JP3138665B2 - Handwritten character recognition method and recording medium - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a handwritten character recognition system and a recording medium, and more particularly, to a handwritten character recognition system for recognizing characters handwritten by a pen input device, and a computer-implemented handwritten character recognition system. The present invention relates to a recording medium storing a program and dictionary data.

[0002]

2. Description of the Related Art A general method of character recognition is to compare an input character pattern with a reference pattern of all characters registered in a character recognition dictionary (reference pattern database), and to determine the number of patterns most similar to each other. Characters are extracted as candidate characters, and their character codes are stored in memory in the order of similarity (an evaluation value indicating the degree of similarity),
A character having the highest similarity among the characters is displayed on the screen as a recognition result, and is called a bitmap method based on pattern comparison of the entire character.

There are two types of handwritten character recognition: off-line character recognition in which characters written in advance on original paper or the like are read and processed by a scanner or the like, and online character recognition in which characters written on a tablet with a pen are processed in real time. . In the former case, the captured characters are treated as a set of points and are compared with the reference pattern.The latter is limited to the bitmap method. Other information can be obtained, and a method using such information has been proposed.
However, there are many individual differences and uncertain factors in these pieces of information, and a bitmap system for performing character recognition by converting a handwriting pattern of the entire character into a bitmap is fundamental.

As one measure for improving the recognition accuracy of handwritten character recognition by the bitmap method, Japanese Patent Laid-Open Publication No.
No. 01 discloses a method in which the pixel value of each pixel of a bitmap is set as multi-valued information representing shading, and a pixel value is determined by performing weighting based on a stroke length or the like.
In general, binary information representing black and white is used for each pixel value of a bitmap.

FIG. 9 is a block diagram showing a basic configuration of a conventional handwritten character recognition system using a bit map. As shown in FIG. 9, a conventional handwritten character recognition method includes a pen input device 1, a data processing device 2b operated under program control, a storage device 3b for storing various information, and a character recognition device used for recognition processing. The data processing device 2b includes a dictionary storage device 4b in which the dictionary unit 41b is stored, and a display device 5. The data processing device 2b checks a handwriting pattern of a character written in the pen input device 1 with the dictionary storage device 4b. ,
The most probable character is displayed on the display device 5 as a recognition result.

For this reason, as shown in FIG. 10, the character recognition dictionary unit 41b of the dictionary storage device 4b associates a character code with bitmap information representing a standard shape (reference pattern) of the character. The data processing device 2b includes a handwriting data acquisition processing unit 21, a character recognition processing unit 22b, and a recognition result display unit 26. The storage device 3b includes handwriting for temporarily storing various information being processed. A data storage unit 31 and a character recognition result storage unit 32b are provided.

The handwriting data acquisition processing unit 21 sequentially acquires handwriting position coordinate data written in the pen input device 1, performs normalization processing such as standardizing the size of characters, and performs handwriting for one character. Is converted into a bitmap, and the bitmap information representing the handwriting pattern is stored in the handwriting data storage unit 31. The character recognition processing unit 22b sequentially compares the handwriting pattern stored in the handwriting data storage unit 31 with the bitmap information stored in the character recognition dictionary unit 41b, and calculates the similarity by quantifying the similarity. Then, a plurality of candidate characters are selected (recognized) in descending order of similarity, and the character codes are stored in the character recognition result storage unit 32b. The recognition result display unit 26
The character code of the character having the highest similarity among the candidate characters stored in the character recognition result storage unit 32b is sent to the display device 5 as the recognition result and displayed.

FIG. 11 is a flowchart showing a processing procedure of the conventional handwritten character recognition system of FIG. Hereinafter, a conventional character recognition operation will be described with reference to FIG.

The handwriting input to the pen input device 1 is acquired by the handwriting data acquisition processing unit 21 (step S2).
1) The bit-mapped handwriting pattern is stored in the handwriting data storage unit 31 (step S22). The character recognition processing unit 22b compares the handwriting pattern for one character with the bitmap information of the character recognition dictionary unit 41b and extracts a plurality of candidate characters in descending order of similarity (step S23). Character recognition result storage 3 for codes in order of similarity
2b (step S24). When the extraction of the candidate characters is completed, the recognition result display unit 26 sends the character code of the first candidate character having the highest similarity to the display device 5 and displays it as a recognition result on the screen (step S25). If the displayed recognition result is incorrect, the next candidate is displayed, and if correct, the information of the corresponding character is deleted from the storage device 3 by the confirmation process.

[0010]

The first problem of the character recognition by the conventional bitmap method described above is that the entire character is converted into a bitmap and the recognition process is performed. It is easy to be misrecognized by a shape character. The bitmap information of the character recognition dictionary is
Since it is created with an average reference pattern, the probability of erroneous recognition increases in the case of handwritten characters such as habit characters having a large individual difference. To improve this, measures such as creating a personalized dictionary for each individual are conceivable.However, in online character recognition, which processes characters written to an input device in real time, it is necessary to obtain a pattern other than the handwriting pattern of the entire character. It is conceivable to use available information effectively.

A second problem is that it takes a long time to perform the character recognition process because all of the characters registered in the character recognition dictionary are compared with the input handwriting pattern and candidate characters are selected. .

An object of the present invention is to focus on the first stroke considered to be most important in inputting handwritten characters, and to perform a recognition process in which information of the first stroke is added in addition to the conventional recognition process based on the pattern of the entire character. Accordingly, it is an object of the present invention to provide a handwritten character recognition processing method capable of securing a higher accuracy recognition rate and further reducing the processing time.

[0013]

According to a first aspect of the present invention, there is provided a handwritten character recognition system for comparing a handwriting pattern of a character input by handwriting with a pen input device with a reference pattern stored in a character recognition dictionary. In the handwritten character recognition method to be performed, a dictionary dedicated to the first stroke in which the reference pattern of the first stroke used for the first stroke of the handwriting is stored in association with the identification information, and one stroke from the input handwriting pattern for one character A first stroke information extracting unit for extracting a handwriting pattern of an eye, and a first stroke for obtaining corresponding identification information by comparing the first stroke handwriting pattern extracted by the first stroke information extracting unit with the first stroke dedicated dictionary; Eye recognition means , stored in the first stroke dedicated dictionary
Identification information is a classification number for classifying a character group,
Character recognition dictionaries contain characters corresponding to the reference pattern of characters.
The classification number is stored in addition to the code, and the character recognition is performed.
The classification number of the candidate character obtained by character recognition
Matched with the classification number obtained by the first stroke recognition means
Candidate ranking correction that sometimes corrects the evaluation rank of the candidate character
And a character recognition unit, wherein character recognition by the character recognition dictionary and recognition by the first stroke recognition unit are used in combination.

According to a second aspect of the present invention , there is provided a handwritten character recognizing method.
Character recognition of handwriting patterns of characters input by handwriting
Characters compared with the reference pattern stored in the dictionary
In the handwriting recognition system that recognizes
Identify the first reference pattern used for each stroke
A dictionary dedicated to the first stroke stored in association with the information
From the handwriting pattern for one character
First stroke information extracting means for extracting the first stroke information extracting means;
The handwriting pattern of the first stroke extracted in the step
First stroke recognizing means for finding the corresponding identification information by comparing with the text
Character recognition by the character recognition dictionary and
In addition to the recognition by the stroke recognition means,
The identification information stored in the dictionary for
Character code group of multiple characters to use
I do.

[0015]

[0016]

According to a third aspect of the present invention, there is provided the recording medium according to the first or second aspect.
A program for realizing the processing of each means described in (1) by a computer, a dictionary for character recognition, and dictionary data of a dictionary dedicated to the first stroke are stored.

[0018]

Next, embodiments of the present invention will be described with reference to the drawings.

FIG. 1 is a block diagram showing the configuration of the first embodiment of the present invention.

In the first embodiment of the present invention, as shown in FIG. 1, a pen input device 1, a data processing device 2 operated by program control, a storage device 3 for storing various information, a recognition process And a display device 5 for displaying the recognition result.

In the first embodiment, in addition to the conventional bitmap type character recognition in which recognition is performed by comparing patterns of one character, only the first stroke (first stroke of writing) of each character is performed. Perform pattern recognition for
It is intended to obtain a highly accurate recognition result in consideration of the recognition result of the first stroke. For this reason, the dictionary storage device 4 includes a first stroke dedicated dictionary unit 42 used for pattern recognition of the first stroke in addition to a character recognition dictionary unit 41 for pattern recognition of normal whole characters. The apparatus 2 includes a handwriting data acquisition processing unit 21, a character recognition processing unit 22, and a recognition result display unit 26 that are used for normal whole pattern recognition of a character, and a first image for performing pattern recognition of a first image. It includes an information extraction processing unit 23, a first stroke recognition processing unit 24, and a first stroke determination processing unit 25 for reflecting the recognition result of the first stroke on the character recognition result. The storage device 3 includes a handwriting data storage unit 31 that stores the handwriting pattern obtained by the handwriting data acquisition processing unit 21 from the pen input device 1 and a character recognition result that stores the candidate characters extracted by the character recognition processing unit 22. A storage unit 32, a first-screen information storage unit 33 for storing information of the first screen extracted by the first-screen information extraction processing unit 23, and a classification number of the first screen recognized by the first-screen recognition processing unit 24. A first stroke recognition result storage unit 34 to be stored is provided.

First, the configuration of each dictionary provided in the dictionary storage device 4 will be described. The first stroke dedicated dictionary unit 42
FIG. 2A is a dictionary in which reference patterns used for pattern recognition of the first stroke are registered. As shown in FIG. 2A, a classification number classified according to the shape and writing position of the first stroke, and the shape of the first stroke , And three types of information, that is, bitmap information representing a reference pattern for recognizing the first stroke and position information indicating the coordinates of the writing start position of the first stroke. The classification number is
As shown as "corresponding character group" on the right side of the figure, characters with the same shape of the first stroke and the same writing position are classified as the same first stroke character group, and the number assigned to each character group is Yes, corresponding to the shape and writing position of the first stroke.

The position information (x, y) is a coordinate value of the writing start position of the first stroke, and indicates a relative position with respect to the size of the entire character. This will be specifically described with reference to FIG. Consider the letter "stone" as in (a). First, a rectangle circumscribing the character "stone" as shown in (b) is extracted. Next, as shown in (c), the height and width of the extracted rectangle are equally divided into a specified number, respectively, and the position information (3, 0) in FIG. Can be

The position information is used as a criterion for recognizing the first stroke of a similar shape. For example, "tree"
When character groups such as "books" and character groups such as "stone" and "thickness" are separated, the shape of the first stroke is both "1". If the degree is increased, it cannot be determined which character group is based on the bitmap information alone. Therefore, by adding the position information, the writing position is also compared so that a character group closer to the writing position can be specified.

As shown in FIG. 3, the character recognition dictionary unit 41 stores a character code, bitmap information for recognizing the character, and a classification number indicating the classification of the first stroke to which the character belongs. Type information is stored. The character code and the bitmap information are the same as those of the conventional character recognition dictionary, but the information of the classification number is newly added.

Next, the operation of each section of the data processing apparatus 2 that performs recognition processing using the dictionaries of the dictionary storage device 4 and the storage device 3 will be described. Note that the handwriting data acquisition processing unit 21 and the recognition result display unit 26 using the same reference numerals as those in the conventional method of FIG. 9 perform the same processing as in FIG.

The handwriting data acquisition processing unit 21 sequentially acquires handwriting position coordinate data written in the pen input device 1, performs normalization processing such as standardizing the character size, and performs handwriting for one character. Is converted into a bitmap, and the bitmap information representing the handwriting pattern is stored in the handwriting data storage unit 31.

The character recognition processing unit 22 reads out the handwriting pattern for one character stored in the handwriting data storage unit 31,
It sequentially compares the bitmap information with the bitmap information stored in the character recognition dictionary unit 41, calculates the similarity by quantifying the degree of similarity of the pattern, and calculates a predetermined number of candidate characters in descending order of similarity. The extracted character codes, the classification numbers, and the reliability, which is a numerical value for evaluating the order, are stored in the recognition result storage unit 32. At this time, the largest numerical value is assigned to the first rank as the reliability, and the lower the rank, the smaller the numerical value.

The first stroke information extraction processing unit 23 extracts only the first stroke handwriting pattern from the stroke patterns of one character stored in the handwriting data storage unit 31, and extracts the first stroke handwriting pattern together with the writing start position information. It is stored in the eye information storage unit 33.

The first stroke recognition processing unit 24 reads the handwriting pattern and the position information of the first stroke stored in the first stroke information storage unit 33, and stores the bitmap information and the position of the first stroke dedicated dictionary unit 41. Compared with the information, the classification number of the first stroke having the highest similarity is acquired and stored in the first stroke recognition result storage unit 34. The first stroke recognition processing unit 24 performs a pattern recognition process based on a bitmap method in the same manner as the character recognition processing unit 22. However, the character recognition processing unit 22 compares the bitmap information of the entire character with a plurality of candidate characters (characters). On the other hand, the first stroke recognition processing unit 24 compares the bitmap information of only the first stroke with the writing position information, and extracts only one classification number having the highest similarity.

The first-stroke determination processing unit 25 determines that a character having the same classification number as the classification number stored in the first-stroke recognition result storage unit 34 among the candidate characters stored in the character recognition result storage unit 32. Judge whether there is, and if so, add a predetermined numerical value to the reliability of the character and correct the reliability,
With the modification of the reliability, the arrangement of the candidate characters is rearranged in the order of the new reliability.

The recognition result display section 26 sends the character code of the candidate character having the highest reliability after rearrangement to the display device 5 as a recognition result and displays it. When a confirmation instruction is issued for the display result, the character code is incorporated as final information in the sentence data, and the storage information on the target character in the storage device 3 is cleared, but when the next candidate instruction is issued, The character code of the candidate character having the next highest reliability among the candidate characters stored in the character recognition result storage unit 32 is sent to the display device 5.

FIG. 4 is a flowchart showing the operation of the first embodiment of the present invention shown in FIG. Hereinafter, FIG.
The processing of the first embodiment will be described with reference to FIG.

The handwriting for one character input from the pen input device 1 is read and normalized by the handwriting data acquisition processing unit 21 (step S1), and the handwriting pattern is stored in the handwriting data storage unit 31 (step S1). S2). The character recognition processing unit 22 sequentially compares the handwriting pattern for one character with the bitmap information of the character recognition dictionary unit 41, extracts a plurality of candidate characters in descending order of similarity (step S3), and extracts the extracted candidate characters. The character code and the classification number of the character are stored in the character recognition result storage unit 32 together with the reliability in which the order is quantified (step S4).

The first stroke information extraction processing unit 23 extracts only the first stroke pattern of the character from the stroke pattern of one character stored in the stroke data storage unit 31 and extracts the first stroke pattern together with the writing start position information. The information is stored in the information storage unit 33 (step S5). Subsequently, the first stroke recognition processing unit 24 reads the handwriting pattern and the position information of the first stroke stored in the first stroke information storage unit 33, and stores the bitmap information and the position information of the first stroke dedicated dictionary unit 42. Then, information having the highest similarity is extracted, the classification number of the first stroke is obtained, and stored in the first stroke recognition result storage unit 34 (step S6).

Next, the first stroke determination processing unit 25 compares the classification number stored in the first stroke recognition result storage unit 34 with the classification number of the candidate character stored in the character recognition result storage unit 32. Then, if the candidate character has the same classification number, the reliability is increased by adding some constant to the reliability (step S7). When the comparison with all the candidate characters is completed, the candidate characters are rearranged in descending order of reliability (step S8). By this processing, the recognition result is obtained from two viewpoints, that is, the recognition result of the entire character and the recognition result of the first stroke, so that the recognition rate can be improved. Finally, the candidate character having the highest reliability is displayed as a recognition result on the recognition result display section 26 (step S9).

Next, with reference to FIG. 5, a description will be given of a specific processing example in a case where the character "sea" is input by handwriting on the pen input device 1. FIG. The input coordinate data is bit-mapped by the handwriting data acquisition processing unit 21 and passed to the character recognition processing unit 22 as a handwriting pattern. Character recognition processing unit 2
2 performs character recognition by sequentially comparing the received handwriting pattern with the bitmap information of the character recognition dictionary unit 41 shown in FIG. As a result, in order of similarity, “regret, contempt, sea,
Consider the case where ten candidate characters of “.
The extracted character codes and classification numbers of the ten candidate characters are stored in the character recognition result storage unit 32 in ascending order of the degree of similarity, along with the reliability that has been converted into constants of 100, 99, 98,. (A). At this time,
The character code “8A43” of “sea” is stored with the reliability “98” and the classification number “3” as the third most likely candidate character.

Next, the first stroke information extraction processing unit 23 extracts the first stroke of the input “sea”, and stores the handwriting pattern and the writing position coordinates of the extracted first stroke in the first stroke information storage unit 33. To save. Subsequently, the first stroke recognition processing unit 24 reads the handwriting pattern and the position information stored in the first stroke information storage unit 33, and reads the first stroke dedicated dictionary unit 4 shown in FIG.
2 is compared with the bitmap information and the position information of No. 2 to extract the classification number. In the case of "sea", the classification number "3" is extracted. The extracted classification number is stored in the first stroke recognition result storage unit 34.

Finally, in the first stroke determination processing section 25,
The classification number of the candidate character stored in the character recognition result storage unit 32 is compared with the classification number of the first stroke stored in the first stroke recognition result storage unit 34. Increase the degree. Here, “5” is added to the reliability. After recalculation of the reliability, the candidate characters are rearranged in the order of the corrected new reliability. The result of the rearrangement is as shown in (B) of the character recognition result storage unit 32. As a result, the sea was the third most reliable
Is the most reliable.

In the conventional character recognition, where the character "regret" is displayed as the recognition result, the character "sea" is correctly displayed as the recognition result by considering the information of the first stroke. become. As described above, it is possible to avoid erroneous recognition of characters having different shapes, and thus an improvement in recognition rate can be expected.

As described in detail above, the handwriting pattern of the entire character is stored in the character recognition dictionary (character recognition dictionary unit 41).
Character recognition means (handwriting data acquisition processing unit 21, handwriting data storage unit 31, character recognition processing unit 22, character recognition result storage unit 32, recognition result display unit 2)
6) In addition to the first stroke dictionary (first stroke dictionary 42)
A first stroke information extraction unit (first stroke information extraction processing unit 23, first stroke information storage unit 33) for extracting a first stroke handwriting pattern from stroke information of handwriting input; A first-order recognition unit (first-order recognition processing unit 24, first-order recognition result storage unit 34) for comparing with the first-only dictionary unit to obtain a classification number; The recognition rate is improved by replacing the order of the candidate characters extracted by the character recognition means by the unit 25).

In the first embodiment described above, when the character recognition processing unit 22 stores the candidate characters extracted by pattern comparison of the entire character in the recognition result storage unit 32, the character recognition processing unit 22 ranks information in a higher rank as information indicating the rank. The first stroke determination processing unit 25 stores a numerical value (reliability) that becomes smaller as the first stroke is reduced, based on the classification number acquired by the first stroke recognition processing unit 24 by comparing the handwriting patterns of the first stroke. The order of candidate characters is changed by correcting the numerical value. This method has an advantage that it can be processed independently of the method of calculating the similarity in the character recognition processing unit 22. However, using the numerical value of the similarity used in the character recognition processing unit 22 as it is, a candidate character whose classification number matches The order may be switched by adding a certain numerical value to the similarity degree of.

FIG. 6 is a block diagram showing the configuration of the second embodiment of the present invention. The difference between the second embodiment and the above-described first embodiment is that in the first embodiment,
The accuracy of recognition is improved by correcting the result of character recognition by pattern comparison of the entire character with the recognition result of only the first stroke.
By performing the recognition processing using only the first stroke before character recognition, the range of the character recognition dictionary to be referred to at the time of character recognition is limited, and the character recognition processing time can be reduced.

Referring to FIG. 6, a second embodiment of the present invention, like the first embodiment shown in FIG. 1, has a pen input device 1, a data processing device 2a, and a storage device 3a.
, A dictionary storage device 4a, and a display device 5.

The data processing apparatus 2a includes a handwriting data acquisition processing unit 21, a first stroke information extraction processing unit 23, a first stroke recognition processing unit 24, and a recognition result display unit that perform the same processing as in the first embodiment. 26, and a dictionary search that limits the search range of the character recognition dictionary unit 41a to be searched by the character recognition processing unit 22a from the classification numbers stored in the first stroke recognition result storage unit 34 so that only the part with the same classification number is performed. Range limit processing unit 2
7 and the handwriting pattern stored in the handwriting data storage unit 31 is compared with only the characters of the character stored in the character recognition dictionary unit 41a which are limited by the dictionary search range limitation processing unit 27. Character recognition result storage unit 3 in descending order of degree
2a for storing candidate characters.

The storage device 3a includes a handwriting data storage unit 31, a first stroke information storage unit 33, a first stroke recognition result storage unit 34 for storing the same data as in the first embodiment, a character recognition processing unit. The character recognition result storage unit 32a stores the candidate character selected in step 22a. The format of the data stored in the character recognition result storage unit 32a is different from the character recognition result storage unit 32 of the first embodiment. There are differences.

The dictionary storage device 4a includes a character recognition dictionary unit 4
1a and a first stroke dedicated dictionary unit 42. While the first stroke dedicated dictionary unit 42 has the same format and the same arrangement as the first embodiment, the character recognition dictionary unit 41a
Has the same format, but the sequences are ordered by classification number.

FIG. 7 is a flowchart showing the operation of the second embodiment of FIG.

The handwriting input to the pen input device 1 is acquired by the handwriting data acquisition processing unit 21 (step S1).
1) The handwriting pattern is stored in the handwriting data storage unit 31 (step S12). The first stroke information extraction processing unit 23
The first handwriting pattern and position information are extracted from the handwriting pattern stored in the handwriting data storage unit 31 and stored in the first stroke information storage unit 33 (step S13). The first stroke recognition processing unit 24 stores the first stroke handwriting pattern and the position information stored in the first stroke information storage unit 33 into a first stroke dedicated dictionary unit 42.
Then, the classification number is acquired by comparing with the first stroke number, and the acquired classification number is stored in the first stroke recognition result storage unit 34 (step S14).

Next, in the dictionary search range limitation processing section 27, the character recognition dictionary section 41a and the first stroke recognition result storage section 3
Compare with the classification number stored in 4. Since the data of the character recognition dictionary unit 41a are arranged in the order of the classification numbers, the start address and the end address of the data having the same classification number as the classification number of the first stroke stored in the first stroke recognition result storage unit 34 are set. Once acquired, the search range can be limited (step S15). The character recognition processing unit 22a compares the handwriting pattern for one character stored in the handwriting data storage unit 31 with the character recognition dictionary unit 41a and extracts a plurality of candidate characters in descending order of similarity. The comparison range may be a range from the start address to the end address acquired by the dictionary search range limitation processing unit 27 (Step S16). For this reason, the number of characters to be compared is smaller than in the past, and the recognition process is faster. Only the character codes of the extracted candidate characters are stored in the character recognition result storage unit 32a (step S17). Finally, the recognition result display unit 26 displays the first candidate character having the highest similarity as the recognition result (step S18).

As described above, in the present embodiment,
First stroke recognition by the first stroke recognition processing unit 24 is performed,
Since the dictionary search range limitation processing unit 27 limits the search range of the character recognition dictionary unit 41a based on the result, the speed of the recognition process by the character recognition processing unit 22a is increased, and the different similar characters in the first stroke are set in the search range. , The erroneous recognition is reduced and the recognition rate is improved.

In the description of the first and second embodiments, the classification number is stored as the information for identifying the first stroke pattern in the first stroke dedicated dictionary unit, and the character code is stored in the character recognition dictionary unit. Together with the classification number. By using this classification number, the configuration of the dictionary for the first stroke is simplified, and the order of candidate characters can be easily corrected and the search range can be limited based on the recognition result of the first stroke, but the classification number is added. Therefore, it is necessary to change the conventional character recognition dictionary unit. On the other hand, if the character codes corresponding to the first-stroke dedicated dictionary unit are enumerated as information for identifying the first-stroke pattern without using the classification number, the process of correcting the rank and limiting the range becomes slightly complicated. The conventional dictionary for character recognition can be used without being changed.

The character recognition dictionary unit has been described as storing only one classification number in correspondence with the character code. However, for characters in which the stroke order of the first stroke is not fixed and is divided by a person, It is preferable to register a plurality of classification numbers.

Further, the dictionary for exclusive use of the first stroke stores bitmap information indicating the shape of the first stroke pattern and position information indicating the writing start position. The position information indicating the writing start position may be deleted as the information including the relative position in the middle.

The handwriting written into the pen input device is converted into a bitmap to form a handwriting pattern by the handwriting data acquisition processing unit. The character recognition processing unit converts the handwriting pattern into bitmap information of the character recognition dictionary unit. Although the collation has been described, the handwriting data acquisition processing unit and the character recognition processing unit may be collectively processed without being separated.

FIG. 8 is a block diagram showing the configuration of the third embodiment of the present invention. The third embodiment includes a pen input device 1, a display device 5, a computer 6 including a data processing unit and a storage unit, and a recording medium 7. The recording medium 7 includes a program necessary for causing the computer 6 to execute the processing performed by the data processing device 2 or 2a in the first or second embodiment, and a character recognition dictionary 41 or 41a. The dictionary data to be stored in the image-specific dictionary section 42 is stored. By reading these into the storage section of the computer 6, the first
The same processing as that described in the second embodiment can be performed.

[0057]

As described above in detail, the handwritten character recognition method of the present invention includes the same character recognition means as the conventional one by the bitmap method, a dictionary dedicated to the first stroke, and handwriting of the first stroke from the handwriting input. A first stroke information extracting unit for extracting a pattern, and a first stroke recognition unit for recognizing the first stroke by referring to a dictionary dedicated to the first stroke; In addition, the probability of erroneously recognizing a different character having a similar shape is reduced and the character recognition rate is improved.

Further, according to the handwritten character recognition method of the third aspect, the recognition of the first stroke is performed before the character recognition and the range of characters to be compared at the time of character recognition is limited, so that the character recognition rate is improved. In addition, there is an effect that the time required for character recognition is reduced.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of a first exemplary embodiment of the present invention.

FIG. 2 is an explanatory diagram showing a format of a character recognition dictionary unit in FIG. 1;

FIG. 3 is an explanatory diagram showing a format of a first-image dedicated dictionary unit in FIG. 1;

FIG. 4 is a flowchart for explaining the operation of the first embodiment of FIG. 1;

FIG. 5 is an explanatory diagram for explaining a specific example of processing according to the first embodiment;

FIG. 6 is a block diagram showing a configuration of a second exemplary embodiment of the present invention.

FIG. 7 is a flowchart for explaining the operation of the second embodiment in FIG. 6;

FIG. 8 is a block diagram showing a configuration of a third exemplary embodiment of the present invention.

FIG. 9 is a block diagram showing a basic configuration of a conventional handwritten character recognition method.

FIG. 10 is an explanatory diagram showing a format of a conventional character recognition dictionary.

FIG. 11 is a flowchart for explaining the operation of the conventional handwritten character recognition method.

[Explanation of symbols]

 Reference Signs List 1 pen input device 2, 2a, 2b data processing device 3, 3a, 3b storage device 4, 4a, 4b dictionary storage device 5 display device 6 computer 7 recording medium 21 handwriting data acquisition processing unit 22, 22a, 22b character recognition processing unit 23 First stroke information extraction processing unit 24 First stroke recognition processing unit 25 First stroke determination processing unit 26 Recognition result display unit 27 Dictionary search range limitation processing unit 31 Handwriting data storage unit 32, 32a, 32b Character recognition result storage unit 33 First stroke information storage unit 34 First stroke recognition result storage unit 41, 41a, 41b Character recognition dictionary unit 42 First stroke dedicated dictionary unit

Claims (3)

(57) [Claims] 1. A character input by handwriting input to a pen input device.
The handwriting pattern is stored in the reference pattern stored in the character recognition dictionary.
Handwritten character recognition system that performs character recognition in comparison with turns
The first stroke used for the first stroke of the handwriting
The first stroke that stores the reference pattern of
A special dictionary and a handwriting pattern for one input character
First stroke information extracting means for extracting a handwriting pattern of the stroke,
The handwriting putter of the first stroke extracted by this first stroke information extraction means
And the corresponding identification information by comparing the
The first stroke recognition means to be sought, and the first stroke
Identification information is a classification number for classifying the character group,
The character recognition dictionary corresponds to the character reference pattern.
The classification number is stored in addition to the character code.
Classification number of candidate characters obtained by character recognition using recognition dictionary
Number matches the classification number obtained by the first stroke recognition means
Candidate order to correct the evaluation order of the candidate character when
Character correction means, and character recognition by the character recognition dictionary.
And recognition by the first stroke recognition means in combination.
Characteristic handwriting recognition method. 2. A character input by handwriting input to a pen input device.
The handwriting pattern is stored in the reference pattern stored in the character recognition dictionary.
Handwritten character recognition system that performs character recognition in comparison with turns
The first stroke used for the first stroke of the handwriting
The first stroke that stores the reference pattern of
A special dictionary and a handwriting pattern for one input character
First stroke information extracting means for extracting a handwriting pattern of the stroke,
The handwriting putter of the first stroke extracted by this first stroke information extraction means
And the corresponding identification information by comparing the
The first stroke recognition means to be sought, and the character recognition dictionary
Character recognition and recognition by the first stroke recognition means
And identification information stored in the first stroke dedicated dictionary
Is the character code of multiple characters that use the first stroke
A handwritten character recognition method characterized by being a group . 3. Each hand according to claim 1 or 2.
A program for realizing the processing of steps by computer
Dictionary data for character and character recognition, and dictionary data for the first stroke
A recording medium characterized by being stored .