JPH06282677A - Candidate character correction system for optical character reader - Google Patents

Abstract

PURPOSE:To simply and fast correct the candidate characters that are wrong recognized by designating plural candidate characters to be united or synthesized together and uniting or synthesizing these designated candidates together. CONSTITUTION:When an image fetching instruction is given to a scanner part 2 from a control part 5, a scanner scans on a slip 1 to optically read the images on the slip 1. These fetched images are sent to a character segmenting part 3 and separated from each other for each character. Then these separated images are sent to a character recognizing part 4 and recognized there. The recognizing results are sent to a display part 7. If a single character is separated and recognized wrong, an operator pushes a 'unit (synthesize)' key to instruct the uniting or synthesization of characters without deleting the wrong character not inputting another character. Then the operator designates the separated candidate characters and corrects these characters by uniting or synthesizing automatically them together.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a candidate character correction system in an optical character reader for correcting a candidate character that is optically read and recognized.

[0002]

2. Description of the Related Art FIG. 16 shows an example of a conventional optical character reading device, and is a block diagram of a part for recognizing characters by taking in a form image.

In FIG. 16, this device has a form 101.
A scanner unit 10 for optically reading image data such as characters written above and converting the image data into an electric signal and outputting the electric signal.
2, a character image is obtained from the image data from the scanner unit 102, a character cutout unit 103 that cuts out an image for each character from the character image, and a character image obtained by the character cutout unit 103 are recognized. , A character recognition unit 104 that outputs a character code corresponding to this, a control unit 105 that controls the entire apparatus, and a keyboard unit that performs an operator's character recognition instruction operation and key input for character correction of the recognition result. 106, a display unit 107 having a screen (not shown) for displaying an image of the recognition result from the character recognition unit 104 according to an instruction from the control unit 105, and an image of an input character input from the keyboard unit 106; It is composed of a mouse 108 connected to the keyboard unit 106 as an input means.

FIG. 17 shows an example of the image data entered on the form 101. Here, the image data is described as "Uchisaiwaicho, Chiyoda-ku, Tokyo".

The operation of reading the image data on the form 101 shown in FIG. 17 with the optical character reading device shown in FIG. 16 will be described below.

First, the image of the form 101 is scanned by the scanner unit 10.
In order to capture in 2, the control unit 105 issues an image capture instruction to the scanner unit 102. Then, form 101
A scanner (not shown) of the scanner unit 102 scans the upper part, and the image on the form 101 is optically read by this scanner.

Further, the captured image is a character cutout portion 1.
It is sent to 03 and separated into the image of each character. Hereinafter, this will be referred to as "character cutout". Further, the cut out image is sent to the character recognizing unit 104 for character recognition, and the result is output to the display unit 107 and displayed on a screen (not shown).

FIG. 18 shows an example of a result cut out by the character cutting section 103 in this character reading device.
In this cutout, the cutout result of "Miyako" is "person" and "▲"
It is separated into "Oozato ▼", and "▲ Oozato ▼" is recognized by "Lord". In addition, the cut-out result of “town” is also erroneously recognized as “ta” and “cho”. For this reason, the character recognition unit 104 and the display unit 107 output an incorrect result, and “Tokyo person, Chiyoda-ku Uchikitada” is displayed on the screen as shown in FIG.

Therefore, as a result of this erroneous recognition, the operator deletes the misread characters "person" and "ba" using the keyboard 106, newly inputs "to" as a candidate character, and It is necessary to delete a certain "Ta" and "Ding" and newly enter "City" as a candidate character to display "Uchisaiwaicho, Chiyoda-ku, Tokyo" as an image. A method of inputting by kana-kanji conversion, a method of displaying and selecting candidate characters and candidate words, etc. are adopted.

FIG. 19 shows another example of the image data entered on the form 101. Here, "Kibayashi Kiyoshi" is written as the image data. FIG. 20 shows an example of the result cut out by the character cutout unit 103 in this character reading device. In this cutout, the cutout result of “clean” is separated into “shi” and “blue”. The character recognition unit 10 is recognized.
4 and the display unit 107 output an erroneous result, and the image is displayed on the screen as "Kobayashi Shiblue" as shown in FIG.

Therefore, as a result of this erroneous recognition, the operator deletes the misread characters "shi" and "blue" by the keyboard portion 106 and newly inputs "quick" as a candidate character and is shown in FIG. It is necessary to correct so that the image is displayed as “Kiyoshi Kobayashi”, but in the correction method here as well, the method of inputting with Kana-Kanji conversion on the keyboard unit 106, and displaying by selecting candidate characters and candidate words The method of making it adopted is adopted.

FIG. 23 shows still another example of the image data entered on the form 101, and here it is assumed that "whole body beauty" is written as the image data.

In the image data on the form 101 shown in FIG. 23, when the image is read by the optical character reading device shown in FIG. 16, the result cut out by the character cutting unit 103 is the result shown in FIG. May be obtained. That is, in the result shown in FIG. 24, “all discerning” is cut out, and in this cutting out, the cutout results of “body” and “beauty” are combined into “discipline”. Therefore, even in this case, the character recognition unit 104 and the display unit 107 output an erroneous result, and “Full discipline” is displayed on the screen as shown in FIG.

Therefore, as a result of this erroneous recognition, the operator deletes the misread character "Tsugi" by the keyboard unit 106, and newly adds "body" and "beauty" as candidate characters.
It is necessary to correct the characters by inputting them so that the image is displayed as “whole body beauty” as shown in FIG. 26. In this modification method as well, there is a method of inputting by kana-kanji conversion on the keyboard unit 106, and candidate characters. The method of displaying and selecting a candidate word is adopted.

[0015]

As described above, according to the conventional candidate character correction method in the optical character reader,
If one character is separated by character cutting,
The characters that have been separated and recognized have to be deleted by the operator and re-entered using a keyboard or the like. On the contrary, when two characters are combined by character cutting, the combined and recognized character must be deleted by the operator and re-input by the keyboard or the like. Therefore,
In any case, there is a problem that the correction takes time.

The present invention has been made in view of the above problems, and an object thereof is a candidate character correction system in an optical character reading device capable of easily and quickly correcting a candidate character that is erroneously recognized. To provide.

[0017]

According to the present invention, an object of the present invention is to provide an optical character reader having means for optically reading characters written on a form and outputting the recognition result. A display unit having a screen for displaying the recognition result as an image, and a candidate character that needs to be merged or combined, out of the plurality of candidate characters recognized and image-displayed on the screen, And a candidate character correction unit for merging or synthesizing the plurality of candidate characters to obtain a new candidate character, the plurality of candidates requiring merging or synthesizing via the designating unit. This is achieved by designating a character and combining or synthesizing the designated candidate character by the candidate character correction means.

Further, according to the present invention, in this invention, in the optical character reading device provided with means for optically reading a character described on a form and outputting the recognition result, the output means is provided. A display unit having a screen for displaying an image of the candidate characters read by and the items necessary for operation, and a plurality of the candidate characters image-displayed on the screen are designated, and the designated candidates are designated. A candidate character correction unit for combining or combining characters to obtain a new candidate character; and an item specifying unit for specifying an item required for the operation and an item requiring combination or combination, respectively. When an item requiring coalescence or composition is designated by the item designation unit after the plurality of candidate characters are designated by the designation unit, the candidate character correction unit Coalescing or synthetic manipulation of the accessory character is achieved by that so as to start.

Further, the object of the present invention is to:
An optical character reading device having means for optically reading a character written on a form and outputting the recognition result, a display means having a screen for displaying the recognition result as an image, and the recognition means. Designation means for designating a candidate character that requires separation among the plurality of candidate characters displayed on the screen as an image, and a candidate for separating the candidate character to obtain a new candidate character This is achieved by including character correction means, designating the candidate characters requiring separation through the designating means, and separating the designated candidate characters by the candidate character correction means.

Further, the object of the present invention is to optically read characters written on a form,
In an optical character reading device having means for outputting the recognition result, a display section having a screen for displaying an image of the candidate character read by the output means and an item necessary for operation, and on the screen. A candidate character correction unit for specifying the candidate character displayed on the screen, separating the specified candidate character to obtain a new candidate character, and an item necessary for the operation and an item requiring separation. An item designating unit for designating each, and when the item designating unit designates an item requiring separation after the candidate character is designated via the designating unit, the candidate character correcting unit This is accomplished by activating the operation of separating the candidate characters.

Further, in the present invention, a further object is to provide a means for optically reading a character written on a form and outputting a recognition result thereof. In the optical character reading device, display means having a screen for displaying the recognition result as an image,
Of the plurality of candidate characters that are recognized and displayed as images on the screen, a designating unit for designating a candidate character that needs to be combined, combined, or separated, and the candidate character is combined, combined, or separated. And a candidate character correction means for obtaining a new candidate character, which specifies the candidate character that needs to be merged or combined or separated via the specifying means, and corrects the specified candidate character as the candidate character. It is achieved by combining or synthesizing or separating by means.

Further, the object of the present invention is to:
In an optical character reader equipped with means for optically reading the characters written on a form and outputting the recognition result, the candidate characters read by the output means and the items necessary for the operation are respectively displayed as images. A display unit having a screen for selecting and the candidate character image-displayed on the screen, and a candidate for obtaining a new candidate character by combining, combining, or separating the specified candidate characters. Character correction means, and item designation means for respectively designating an item required for the operation, an item that needs to be merged or combined and an item that needs to be separated, and the plurality of candidates through the designation means When the item designating unit designates an item that needs to be merged or combined or separated after the character is designated, the candidate character correction unit Body or synthetic or separation operation is achieved by that so as to start.

[0023]

According to this, when one character is cut out and recognized in a separated state, or when one character is cut out and recognized in a combined state, these error Even if the recognized candidate characters are not deleted and input again, if the merged or combined or separated character is specified and the candidate character correction means is activated, the specified candidate character is combined or combined or combined. The separation can be done automatically and corrected.

[0024]

Embodiments of the present invention will be described in detail below with reference to the drawings. FIG. 1 is a flow chart for explaining correction of candidate characters, and FIG. 2 is a main part configuration of an optical character reading device to which an embodiment of the present invention is applied, particularly a part for working on a form image and recognizing characters. 3 is a block diagram showing the configuration of FIG.

In FIG. 2, this device is similar to the conventional device in that a scanner unit 2 for reading image data such as characters written on a form 1 and a character image is obtained from the image data from the scanner unit 2. A character cutout unit 3 that cuts out images of each character, a character recognition unit 4 that recognizes the character image obtained from the character cutout unit 3 and outputs a character code, and a control unit 5 that performs recognition control of the entire apparatus. And a keyboard unit 6 for performing an instruction operation of character recognition by the operator and a key input for correcting a character of the recognition result, and a recognition result from the character recognition unit 4 is displayed by an instruction of the control unit 5, or from the keyboard unit 6. A display unit 7 having a screen for displaying the input characters that have been input, a character combination table 5A stored in the control unit 5, and a mouse connected to the keyboard unit 6 as an auxiliary input means. It is composed of a 8.

Next, the operation of this optical character reader will be described. First, in the description of the operation, it is assumed that, on the form 1, as in the form 101 shown in FIG. 17, “Uchisaiwaicho, Chiyoda-ku, Tokyo” is described as image data.

Then, in order to capture the image of the form 1 by the scanner section 2, the control section 5 issues an image capture instruction to the scanner section 2. Then, a scanner (not shown) scans the form 1, and the image on the form 1 is optically read by this scanner. In addition, the captured image is sent to the character cutting unit 3 and separated into images of each character. Further, the cut-out image is sent to the character recognizing section 4 for recognition, and the result is output to the display section 7.

Here, when the result of being cut out by the character cutting section 3 in this character reading device is cut out as shown in FIG. The cut-out results of "person" and "▲ Oozato ▼" are separated, and "▲ Oozato ▼" is recognized by "Lord". The character recognition unit 4 and the display unit 7 have been separated.
If an incorrect result is output, the correction of this candidate character is performed in the following processing example 1 according to the flowchart shown in FIG.
It can be processed as ~ 3.

(Processing Example 1) First, the operator operates the display unit 7
When the recognition result displayed on the screen of is judged to be incorrect (step S1),
The operator presses a "combination (combination) key" (not shown) that indicates the combination or combination of characters to instruct correction of candidate characters.

Then, the cursor on the screen of the display unit 7 is moved onto the candidate character "person" obtained as an erroneous recognition result, and the return key is used to specify the starting point of the candidate character that needs correction. Then, the cursor is moved to the above "Ura" and the end point of the candidate character requiring correction is designated by the return key (step S2). Then, the "person" and the "girl" are displayed in black and white, for example, and this is designated as an incorrect character.

When the return key designates the end point of the candidate character that needs correction, the program operation of word combination (synthesis) is started (step S3).

Then, the control unit 5 in this case is configured to perform a character combination table 5 obtained in advance by combining various characters.
If you have A, first of all, in the case of this example, the candidate character is "person".
And the data of "Uba", what field of which form, what number of character / character group, and data of whether or not it is possible to synthesize (combine) this candidate character from the character combination table 5A. Is taken out and sent to the character cutout unit 3 for judgment (step S4).

The above character combination table 5A
There are various combinations such as combinations of letters and characters, combinations of sides and structures, combinations of crowns and feet, and sides, structures, crowns,
Candidate characters can also be corrected by assigning an external character code to the character code such as foot.

If it is determined that the word composition is impossible, the process proceeds to step S5, and further, in step S7, a message to that effect is displayed on the screen of the display unit 7. Then, the operator who sees this recognizes that the word composition is impossible, and the operator deletes the character by himself and causes the keyboard unit 6 to re-input.

On the other hand, if it is determined that word synthesis is possible, the process proceeds to step S6, and this is corrected into image data for one character to perform word combination (synthesis). In the case of this example, as shown in FIG. 4, “person” and “girl” are combined and replaced with “to”, and the corrected image data is sent to the character recognition unit 4. Then, the character recognition unit 4 displays an image of this on the display unit 7 (step S7).
At this time, the control unit 5 does not need to operate the character cutout unit 3 and the character recognition unit 4.

Subsequently, the operator corrects the erroneously recognized candidate characters "Ta" and "Ding". Here, as in the case of correcting the "person" and the "regular character", the operator first presses the "combine (combine) key" for instructing the combination or combination of characters to instruct the candidate character correction.

After that, the cursor on the display unit 7 is moved to "Ta" as an incorrect recognition result, the start point of the candidate character requiring correction is designated by the return key, and then "Ding". Move the cursor to and specify the end point of the candidate character that needs correction with the return key. Then, "Ta" and "Ding" are highlighted in black and white, which is designated as an incorrect character. At the same time, the program operation of word combination (synthesis) is started, and the following cutout characters "Ta" and "Ding" are combined and replaced with the candidate character "machi" as shown in FIG. Sent to 4. As a result, on the screen of the display unit 7, as shown in FIG. 5, the correctly corrected characters “Uchisaiwaicho, Chiyoda-ku, Tokyo” are displayed as an image.

On the other hand, if all the recognition results are not incorrect in step S1, the process ends without doing anything.

Therefore, according to the above processing example 1, when one character is cut out in a separated state and is erroneously recognized, the characters are merged or combined without having to delete them and input them again. After the operator presses the "combine (combine) key" that indicates, the separated characters can be specified by this operation, and the characters separated by this operation can be automatically combined or combined and corrected. .
As a result, it is possible to speed up the correction of the candidate character.

(Processing Example 2) First, the operator operates the display unit 7
When the recognition result displayed on the screen of is judged to be incorrect (step S1),
Press the "character correction key" to instruct correction of the candidate character. Then, this starts the candidate character correction program.

Subsequently, the cursor on the screen of the display unit 7 is moved to "person" as an erroneous recognition result, the start point of the candidate character requiring correction is designated by the return key, and then "wa". Move the cursor to and press the return key to specify the end point of the candidate character that needs correction (step S
2). Then, the "person" and the "girl" are displayed in black and white, for example, and this is designated as the erroneously recognized character.

Next, when the "combination (combination) key" for starting the combination or combination of the designated characters is pressed, the program operation of word combination (combination) is activated (step S3), and the processing after step S4. Is processed in the same manner as in the processing example 1.

Then, in step S3, the words are combined (synthesized).
When the above program is started, it is first determined whether it is possible to combine "person" and "ba" with the character combination table 5
Judgment is made from A (step S4), and when it is judged that word composition is possible, the process proceeds to step S6 and word combination (composition)
Then, as shown in FIG. 4, "person" and "girl" are combined and replaced with "city", which is displayed on the display unit 7 as an image (step S7).

On the other hand, if it is determined in step S4 that the word composition is impossible, the process proceeds to step S5, and further in step S7 that fact is displayed on the screen of the display unit 7. Then, the operator who sees this recognizes that the word composition is impossible, deletes the character by himself, and causes the keyboard unit 6 to re-input.

Then, the operator corrects the erroneously recognized candidate characters "Ta" and "Ding" in the same manner as in the case of "person" and "ba".

Therefore, even in the case of the above processing example 2, 1
If two characters are cut out and are mistakenly recognized, you can delete them and press the "correct character key" to correct the candidate characters without inputting them again. By designating the required character with the return key and then pressing the "combine (composite) key", the characters separated by this operation can be automatically combined or combined and corrected. As a result, it is possible to speed up the correction of the candidate character.

(Processing Example 3) First, the operator operates the display unit 7
When the recognition result displayed on the screen of is judged to be incorrect (step S1),
A "combine (combine) key" for instructing to combine or combine candidate characters is pressed. Then, this starts the candidate character correction program.

Subsequently, the cursor on the screen of the display unit 7 is moved to "person" as an erroneous recognition result, and the return key is used to specify the starting point of the candidate character that needs to be corrected. Move the cursor to and press the return key to specify the end point of the candidate character that needs correction (step S
2). Then, the "person" and the "girl" are displayed in black and white, for example, and this is designated as the erroneously recognized character. At the same time when the end point is designated, a program operation for word combination (synthesis) is started (step S3), and step S4.
Subsequent processing is performed in the same manner as in the processing examples 1 and 2.

Therefore, even in the case of the above processing example 3, 1
When two characters are cut out in a separated state and are mistakenly recognized, the "composite (combine) key" is used to instruct correction of candidate characters without deleting and reentering them.
After pressing, the start and end points of the character requiring correction can be specified with the return key, and the characters separated by this operation can be automatically combined or combined and corrected. As a result, it is possible to speed up the correction of the candidate character.

In the above embodiment, the return key on the keyboard 6 is used for the operator's correction input. However, the input means such as a mouse 8 or a handwriting tablet or a pen input board can be used. There is no problem even if there is. In addition, although an explanation has been given by taking an example of a right and left clipping error, when the upper and lower characters come close to each other, the vertical clipping may be erroneous. If the upper and lower merging (combining) is instructed (key pressed) using the keyboard, the upper and lower merging (combining) is also possible.

Next, as in the case of the form 101 shown in FIG. 23, when the image data entered on the form 1 is described as "whole body beauty", the result cut out by the character cutting section 3 When the result shown in Fig. 7 is obtained,
That is, when “cutout” is cut out and the cutout results of “body” and “beauty” are combined with “discipline” in this cutout, the following processing examples 4 to 4 are performed according to the procedure of the flowchart shown in FIG. It can be processed as in 6.

(Processing Example 4) First, the operator operates the display unit 7
If it is determined that the cutout results of "body" and "beauty" are combined and erroneously recognized as "discipline" by looking at the recognition result displayed in step S11, character separation is instructed. The operator presses the "separation key" to instruct correction of candidate characters.

Then, the cursor on the screen of the display unit 7 is changed to
The character is moved to "training" as an incorrect recognition result, the return key is pressed, and this is designated as an incorrect character (step S12).

When the return key is designated, the program operation for word separation is activated (step S13). Then, the control unit 5 first refers to the data of the candidate character “training”, what field in what form and what character / character group, and the candidate character from the character combination table 5A. The data indicating whether or not the data can be separated is taken out and sent to the character cutout unit 3 for judgment (step S14).

If it is determined that the word composition is impossible, the process proceeds to step S15, and in step S17, the fact is displayed on the screen of the display unit 7. Then, the operator who sees this recognizes that the word separation is impossible, and the operator deletes the character by himself and causes the keyboard unit 6 to re-input.

On the other hand, if it is determined that word separation is possible, the process proceeds to step S16, and this is corrected into image data for two characters to perform word separation. In the case of this example, as shown in FIG. 4, the candidate characters “body” and “beauty” are separated and replaced with “training”, and the corrected image data is sent to the character recognition unit 4. Then, the character recognition unit 4
Then, this is displayed as an image on the display unit 7 (step S1).
7). At this time, the control unit 5 does not need to operate the character cutout unit 3 and the character recognition unit 4.

On the other hand, if all the recognition results are correct, the process ends without doing anything.

Therefore, according to the processing example 4, when one character is cut out in a combined state and is erroneously recognized, the character separation is instructed without deleting each character and inputting again. After the operator presses the "separation key", the combined character can be specified by the return key, and the combined character can be easily separated and corrected automatically. As a result, it is possible to speed up the correction of the candidate character.

(Processing example 5) First, the operator operates the display unit 7
When it is determined that the cutout result of "body" and "beauty" are combined and recognized as "discipline" by looking at the recognition result displayed in (step S11), the correction of the candidate character is instructed. Press the "Character correction key" for. Then, this starts the candidate character correction program.

Then, the cursor on the screen of the display unit 7 is moved to "train" as an incorrect recognition result and the return key is pressed to designate this as an incorrect character (step S12). Then, this "discipline" is displayed in black and white inversion, for example, and this is designated as a character that has been erroneously recognized.

Next, when the "separation key" for starting the separation of the designated characters is pressed, the word separation program is activated (step S13), and step S1
The processes after 4 are processed in the same manner as the case of the processing example 4.

Therefore, according to the above processing example 5, when one character is cut out in a combined state and is erroneously recognized, it is not necessary to delete each character and input again. After pressing the designated "character correction key", the return key is used to specify the character that needs to be corrected, and then the "separation key" is pressed, so that the combined characters can be easily separated. This allows
It is possible to speed up the correction of candidate characters.

(Processing Example 6) First, the operator operates the display unit 7
When it is determined that the cutout results of "body" and "beauty" are combined and are erroneously recognized as "discipline" by looking at the recognition result displayed in (step S11), the separation of candidate characters is instructed. Press the "separation key" to do this. Then, this starts the candidate character correction program.

Subsequently, the cursor on the screen of the display unit 7 is moved to "train" as an incorrect recognition result and the return key is pressed to designate this as an incorrect character (step S12). Then, this "discipline" is designated as a character that has been erroneously recognized. At the same time as the designation is made, the word separation program is started (step S13), and the processing after step S14 is processed in the same manner as in the processing examples 4 and 5.

Therefore, according to the above processing example 6, when one character is cut out in a combined state and is erroneously recognized, the candidate characters are separated without deleting each character and inputting again. After pressing the "separation key" to instruct, by designating the character requiring correction with the return key, the combined characters can be easily separated and corrected automatically. As a result, it is possible to speed up the correction of the candidate character.

In each of the above-mentioned embodiments, the "character correction key", the "combination (combination) key" and the "separation key" are not provided as independent keys. "Combined (composite) key",
There is no problem even if they are allotted to the "separation keys" and they are pressed.

In each of the above embodiments, the case of combining words and the case of separating words have been described separately. However, each program is installed in the same device, and a "combine (composite) key", It goes without saying that it is possible to switch between the word combination (synthesis) program and the word separation program in conjunction with the operation of the "separation key".

Furthermore, in each of the above-mentioned embodiments, Chinese characters have been described as an example, but other than this, alphabetic characters, katakana, Hangul characters, Arabic characters, etc. can be used regardless of the language.

Next, FIGS. 10 and 11 show a modification of the screen according to the present invention. 10 and 11
In FIG. 1, the same reference numerals as those in FIGS. 1 to 9 indicate the same parts as those in FIGS.

In each of the above embodiments, when the candidate characters are combined, combined, or separated, the "character correction key", the "combination (combine) key", the "separation key", etc. are pressed. Although the structure is disclosed, in this modified example, in addition to the image display of characters on the screen 7a of the display unit 7, items necessary for the correction operation, here, “character correction”, “separation”, and “combination” items are included. Can be displayed, and this item can be selected by the mouse 8 or the cursor movement key. In addition, the word combination (synthesis) program that can be processed according to the flowchart shown in FIG. 1 and the flowchart shown in FIG. A word separation program capable of processing is incorporated so as to be switchable from each other.

The operation of this modified example will be described below. First, on the form 1, as shown in FIG. 12, "Kibayashi Kiyoshi" is written as image data.

Then, when the result cut out by the character cutting unit 3 is cut out as "Kobayashi shi blue" as shown in FIG. 13, that is, the cut result of "sei" is "shi" and "blue".
It is assumed that the character recognition section 4 and the display section 7 are output as erroneous results because they are separated into and.

Here, the operator selects the screen 7a on the display unit 7.
If you look at the recognition result displayed above and determine that this recognition result is incorrect, select the item "correct character" on the screen 7a with the mouse 8 and click, or select with the cursor movement key. Confirm with the return key. Then, this starts the candidate character correction program.

Subsequently, the cursor on the screen 7a of the display unit 7 is moved to "SI" as an erroneous recognition result, and the return key is used to specify the start point of the candidate character that needs correction,
Then, move the cursor over "blue" and specify the end point of the candidate character requiring correction with the return key. Then, the "shi" and "blue" are displayed in black and white, for example, and this is designated as the erroneously recognized character.

Then, the item "combination" on the screen 7a for starting the combination or combination of the designated characters is selected with the mouse 8 or clicked, or selected with the cursor movement key and determined with the return key. Then, the program operation of word combination (synthesis) is started.

First, it is judged from the character combination table 5A whether or not "shi" and "blue" can be combined. If it is judged that word combination is possible, word combination is performed and re-recognition is performed. The result is replaced with the first candidate, for example, "Qing", and this is displayed as an image on the screen 7a of the display unit 7, and the recognition result is correctly corrected to "Qing".

If the candidate characters recognized again in the higher rank are different, the lower rank candidate characters can be sequentially replaced as candidate characters.

When it is determined that the word composition is impossible, the fact is displayed on the screen 7a of the display unit 7. Then, the operator who sees this recognizes that the word composition is impossible, and the operator deletes the character by himself and causes the keyboard unit 6 to re-input.

Next, as in the case of the form 101 shown in FIG. 23, when the image data entered on the form 1 is described as "whole body beauty", the result cut out by the character cutting section 3 When the result shown in Fig. 7 is obtained,
That is, when "cutout" is cut out and the cutout results of "body" and "beauty" are combined with "discipline" in this cutout, the operator first sets the item "character" on the screen 7a of the display unit 7. Select "Modify" with the mouse 8 and click, or select with the cursor movement key and press the return key. Then, this starts the candidate character correction program.

Then, the cursor on the screen 7a of the display unit 7 is moved to "train" as an incorrect recognition result and the return key is pressed to designate this as an incorrect character. Then, this "discipline" is displayed in black and white inversion, for example, and this is designated as a character that has been erroneously recognized.

Then, the item "separation" for starting the separation of the designated character is selected with the mouse 8 and clicked, or selected with the cursor movement key and determined with the return key. Then, the word separation program is started. Then, the control unit 5 determines whether it is possible to separate the “training” from the character combination table 5A obtained in advance by separating various characters, and if it is determined that the word separation is possible, the word is separated. Perform separation. In the case of this example, as shown in FIG. 9, “training” is replaced with the candidate characters “body” and “beauty”.
And is displayed on the screen 7a of the display unit 7.

On the other hand, when it is determined that the word synthesis is impossible, the fact is displayed on the screen 7a of the display unit 7. Then, the operator who sees this recognizes that the word separation is impossible, and the operator deletes the character by himself and causes the keyboard unit 6 to re-input.

Therefore, in the case of this modified example, when one character is cut out in a separated state and is erroneously recognized, it is instructed to correct the candidate character without deleting each character and inputting it again. After selecting the item "character correction" to be performed, the character requiring correction is designated by the return key, and then the item "combine" is pressed, so that the separated characters can be easily combined or combined. Also, if one character is cut out in a synthesized state and is mistakenly recognized, delete it one by one, and select the item "correct character" to instruct correction of the candidate character without inputting again. , "Separation key" that specifies the characters that need to be corrected with the return key and then instructs the separation of the characters
After the operator presses, the combined character can be specified by the return key, and the combined character can be easily separated and corrected automatically. This allows
It is possible to speed up the correction of candidate characters.

[0084]

As described above, according to the present invention,
When one character is cut out and recognized in a separated state, or when one character is cut out and recognized in a combined state, these misrecognized candidate characters are deleted one by one. However, even if you do not input it again, if you start the candidate character correction means by specifying the combined or combined or separated characters, the specified candidate characters will be automatically combined or combined or separated. Since it can be corrected, an effect such as speeding up the correction can be expected.

[Brief description of drawings]

FIG. 1 is a flowchart showing a candidate character correction procedure of the present invention.

FIG. 2 is a block diagram showing a main part configuration of an optical character reading device to which the present invention is applied.

FIG. 3 is a diagram showing an example of erroneous recognition.

FIG. 4 is a diagram illustrating a method of correcting a character that is erroneously recognized.

FIG. 5 is a diagram showing a display state of corrected image data.

FIG. 6 is a flowchart showing another candidate character correction procedure of the present invention.

FIG. 7 is a diagram showing an example of erroneous recognition.

FIG. 8 is a diagram illustrating a method of correcting a character that is erroneously recognized.

FIG. 9 is a diagram showing a display state of corrected image data.

FIG. 10 is a diagram for explaining a modification of the screen according to the present invention.

FIG. 11 is a diagram for explaining a modification of the screen according to the present invention.

FIG. 12 is a diagram showing another example of image data on a form.

FIG. 13 is a diagram showing an example of erroneous recognition.

FIG. 14 is a diagram showing an example of image data which is erroneously recognized and displayed on the screen.

FIG. 15 is a diagram showing an example of image data which is corrected and displayed on the screen.

FIG. 16 is a block diagram of a main part configuration in a conventional optical character reading device.

FIG. 17 is a diagram showing an example of image data on a form.

FIG. 18 is a diagram showing an example of erroneous recognition.

FIG. 19 is a diagram showing another example of image data on a form.

FIG. 20 is a diagram showing an example of erroneous recognition.

FIG. 21 is a diagram showing an example of image data which is erroneously recognized and is displayed on the screen.

FIG. 22 is a diagram showing an example of image data which is corrected and displayed on the screen.

FIG. 23 is a diagram showing still another example of image data on a form.

FIG. 24 is a diagram showing an example of erroneous recognition.

FIG. 25 is a diagram showing an example of image data which is erroneously recognized and is displayed on the screen.

FIG. 26 is a diagram showing an example of image data which has been corrected and displayed on the screen.

[Explanation of symbols]

 1 form 2 scanner section 3 character cut-out section 4 character recognition section 5 control section 5A character combination table 6 keyboard section 7 display section 7a screen 8 mouse

Claims (6)

[Claims] 1. An optical character reading device comprising means for optically reading a character written on a form and outputting the recognition result, wherein a display having a screen for displaying the recognition result as an image. Means, specifying means for specifying a candidate character that needs to be merged or combined from among the plurality of candidate characters that are recognized and displayed as an image on the screen, and combine or combine the plurality of candidate characters And a candidate character correction unit for obtaining a new candidate character, designating the plurality of candidate characters that need to be merged or combined through the designating unit, and designating the designated candidate character as the candidate character. A candidate character correction method for an optical character reading device, characterized in that the correction means merges or combines the characters. 2. An optical character reading device having means for optically reading a character written on a form and outputting a recognition result thereof, wherein the candidate character read by the output means and necessary for operation. A display unit having a screen for displaying an image of each item, and a plurality of the candidate characters image-displayed on the screen are designated, and the designated candidate characters are combined or combined to create a new candidate character. A candidate character correction means for obtaining, and an item designating means for designating an item required for the operation and an item that needs to be merged or combined, respectively, and the plurality of candidate characters are assigned via the designating means. When the item designating unit designates an item that needs to be merged or combined after the designation, the candidate character correction unit causes the candidate character to be merged or combined. Candidate character modification method in an optical character reading apparatus, characterized in that the the to. 3. An optical character reading device comprising means for optically reading a character written on a form and outputting a recognition result thereof, a display having a screen for displaying the recognition result as an image. Means, a designation means for designating a candidate character that needs to be separated from among the plurality of candidate characters that are recognized and image-displayed on the screen, and a new candidate character that separates the candidate character And a candidate character correction means for obtaining the candidate character, the candidate character requiring separation is specified through the specifying means, and the specified candidate character is separated by the candidate character correction means. Candidate character correction method for a characteristic optical character reader. 4. An optical character reading device having means for optically reading a character written on a form and outputting a recognition result thereof, wherein the candidate character read by said output means and necessary for operation. A display unit having a screen for displaying an image of each item, and a candidate for designating the candidate character displayed on the screen and separating the designated candidate character to obtain a new candidate character A character correction unit; and an item designation unit for designating an item required for the operation and an item required to be separated, respectively, and the item designation unit after the candidate character is designated through the designation unit. When an item requiring separation is specified by, the operation for separating the candidate characters by the candidate character correction means is activated. Character modification method. 5. An optical character reading device equipped with means for optically reading a character written on a form and outputting the recognition result, a display having a screen for displaying the recognition result as an image. Means, a designation means for designating a candidate character that needs to be merged or combined or separated from among the plurality of candidate characters that are recognized and image-displayed on the screen; and the candidate character is combined or combined. Alternatively, a candidate character correction unit for separating and obtaining a new candidate character is provided, and the candidate character that needs coalescing, combining, or separating is specified through the specifying unit, and the specified candidate character is A candidate character correction method in an optical character reading device, characterized in that the candidate character correction means merges, combines, or separates. 6. An optical character reading device having means for optically reading a character written on a form and outputting a recognition result thereof, wherein the candidate character read by the output means and necessary for operation. A display unit having a screen for displaying an image of each item, and the candidate character image-displayed on the screen are designated, and the designated candidate character is merged or combined or separated to create a new candidate character. And an item designating unit for designating an item required for the operation, an item that needs to be united or combined and an item that needs to be separated, respectively, via the designating unit. After designating the plurality of candidate characters, the item designating means designates an item that needs to be merged, combined or separated, and the candidate character correcting means determines Candidate character modification method in an optical character reading apparatus characterized by coalescence or synthetic or separation operation of said candidate character is to boot.