JP2000029983A - Document reader device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a document reader device for accurately reading a non-standard slip in which plural item names or data are described in one character frame. SOLUTION: The image data of a slip read by an image reader 21 are stored in an image memory 22, and read to a ruled line extracting pat 23 so that the ruled lines can be extracted, and plural character frames are extracted by a character frame extracting part 24. Image data such as character strings in the character frame are recognized by a character recognizing part 25, and an item description frame in which item names are described is detected by an item frame detecting part 27 based on an item name dictionary 29. Also, a data description frame in the same row as the item description frame is detected by a data frame detecting part 28. The item description frame is made to correspond to the data description frame by a data identifying part 30, and when plural lines or plural character strings divided by division codes are present, the item names and the data are made to correspond to each other and identified in the division order.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] 1. Field of the Invention [0002] The present invention relates to a document reading apparatus for reading characters and the like written on a document, and more particularly to a technique for reading an irregular form in which the positions of read items are not fixed.

[0002]

2. Description of the Related Art Forms are classified into standard forms and non-standard forms. In the standard form, for example, the position of the item to be described is printed in advance, such as a `` transfer request form '' provided at the bank counter, and the specified data is described in the predetermined position. It is a form that is. In reading such a fixed form, a predetermined position of the form (for example,
The form code of the form is pre-printed on the upper right, etc., and the position of the data for each item can be known based on the form code. On the other hand, in the non-standard form, the items to be read are determined in advance, but the description positions of these items are different for each form. As an example of such a non-standard form, a form created by a customer in a free format, such as a purchase order for a product, can be given.

FIG. 2 is a diagram showing an example of a conventional atypical form. In this non-standard form, for example, items to be read such as "product name", "product code", and "quantity" are predetermined, but the description positions of these items are different for each form. Therefore, the form code of the form is not described, and it is necessary to identify data corresponding to each item from the data described in the form and make correspondence. For example, when reading a non-standard form as shown in FIG. 2, a conventional document reading apparatus performs reading by the following method. First, an atypical form is decomposed into black and white binary pixels by an image scanner or the like and read, and the read image data is temporarily stored in an image memory. Then, the image data stored in the image memory is scanned and read in the vertical and horizontal directions, and a histogram indicating the vertical and horizontal appearance frequencies of the black pixels is generated. Based on the generated histogram, the positions of the ruled lines are extracted, and a plurality of rectangular character frames surrounded by vertical and horizontal ruled lines are extracted.

Next, a character or a character string described in the extracted character frame is recognized and converted into a character code, and the character code is registered as an item name to be read by referring to an item name dictionary. Search whether it is. If the character or character string in the character frame is registered in the item name dictionary, the character frame is determined to be an item description frame, and if not registered, the character frame is determined to be a data description frame. In the case of the form shown in FIG. 2, based on the item name dictionary, a character frame having character strings such as “product name”, “product code”, “quantity”, and “unit price” is determined to be an item description frame. Is done. on the other hand,
Since “voice LSI”, “M9841”, “100”, “4500”, and the like are not registered in the item name dictionary, a character frame having these character strings is determined to be a data description frame. Then, it is determined whether or not a data description frame is arranged on the right or lower side of the item description frame having the character string described as “product name”. In this case, since the character frame on the right side is a data description frame, it is determined that the “voice LSI” described in the data description frame is data corresponding to the item name “product name”.

Similarly, it is determined whether or not a data description frame is arranged on the right or lower side of an item description frame having a character string described as "product code". In this case, the character frame on the right side is the item description frame, but the lower character frame is the data description frame.
9841 "is data corresponding to the item name" product code ". In addition, although there is no character frame on the right side of the item description frame having the character string described as “quantity”, since the lower character frame is the data description frame, “ It is determined that “100” is data corresponding to the item name “quantity”. Furthermore, although there is no character frame below the item description frame having the character string described as “unit price”, the character frame on the right side is the data description frame. It is determined that “4500” is data corresponding to the item name “unit price”. in this way,
An item name is extracted based on the item name dictionary, and the non-standard form is read in association with data described adjacently to the right or below the item name.

[0006]

However, the conventional document reading apparatus has the following problems. FIG.
FIG. 11 is a diagram illustrating another example of an irregular form for explaining a problem of the conventional document reading apparatus. For example, as shown in FIG. 3, two item names “product name” and “product code” are described in two lines in one item description frame 11 of the form 10, and two item names are described in a data description frame 12 therebelow. Data "Sound L
When "SI" and "M9841" were described over two lines, two items could not be reliably identified and read.

[0007] In addition, two item names "quantity" and "unit price" are described in one item description frame 13 of the form 10 separated by a slash "/" as a delimiter, and a data description frame 14 therebelow. Two data "100" and "450"
When "0" was described separated by a slash "/", it was not possible to reliably identify and read two items. An object of the present invention is to provide a document reading apparatus which solves the problem of the conventional technique and can reliably read an irregular form in which a plurality of item names and data are described in one character frame. It is.

[0008]

According to a first aspect of the present invention, there is provided a document reading apparatus for converting image data of a document having a plurality of character frames delimited by ruled lines. Image reading means for decomposing and reading pixels; ruled line extracting means for extracting vertical and horizontal ruled lines from image data read by the image reading means; and vertical and horizontal ruled lines extracted by the ruled line extracting means. Character frame extracting means for extracting a character frame surrounded by a ruled line in a direction, and character recognition for recognizing a character or a character string described in the character frame extracted by the character frame extracting means and converting the character or character string into a character code. Means,
Referring to an item name dictionary in which the name of the item to be read is registered as the item name, the character or character string in the character frame recognized by the character recognition unit is registered in the item name dictionary. Item frame detecting means for detecting a description frame. Further, in this document reading device, data of characters or character strings that are not registered in the item name dictionary are described in the right or lower character frame in the same arrangement as the item description frame detected by the item frame detection unit. Data frame detecting means for detecting the data frame described, and associating the item description frame detected by the item frame detecting means with the data description frame detected by the data frame detecting means, When a plurality of item names are described and a plurality of data are described in the data description frame, data identification means for identifying the item names and the data in association with the described order is provided. Is provided.

According to the first aspect, since the document reading apparatus is configured as described above, the following operation is performed. Image data of a document having a plurality of character frames separated by ruled lines is read by being decomposed into pixels by image reading means,
The ruled line extracting means extracts vertical and horizontal ruled lines from the image data. A character frame surrounded by the extracted vertical and horizontal ruled lines is extracted by character frame extracting means,
Characters or character strings described in this character frame are recognized by the character recognition means. In the item frame detecting means,
The item name dictionary in which the item names are registered is referred to, and the item description frame in which the recognized character or character string in the character frame is registered in the item name dictionary is detected. Further, in the right or lower character frame in the same row as the detected item description frame, a data description frame in which data of a character or a character string not registered in the item name dictionary is described is detected by the data frame detecting means. You. Then, the data identification unit associates the item description frame with the data description frame, and a plurality of item names are described in the item description frame, and a plurality of data are described in the data description frame. , Item names and data are associated with each other and identified according to the described order.

According to a second aspect of the present invention, the data identifying means according to the first aspect of the invention is arranged such that, when the item description frame and the data description frame describe characters or character strings divided into a plurality of lines, Alternatively, a character string is identified as a plurality of independent item names and data, and the item names and the data are associated with each other in the order described. According to the second invention, the following operation is performed in the data identification means in the first invention. When characters or character strings divided into multiple lines are described in both the item description frame and the data description frame, it is recognized that the characters or character strings in each line are independent item names and data. Then, the item names and data are associated with each other according to the described order.

In a third aspect of the present invention, the data identification means in the first aspect of the present invention is arranged such that, when a plurality of characters or character strings are described in the item description frame and the data description frame by a delimiter, The identified characters or character strings are identified as a plurality of independent item names and data, and the item names and the data are associated with each other in the order described. According to the third aspect, the first aspect
The following operation is performed by the data identification means in the invention. When a plurality of characters or character strings separated by delimiters are described in both the item description frame and the data description frame, the delimited characters or character strings are independent item names and data. , And the item name and the data are associated with each other in the described order.

According to a fourth aspect of the present invention, the data identifying means according to the first aspect of the present invention is arranged such that, when a plurality of the data description frames are continuously arranged below the item description frame, The configuration is such that the item names are associated with the data in the data description frames. According to the fourth invention, the following operation is performed by the data identification means in the first invention. When a plurality of data entry frames are continuously arranged below the item entry frame, the data in each data entry frame is associated with the item name in the common item entry frame.

[0013]

FIG. 1 is a block diagram of a document reading apparatus showing an embodiment of the present invention. As shown in FIG. 3, for example, this document reading apparatus has an item description frame 1 separated by ruled lines.
An image reading unit (for example, an image reading unit) 21 that reads image data of a document such as an atypical form 10 having a plurality of character frames such as data frames 1 and 13 and data description frames 12 and 14 is provided. The image reading unit 21 is configured by an image scanner or the like, and reads image data described in a document by decomposing the image data into black and white binary pixels. An image memory 22 for temporarily storing read image data is connected to an output side of the image reading unit 21. In the image memory 22,
A ruled line extracting means (for example, a ruled line extracting unit) 23 is connected. The ruled line extraction unit 23 scans and reads the image data stored in the image memory 22 in the vertical and horizontal directions, and generates a histogram indicating the vertical and horizontal appearance frequencies of the black pixels. Further, the ruled line extracting section 23 has a function of extracting the positions of the ruled lines in the vertical and horizontal directions based on the generated histogram. On the output side of the ruled line extracting unit 23, a character frame extracting means (for example, a character frame extracting unit) 24
Is connected. The character frame extracting unit 24 includes the ruled line extracting unit 2
Based on the vertical and horizontal ruled lines extracted in step 3, a plurality of rectangular character frames surrounded by these ruled lines are extracted. On the output side of the character frame extraction unit 24 and the image memory 22,
A character recognition unit (for example, a character recognition unit) 25 is connected.

The character recognizing unit 25 cuts out image data from the image memory 22 for each character frame extracted by the character frame extracting unit 24, refers to the character dictionary 26, and writes a character described in the character frame. Alternatively, a character string is recognized and converted into a character code. An output side of the character recognition unit 25 is connected to an item frame detection unit (for example, an item frame detection unit) 27 and a data frame detection unit (for example, a data frame detection unit) 28. The item frame detection unit 27 searches the item name dictionary 29 in which the item name to be read is registered in advance, and when the character or character string recognized by the character recognition unit 25 is registered, the character frame is detected. It is determined to be the item description frame. If the character or character string in the right or lower character frame in the same row as the item description frame detected by the item frame detection unit 27 is not registered in the item name dictionary 29, the data frame detection unit 28 The character frame is determined as the data description frame. On the output side of the item frame detection unit 27 and the data frame detection unit 28, a data identification unit (for example, a data identification unit) 30 is provided.
Is connected. The data identification unit 30 associates the item description frame detected by the item frame detection unit 27 with the data description frame detected by the data frame detection unit 28. Further, when a plurality of item names are described in the item description frame and a plurality of data are described in the data description frame, the data identification unit 30 determines the item name and the data according to the described order. Have the function of identifying them.

FIG. 4 is a flowchart showing the operation of the document reading apparatus of FIG. Next, the operation of FIG. 1 will be described with reference to FIGS. When the reading of the non-standard form 10 of FIG. 3 is started, in step S1 of FIG.
The form 10 is decomposed into black and white binary pixels by an image scanner or the like of the image reading unit 21 in FIG. 1 and read as image data. In step S2, the image data of the form 10 is temporarily stored in the image memory 22. In step S3, the ruled line extraction unit 23 sets the image memory 22
The image data inside is read out by scanning in the vertical and horizontal directions,
A histogram indicating the vertical and horizontal appearance frequencies of the black pixels is generated. Further, ruled lines are extracted based on the generated histogram. In step S4, the character frame extracting unit 24 extracts a plurality of rectangular character frames 11 to 14 surrounded by vertical and horizontal ruled lines.

In step S5, the character dictionary 26 is referred to by the character recognizing unit 25, and character strings such as "product name" and "product code" described in the character boxes 11 to 14 are recognized and converted into character codes. Is converted. In step S6, the item frame detection unit 27 refers to the item name dictionary 29, and searches whether or not the characters or character strings in the character frames 11 to 14 are registered in the item name dictionary 29. At this time, when characters or character frames are described in a plurality of lines in the character frames 11 to 14, or when the characters or character frames are described by being separated by a delimiter such as a slash "/", The item name dictionary 29 is searched separately as a character or a character string. As a result, the character frame 11 of the form 10 in FIG. 3 is determined to be the item description frame 11 in which the two item names “product name” and “product code” are described. The character frame 13 is determined to be the item description frame 13 in which two item names, “quantity” and “unit price” are described.

In step S7, the data frame detection unit 2
8 detects character or character strings other than the item name in the right and lower character frames (for example, character frames 12 and 14) in the same arrangement as the item description frames 11 and 13. At this time, if a character or a character frame is described in a plurality of lines in the detected character frame, or if the character or the character frame is described by a delimiter such as a slash "/", Identified as a character or string. Thus, the character frame 12 of the form 10 in FIG.
It is determined that this is the data description frame 12 in which two data of “voice LSI” and “M9841” are described. Also,
The character frame 14 is determined to be the data description frame 14 in which two data “100” and “4500” are described. In step S8, the data identification unit 30 associates the item description frames 11, 13 with the data description frames 12, 14, as follows. Since the item description frame 13 exists on the right side of the item description frame 11 and the data description frame 12 exists below the item description frame 11, the item description frame 11 and the data description frame 12 are provided.
Are associated with each other. In addition, since no character frame exists on the right side of the item description frame 13 and the data description frame 14 exists below the item description frame 13, the item description frame 13 and the data description frame 14 are associated with each other.

In step S9, it is checked whether the number of item names in the item description frame is singular or plural. If the number is singular, the process proceeds to step S10. If the number is plural, the process proceeds to step S11. In step S10, the item name in the item description frame is associated with the corresponding data in the data description frame. On the other hand, in step S11, a plurality of item names in the item description frame and a plurality of data in the corresponding data description frame are sequentially associated with each other in the order and identified. For example, in the item description frame 11, two item names “product name” and “product code” are described in two lines in this order, and in the data description frame 12, two data “audio LSI” are written. And “M9841” are described in two lines in this order. Therefore, the data of the item name “product name” is “voice LS
I "and the data of the item name" product code "is" M
9841 ".

Further, in the item description box 13, two item names “quantity” and “unit price” are described in this order, separated by a delimiter slash “/”.
Describes two pieces of data “100” and “4500” separated in this order by a slash “/”. Therefore, the data of the item name “quantity” is identified as “100”,
The data of the item name “unit price” is identified as “4500”.
When the processes of steps S7 to S11 are completed for all the item description frames detected by the item frame detection unit 27,
The operation of the form 10 ends. As described above, according to the document reading apparatus of the present embodiment, the characters or character strings in the character frames such as the item description frames 11 and 13 and the data description frames 12 and 14 are described in a plurality of lines, In the case where data items are described separately, a data identification unit 30 that identifies the data items as a plurality of item names or data and associates the data items in accordance with the described order is provided. As a result, the present invention is not limited to an irregular form in which one item is associated with one character frame,
An irregular form in which a plurality of items are described in one character frame can be reliably read, and there is an advantage that the degree of freedom in form design is increased.

The present invention is not limited to the above embodiment, but can be variously modified. For example, there are the following modifications (a) to (f). (A) The delimiter in the character frame is not limited to the forward slash “/”, but may include parentheses “()”, midpoint “•”, comma “,”,
Delimiters such as colon ":" or semicolon ";" can be used as well. (B) The number of item names and data described in the character frame is not limited to two, and a single item or a plurality of three or more items can be similarly applied. (C) If there is anything other than the item name in the character or character string in the item description frame, the item frame detection unit 27 may ignore it. (D) If the number of data in the corresponding data box is larger than the number of item names in the box,
After the data identification unit 30 associates the item names with the data in order, the remaining data may be ignored. (E) When a plurality of character description frames of the same form are continuously arranged below the item description frame as in a tabular form, the data identification unit 30 uses The data may be associated with the item names in the common item description frame. (F) In the document reading apparatus of FIG. 1, the ruled line extracting unit 23, the character frame extracting unit 24, the character recognizing unit 25, the item frame detecting unit 27,
Although the data frame detection unit 28 and the data identification unit 30 are configured as independent processing units, they may be configured to perform the functions of these processing units under program control using a computer.

[0021]

As described in detail above, according to the first aspect, an item frame for detecting an item description frame whose character or character string is an item name with reference to the item name dictionary. Detection means, and data frame detection means for detecting the right or lower data description frame in the same row as the detected item description frame,
When the item description frame and the data description frame are associated with each other, and when a plurality of item names are described in the item description frame and a plurality of data are described in the data description frame,
There is a data identification means for associating the item names with the data in the described order. This gives 1
An advantage is that an irregular form in which a plurality of item names and data are described in one character frame can be reliably read. According to the second aspect, the character or character string divided into a plurality of lines is identified as a plurality of item names and data by the data identification means. This gives 1
Even if a plurality of lines of characters or character strings are described in one character frame, there is an effect that these can be reliably associated with each other.

According to the third aspect, the character or character string delimited by the delimiter is identified as a plurality of item names and data by the data identification means. Thereby, even if a plurality of characters or character strings are described in one character frame, there is an effect that these can be surely associated with each other. According to the fourth aspect, when a plurality of data description frames are continuously arranged below the item description frame, the data in each of the data description frames is shared by the data identifying means. It is made to correspond to the item name in the description frame. As a result, there is an effect that item names and data can be surely associated with each other even in a tabular form.

[Brief description of the drawings]

FIG. 1 is a configuration diagram of a document reading apparatus according to an embodiment of the present invention.

FIG. 2 is a diagram showing an example of a conventional non-standard form.

FIG. 3 is a diagram showing another example of a non-standard form for explaining the problem of the conventional document reading apparatus.

FIG. 4 is a flowchart illustrating an operation of the document reading apparatus of FIG. 1;

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 10 Non-standard form 11,13 Item description frame 12,14 Data description frame 21 Image reading part 22 Image memory 23 Ruled line extraction part 24 Character frame extraction part 25 Character recognition part 26 Character dictionary 27 Item frame detection part 28 Data frame detection part 29 Item name dictionary 30 Data identification part

Claims (4)

[Claims] An image reading unit that separates image data of a document having a plurality of character frames separated by ruled lines into pixels, and reads the image data from the image data read by the image reading unit in a vertical direction and a horizontal direction. Ruled line extracting means for extracting a ruled line, character frame extracting means for extracting a character frame surrounded by vertical and horizontal ruled lines extracted by the ruled line extracting means, and characters extracted by the character frame extracting means Character recognition means for recognizing a character or a character string described in a frame and converting the character or character string into a character code; and an item name dictionary in which names of items to be read are registered as item names. Item frame detection means for detecting an item description frame in which the character or character string in the character frame recognized by the means is registered in the item name dictionary; and the same as the item description frame detected by the item frame detection means. Right of line Or a data frame detecting means for detecting a data description frame in which data of a character or a character string not registered in the item name dictionary is described in a lower character frame, and an item description detected by the item frame detecting means. While associating the frame with the data description frame detected by the data frame detection means, a plurality of item names are described in the item description frame,
And, when a plurality of data are described in the data description frame, a data identification means for identifying the item name and the data in association with the described order is provided. Document reading device. 2. When the character or character string divided into a plurality of lines is described in the item description frame and the data description frame, the data identification means separates the characters or character strings of each line into a plurality of independent lines. 2. The item name and the data are identified as the item name and the data are associated with each other in the order described.
Document reader as described. 3. The data identification unit, when a plurality of characters or character strings are described in the item description frame and the data description frame with a delimiter, respectively, separates the separated characters or character strings. 2. The document reading apparatus according to claim 1, wherein the document reading apparatus is configured to identify a plurality of independent item names and data and associate the item names with the data according to the described order. 4. When a plurality of the data description frames are continuously arranged below the item description frame, the data identification unit determines the item name in the item description frame and the description of each of these data. 2. The document reading apparatus according to claim 1, wherein data in the frame is associated with each other.