JPH08263595A - Device and method for character recognition - Google Patents

Abstract

PURPOSE: To provide the device and method for character recognition which can specify one standard pattern even when inputted characters, symbols, etc., are large in the degree of deformation. CONSTITUTION: An optical character reader 1 (OCR1) which is connected to a terminal 1a for control is equipped with a recognition part 7 which recognizes a read pattern and outputs the recognition result to the terminal 1a for control and a dictionary memory 9 which stores an unillustrated dictionary. The recognition part 7 is equipped with a memory 7a which contains an unillustrated ranked character set table. In the dictionary, ranked categories are registered, and one standard pattern is set for each category. A category selected out of the dictionary is set in the character set table through the input part 1d of the terminal 1a for control.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a character recognition device and a character recognition method for collating a reading pattern obtained by reading characters / symbols with a standard pattern registered in a dictionary memory to identify a standard pattern matching the reading pattern. Regarding

[0002]

2. Description of the Related Art Conventionally, an optical character reading device has a dictionary memory, and a plurality of standard patterns are registered in the dictionary memory. These standard patterns are collated with the reading patterns of handwritten characters and symbols input to the device, and if they match the reading patterns, the matching standard patterns are output to the control terminal connected to the device and displayed on the screen. . However, when the degree of deformation of the read pattern is large and there are a plurality of standard patterns that match the read pattern, one of the standard patterns cannot be selected on the device side, that is, the standard pattern cannot be specified as one. It was outputting to the control terminal.

By the way, depending on the place where such an optical character reader is used, the characters and symbols to be input are limited, and therefore the standard pattern of the dictionary memory to be collated with the read pattern is also limited. It was That is, when the read pattern is only alphanumeric characters, the standard patterns such as kanji, hiragana, katakana, etc. stored in the dictionary memory are not used in the collation. Therefore, a category that identifies individual characters / symbols, etc. is attached to the standard pattern, and only the categories of a plurality of alphanumeric standard patterns are preset in the device as a character set. When the category is collated with the category of the character set and one category of the standard pattern that matches the category of the character set can be specified, this standard pattern is recognized and the unread rate is reduced.

[0004]

In the conventional optical character reader, the character set is set to reduce the non-reading rate. However, even if the character set is set, the degree of deformation of the reading pattern is reduced. In some cases, the standard pattern cannot be specified as one when the value is large. That is, the device side cannot recognize the reading pattern, and as a result, the operator has to input to the optical character reading device using the control terminal, etc., and if there are many unreads, it takes time to input and the whole Took a long time to process.

[0005]

In order to solve the above-mentioned problems, according to the present invention, a plurality of standard patterns of a plurality of forms are registered corresponding to categories for specifying individual characters, symbols, etc. A dictionary memory in which standard patterns are divided into a plurality of ranks according to the degree of deformation and registered, a character set in which a category to be recognized in each rank is set, and a predetermined rank The rank selecting means for selecting, the matching means for matching the reading pattern and the standard pattern for each selected rank, and the character set corresponding to the selected rank are referred to, and the category of the standard pattern matching the reading pattern is selected. The first determining means for determining whether or not the character set is set, and the reading pattern match with the character set, Reference pattern Li has been set, the determines one or not in ranks selected 2
And a character / symbol or the like based on the standard pattern category identified by the second determining means.

[0006]

The reading pattern is collated with the standard pattern in the dictionary memory. When the standard pattern matches the reading pattern, the rank selecting means selects the rank of the character set corresponding to the rank of the standard pattern, and the matching means matches the rank. The category of the standard pattern is compared with the category of the character set of the rank selected by the rank selecting means. Then, the first determining means can determine whether the category of the standard pattern and the character set category match, and the second determining means can determine that this standard pattern is one of the selected ranks. Characters and symbols are recognized based on the category of this standard pattern.

[0007]

Embodiments of the present invention will now be described in detail with reference to the drawings. In the present embodiment, as the character recognition device, an optical character reading device to which a control terminal is connected will be described as an example. The elements common to the drawings are given the same reference numerals. FIG. 1 is a schematic configuration diagram showing an optical character reading device according to an embodiment of the present invention and a control terminal connected thereto, FIG. 2 is a configuration diagram of a dictionary memory of the optical character reading device according to the embodiment, and FIG.
FIG. 3 is a configuration diagram of a memory provided in a recognition unit of the optical character reading device according to the embodiment.

First Embodiment A first embodiment will be described with reference to FIGS.

An optical character reader 1 (hereinafter referred to as "
OCR1) is connected to the control terminal 1a,
Information such as characters and symbols read by the OCR 1 is output to the control terminal 1a. The control terminal 1a has a control unit 1b, and the control unit 1b includes a display unit 1c and an input unit 1d.
Is connected.

The OCR 1 is a light source (not shown) that illuminates the surface of the medium 2 such as a form, a lens 3 that collects the reflected light from the surface of the medium 2, and a CCD that photoelectrically converts the light collected by the lens 3. It has a sensor 4. An image memory 5, a cutout unit 6, and a recognition unit 7 are sequentially connected to the CCD sensor 4. The image memory 5 captures image data (hereinafter referred to as a read pattern) which is photoelectrically converted by the CCD sensor 4 and becomes an electric signal, and the cutout unit 6 cuts out the read pattern captured in the image memory 5 character by character and recognizes it. Output to the unit 7. The recognition unit 7 recognizes the read pattern and outputs the recognition result to the control terminal 1a.
The control unit 1b of b displays the recognition result on the display unit 1c. The recognition unit 7 includes a memory 7a, and the recognition unit 7
A control unit 8 for OCR and a dictionary memory 9 are connected to. A dictionary 9a having the structure shown in FIG. 2 is registered in the dictionary memory 9.

The dictionary 9a has a structure in which a plurality of categories 1 to n (n> 0) indicating codes such as characters and symbols are ranked from 0 to N (N> 0) according to the degree of transformation. The higher the rank value, the greater the degree of deformation of the character. A plurality of corresponding standard patterns are set in each category, and the recognition unit 7 collates the standard pattern with the reading pattern photoelectrically converted by the CCD sensor 4 to read the characters and symbols read. recognize.

A character set table 7b is stored in the memory 7a shown in FIG. The character set table 7b has a character set provided corresponding to each rank of the dictionary 9a. These character sets are a set of categories shown in the dictionary 9a,
By comparing the character set of the rank selected by the recognition unit 7 with the category of the standard pattern, the recognized characters / symbols are limited. The character set table 7b is displayed on the input unit 1 of the control terminal 1a by the operator.
It is set via d. That is, the categories and standard patterns of the dictionary 9a are displayed on the display unit 1c, and the operator selects, for example, a category of a standard pattern that is frequently used, and the selected category has the same character set table 7b as the rank to which the category belongs. Store in rank. The above-mentioned category is selected when, for example, alphanumeric characters are written on the medium 2, and particularly when a large number of numbers are entered, it is expected that the number of input numbers will be deformed. Numerical categories will be stored in the higher ranks of. It should be noted that the operator is not limited to such a case, and is set by the operator according to the usage environment of the medium 2, the actual erroneous reading or the unread status, and the like.

Next, the character recognition operation of the first embodiment will be described with reference to FIGS. 1, 4, 5, 6, and 7. 4 is a diagram showing an example of read characters of the embodiment, FIG. 5 is a diagram showing an example of a dictionary of the embodiment, FIG. 6 is a memory configuration diagram of the first embodiment with respect to the dictionary example of FIG. 5, and FIG. 7 is a first embodiment. It is a flowchart which shows the example character recognition operation. In the first embodiment, the characters and numbers read by the OCR 1 are 4 of "A, B, 4, 8" shown in FIG.
Describe the type with alphanumeric characters. Therefore, as shown in FIGS. 5 and 6, the categories of the dictionary 9a and the respective character sets are
Only categories corresponding to “A, B, 4, 8” in FIG. 4 are shown, and other categories are omitted. Further, the allowable range of the degree of deformation of the characters to be read is up to rank 2. Therefore, the rank of the standard pattern of the dictionary 9a is rank 0 to rank 2, and the other ranks are omitted.

First, the light source is "A, B, 4, 8" of the medium 2.
The part is irradiated with light and the reading operation is started. This light is reflected by the medium 2, condensed by the lens 3, sent to the CCD sensor 4, and photoelectrically converted. Then, the patterns corresponding to “A, B, 4, 8” which are photoelectrically converted into electric signals are sent to the image memory 5. The cutout unit 6 cuts out one pattern from the read pattern captured in the image memory 5 and outputs it to the recognition unit 7. When one pattern is input, the recognizing unit 7 starts collation between the read pattern input in step S1 and the standard pattern of rank 0 in the dictionary memory 9. At this time, the size of the rank to be matched is checked in step S2. That is, the rank to be collated is compared with the largest rank N in the dictionary 9a. Since the size of the rank to be checked now is "0", the process proceeds to step S3. In step S3, the read pattern is compared with the standard pattern of rank 0 one by one. Then, in step S4, when there is a standard pattern that matches the read pattern, the recognition unit 7 assigns a rank having the same size as the rank of the matching standard pattern to the character set table 7
Select from b, and proceed to step S5. Step S5
The recognition unit 7 then recognizes the category of the standard pattern of rank 0 that matches the read pattern, and the character set table 7b.
When it is determined that the category that matches the category of the dictionary 9a is in the character set by collating with the character set of rank 0 of, the process proceeds to step S6. And the recognition unit 7
Determines that there are two or more standard patterns that match the read pattern, and that each character set has a category that matches the category of each standard pattern, step S
When it is determined that the character set has one or two or more standard patterns but the category that matches the category of the standard pattern is only one, the process proceeds to step S8.

When the process proceeds to step S8, the recognition unit 7 determines that the standard pattern matching the read pattern has been specified, and sends the category of this standard pattern to the control unit 8. The control unit 8 outputs this category to the control unit 1b of the control terminal 1a, and when the control terminal 1b inputs the category, the control unit 1b displays characters or numbers corresponding to this category on the display unit 1c.

Here, the operation when the standard pattern that matches the read pattern cannot be specified at rank 0, that is, when the process proceeds from step S6 to step S7, will be described by taking the numeral "4" shown in FIG. 4 as an example. The degree of deformation of the number "4" is rank 1.

In step S7, the recognition section 7 collates the standard pattern of rank 1 next to rank 0 with the read pattern. At this time, it is checked in step S2 whether or not the size of the rank to be matched is N or less. Since the size of the rank to be checked now is "1", step S3
Go to. In step S3, the recognition unit 7 compares the read pattern with the standard pattern of rank 1 one by one.
The read pattern matches the standard pattern of the category “4”, and therefore, in step S4, the recognition unit 7 determines that there is a match and proceeds to step S5. In step S5, the recognition unit 7 collates the category "4" with the categories "4" and "8" of rank 1 of the character set table 7b. The category "4" matches the category "4" of the character set table 7b, and therefore the process proceeds to step S6. Since there is one category in the character set table 7b that matches the category, the recognition unit 7 determines in step S8 that the standard pattern that matches the read pattern has been identified, and outputs category "4" to the control terminal 1a. To do.

The above-mentioned character recognition operation has been described in the case where the deformation degree of the numeral "4" is rank 1, but when the deformation degree is rank 2 or more, the operation from step S2 to step S7 is repeated to perform the character recognition operation. Is done.

Next, the character recognition operation when there are two or more standard patterns in the dictionary 9a that match the read pattern will be described by taking the numeral "8" shown in FIG. 4 as an example. The deformation degree of the number "8" is rank 2. Since the operation from step S1 is the same as the above-mentioned operation, the operation of matching the read pattern with the standard pattern of rank 2 of the dictionary 9a will be described.

In step S7, the recognition section 7 collates the standard pattern of the rank 2 next to the rank 1 with the read pattern. At this time, it is checked in step S2 whether or not the size of the rank to be matched is N or less. Since the size of the rank to be checked now is "2", the process proceeds to step S3. In step S3, the recognition unit 7 collates the read pattern with the standard pattern of rank 2 one by one. The read patterns match the two categories "B" and "8". Therefore, in step S4, the recognition unit 7 determines that there is a match and proceeds to step S5. In step S5, the recognition unit 7 collates the categories "B" and "8" with the category "8" of rank 2 in the character set table 7b. The category "B" does not match the category "8" in the character set table 7b, and the category "8" matches the category "8" in the character set table 7b. Therefore, the process proceeds to step S6. Since there is one category in the character set table 7b, the recognition unit 7 determines that the standard pattern that matches the read pattern has been identified, and the step S
Proceed to 8. Since the operation of step S8 is the same as that described above, description thereof will be omitted.

By the way, when the deformation of the read pattern is so large that it does not match any of the standard patterns of ranks 0 to N, or the match is found but the rank of the dictionary 9a that matches but the rank of the character set table 7b corresponding to the category. If there is not, that is, if the size of the rank to be matched in step S2 is larger than N, the recognition unit 7 determines that it is unreadable in step S9.

In the first embodiment, the operator examines the predicted degree of deformation of the read pattern in advance from past data and the like, and further determines the allowable range of deformation based on the result of the examination. By setting the category of the rank of the dictionary 9a corresponding to the range of deformation as the character set in association with the rank of the dictionary 9a, it is possible to read a reading pattern having a large degree of deformation. Further, since the character set can be set for each rank based on past misreading or unreading data, misreading or unreading of similar characters and numbers such as "B" and "8" can be prevented. .

Second Embodiment Next, a second embodiment will be described with reference to FIGS. 1, 4, 5, 8 and 9. FIG. 8 is a block diagram of the memory of the second embodiment, and FIG. 9 is a flow chart showing the character recognition operation of the second embodiment.

In the first embodiment, all categories to be recognized are set in the character set for each rank, but generally, the category set in a certain rank is smaller than the set rank (lower rank). Since the rank is also set, in the second embodiment, in a category set to two or more ranks, the category is set to the highest rank, which is the highest numerical value. I have to. That is, as shown in FIG. 8, category "4" set to rank 0 and rank 1 in the first embodiment.
Will be set only to the highest rank 1 of the two character sets of rank 0 and rank 1, and the category “8” set to rank 0, rank 1 and rank 2 in the first embodiment. Will be set only to the highest rank 2 of the three character sets.

In the second embodiment, the letters and numbers read by the OCR 1 are shown in FIG.
B, 4, 8 ”will be described. Further, the allowable range of the degree of deformation of the characters to be read is up to rank 2. Therefore, the rank of the standard pattern of the dictionary 9a is rank 0 to rank 2, and the other ranks are omitted.

Since the other structure is the same as that of the first embodiment, its explanation is omitted.

Next, the character recognition operation of the second embodiment will be described. In the second embodiment, the recognition unit 7 determines the category of the dictionary 9a that matches the read pattern as the character set table 7
When the category of the character set table 70b that matches the category of the dictionary 9a can be identified by comparing with all the categories of b, the rank size of the category of the dictionary 9a is compared with the rank size of the category of the character set table 70b. If the category rank size is equal to or smaller than the category rank size of the character set table 70b, it is determined that the standard pattern that matches the read pattern has been identified. Since the operation of photoelectrically converting the characters on the medium 2 by the sensor 4 and sending one pattern to the recognition unit 7 is the same as that of the first embodiment, the description thereof will be omitted.

First, the operation of recognizing the numeral "4" among the characters of the medium 2 shown in FIG. 4 will be described as an example. Number "4"
Degree of deformation is rank 1. Therefore, since the operations from step S1 to step S4 and from step S4 to step S7 are the same as those in the first embodiment, the description thereof will be omitted, and the operation of comparing the size of rank 1 in step S2 will be described.

Since the size of the rank to be checked in step S2 is "1", the process proceeds to step S3. Step S
In 3, the recognition unit 7 collates the read pattern and the standard pattern of rank 1 one by one. The read pattern matches the standard pattern of category "4", so step S
In step 4, the recognition unit 7 determines that there is a match, and proceeds to step S50. In step S50, the recognition unit 7 determines that the category is "4",
The character set table 70b is compared with each rank category. The category "4" matches the category "4" of rank 1 of the character set table 70b, and therefore the process proceeds to step S60, where the recognition unit 7 determines the size of rank 1 of the category "4" and the character set table 70b.
The size of the rank 1 of the category “4” is compared. Since both ranks have a size of 1, the process proceeds to step S6.
Since there is one category in the character set table 70b that matches the category, the process proceeds to step S8, and in step S8, the recognition unit 7 determines that the standard pattern that matches the read pattern can be identified, and the category “4” is assigned to the control terminal. Output to 1a.

Here, a case where the deformation degree of the numeral "4" on the medium 2 is rank 0 will be described. Step S1
The operations from to step S4 are the same as those in the first embodiment, and therefore the description thereof will be omitted and the operation will be described from step S50. In step S50, the recognition unit 7 matches the category “4” of the dictionary 9a with the category of each rank of the character set table 70b. Since the category “4” matches the category “4” of rank 1 of the character set table 70b, the process proceeds to step S60, and the recognition unit 7 determines the size of rank 0 of the category “4” and the category “4” of the character set table 70b. The size of Rank 1 is compared. Dictionary 9
Since the rank on the a side is smaller than the rank on the character set table 70b side, the process proceeds to step S6. Since there is one category in the character set table 70b that matches the category, the process proceeds to step S8. Since the operation of step S8 is the same as that described above, description thereof will be omitted.

Next, the dictionary 9 that matches the reading pattern
The character recognition operation when there are two or more standard patterns of a will be described by taking the numeral "8" shown in FIG. 4 as an example. Number "8"
The degree of deformation of is rank 2. Since the operation from step S1 is the same as the above-mentioned operation, the operation of moving from rank 1 to rank 2 of the dictionary 9a in step S7 will be described.

In step S7, the recognition section 7 collates the standard pattern of the rank 2 next to the rank 1 with the read pattern. At this time, it is checked in step S2 whether or not the size of the rank to be matched is N or less. Since the size of the rank to be matched is "2", the process proceeds to step S3, and the read pattern is matched with the standard pattern of rank 2 one by one. The read patterns match with the two categories "B" and "8". Therefore, in step S4, the recognition unit 7 determines that there is a match, and proceeds to step S50. In step S50, the recognition unit 7 matches the categories "B" and "8" with all the categories in the character set table 7b. Category "B" is the character set table 70b
Matches the category "B" of rank 0, and the category "8"
Matches the category “8” of rank 2 in the character set table 70b. Therefore, the process proceeds to step S60.

In step S60, the recognizing unit 7 compares the rank sizes of the category "B" and the category "B". The rank on the dictionary 9a side is the character set table 7
Although the rank is larger than the rank on the b side, the rank of the dictionary 9a side is larger when the rank size of the category “8” is compared with the rank size of the category “8” of the character set table 70b, so the process proceeds to step S6. Since there is one category in the character set table 70b that matches the category,
The recognition unit 7 determines that the standard pattern that matches the read pattern has been identified, and proceeds to step S8.

By the way, when the deformation of the read pattern is so large that it does not match any of the standard patterns of ranks 0 to N, or the match is found, but the rank of the dictionary 9a that matches but the rank of the character set table 7b corresponding to the category is classified. If there is not, that is, if the size of the rank to be matched in step S2 is larger than N, the first
Similar to the embodiment, the recognition unit 7 determines non-reading in step S9.

In the second embodiment, the categories of the character set table 70b which are set to two or more ranks are:
Since the category is set to the highest rank to be set, it is not necessary for the operator to set all categories to each rank when setting the character set table 70b in the memory 7a. Therefore, the setting of the character set table 7b becomes simple and easy. Further, since the number of categories set in the character set table 70b is smaller than that in the first embodiment, the capacity of the memory 70b can be made smaller than that in the first embodiment.

In the present embodiment, the characters / symbols to be read are greatly deformed, and when the recognition unit 7 recognizes the read pattern, the characters / symbols that may be erroneously recognized should not be recognized. It is also possible to set a character set for each rank in the character set table so as to make it unreadable.

In the present embodiment, the optical character reading device has been described by taking an example in which a dictionary is formed for each rank when setting categories in the recognition process, but another recognition having a standard pattern for each rank is described. This embodiment can be applied to a system having a recognition process of voice, fingerprint, image, etc. as a system.

[0038]

As described in detail above, the present invention provides
A dictionary memory in which standard patterns in a plurality of forms are registered in correspondence with categories that specify individual characters, symbols, etc., and the entire standard pattern is ranked and registered according to the degree of deformation, , A character set which is provided corresponding to each rank and in which a category to be recognized in the rank is set, a rank selecting means for selecting a predetermined rank, and a reading pattern and a standard pattern for each selected rank. A collating means for collating, a first judging means for judging whether or not a category of a standard pattern matching the reading pattern is set in the character set by referring to a character set corresponding to the selected rank; A standard pattern that matches the pattern and whose category is set in the character set is 1 in the selected rank. The character recognizing device is provided with a second determining means for determining whether or not the read pattern is deformed by recognizing a character, a symbol, or the like based on the category of the standard pattern specified by the second determining means. Even if the degree is large, it is possible to specify one standard pattern. Therefore, the occurrence of non-reading is eliminated, and the labor of the operator's input is saved and the entire processing time can be shortened.

[Brief description of drawings]

FIG. 1 is a schematic configuration diagram showing an optical character reading device according to an embodiment of the present invention.

FIG. 2 is a configuration diagram of a dictionary memory of the optical character reading device according to the embodiment.

FIG. 3 is a configuration diagram of a memory according to an embodiment.

FIG. 4 is a diagram showing an example of read characters in the embodiment.

FIG. 5 is a diagram showing an example of a dictionary of an embodiment.

FIG. 6 is a configuration diagram of a memory according to the first embodiment.

FIG. 7 is a flowchart showing a character recognition operation of the first embodiment.

FIG. 8 is a configuration diagram of a memory according to a second embodiment.

FIG. 9 is a flowchart showing a character recognition operation of the second embodiment.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Optical character reader 8, 1b Control part 1d Input part 7 Recognition part 7a Memory 7b, 70b Character set table 9 Dictionary memory 9a Dictionary

Claims (4)

[Claims] 1. A character recognition device for recognizing characters / symbols by reading the characters / symbols and collating the read pattern with a standard pattern registered in a dictionary memory to identify each character / symbol, etc. Multiple types of standard patterns are registered for each category, and a dictionary memory in which the entire standard pattern is classified into multiple ranks according to the degree of deformation and registered is provided for each rank. The character set in which the category to be recognized by the rank is set, the rank selecting means for selecting a predetermined rank, the matching means for matching the reading pattern and the standard pattern for each selected rank, and the selected Whether the category of the standard pattern that matches the read pattern is set in the character set by referring to the character set corresponding to the rank And a second determination means for determining whether or not the standard pattern that matches the reading pattern and has the category set in the character set is one of the selected ranks. Means for recognizing characters, symbols, etc. based on the category of the standard pattern identified by the second determining means. 2. The character recognition device according to claim 1, wherein the rank selecting means selects the ranks in order from a rank having a smaller degree of deformation. 3. A character recognition device for reading characters / symbols, etc. and recognizing the characters / symbols etc. by collating this read pattern with a standard pattern registered in a dictionary memory to specify individual characters / symbols etc. Multiple types of standard patterns are registered for each category, and a dictionary memory in which the entire standard pattern is classified into multiple ranks according to the degree of deformation and registered is provided for each rank. The character set in which the category to be recognized in the rank is set so that the category set in the higher rank having a large degree of deformation is not set in the lower rank having a smaller degree of deformation than the rank. Rank selection means for selecting ranks, matching means for matching the read pattern with the standard pattern for each selected rank, and characters for all ranks. First determining means for determining whether or not the category of the standard pattern that matches the read pattern is set in the character set, and the character set in which the category of the standard pattern that matches the read pattern is set And a second rank for determining whether or not the rank selected at that time is the same rank as the rank detected by the rank detection means or a higher rank having a larger degree of deformation than that. And a third determining means for determining whether or not the standard pattern determined to be the same or higher rank by the second determining means is one in the selected ranks. A character recognition device characterized by recognizing characters, symbols, etc. based on the category of the standard pattern specified by the determination means of No. 3. 4. A display unit for displaying the categories registered in the dictionary memory, an input unit for inputting a desired category from the categories displayed on the display unit, and a category input from the input unit. The character recognition device according to claim 1 or 3, further comprising a processing unit for registering the character set in accordance with the rank.