JP2015022520A - Business form reader and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To improve the accuracy of reading when characters are read from a business form in which fields where the characters to be read are printed are fixed in advance.SOLUTION: A business form reader of an embodiment has a first character recognition unit, a re-recognition field determination unit, and a second character recognition unit. The character recognition unit character-recognizes an image provided in a field of a tabular format business form in which a plurality of fields correlated for each item are provided. The re-recognition field determination unit determines, on the basis of the character attribute of the image character-recognized by the first character recognition unit, a re-recognition field including an image having a character attribute different from other characters in the plurality of fields. The second character recognition unit obtains, with respect to the re-recognition field determined by the re-recognition field determination unit, a character recognition result in which unclear characters are excluded by character-recognizing on the basis of the character attribute.

Description

  Embodiments described herein relate generally to a form reading apparatus and a program.

  For ordering and receiving orders for merchandise, a form with a unified form format (format) such as a field (field) for printing a product name and product code called a unified slip is used. An example of a unified slip is a purchase slip. For such unified slip processing, a form reading device that uses a scanner to read a form image and recognize characters is used.

  When the form reading device reads the characters printed on the unified slip, the form format (format) is defined in advance as the field to be read (reading range) where the product name or product code on the form is printed. By doing so, a character image is detected from a portion of the form image corresponding to the predefined reading field from the form image read by the scanner, and the character image is recognized.

Japanese Patent Laid-Open No. 9-305715

  In the unified slip, the form format of fields (fields) etc. for printing product names, product codes, etc. is unified, but it is used where the product name, product code, etc. are printed in the range of those fields (fields). It is not unified by the person.

  There are various cases, such as when a product name or product code is printed in the vicinity of the top of the corresponding column (field), or when a product name or product code is printed in the left end of the corresponding column (field).

  In addition, the number of characters (number of digits in the case of numbers) such as product names and product codes to be printed is not uniform. In the unified slip, the form format of the field (field) that prints the product name, product code, etc. is only unified. About the number of characters of the product name to be printed, the number of digits of the product code, and the characters to be printed such as the font Nothing is unified.

  When the form reading device performs character recognition of such a unified form, characters of unknown number of characters existing at unknown positions from the form image of the part corresponding to the field to be read (reading range) defined in advance. The image is detected and cut out for character recognition.

  However, in such a conventional form reading apparatus, when an unknown character (so-called noise that is not recognized as a character) straddling a field to be read or between adjacent fields is detected, the character recognition result in that field is erroneously read. It becomes unread and the reading accuracy of the character is lowered.

  For example, check marks handwritten by the checker after printing characters on the slip, characters that protrude from the definition area of the slip, etc. become noise, or the ruled lines themselves become noise because the characters printed on the slip overlap the ruled lines around the characters. As a result, misreading and rejection of characters occur, and the character reading accuracy decreases.

  In addition, since character reading is performed on all reading target fields (reading ranges) in the slip, noise existing in empty fields that are not printed may be misread as characters.

  The problem to be solved by the present invention is to provide a form reading apparatus and program capable of improving the reading accuracy when reading a character from a form in which a field on which a character to be read is printed is fixed in advance. It is in.

  The form reading apparatus according to the embodiment includes a first character recognition unit, a re-recognition field determination unit, and a second character recognition unit. The first character recognizing unit character-recognizes an image provided in a tabular form field provided with a plurality of fields associated with each item. The re-recognition field determination unit determines a re-recognition field including an image having a character attribute different from other characters in a plurality of fields based on the character attribute of the image recognized by the first character recognition unit. The second character recognition unit recognizes the character based on the attribute of the character for the re-recognition field determined by the re-recognition field determination unit.

It is a figure which shows the structure of the form reading system of embodiment. It is a figure which shows an example (form sample) of a form. It is a flowchart which shows operation | movement of a form reading system. It is a figure which shows the recognition result containing the location which became unknown by the character recognition process of the 1st step. It is a figure which shows the attribute of a character. It is a figure which shows the recognition result from which the unknown location was excluded by the two-stage character recognition process.

Hereinafter, embodiments will be described in detail with reference to the drawings. FIG. 1 is a block diagram illustrating a configuration of a form reading system in an implementation state.
As shown in FIG. 1, the form reading system of this embodiment includes a display unit 3 such as a scanner 1, a computer 2 (hereinafter referred to as “PC2”), and a monitor.

  The PC 2 includes, for example, a disk playback device as a playback device for playing back a recording medium such as a CPU, a memory, a hard disk device, a CD-ROM, and a DVD-ROM. In the PC 2, the CPU reads the control software installed in the hard disk device from the recording medium set in the disk reproducing device onto the memory, and executes the processing of the software, whereby the PC 2 functions as a character recognition device.

  The PC 2 functioning as a character recognition device includes an image storage unit 21, a form definition information storage unit 22, a first character recognition processing unit 23, a recognition information storage unit 24, a read item character information processing unit 25, and a second character recognition processing unit 26. The output unit 27 that outputs the read result to the display unit 3 is provided.

  The scanner 1 optically reads a slip 30 (see FIG. 2) as a form to be read, and stores the read image (image) in the image storage unit 21 of the PC 2.

  As shown in FIG. 2, the slip 30 is referred to as a “unified slip” used for ordering of products sold in, for example, a supermarket or a large-scale retail store, and includes a product code, a unit price, a quantity, and a total. The item names are arranged in a line horizontally, and print fields 31a to 31d, 32a to 32d, 33a to 33d, and 34a to 34d are provided in a vertical line below each item name. The print fields 31a to 31d, 32a to 32d, 33a to 33d, and 34a to 34d are character recognition target reading fields corresponding to the item names of product code, unit price, quantity, and total, respectively.

  “A000”, “B000”, and “C000” are printed in the print columns 31a to 31c. In the print fields 32a to 32c, “¥ 1,000”, “¥ 500”, and “¥ 1,500” are printed. “20”, “10”, and “120” are printed in the print columns 33a to 33c.

  In the print fields 34a to 34c, “¥ 20,000”, “¥ 5,000”, and “¥ 180,000” are printed. It is assumed that noise such as check marks 35 and 36 is handwritten (described) in the print field 31c and the print field 32d.

  The image storage unit 21 stores (temporarily saves) an image of the slip 30 read by the scanner 1. The image storage unit 21, the form definition information storage unit 22, the recognition information storage unit 24, and the like are realized by a hard disk device or the like. The form definition information storage unit 22 stores position information of a reading field that is a target of character recognition of the slip 30 as form definition information. In the case of the slip 30, the print fields 31 a to 31 d, 32 a to 32 d, 33 a to 33 d, and 34 a to 34 d are reading fields that are objects of character recognition.

  The first character recognition processing unit 23 prints the character to be read from the image stored (temporarily saved) in the image storage unit 21 with reference to the form definition information stored in the form definition information storage unit 22. The image of the reading field to be cut out is cut out, and the cut out image is recognized as characters.

  Further, the first character recognition processing unit 23 obtains character attributes from the character-recognized image. Character attributes include character size, pitch, position, character shape, shading, aspect ratio, ruled line, and the like. The first character recognition processing unit 23 stores the result of character recognition (character code), the attribute of the character, and the character recognized image in the recognition information storage unit 24 for each corresponding reading field in association with each other.

  That is, the first character recognition processing unit 23 recognizes characters in each field from an image optically read from a tabular form slip 30 having a field in which characters are printed for each item (first-stage character recognition processing).

  The read item character information processing unit 25 reads the character attributes associated with the character recognition results in all the read fields corresponding to the item names of the slip 30 from the recognition information storage unit 24, and sets the character attributes of these characters. A mutual relationship is determined, and a reading field (re-recognition field) to be recognized again is determined based on the determination result.

  The re-recognition field is a field to be read that includes a character having an attribute different from that of the other field, although it is a field to be read corresponding to the same item name. Since the re-recognition field includes characters having attributes different from those of the other fields, the character recognition result by the first character recognition processing unit 23 in the reading field determined to be the re-recognition field includes misread characters and unreadable characters. This is because there is a high possibility that

  The second character recognition processing unit 26 performs character recognition (second stage) of the re-recognition field using the correlation between the character attributes determined by the read item character information processing unit 25, and the character recognition result is displayed as the first character. The result is merged with the character recognition result by the recognition processing unit 23 and stored in the recognition information storage unit 24. That is, the character recognition result by the first character recognition processing unit 23 stored in the recognition information storage unit 24 is replaced (overwritten) with the character recognition result by the second character recognition processing unit 26. The output unit 27 outputs the character recognition result after merging by the second character recognition processing unit 26 to the display unit 3 and displays it on the screen of the display unit 3.

Next, the operation of the form reading system of this embodiment will be described with reference to FIGS.
In the case of this form reading system, when the slip 30 to be recognized set in the scanner 1 is optically read, that is, scanned (step S101 in FIG. 3), the scanned image of the slip 30 is the image of the PC 2. It is stored in the storage unit 21 (step S102).

  The first character recognition processing unit 23 specifies a reading target field for the image of the slip 30 stored in the image storage unit 21 based on the form definition information read from the form definition information storage unit 22, and for each field The first stage character recognition processing is executed in order (step S103).

  Further, the first character recognition processing unit 23 obtains character attributes from the character-recognized image. The first character recognition processing unit 23 stores the result of character recognition (character code), the attribute of the character, and the character recognized image in the recognition information storage unit 24 for each corresponding reading field in association with each other.

  Specifically, the recognition result 50 (character) as shown in FIG. 4 is obtained by the first-stage character recognition processing by the first character recognition processing unit 23. In the figure, the character recognition result in the column 51c is “C000?”, And the character recognition result in the column 52d is “?”. Here, “?” Represents an unreadable character (reject character) that could not be recognized.

The read item character information processing unit 25 reads the character attribute associated with the character recognition result in all the read fields corresponding to each item name of the slip 30 from the recognition information storage unit 24,
The mutual relationship between these character attributes is determined, and based on the determination result, a reading field (re-recognition field) to be recognized again is determined (step S104).

  The character attributes will be described with reference to FIG. The character attributes shown in FIG. 5 are examples, and this is not all. Further, it is not necessary to determine the re-recognition field using all the character attributes exemplified in FIG. The re-recognition field may be determined using the attribute of at least one of the characters.

  The attribute “shading” is shading information of the character image cut out from the background image, and can be used when the image is expressed in multiple values. The attribute “character size” is the size of the character, and includes two pieces of information of height and width. The attribute “character pitch” is a distance between adjacent characters.

  The attribute “character aspect ratio” is a ratio between the height and the width of the character. The attribute “position” is 3 in the character string (including the case where there is only one character) recognized in the reading field: the leftmost character position, the rightmost character position, and the decimal point position included in the character string. Contains one piece of information. The attribute “character shape” is a type of character font. The attribute “ruled line” is an attribute given when the recognition target character is in contact with the vertical ruled line or the horizontal ruled line.

  Here, FIG. 4 shows the result of the first character recognition processing unit 23 recognizing the slip 30 shown in FIG. Character recognition results in the reading field corresponding to the item name “product code” of the slip 30 are “A000”, “B000”, and “C000?”.

  In this case, the read item character information processing unit 25 determines from the character attributes that the left end position of the character string is the same in the character recognition results of these three read fields. On the other hand, the read item character information processing unit 25 determines that the character position (right end) of the character attribute is different only by the character recognition result of the read field corresponding to the print field 31c.

  That is, it is determined that the position of the right end of the character string is shifted from the position of the right end of the character string of the other reading field by the reject character “?”. From this, the read item character information processing unit 25 determines that the read field corresponding to the print field 31c is the re-recognition field.

  In FIG. 4, the character recognition results in the reading field corresponding to the item name “unit price” of the slip 30 are “1000”, “500”, “1500”, “?”. In this case, the read item character information processing unit 25 indicates that the character recognition result “?” Of the read field corresponding to the print field 32d is different from the character attribute and character size and character shape of the other read fields from the character attributes. Judge that From this, the read item character information processing unit 25 determines that the read field corresponding to the print field 32d is the re-recognition field.

  The second character recognition processing unit 26 recognizes again the read field corresponding to the print field 31c and the read field corresponding to the print field 32d, which the read item character information processing unit 25 determines to be the re-recognition field in the slip 30. (Step S105).

  The character or character string (text) resulting from the second-stage character recognition process is replaced (overwritten) with the character or character string at the corresponding location (corresponding column) in the first-stage character recognition result.

Then, the output unit 27 outputs to the display unit 3 a character recognition result in which the result of the second-stage character recognition process is reflected in the first-stage character recognition result (step S106).
As a result, as shown in FIG. 6, a recognition result 50 is obtained in which the noise 35 in the print field 31c and the noise 36 in the print field 32d shown in FIG. 2 are removed.

  Here, referring to FIG. 5, the second character recognition processing unit 26 performs the character recognition processing (second stage) of the re-recognition field using the correlation of the character attributes determined by the read item character information processing unit 25. Will be explained.

(Number 1) Shading The second character recognition processing unit 26 recognizes only characters whose shading of the character to be recognized is within a preset threshold range, and does not recognize characters outside the threshold range. Characters that are too dark or light that are outside the range of the light and shade threshold are excluded from recognition targets.

(Number 2) Character size The second character recognition processing unit 26 recognizes only characters in which one or both of the height and width of the character to be recognized are within a preset threshold range, and is outside the threshold range. Does not recognize characters.

(Number 3) Character pitch The second character recognition processing unit 26 recognizes only characters whose character pitch of the character string to be recognized is within a preset threshold range, and does not recognize characters outside the threshold range. .

(Number 4) Character Aspect Ratio The second character recognition processing unit 26 recognizes only characters that are within the threshold range in which the aspect ratio of the character to be recognized is set, and does not recognize characters outside the threshold range.

(No. 5) Character position The second character recognition processing unit 26 is configured such that any one of the left end position, the right end position, and the decimal point position of the character string to be recognized in the re-recognition field or a combination thereof is the same. If it is different from the position of the character string in the other reading field corresponding to the item name, only the character with the different position is not recognized.

(No. 6) Character shape of character When the character shape of the character to be recognized in the re-recognition field is different from the character shape of other characters to be recognized in the same re-recognition field, the second character recognition processing unit 26 Do not recognize characters with different glyphs. When recognizing characters, each character font has a character recognition dictionary (standard pattern) so that the character shape of the character to be recognized (difference from other character fonts) can be determined. When character recognition is performed on non-character images such as noise, the character shape cannot be determined, and thus the generation of rejected characters can be prevented by not recognizing characters.

(No. 7) Ruled Line The second character recognition processing unit 26 does not recognize the recognition target character when the recognition target character overlaps the vertical ruled line or the horizontal ruled line. Such a case in which character recognition is not performed is a case where the character string printed on the slip 30 is positioned and printed so as to overlap the ruled line, or a case where a non-character image such as noise overlaps the ruled line.

  As described above, according to this embodiment, assuming that characters printed on a slip are recognized, it is possible to avoid character recognition of non-printed images (noise) such as dirt and handwritten check marks. The generation of unnecessary reject characters can be avoided, and the character recognition accuracy can be improved. When a reject character is generated, a reject correction process is required in which the operator refers to the original slip and manually inputs the correct character in a later process. According to this embodiment, unnecessary reject correction processing can be avoided.

  Although the embodiment of the present invention has been described, this embodiment is presented as an example and is not intended to limit the scope of the invention. The novel embodiment can be implemented in various other forms, and various omissions, replacements, and changes can be made without departing from the scope of the invention. This embodiment and its modifications are included in the scope and gist of the invention, and are included in the invention described in the claims and the equivalents thereof.

  Furthermore, in the above embodiment, each component is realized by a program installed in a storage such as a hard disk device of a computer. However, the above program is stored in a computer-readable electronic medium: electronic media, and the program is stored in the electronic medium. The computer may realize the functions of the present invention by causing the computer to read the information. Examples of the electronic medium include a recording medium such as a CD-ROM and a DVD-ROM, a flash memory, a removable media, and the like. Further, the configuration may be realized by distributing and storing components in different computers connected via a network, and communicating between computers in which the components are functioning.

  DESCRIPTION OF SYMBOLS 1 ... Scanner, 2 ... PC (computer), 3 ... Display part, 21 ... Image storage part, 22 ... Form definition information storage part, 23 ... 1st character recognition part, 24 ... Recognition information storage part, 25 ... Read item character Information processing section, 26 ... second character recognition processing recognition section, 27 ... output section, 30 ... slip, 31a to 31d, 32a to 32d, 33a to 33d, 34a to 34d ... field, 35, 36 ... noise, 50 ... recognition Results (table), 51, 52... Columns including unknown characters as recognition results, 51a, 52b... Columns where unknown characters are excluded as recognition results.

Claims (2)

A first character recognition unit for character recognition of an image provided in the field of a tabular form provided with a plurality of fields associated with each item;
A re-recognition field determination unit that determines a re-recognition field including an image having a character attribute different from other characters in the plurality of fields based on the character attribute of the image recognized by the first character recognition unit; ,
A second character recognition unit that recognizes characters based on attributes of the characters for the re-recognition field determined by the re-recognition field determination unit;
A character recognition device comprising: In a program that causes a computer to execute processing,
The computer,
A first character recognition unit for character recognition of an image provided in the field of a tabular form provided with a plurality of fields associated with each item;
A re-recognition field determination unit that determines a re-recognition field including an image having a character attribute different from other characters in the plurality of fields based on the character attribute of the image recognized by the first character recognition unit; ,
A program that causes the re-recognition field determined by the re-recognition field determination unit to function as a second character recognition unit that recognizes characters based on the character attributes.

Images