JPH07117983B2 - Character correction method in character recognition device - Google Patents

Description

Description: TECHNICAL FIELD OF THE INVENTION The present invention relates to a character correction method for correcting an erroneously recognized character when a character recognized by a character recognition device such as a Chinese character is erroneous.

[Background of technology]

Generally, a character recognition device optically observes a character written on a paper or the like to extract the feature of the character, and then collates the feature of the character with the content of a predetermined recognition dictionary to detect the character. Is recognized, and the recognition result is displayed on the display unit. However, in such a character recognition device, there is a possibility that the characters may not be recognized correctly, especially when the characters written on the paper or the like are messy or quirky. Therefore, it is necessary to correct the erroneously recognized display character to the correct one.

[Conventional technology and problems]

Conventionally, as a character correction method of this kind, for example, a plurality of candidate character groups including recognized characters in order of high similarity to observed characters are preselected from the recognition dictionary, and the candidate character groups are displayed on the display unit. Displayed one by one, and select the correct one from them to correct the misrecognized character, or
By inputting information for specifying the correct answer character, for example, information such as kana reading of the correct answer character and the number of strokes with a keyboard device, a correction character group based on the key input is selected from a predetermined dictionary, and these There is one in which the correction character group is sequentially displayed on the display unit, and the correct character is selected from the correction characters to correct the erroneously recognized character.

However, in such a conventional character correction method,
In the former case, since the candidate character group has a high degree of similarity to the observed character, the correct character is not always included in the candidate character group. For this reason, when the correct character is not included in the candidate character group, there arises a problem that the method cannot correct the erroneously recognized character. In the latter case,
Since the correct character is always included in the above-mentioned correction character group, the problem like the former does not occur, but if there are many correction character groups, for example, in the correction character group, It takes time to select the correct answer character from
This causes a problem that the correction speed of characters becomes faster.

[Object of the Invention]

This invention was made from the above viewpoints,
It is an object of the present invention to provide a character correction method in a character recognition device which can correct an erroneously recognized character quickly and always accurately.

[Structure of Invention]

In order to achieve the above-mentioned object, the present invention selects from the recognition dictionary a plurality of candidate characters ranked in descending order of similarity to the observed character, and selects the first candidate character group having the first rank. Assuming a character recognition device that displays on the display unit as a recognition result, paying attention to the relatively high probability that the correct character is included in the candidate character group, when correcting a display character that is erroneously recognized. , A correction character group including the correct answer character is selected from a predetermined dictionary based on information for identifying the correct answer character, and when the candidate character is included in the correction character group, After the correction characters that match the candidate characters are arranged higher than the other correction character groups according to the order of the above candidate character groups, the correction characters that match the correct answer character from the correction character group having this new sequence rank are arranged. Select characters , In which the modified character was to reappear on the display unit in place of the misrecognized characters.

Example of Invention

 Hereinafter, the present invention will be described in detail with reference to the accompanying drawings.

FIG. 1 is a block diagram showing an embodiment of a character recognition device for realizing the character correction method according to the present invention. In the figure, 1 is an observing section for optically observing characters such as Chinese characters written on paper or the like, 2 is a feature extracting section for extracting features of input characters based on the observation data from the observing section 1, 3 Is a recognition dictionary in which standard feature patterns of characters to be recognized are stored in advance, and 4 is a feature pattern of the input characters transferred from the feature extraction unit 2 and the standard feature pattern in the recognition dictionary 3 are collated. A collation identification unit that identifies the characters in the recognition dictionary 3 as input characters in order from the highest similarity, 5
Represents each candidate character (for example, 5
Character), a candidate storage memory for storing in descending order of similarity, 6
Is a kana-kanji conversion dictionary in which the correction character group corresponding to each kana input and the number of the correction character group are stored in advance, 7 is a correction character storage table for storing the correction character group corresponding to a predetermined kana input, 8 is a display unit, 9 is a display output unit for displaying the first rank candidate character in the candidate storage memory 5 on the display unit 8, and a correction character group corresponding to a predetermined Kana input is the Kana-Kanji conversion dictionary 6 After that, the correction character group is read and stored in the correction character storage table 7, and then the arrangement order of the correction character group in the correction character storage table 7 is changed according to a predetermined rule. A control unit 10 is provided with a character correction unit for outputting to the display unit 8 in order from the one, and a keyboard device 10 for giving a predetermined key input such as kana input to the control unit 9.

In this embodiment, the character correction unit of the control unit 9 is the second
It operates according to the flowchart shown in the figure. Now, as a result of observing a certain input character, as shown in FIGS. 1 and 3 (a), a candidate character group ranked in the candidate storage memory 5 (displayed as CAN in FIGS. 2 and 3) It is assumed that A1 to AN are stored and that the first-ranked candidate character A1 as the recognition result is different from the input character.

In this case, as shown in FIG. 1, when a kana input corresponding to an input character is given from the keyboard device 10 (FIG. 2), the character correction section causes the correction character group B1 corresponding to the kana input. To BM and the number M thereof are read from the Kana-Kanji conversion dictionary 6 (Fig. 2), and the Kana-Kanji conversion dictionary 6 is read.
The correction character groups B1 to BM are stored in the correction character storage table 7 (indicated as T in FIGS. 2 and 3) with the arrangement of (see FIG. 3B).

After that, the character correction unit refers to the candidate character groups A1 to AN and refers to the candidate character groups A1 to AN to correct the character groups for correction B1 to B1 in the corrected character storage table 7.
Change the sequence order of BM. At this time, the character correction unit is provided with a pointer P for supporting the candidate ranks of the candidate characters A1 to AN.
1. A pointer P2 for instructing the arrangement order of the correction character groups B1 to BM before the change and a pointer P for instructing the advance order of the correction characters are provided, and the pointer P is 1
The pointers P1 and P2 are initialized to 0 respectively (FIG. 2). In this state, the character correction unit first sets the pointer P1 to 1 and the pointer P2 to 1, and then the candidate character CAN (P1) supported by the pointer P1 is corrected by the correction character T (supported by the pointer P2. P2), and if they do not match, the pointer P2 is sequentially updated from 1 to M while the pointer P1 is set to 1, and the pointer P1 is changed from 2 to N. While updating, the pointer P2 is sequentially updated from 1 to M, and the above determination operation is repeated (FIG. 2). In the determination operation process, when the candidate character CAN (P1) supported by the pointer P1 first matches the correction character T (P2) pointed by the pointer P2, the character correction unit determines As shown in FIG. 3 (c), after the correction character T (P2) designated by the pointer P2 is transferred by the work area Wk (FIG. 2), the pointer P2 is replaced with the pointer P3 and the pointer P3 is replaced. Is updated in the negative direction, the order of arrangement of each correction character arranged higher than the correction character T (P2) in the correction character storage table 7 is lowered by one, and the arrangement indicated by the pointer P is performed. The storage area of the correction character storage table 7 corresponding to the rank (first rank) is set to a blank state (FIG. 2), and then the correction character T (P2) stored in the work area Wk is set to T.
(P) is stored in the blank storage area (FIG. 2). After that, the character correction unit updates the pointer P (FIG. 2), and thereafter repeats the same determination operation,
Candidate character CAN (P1) indicated by pointer P1 is pointer P
When it matches with the correction character T (P2) designated by 2, the correction character T is stored in the correction character storage table 7.
(P2) is moved up to the array order pointed by the pointer P.

Upon completion of such a series of operations, the character correction unit displays the corrected character group of the corrected character storage table 7 on the display unit 8 in accordance with the new arrangement order (FIG. 2), and the correct corrected character displayed is displayed. Is selected by the key operation of the keyboard device 10, the correction character is displayed again on the display section 8 in place of the erroneously recognized character A1 (FIG. 2). At this stage, the correction of the erroneously recognized character A1 is completed.

Next, the character correction process as described above will be described with a specific example.

Now, assume that the recognition result of the character recognition device for the input form as shown in FIG. 4 (a) is obtained as shown in FIG. 4 (b). In this case, since the observation character “unusual” is erroneously recognized as “love” and the observation character “thing” is erroneously recognized as “brush”, the operator indicates these erroneously recognized characters with the keyboard device 10, You have to correct these letters.

First, in correcting the erroneously recognized character “love”, the operator uses the keyboard device 10 to key in “Hen” as the reading of the correct character “Hen”. Then, as shown in FIG. 5 (b), the character correction unit of the control unit 9 uses the "Kana, Kanji, Kanji, Kanji, Kanji, Kanji, Kanji" dictionary 7
The characters are read out, and the above-mentioned correction character group is stored in the correction character storage table 7 with this arrangement order. Next, the character correction unit causes the candidate characters in the candidate storage memory 5 shown in FIG. With reference to, the “unusual” existing in the candidate character is moved up to the top of the modified character storage table 7, and the modification character group in the modified character storage table 7 is defined as “unusual, piece, side, return, bias, Um,
Then, the correction character groups are displayed in order or collectively on the display unit 8 in accordance with the arrangement order. At this stage, if the operator selects the corrective character "Imperial" that matches the correct character by operating the keys of the keyboard device 10, the character "Immediately" is redisplayed on the display unit 8 instead of the misrecognized character "Koi". , Character correction is complete. At this time, since the correction character "Hen" that matches the correct answer character is positioned at the top of the correction character group, the operator can relatively easily select the correction character.

Next, in correcting the erroneously recognized character "brush", the operator inputs "koto" as a reading of the correct character "thing" by the keyboard device. Then, as shown in FIG. 6 (b), the character correction unit of the control unit 9 reads out the four characters of the collation identification unit of "Koto, Koto, Knot, Oto" from the Kana-Kanji conversion dictionary 6 and arranges them in this array. The rank is stored in the corrected character storage table 7 as it is. Next, the character correction unit is used for the candidate character in the candidate storage memory 5 shown in FIG. Is attempted to change the arrangement of the correction character group in the correction character storage table 7, but since the correction character does not exist in the candidate character group, the correction character portion is The display unit 8 displays the correction character groups in order or collectively according to the order of arrangement without changing the arrangement.
To display. At this stage, when the operator selects the correction character “thing” that matches the correct answer character by the key operation of the keyboard device 10, the character “thing” is displayed again on the display unit 8 instead of the misrecognized character “brush”. Then, the characters are corrected.

In the above embodiment, the reading of the character is used as the information for identifying the correct answer character, but the number of strokes of the character may be used. Further, the specific configuration of the character correction unit is not limited to that shown in the above embodiment,
Of course, the design can be changed as appropriate.

〔The invention's effect〕

As described above, according to the character correction method in the character recognition device according to the present invention, when correcting an erroneously recognized character, the correct character is selected from the correction character group based on the information for identifying the correct character. I arranged the ones with a high probability to be higher in the correction character group and selected the correction character that matches the correct answer character from this correction character group.
Not only can you correct the characters, but tentatively,
Even if the number of characters in the correction character group is large, it is possible to select the correct character from the character group relatively easily, and the character correction speed can be improved accordingly.

[Brief description of drawings]

FIG. 1 is a block diagram showing an embodiment of a character recognition device for carrying out the character correction method according to the present invention, FIG. 2 is a flowchart showing the operation of a control unit according to the embodiment, and FIG. 3 (a). Is a schematic diagram showing the operation of the control unit for the candidate storage memory according to the embodiment, FIGS. 3B and 3C are schematic diagrams showing the operation of the control unit for the corrected character storage table according to the embodiment, and FIG. FIG. 5 (b) is an explanatory view showing an example of an input form including observation characters and an example of a recognition result thereof, FIG. 5 (a) is an explanatory view showing an example of contents of a candidate storage memory for the observation character “unusual”, and FIG.
Is an explanatory diagram showing an example of the contents of the corrected character storage table for reading the kana of the observation character "unusual". Fig. 5 (c) is an explanatory diagram showing an example of the contents of the corrected character storage table controlled by the control unit. (A), (b) and (c) are observation characters "things"
It is explanatory drawing corresponded to FIG. 5 (a) (b) and (c) with respect to FIG. 1 ... Observation unit 2 ... Feature extraction unit 3 ... Recognition dictionary 4 ... Collation identification unit 5 ... Candidate storage memory 6 ... Kana-Kanji conversion dictionary 7 ... Corrected character storage table 8 ... Display unit 9 ... Control unit 10: Keyboard device

Claims (1)

[Claims] 1. A plurality of candidate characters ranked in descending order of similarity to observed characters are selected from a recognition dictionary, and the first candidate character of the candidate character group is displayed as a recognition result on a display unit. In the character recognition device thus configured, when correcting the display character that is erroneously recognized, a correction character group including the correct answer character is selected from a predetermined dictionary based on the information for identifying the correct answer character, and When the above-mentioned candidate character is included in the correction character group, after the correction character matching the candidate character is arranged above the other correction character group according to the order of the above-mentioned candidate character group, It is characterized in that a correction character that matches the correct answer character is selected from the correction character group having a new arrangement rank, and the correction character is replaced with the misrecognized character and redisplayed on the display unit. Character recognition device Definitive character correction method.