According to a first aspect of the present invention, there is provided an input character processing apparatus for inputting characters by handwriting, and inputting from the handwritten character inputting means. Stroke detection means for detecting each stroke of the handwritten character in an online manner; intersection stroke detection means for detecting the state of intersection between the two strokes detected by the stroke detection means; and Intersecting stroke overlapping determining means for determining whether the two strokes should originally intersect or be separated based on the state of intersection between the two strokes, and the intersecting stroke overlapping determining means intersect. When it is determined that the two strokes overlap, the two intersecting strokes It has a structure which is characterized in that a cross-stroke separating means for separating.

In the input character processing apparatus having this configuration, the stroke detecting means detects each stroke of the handwritten character input from the handwritten character input means in an online manner, and the intersection stroke detecting means detects the two strokes detected by the stroke detecting means. The intersection stroke detection unit detects an intersection state between the two strokes based on the intersection state between the two strokes detected by the intersection stroke detection unit, and determines whether the two strokes should originally intersect. The intersecting stroke separation means separates the two intersecting strokes when the intersecting stroke overlap determining means determines that the two intersecting strokes overlap. .

According to a second aspect of the present invention, in addition to the configuration of the first aspect, it is determined whether or not the stroke detected by the stroke detecting means is a straight line. A stroke type determination unit; and an intersection angle measurement unit that measures an intersection angle between the two intersecting strokes. The intersection stroke detection unit determines that the stroke type determination unit determines that the stroke is a straight line. When an intersection between two strokes is detected and the intersecting stroke detecting means detects the presence or absence of an intersecting between the two strokes, the intersecting stroke overlap judging means detects two intersections based on the detection result of the intersecting angle measuring means. The configuration is characterized in that it is determined whether or not the stroke is an overlap in which what should be separated originally overlaps.

[0008] In the input character processing device having this configuration, in addition to the operation of the input character processing device according to the first aspect, the stroke type determining means determines whether the stroke detected by the stroke detecting means is a straight line, and determines whether the stroke is a straight line. The angle measuring means measures the intersection angle between the two intersecting strokes, and the intersection stroke detecting means detects the intersection between the two strokes determined by the stroke type determining means to be a straight line, and When the detecting means detects the presence or absence of the intersection between the two strokes, the intersection stroke overlap judging means judges that the two strokes are overlaps which should be separated based on the detection result of the intersection angle measuring means. It is determined whether or not.

According to a third aspect of the present invention, in addition to the configuration of the input character processing apparatus of the first aspect, it is determined whether or not the stroke detected by the stroke detecting means is a straight line. A stroke type determining unit, and an inner product calculating unit that calculates an inner product of vectors of two intersecting strokes, wherein the intersecting stroke detecting unit determines that the stroke type determining unit determines that the stroke is a straight line. When the presence or absence of the intersection between the strokes is detected, and the intersection stroke detection unit detects the intersection between the two strokes, the intersection stroke overlap determination unit determines whether the two strokes are based on the detection result of the inner product calculation unit. Originally, it is characterized in that it is determined whether or not overlaps are to be separated.

[0010] In the input character processing device having this configuration, in addition to the operation of the input character processing device according to the first aspect, the cross stroke detecting means includes two strokes whose stroke type determining means determines that the stroke is a straight line. When the presence or absence of an intersection between the two strokes is detected and the intersection stroke detection means detects the intersection between the two strokes, the intersection stroke overlap determination means determines whether the two strokes should be separated based on the detection result of the inner product calculation means. It is determined whether the kimono is an overlapping part or not.

According to a fourth aspect of the present invention, in addition to the configuration of the first aspect, the stroke amplifying means for increasing the width of the stroke detected by the stroke detecting means. A stroke area measuring means for measuring an area of a stroke in which the stroke amplifying means widens a stroke, and an inter-stroke overlapping area for measuring an overlapping area between two strokes in which the stroke amplifying means widens the stroke. Measuring means, comprising overlapping area occupancy calculating means for calculating the ratio of any of the two stroke areas measured by the stroke area measuring means and the overlapping area measured by the inter-stroke overlapping area measuring means, The intersecting stroke overlap judging means determines two strokes based on the measurement result of the overlapping area occupancy calculating means. Over click originally has a structure which is characterized by determining whether a duplicate overlapping those to be separated.

[0012] In the input character processing device having this configuration, in addition to the operation of the input character processing device according to the first aspect, the stroke amplifying means widens the width of the stroke detected by the stroke detecting means, and the stroke area measuring means includes: The stroke amplifying means measures the area of the stroke whose stroke width is widened, and the inter-stroke overlapping area measuring means measures the overlapping area between the two strokes whose stroke amplifying means widens the stroke, and calculates the overlapping area occupancy. The means calculates a ratio of one of the two stroke areas measured by the stroke area measuring means to the overlapping area measured by the inter-stroke overlapping area measuring means, and the intersection stroke overlapping determining means calculates the ratio of the overlapping area occupancy calculating means. Based on the results, it is determined whether or not the two strokes are overlaps that should be separated. To.

According to a fifth aspect of the present invention, in addition to the configuration of the input character processing apparatus of the first aspect, in addition to the configuration of the input character processing apparatus, a character enlargement for enlarging a handwritten character input from the handwritten character input means. Means for detecting whether there is an intersection between two strokes based on the character enlarged by the character enlarging means.

[0014] In the input character processing device having this configuration, in addition to the operation of the input character processing device according to the first aspect, the intersection stroke detecting means detects the distance between two strokes based on the character enlarged by the character enlarging means. Detect the presence or absence of intersection.

According to a sixth aspect of the present invention, in addition to the configuration of the input character processing device according to any one of the first to fifth aspects, the stroke and the intersection detected by the stroke detecting means are provided. An off-line character recognizing means for recognizing a handwritten character using the whole shape of the character based on the stroke separated by the stroke separating means and converting it into a character code is provided.

In the input character processing device having this configuration, in addition to the operation of the input character processing device according to any one of claims 1 to 5, the off-line character recognizing means can separate the stroke and the cross stroke detected by the stroke detecting means. The means recognizes the handwritten character using the entire shape of the character based on the separated stroke and converts it into a character code.

According to a seventh aspect of the present invention, in addition to the configuration of the input character processing apparatus according to any one of the first to sixth aspects, the stroke and the intersection detected by the stroke detecting means are provided. The stroke separation means includes a display means for displaying the entire shape of the character based on the separated stroke.

In the input character processing device having this configuration, in addition to the operation of the input character processing device according to any one of claims 1 to 6, the display means may include a stroke detected by the stroke detection means and a cross stroke separation means. Displays the overall shape of the character based on the stroke.

According to an eighth aspect of the present invention, in addition to the configuration of the input character processing device according to any one of the first to seventh aspects, the intersection stroke detecting means intersects. The configuration is characterized by detecting a state of intersection of two strokes.

In the input character processing device having this configuration, in addition to the operation of the input character processing device according to any one of claims 1 to 7, the intersecting stroke detecting means may detect the intersection of two intersecting strokes. Is detected.

[0021]

DESCRIPTION OF THE PREFERRED EMBODIMENTS One embodiment of the present invention will be described below with reference to the drawings.

FIG. 1 is a block diagram showing a schematic configuration of the character recognition device 1, and FIG. 2 is a plan view of a tablet 4 for inputting handwritten characters.

As shown in FIG. 1, a character recognition apparatus 1 according to a first embodiment of the present invention includes a RAM 2 for temporarily storing data, a RAM 8 for temporarily storing data,
ROM 7 storing programs executed by PU 2
And an I / O interface 3 that mediates input / output of data to / from the CPU 2. The I / O interface 3 is connected to a handwritten character input tablet 4 for inputting characters handwritten by the pen 5 and a display device 6 including a liquid crystal display device.

Next, the storage area of the ROM 7 will be described. As shown in FIG. 1, the ROM 7 has an off-line recognition program storage area 7a, an off-line recognition dictionary storage area 7b, and a character extraction program storage area 7c.
A stroke recognition program storage area 7d, a basic stroke dictionary storage area 7e, an intersection stroke detection program storage area 7f, an overlap determination value calculation program storage area 7g, an overlap determination threshold storage area 7h,
A cross stroke separation program storage area 7i is provided.

In the off-line recognition program storage area 7a, there is stored an off-line recognition program for causing the CPU 2 to execute an off-line recognition process for recognizing using the information on the entire shape of one character input to the handwritten character input tablet 4. It is remembered. Further, an off-line recognition dictionary storage area 7b stores an off-line recognition dictionary storing feature data in the form of an entire character referred to in the off-line recognition processing. The character extraction program storage area 7c stores a character extraction program that causes the CPU 2 to execute processing of identifying one character from the data input from the handwritten character input tablet 4 and extracting the data as one character data.

Further, a stroke recognition program for causing the CPU 2 to execute a stroke type determination process for determining the type of stroke such as whether or not the stroke input from the handwriting character input tablet 4 is a straight line is stored in the stroke recognition program storage area 7d. Is stored. Also,
The basic stroke dictionary storage area 7e stores data of the direction sequence of the basic stroke referred to in the stroke type determination processing. Further, the cross stroke detection program storage area 7f stores a cross stroke detection program for causing the CPU 2 to execute a cross stroke detection process for detecting a crossing state between strokes.
Also, the duplication determination value calculation program storage area 7g has
An overlap determination value calculation program for causing the CPU 2 to execute an overlap determination value calculation process of calculating an intersection angle between strokes or an inner product, which is an overlap determination value for determining overlap between strokes, is stored.

Further, a threshold for determining the overlap between strokes referred to in the overlap determination value calculation processing is stored in the overlap determination threshold storage area 7h. The cross stroke separation program storage area 7i stores a cross stroke separation program that causes the CPU 2 to execute a process of separating intersecting strokes determined to have overlapping strokes in the overlap determination value calculation process. I have.

Next, the storage area of the RAM 8 will be described. As shown in FIG. 1, an input stroke data storage area 8 a for storing strokes input to the handwritten character input tablet 4, and a result of offline recognition of characters input to the handwritten character input tablet 4 are stored in the RAM 8. Recognition result storage area 8b for storing character codes
And an input stroke type information storage area 8c for storing the result of the stroke type determination processing, an intersection stroke presence / absence information storage area 8d for storing the result of the intersection stroke detection processing, and a stroke determined by the overlap determination value calculation processing. And a working area 8f for storing data in various processes performed by the CPU 2.

Next, the structure of the handwritten character input tablet 4 will be described with reference to FIG. As shown in FIG. 2, the handwritten character input tablet 4 is formed in a horizontally long rectangle, and a first square formed for inputting handwritten characters is formed from a substantially center to a lower part of the surface of the handwritten character input tablet 4. An input area 4a, a second input area 4b,
The third input area 4c is provided adjacent to one horizontal row. When characters are written in any of the first input area 4a to the third input area 4c with the pen 5, the input stroke data is transferred to the CP via the I / O interface 3.
It is sent to U2.

The inside of the first input area 4a to the third input area 4c of the handwritten character input tablet 4 is expressed by the following expression 1 in parallel with the surface of the handwritten character input tablet 4.
Two conductive films having a resistance of 00Ω are stretched with a slight gap therebetween. When the surface of the handwriting character input tablet 4 is pressed with the tip of the pen 5, the two conductive films come into contact with each other and the X coordinate and It has a well-known structure from which the Y coordinate can be read.

A display device 6 composed of a liquid crystal display (LCD) is provided above the handwritten character input tablet 4. In the display device 6, handwritten input is performed in the first input area 4a to the third input area 4c.
Characters recognized by the PU2 are displayed in the order of input. In addition, handwriting character input tablet 4
The first input area 4a to the third input area 4c of the
A liquid crystal tablet in which the input stroke is displayed by incorporating D may be used.

The operation of the character recognition device 1 configured as described above will be described with reference to FIGS. FIG. 3 is a flowchart of main control of character recognition processing performed by the character recognition device 1, FIG. 4 is a flowchart of stroke type determination processing, FIG. 5 is a flowchart of stroke intersection detection processing, and FIG. FIG. 7 is a schematic diagram of a handwritten character input, FIG. 8 is a schematic diagram of an intersection stroke, and FIG. 9 is a schematic diagram of a state where the starting points of the intersection stroke are aligned. FIG. 10 is a schematic view showing a state in which the intersecting stroke is separated, FIG. 11 is a view showing a state in which the recognized character is displayed, and FIGS. It is a conceptual diagram, FIG.
0) is an example of a basic stroke.

Hereinafter, a case where the user inputs the character "Old" using the pen 5 will be described with reference to the flowchart shown in FIG. First, as shown in FIG. 7, when the user inputs one of the strokes of the "old" character in the first input area 4a of the handwritten character input tablet 4 (S1), the character is input. It is determined whether the input of all strokes has been completed (S2). This judgment is made in ROM 7
According to the character extraction program stored in the character extraction program storage area 7c, it is determined whether or not the input of all strokes of one character is completed (S2). Specifically, it is determined whether or not the input of one character is completed, for example, when the time until the input of the next stroke is opened for a predetermined time (for example, “1 second” or “2 seconds”), or when the first stroke is input. When the input to the input area 4a starts inputting to the second input area 4b, or when the input to the second input area 4b starts inputting to the third input area 4c,
It is determined that the input of all strokes of one character has been completed (S
2: YES). If the input of all strokes of one character has not been completed (S2: NO), the process proceeds to S3.

In step S3, a stroke type determination process is performed. This stroke type determination process will be described with reference to a subroutine of the stroke type determination process shown in FIG. 4, a schematic diagram of the stroke type determination process shown in FIG. 12, and a conceptual diagram of a stroke vector in 16 directions shown in FIG. .

First, when a handwritten stroke a as shown in FIG. 12 (1) is inputted, as shown in FIG. It is divided into a direction sequence b, which is a 16-segment vector oriented in one of the directions (see FIG. 13) (S21). Next, the direction sequence of the input direction shown in FIG. 12 (3) is matched with the direction sequence of the basic stroke (see FIG. 12 (4)) corresponding to the basic stroke shown in FIG. 12 (5) (S22). The basic stroke having a small difference is determined as the stroke type of the handwritten stroke a (S22) and stored in the input stroke type information storage area 8c of the RAM 8 (S22).

The data of the direction sequence of the basic stroke is as follows.
It is stored in a basic stroke dictionary storage area 7e of the ROM 7. Here, an example of the type of the basic stroke is shown in FIG.
This will be described with reference to FIG. FIGS. 14 (1) to (10) are examples of basic strokes. (1) to (10) of FIG.
Among the basic strokes shown in (1), the strokes shown in (1) to (4) are linear strokes. Therefore, all the constituent lines such as the “old” stroke A and the stroke B of the handwritten character shown in FIG. 7 are determined to be straight strokes.

In the determination process of S4 in the flowchart of FIG. 3, the determination is made in the process of S22, and it is determined whether or not the type of the input stroke stored in the input stroke type information storage area 8c of the RAM 8 is a straight line (S4). . If the type of the input stroke is a straight line (S4: YES), an intersection detection process between strokes (S5) is performed. If the type of the input stroke is not a straight line (S4: NO), the process returns to S2.

In the process of detecting intersection between strokes (S5), the stroke determined to be a straight stroke in S4 is compared with each of the already input straight strokes to determine whether there is an intersection between the strokes. Or to detect. Specifically, a stroke intersection detection process shown in FIG. 5 is performed.

For example, the stroke B shown in FIG.
Attention is paid, and when it is determined that the stroke B is a straight stroke in the determination processing of S4, first, the coordinates of the middle point between the characteristic points of the stroke B are obtained (S31). Specifically, until the number of midpoints becomes one,
A process of connecting the midpoints of the direction sequence, which is a vector of 16 segments oriented in any of the six directions, and further obtaining the midpoint is performed (S32: NO, S33). When the number of midpoints becomes one (S32: YES), Y of the obtained midpoint is obtained.
The value of the coordinates is compared with the value of the Y coordinate of the upper end and the value of the Y coordinate of the lower end of the linear stroke A input before the stroke B (S34).

If the value of the Y coordinate of the middle point obtained in S33 is between the values of the Y coordinate at the upper end and the Y coordinate at the lower end of the straight stroke A (S35: YES), then
The value of the X coordinate of the middle point obtained in S33 is compared with the value of the X coordinate of the left end and the value of the X coordinate of the right end of the straight stroke A input before the stroke B (S36). S
If the value of the X coordinate of the midpoint obtained in 33 is present between the value of the X coordinate of the left end and the value of the X coordinate of the right end of the straight stroke A (S37: YES), the stroke B and the stroke A Can be determined to intersect (S38), and information indicating that there is an intersection between the stroke B and the stroke A is RA
It is stored in the information storage area 8d for the presence or absence of the crossing stroke of M8 (S38). If the determination in S35 is NO and the determination in S36 is NO, the information indicating that there is no intersection between the stroke B and the stroke A is stored in the RAM 8 in the presence / absence information of the intersection stroke. It is stored in the area 8d (S38). Thereafter, the process returns to S6 of the flowchart shown in FIG.

In the determination process of S6, if the existence of the intersection between the stroke B and the stroke A is stored in the intersection stroke presence / absence information storage area 8d of the RAM 8, FIG. As shown in FIG. 9, the stroke B or the stroke A is translated so that the starting points of the stroke B and the stroke A coincide with each other, as shown in FIG. θ or the inner product of the vector of stroke B (Bx, By) and the vector of stroke A (Ax, Ay) (inner product = AxBx + AyBy or inner product = | A || B | cos θ)
(S7). The intersection angle θ or the inner product becomes an overlap determination value and is stored in the overlap determination value storage area 8e of the RAM 8 (S7).

It is determined whether or not the duplication judgment value obtained in S7 is equal to or smaller than a predetermined threshold stored in the duplication judgment threshold storage area 7h of the ROM 7. For example, if the intersection angle θ is equal to or smaller than a predetermined angle (for example, 10 degrees) which is a threshold value, it means that the stroke A and the stroke B intersect at a very shallow angle. Therefore, the two strokes intersect at a shallow angle that is impossible with kanji or kana, and what should not originally intersect is determined to be an overlapped stroke that intersected due to random handwriting. (S8: YES). When the inner product of the stroke vectors is used as the overlap determination value, the threshold value differs depending on the length of the stroke. Therefore, the threshold value of the character input area (character frame) of the handwritten character input tablet 4 into which the stroke is input is set. It is stored in the overlap determination threshold storage area 7h of the ROM 7 according to the size.

In the determination process of S8, if it is determined that what should not intersect should be an intersecting overlapping stroke due to random handwriting (S8: YES),
The cross stroke separation process is performed (S9). Also, S6
If it is determined that there is no intersection (S6: NO),
If it is determined in S8 that the stroke is not an overlapping stroke (S8: NO), the process returns to S2.

In the cross stroke separation process (S9), as shown in FIG. 10, for example, the stroke A is separated from the stroke B by a parallel movement so as to be separated from the stroke B by a predetermined interval (at least one dot). The predetermined interval is determined in advance so that the two strokes are separated best by an interval in which erroneous recognition does not easily occur during off-line character recognition.
In the intersecting stroke separation process (S9), information on the separated stroke is stored in the input stroke data storage area 8a of the RAM 8, and the information on the intersecting stroke is corrected (S9).

Thereafter, the separation processing of the cross stroke is completed (S9), and the input of all the strokes is completed (S2:
YES), and proceeds to off-line character recognition processing (S10). The online character recognition processing will be described with reference to the flowchart of the online character recognition processing subroutine shown in FIG.

In the off-line character recognition processing, first, all strokes of one character stored in the input stroke data storage area 8a are read (S51), and the shape of the character is extracted (S52). And the offline recognition dictionary storage area 7 in the ROM 7
b until pattern matching with all registered characters registered in the offline recognition dictionary stored in
53: NO), information on the character shape stored in the offline recognition dictionary and the input stroke data storage area 8a
The character is identified by pattern matching (shape comparison) with the information on the shape of the character composed of the strokes stored in (1) (S54).

Then, the character candidates having a high degree of similarity in the character shape identified in the processing of S54 are stored in the recognition result storage area 8b of the RAM 8 (S55). Next, returning to the determination processing of S53, when pattern matching with all registered characters registered in the offline recognition dictionary stored in the offline recognition dictionary storage area 7e of the ROM 7 is completed (S53).
53: YES), and perform post-processing (S56). Post-processing (S
In 56), the grammatical connection of the character candidates stored in the recognition result storage area 8b of the RAM 8 is determined in the processing of S54, and the RAM 8 is determined in accordance with the grammatical connection. The order of the character candidates stored in the recognition result storage area 8b is changed (S56). Thereafter, the process returns to the processing of S11 shown in FIG. 3, and the post-processing (S56)
The first candidate of the selected character candidate is stored in the working area 8f of the RAM 8 functioning as a display buffer (S
11) Display on the display device 6 (S12).

As described above, in the character recognition apparatus 1 according to the first embodiment, when performing character recognition by the off-line method, even if strokes that originally do not intersect are written by overlapping, the characters are not overlapped. Offline character recognition is performed after strokes are detected and separated.
Input characters including overlapping strokes can be recognized correctly. As shown in the first input area 4a of FIG. 2, the character "Old" in which the stroke is duplicated and input is correctly recognized by the offline recognition and displayed on the display device 6 as "Old" as shown in FIG. A correct recognition result can be displayed.

Next, a second embodiment of the present invention will be described with reference to FIGS. In the second embodiment, the same members as those in the first embodiment are denoted by the same reference numerals and described. FIG. 15 is a block diagram illustrating a schematic configuration of the character recognition device 1 according to the second embodiment.
FIG. 17 is a flowchart of control of the character recognition device 1 of the second embodiment, and FIG. 17 is a conceptual diagram illustrating separation processing of overlapping strokes of the character recognition device 1 of the second embodiment.

As shown in FIG. 15, a character recognition device 1 according to a first embodiment of the present invention includes a RAM 8 for temporarily storing data in a CPU 2 for controlling the character recognition device 1;
ROM 7 storing programs executed by CPU 2
And an I / O interface 3 that mediates input / output of data to / from the CPU 2. The I / O interface 3 is connected to a handwritten character input tablet 4 for inputting characters handwritten by the pen 5 and a display device 6 including a liquid crystal display device.

Next, the storage area of the ROM 7 will be described. As shown in FIG. 15, the ROM 7 has an off-line recognition program storage area 70a, an off-line recognition dictionary storage area 70b, a character cutout program storage area 70c, and a stroke recognition program storage area 70a.
d, an expanded width value storage area 70e for stroke amplification,
An overlap area measurement program storage area 70f, a stroke area measurement program storage area 70g, an overlap area occupancy calculation program storage area 70h, an overlap area occupancy threshold storage area 70i, and an intersection stroke separation program storage area 70j are provided. ing.

In the off-line recognition program storage area 70a, there is stored an off-line recognition program for causing the CPU 2 to execute an off-line recognition process for performing recognition using information on the entire shape of one character input to the handwritten character input tablet 4. It is remembered. Further, the offline recognition dictionary storage area 70b stores an offline recognition dictionary that stores feature data in the form of a single character referred to in the offline recognition processing. The character extraction program storage area 70c stores a character extraction program that causes the CPU 2 to execute a process of identifying one character from the data input from the handwritten character input tablet 4 and extracting the data as one character data.

Further, in the stroke width amplification program storage area 70d, a process for amplifying the width of the stroke input from the handwriting character input tablet 4 is performed by the CPU 2.
Is stored. Also, the stroke width expansion width value storage area 70
In e, an enlarged width value indicating how much the stroke width is amplified in the process of amplifying the stroke width is stored. Further, the overlapping area measurement program storage area 70f
Stores an overlap area measurement program for causing the CPU 2 to execute an overlap area measurement process for measuring an overlap area of two strokes having amplified stroke widths.

In the stroke area measurement program storage area 70g, a stroke area measurement process for measuring the area of two strokes having amplified stroke widths is stored in C.
A stroke area measurement program to be executed by PU2 is stored. Further, an overlapping area occupancy calculation program storage area 70h stores an overlapping area occupancy calculating program for causing the CPU 2 to execute a process of calculating an overlapping area occupancy described later.

The overlapping area occupancy threshold storage area 70
In i, an overlap area occupancy threshold, which is a reference for determining overlap between two strokes whose overlap area occupancy has been calculated, is stored. Further, in the cross stroke separation program storage area 70j, the CPU 2 compares the overlap area occupancy with the overlap area occupancy threshold and separates the intersecting strokes determined to have overlapped strokes. A cross stroke separation program to be executed is stored.

Next, the storage area of the RAM 8 will be described. As shown in FIG. 15, the RAM 8 has an input stroke data storage area 80a for storing strokes input to the handwritten character input tablet 4, and a result of offline recognition of characters input to the handwritten character input tablet 4, as shown in FIG. Recognition result storage area 8 for storing character codes
0b, an overlap area measurement value storage area 80c for storing the result of the overlap area measurement processing, and a stroke area measurement value storage area 80d for storing the result of the stroke area measurement processing.
And an overlapping area occupancy calculation result storage area 80e for storing the overlapping area occupancy, and a working area 80f for storing data in various processes performed by the CPU 2.

Note that, also in the second embodiment, the structure of the handwritten character input tablet 4 has the same structure and function as the structure of the handwritten character input tablet 4 shown in FIG.

The operation of the character recognition apparatus 1 according to the second embodiment having the above-described configuration will be described with reference to FIGS.

First, as shown in FIG. 16, the user operates the first input area 4 of the tablet 4 for inputting handwritten characters.
When one of the strokes of the character is input to a (S61), it is determined whether or not the input of all the strokes of the character has been completed (S62). This determination is made according to the character extraction program stored in the character extraction program storage area 70c of the ROM 7 to determine whether or not the input of all strokes of one character has been completed (S62).

Specifically, it is determined whether or not the input of one character is completed, for example, when the time until the input of the next stroke is opened for a predetermined time (for example, “1 second” or “2 seconds”). For example, when an input to the first input area 4a starts inputting to the second input area 4b, an input to the second input area 4b starts inputting to the third input area 4c, etc. It is determined that the input of all strokes of one character is completed (S62: YES). If the input of all strokes of one character has not been completed (S62: NO),
The process proceeds to S63.

In the process of S63, a process of amplifying the stroke width is performed. FIG. 1 shows the stroke width amplification process.
This will be described with reference to a conceptual diagram of stroke width amplification shown in FIG. First, when a handwritten stroke of “x” having a large intersection angle shown in FIG. 17 (1) or “x” having a small intersection angle as shown in FIG. 17 (2) is input, respectively, FIG. 17 (3) ,
As shown in (4), the width of the stroke is expanded (amplified) by a predetermined value stored in advance in a stroke amplification expanded width value storage area 70e of the ROM 7 (S63). More specifically, the enlarged width value storage area 70e for stroke amplification
For example, when “8” is stored, the stroke is set to a width as shown in FIGS. 17 (3) and (4) by, for example, painting 8 dots around one dot of the stroke. It is expanded (amplified) (S63).

Next, it is determined whether or not there is a portion where the stroke whose stroke width has been enlarged overlaps another stroke whose stroke width has been enlarged (S64). In particular,
Attention is paid to one dot of the stroke in which the stroke width of interest is enlarged, and it is determined whether or not there is one dot coordinate data of another stroke overlapping with the (x, y) coordinate data of the one dot. If there is coordinate data of another stroke, it is determined that there is an overlapping portion (S64: Y
ES). This determination is made for all the coordinate data of the stroke in which the currently focused stroke width is enlarged (S6).
4).

When there is a portion where the stroke whose stroke width is enlarged overlaps another stroke whose stroke width is enlarged (S64: YES), first, the area of each stroke whose stroke width is enlarged is obtained (S64: YES). S6
5). For example, in the example shown in FIG. 17C, the areas of the strokes C and D whose stroke widths are enlarged are obtained. In the example shown in FIG. 17D, the areas of the strokes E and F whose stroke widths are enlarged are obtained (S
65). The obtained area is stored in the stroke area measured value storage area 80d of the RAM 8 (S65).

Next, the overlapping area between the strokes having the increased stroke width is determined (S66). Specifically, FIG.
In the example shown in FIG. 7 (3), the overlap area of the strokes C and D with the increased stroke width (the area of the hatched portion in the figure) is obtained. In the example shown in FIG. 17 (4), the area of the overlap area of the strokes E and F with the increased stroke width (the area of the hatched portion in the figure) is obtained (S66). The obtained overlapping area is stored in the overlapping area measured value storage area 80c of the RAM 8 (S66).

Next, the overlapping area occupancy is calculated (S6).
7). Specifically, it is obtained based on the following formula. “Overlapping area occupancy = overlapping area / stroke area”. In the case of FIG. 17C, the stroke area of the denominator is
The smaller area of the areas of the two strokes C or D obtained in S65 is used. Even in the case of FIG. 17D, the value of the area of the two strokes E or F obtained in S65 is used. Use the smaller area of. The overlapping area occupancy is larger between strokes having a smaller intersection angle than that having a larger intersection angle. The obtained overlapping area occupancy is
It is stored in the overlapping area measurement value storage area 80c (S6).
7).

Next, the overlapping area occupancy obtained in S67 is R
It is determined whether the threshold value is equal to or larger than the threshold value stored in the overlapping area occupancy threshold value storage area 70i of the OM 8 (S68). For example, when “0.5” is stored as the threshold in the overlapping area occupancy threshold storage area 70i, it is determined that the overlapping area occupancy is less than the threshold in the example illustrated in FIG.
68: NO), in the example shown in FIG. 17D, the overlapping area is large and the overlapping area occupancy is determined to be equal to or greater than the threshold (S6).
8: YES). If it is determined that the overlapping area occupancy is equal to or larger than the threshold value (S68: YES), it means that the stroke E and the stroke F intersect at a very shallow angle. Therefore, the two strokes intersect at a shallow angle that cannot be achieved with kanji or kana, and what should not intersect is the overlapping stroke that intersected due to random handwriting. A stroke separation process is performed (S69). Further, when it is determined in S64 that there is no overlap (S64: NO), or when it is determined in S68 that the overlapping area occupancy is less than the threshold (S68: N).
In O), the process returns to S62. Accordingly, as shown in FIG. 17 (5), the stroke C and the stroke D are not originally separated as intersecting strokes.

Next, separation processing of the intersection stroke (S6)
In 9), as shown in FIG. 17 (6), the original strokes E and F, whose stroke widths are not enlarged, are separated by parallel movement so as to be separated by a predetermined interval (at least one dot or more). The predetermined interval is determined in advance so that the two strokes are separated best by an interval in which erroneous recognition does not easily occur during off-line character recognition. In the intersecting stroke separation processing (S69), information on the separated stroke is stored in the input stroke data storage area 80a of the RAM 8, and the information on the intersecting stroke is corrected (S69).

Thereafter, the separation processing of the cross stroke is completed (S69), and the input of all the strokes is completed (S6).
2: YES), an off-line character recognition process is performed in the same manner as in the first embodiment (S70), and the first candidate of the off-line recognized character candidate functions as a display buffer.
Stored in the working area 80f of AM8 (S7
1) Display on the display device 6 (S72).

As described above, in the character recognition device 1 according to the second embodiment, when character recognition is performed by the off-line method, even if strokes that originally do not intersect are written by overlapping, the characters are not overlapped. Offline character recognition is performed after the stroke width of the stroke is amplified and the overlapping stroke is detected and separated using the overlapping area occupancy, so that input characters including the overlapping stroke can be correctly recognized.

Next, a third embodiment of the present invention will be described with reference to FIGS. FIG. 18 is a flowchart of control of the character recognition device according to the third embodiment, and FIG. 19 is a conceptual diagram illustrating a character enlargement process of the character recognition device of the third embodiment.

As shown in FIG. 18, the character recognition device according to the third embodiment of the present invention is characterized in that the input stroke is enlarged and the cross stroke is separated. For the detection processing and the separation processing of the intersection stroke, any of the methods of the first embodiment and the second embodiment described above may be employed.

A description will now be given, with reference to FIGS. 18 and 19, of a process for recognizing handwritten characters by the character recognition apparatus according to the third embodiment. In the character recognition device of the third embodiment,
As shown in FIG. 19A, when the input character "three" is input (S81), the character is enlarged as shown in FIG. 19B (S82). For example, if the character input frame is 10 dots × 10 dots, this enlargement is multiplied by 5
The image is enlarged to 0 dots × 50 dots (S82). The enlargement of each stroke is shown in FIG. 1 on the corresponding coordinates of the enlarged character frame.
Enlarge as shown in 9 (2). Next, based on the stroke expanded in S82, detection and separation processing of an overlapping cross stroke is performed (S83). The cross-slot talk separation process in S83 uses the method according to the first embodiment or the second embodiment.

When the process of detecting and separating overlapping intersection strokes (S83) is completed, the feature of each stroke is detected (S84), and then off-line character recognition is performed (S85). Post-processing is performed to check the grammatical connection relations of the results of the off-line character recognition and rearrange them (S86). The recognition result after the post-processing is stored (S8
7) After that, the first candidate of the character candidates recognized off-line is displayed on the display device 6 shown in FIG. 2 (S88), and the process ends.

As described above, in the character recognition apparatus according to the third embodiment, when performing the character recognition by the off-line method, first, the stroke of the input character is enlarged to detect the overlapped stroke. Offline character recognition is performed after separation, so that even if handwritten characters are input from a handwritten character input tablet with a small input frame, input characters including overlapping strokes can be correctly recognized.

It should be noted that the present invention is not limited to the above-described embodiment, and it goes without saying that the present invention can be implemented in various modes without departing from the gist of the present embodiment. For example,
In the example of the first embodiment, (1) to (4) in FIG. 14 are straight lines, but strokes including straight lines such as (5), (8), (9), and (10) are also The processing of the present proposal can be performed by regarding the straight line portion as one stroke. Further, the stroke area calculation processing (S65) and the overlap area calculation processing (S66) of the second embodiment are as follows.
Either one may be performed first, or parallel processing may be performed.

[0076]

As described in detail above, in the input character processing apparatus according to the first aspect of the present invention, the stroke detecting means detects each stroke of the handwritten character input from the handwritten character input means in an online manner, The stroke detecting means detects the state of the intersection between the two strokes detected by the stroke detecting means, and the intersection stroke overlapping determining means
Based on the state of intersection between the two strokes detected by the intersection stroke detection means, it is determined whether the two strokes should originally intersect or should be separated, and the intersection stroke overlap determination means When it is determined that the two intersecting strokes overlap, the intersecting stroke separating means can separate the two intersecting strokes. Therefore, it is possible to separate overlapping strokes in which strokes that do not originally intersect due to being input by handwriting overlap.

In the input character processing device according to the second aspect of the present invention, in addition to the effect of the input character processing device according to the first aspect, the stroke type determination means determines whether or not the stroke detected by the stroke detection means is a straight line. The intersection angle measurement means measures the intersection angle between the two intersecting strokes, and the intersection stroke detection means determines the intersection between the two strokes determined by the stroke type determination means as a straight line. When the intersection stroke detection means detects the presence or absence of intersection between the two strokes, the intersection stroke overlap determination means determines whether the two strokes should be separated based on the detection result of the intersection angle measurement means. It can be determined whether or not the overlapping strokes are overlapping.

In the input character processing apparatus according to the third aspect of the present invention, in addition to the effect of the input character processing apparatus according to the first aspect, in the cross stroke detection means, the stroke type determination means is a straight stroke. When the intersection stroke detection means detects the intersection between the two strokes, the intersection stroke overlap determination means detects two intersections based on the detection result of the inner product calculation means. It can be determined whether or not the stroke is an overlap in which what should be separated overlaps.
Therefore, the overlapped strokes can be determined by the inner product between the strokes.

In the input character processing device according to the fourth aspect of the present invention, in addition to the effect of the input character processing device according to the first aspect, the stroke amplifying means increases the width of the stroke detected by the stroke detecting means. The stroke area measuring means measures the area of the stroke in which the stroke amplifying means widens the stroke, and the inter-stroke overlapping area measuring means measures the overlapping area between the two strokes in which the stroke amplifying means widens the stroke width. The overlapping area occupancy calculating means calculates the ratio of one of the two stroke areas measured by the stroke area measuring means to the overlapping area measured by the inter-stroke overlapping area measuring means, and the intersection stroke overlapping determining means calculates the overlapping area. Based on the measurement result of the occupancy calculation means, the two strokes are overlapped because they should be separated It can be determined whether a stroke.

Further, in the input character processing device according to the fifth aspect of the present invention, in addition to the effect of the input character processing device according to the first aspect, the intersection stroke detecting means is based on the character enlarged by the character enlarging means. , The presence or absence of the intersection between the two strokes can be detected. Therefore, since the input character is enlarged even in the input character processing device having a small input frame for handwritten characters, it is possible to accurately determine whether or not two strokes are overlapped strokes in which what should be separated overlaps. it can.

Further, in the input character processing device according to the present invention, in addition to the effect of the input character processing device according to any one of the first to fifth aspects, the off-line character recognizing means includes a stroke detecting means for detecting the off-line character. The handwritten character can be recognized and converted into a character code using the entire shape of the character based on the separated stroke and the crossed stroke separating means. Therefore, since the off-line character recognition is performed after separating the overlapped strokes in which the strokes that do not intersect due to being input by handwriting overlap, the input characters having the overlapped strokes can be accurately recognized.

In the input character processing device according to the present invention, in addition to the effects of the input character processing device according to any one of the first to sixth aspects, the display means may include a stroke detected by the stroke detection means. The entire shape of the character based on the stroke separated by the intersecting stroke separating means can be displayed.

In the input character processing device according to the invention of claim 8, in addition to the effect of the input character processing device according to any one of claims 1 to 7, the cross stroke detecting means crosses. A situation where two strokes intersect can be detected.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a schematic configuration of a character recognition device 1;

FIG. 2 is a plan view of the handwritten character input tablet 4;

FIG. 3 is a flowchart of main control of a character recognition process performed by the character recognition device 1.

FIG. 4 is a flowchart of a stroke type determination process.

FIG. 5 is a flowchart of a stroke intersection detection process.

FIG. 6 is a flowchart of an off-line character recognition process.

FIG. 7 is a schematic diagram of handwritten character input.

FIG. 8 is a schematic diagram of an intersection stroke.

FIG. 9 is a schematic diagram of a state where the starting points of the intersection strokes are aligned.

FIG. 10 is a schematic diagram of a state where a cross stroke is separated.

FIG. 11 is a diagram illustrating a state where recognized characters are displayed;

FIG. 12 is a conceptual diagram of a stroke type determination process.

FIG. 13 is a conceptual diagram of a direction row.

FIGS. 14 (1) to (10) are examples of basic strokes.

FIG. 15 is a block diagram illustrating a schematic configuration of a character recognition device 1 according to a second embodiment.

FIG. 16 is a flowchart of control of the character recognition device 1 according to the second embodiment.

FIG. 17 is a conceptual diagram illustrating a separation process of overlapping strokes of the character recognition device 1 according to the second embodiment.

FIG. 18 is a flowchart of control of the character recognition device according to the third embodiment.

FIG. 19 is a conceptual diagram illustrating a character enlarging process of the character recognition device according to the third embodiment.

[Explanation of symbols]

 1 Character Recognition Device 2 CPU 3 I / O Interface 4 Handwriting Character Input Tablet 4a First Input Area 4b Second Input Area 4c Third Input Area 5 Pen 6 Display 7 ROM 7a Offline Recognition Program Storage Area 7b Offline Recognition Dictionary Storage Area 7c Character extraction program storage area 7d Stroke recognition program storage area 7e Basic stroke dictionary storage area 7f Cross stroke detection program storage area 7g Duplication determination value calculation program storage area 7h Duplication determination threshold storage area 7i Cross stroke separation program storage area 8 RAM 8a Input stroke data storage area 8b Recognition result storage area 8c Input stroke type information storage area 8d Cross stroke presence / absence information storage area 8e Duplicate judgment value storage area Rear 8f working area

Claims (8)

[Claims] 1. A handwritten character input unit for inputting a character by handwriting, a stroke detection unit for detecting each stroke of a handwritten character input from the handwritten character input unit in an online manner, An intersecting stroke detecting means for detecting an intersecting state between the two strokes, and the two strokes should intersect based on an intersecting state between the two strokes detected by the intersecting stroke detecting means. And an intersecting stroke overlap judging means for judging whether the two strokes intersect when the intersecting stroke overlapping judging means judges that the two intersecting strokes overlap. And an intersecting stroke separating means for separating the input character. 2. The apparatus according to claim 1, further comprising: a stroke type determination unit configured to determine whether a stroke detected by the stroke detection unit is a straight line; and an intersection angle measurement unit configured to measure an intersection angle between two intersecting strokes. Intersecting stroke detecting means detects an intersection between the two strokes that the stroke type determining means has determined that the stroke is a straight line, and when the intersecting stroke detecting means detects the presence or absence of an intersection between the two strokes, 2. The method according to claim 1, wherein the intersection stroke overlap determination unit determines whether or not the two strokes are overlaps where two strokes that should be separated overlap each other based on a detection result of the intersection angle measurement unit. Input character processing device as described. 3. A method according to claim 1, further comprising: a stroke type determining unit configured to determine whether the stroke detected by the stroke detecting unit is a straight line; and an inner product calculating unit configured to calculate an inner product of two intersecting stroke vectors. The stroke detecting means detects the presence or absence of an intersection between the two strokes that the stroke type determining means has determined that the stroke is a straight line, and when the intersection stroke detecting means detects the intersection between the two strokes, 2. The cross stroke overlap judging unit judges whether or not two strokes are overlaps in which two strokes which should be separated overlap each other, based on a detection result of the inner product calculation unit. Input character processing device. 4. A stroke amplifying means for increasing the width of the stroke detected by the stroke detecting means, a stroke area measuring means for measuring an area of the stroke whose stroke width is increased by the stroke amplifying means, and the stroke amplifying means. Is an inter-stroke overlapping area measuring means for measuring an overlapping area between two strokes having a widened stroke, and any one of the two stroke areas measured by the stroke area measuring means and the inter-stroke overlapping area measuring means. An overlapping area occupancy calculating means for calculating a ratio of the overlapping area to the measured overlapping area, wherein the intersection stroke overlap determining means is configured to separate two strokes based on the measurement result of the overlapping area occupancy calculating means. 2. A method according to claim 1, wherein it is determined whether or not there is an overlap. Input character processing apparatus according. 5. A character enlarging unit for enlarging a handwritten character input from the handwritten character input unit, wherein the intersection stroke detecting unit detects an intersection between two strokes based on the character enlarged by the character enlarging unit. 2. The input character processing device according to claim 1, wherein presence / absence of a character is detected. 6. Off-line character recognition means for recognizing a handwritten character by using the entire shape of a character based on the stroke detected by the stroke detection means and the stroke separated by the intersecting stroke separation means and converting it into a character code. The input character processing device according to any one of claims 1 to 5, further comprising: 7. A display device according to claim 1, further comprising display means for displaying the entire shape of the character based on the stroke detected by said stroke detection means and the stroke separated by said cross stroke separation means. An input character processing device according to any of the claims. 8. The input character processing device according to claim 1, wherein said intersection stroke detection means detects an intersection state of two intersecting strokes.