JP2740335B2 - Table reader with automatic cell attribute determination function - Google Patents

Description

Description: BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a table reading apparatus that automatically extracts a ruled line from a document image in a table format and reads characters in cells surrounded by the ruled line. The present invention relates to a table reader having a function of automatically determining a cell attribute.

[Conventional technology]

The conventional table reader is configured as shown in FIG.

This conventional apparatus includes an image input unit 1 for optically operating a tabular document to perform photoelectric conversion, an image data storage unit 2 for storing binary image data output from the image input unit as image data, A cell extraction unit 3 for extracting a ruled line from the image data stored in the image data storage unit 2 and calculating a table structure and a position coordinate of a cell from the extracted ruled line information; The character recognition unit 4 recognizes characters in the cell based on the position coordinates of the cell and converts the character into a character code.

Here, the character recognizing unit 4 can specify attributes such as the character type and the number of digits of the characters to be recognized. By specifying the attributes in detail, the characters to be recognized are limited.
The recognition performance can be improved. For example, if it is known in advance that a character to be recognized is only a number, a character having a similar character shape other than a number (such as a kanji or an alphabet) can be obtained by specifying a number as an attribute and performing recognition.
The character recognition rate can be improved by avoiding erroneous reading.

[Problems to be solved by the invention]

However, in the above-described conventional table reading apparatus, the operator sets an attribute of a character in a cell (hereinafter, referred to as a cell attribute).
Can be improved for each cell, but in order to specify the cell attributes, the operator must determine the cell attributes by looking at the contents of the document and specify them in the table reader. However, there is an unsolved problem that it takes enormous effort and time to specify attributes of all cells. For example, since there are 100 cells in a table with 10 rows and 10 columns, it is necessary to set each attribute for 100 cells. In this case, a method for easily setting the cell attribute by batch, column, or all designations has been proposed, but when the cell attributes of the rows or columns are different, the designation of the cell attributes becomes complicated.

Further, since the operator specifies the cell attribute, there is an unsolved problem that an erroneous specification may occur.

Therefore, the present invention has been made in view of the unsolved problem of the conventional example, and provides a table reading device capable of automatically determining a cell attribute and efficiently reading a tabular document. It is an object.

[Means for solving the problem]

In order to achieve the above object, a table reader having an automatic cell attribute determining function according to the present invention stores an image input unit for inputting a tabular document as image data, and stores the image data input by the image input unit. An image data storage unit, a cell extraction unit that extracts a ruled line from the image data stored by the image data storage unit and recognizes a cell surrounded by the ruled line, and an item column among the cells extracted by the cell extraction unit. Item column character recognizing means for recognizing a character in a cell of a corresponding cell; cell attribute determining criterion storing means for storing a criterion for determining a cell attribute;
A cell attribute determining unit for determining a cell attribute other than an item column from an item column character recognized by the item column character recognizing unit based on a determination criterion of the cell attribute determining criterion storage unit; and a cell determined by the cell attribute determining unit. Character recognition means for recognizing characters in each cell according to the attribute. Here, the cell attribute includes format information including the number of digits, units, etc., in addition to the character type.

[Action]

In the present invention, the characters in the cell corresponding to the item column are recognized by the item column character recognizing device from the cells extracted by the cell extracting device from the image data of the tabular document, and the characters are obtained from the item names obtained as the recognition result. The cell attribute of the data column corresponding to the item column is automatically determined based on the criterion for determining the cell attribute stored in the cell attribute criterion storage unit, and the character of the data column is recognized using the determined cell attribute. I do. In this way, by automatically and accurately determining the cell attribute, the trouble of designating the cell attribute can be omitted, and the recognition rate can be improved.

〔Example〕

 Hereinafter, embodiments of the present invention will be described with reference to the drawings.

FIG. 1 is a block diagram showing one embodiment of the present invention.

In the figure, 1 is an image input unit, 2 is an image data storage unit, 3 is a cell extraction unit, and 4 is a character recognition unit. These components have the same configuration as the above-mentioned conventional example. According to the present invention, recognition of a character in a cell corresponding to an item column of a table is performed between a ruled line extracting unit 3 and a character recognizing unit 4 based on a table structure output from the cell extracting unit 3 and cell position coordinates. And a cell attribute determining unit 6 for determining a cell attribute of a data column corresponding to the item name based on the item name recognized by the item column character recognizing unit 5. And a cell attribute determination criterion storage unit 7 for referring to a criterion for determining the attribute of the data column in the cell attribute determination unit 6.

Here, as shown in FIG. 3, the item column character recognition unit 5 normally uses the item column 41 for a tabular document such as a numerical table or a list.
And a data column 42 following this, generally the first in the table.
An item column 41 in row is formed, the cell is surrounded by the horizontal lines l _X0 and l _X1 of the first row and the vertical ruled line l _Y0 to l _Y3 40
The characters in the cells printed in 1 to 403 are recognized, and the item names “name”, “address” and “telephone number” are obtained as the recognition result.

The cell attribute determination criterion storage unit 7 previously stores cell attribute information corresponding to the item name as an attribute table, and this cell attribute information includes, for example, “name = personal name kanji / kana”, “address = kanji / kana”.・ Number ”,“ Amount = Number ”,
In addition to information such as “age = three digits”, for example, the address “4-chome 2-79” also includes format information that allows the simplified display “4-2-79”. This cell attribute information may be added or changed later in a learning manner.

Further, in the cell attribute determining unit 6, the item column character recognizing unit 5
The attribute determination of the data column 42 corresponding to the item column 41 is performed with reference to the attribute table of the cell attribute determination criterion storage unit 7 based on the item names output from the. Therefore, when the item name is, for example, “name”, the corresponding data column
As the attribute of 404, it is determined that the name is Kanji / Kana, and the cell attribute of this determination result is specified to the character recognition unit 4.

Next, the operation of the above embodiment will be described with reference to the flowchart showing the processing procedure of FIG.

First, the image input unit 1 is activated in a step,
For example, the tabular document shown in FIG. 3 is optically scanned and read as image data, and this image data is stored in the image data storage unit 2.

Next, in a step, the cell extracting unit 3 is operated,
Extracting horizontal ruled l _X0 to l _X2 and vertical lines l _Y0 to l _Y3 from image data stored in the image data storage unit 2, the extracted tables to calculate the intersection of the ruled line structure (the number of rows and number of columns )
And the position coordinates of the cell surrounded by each ruled line.

Next, in a step, the item column character recognition unit 5 is operated to extract the item column 41 from the position coordinates of the extracted cell,
The characters “name”, “address”, and “telephone number” in each of the cells 401 to 403 of the item column 41 are recognized.

Next, in a step, referring to the attribute table stored in the cell attribute determination criterion storage unit 7 based on the recognition result of the item column character recognizing unit 5, the data columns 404 to 404 corresponding to the respective item column cells 401 to 403 are referred to. The cell attribute of 406 is determined. here,
Since the character of the item column cell 401 is "name", the data column 4
As the cell attribute of 04, it is determined that the name is Kanji or Kana, and the character in the item column cell 402 is “address”.
Are determined to be kanji, kana, numbers, and format information. Since the character in the item column cell 403 is "telephone number", the cell attribute in the data column 406 is determined to be numbers and symbols.

Next, the character recognition unit 4 is activated in a step, the cell attribute determined by the cell attribute determination unit 6 is specified to the character recognition unit 4, and the characters in the cells 404 to 406 of the data column 42 are recognized. This is converted to a character code and output.

As described above, according to the above-described embodiment, the character in the cell of the item column is recognized, the cell attribute of the data column corresponding to the item column is determined by the cell attribute determining unit 6, and the cell attribute of the determination result is determined by the character. Since the designation is made to the recognizing unit 4, the cell attribute of the character in the cell can be automatically designated without any trouble, and the character recognition rate is not reduced by erroneous designation of the cell attribute. A highly accurate table reading function can be realized by reducing erroneous determination of the character shape and erroneous format.

In the above embodiment, the case where the character in the item column cell is recognized by the normal character recognition in the item column character recognizing unit 5 has been described. However, the present invention is not limited to this, and is used in the item column. Since the number of words is limited, the item name may be recognized by collating words. In this case, the recognition rate of characters in a cell can be improved.

Further, the present invention can be applied not only to a table surrounded by solid ruled lines but also to a table surrounded by virtual ruled lines formed by blank spaces.

〔The invention's effect〕

As described above, according to the present invention, among the cells extracted by the cell extracting means, the character in the cell of the cell corresponding to the item column is recognized by the item column character recognizing means, and the cell is recognized based on the recognition result. The attribute determination means refers to the determination criteria stored in the cell attribute determination criteria storage means to determine the cell attribute of the data column other than the item column, designates the determination result to the character recognition means, and uses this character recognition means. Since the character in each cell is configured to be recognized according to the cell attribute, the cell attribute of the character in the cell can be automatically specified without bothering the user, and the character recognition rate is reduced due to incorrect specification of the cell attribute. Therefore, it is possible to achieve an effect that a highly accurate table reading function can be realized by reducing erroneous determination of a similar character shape and erroneous format.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an embodiment of the present invention, FIG. 2 is a flowchart showing an example of a processing procedure of the present invention, FIG. 3 is an explanatory diagram showing a tabular document, and FIG. 4 shows a conventional example. It is a block diagram. In the figure, 1 is an image input unit, 2 is an image data storage unit, 3 is a cell extraction unit, 4 is a character recognition unit, 5 is an item column character recognition unit, 6 is a cell attribute determination unit, and 7 is a cell attribute determination reference storage. Department.

Claims (1)

(57) [Claims] An image input means for inputting a tabular document as image data, an image data storage means for storing the image data input by the image input means,
A cell extracting means for extracting a ruled line from the image data stored by the image data storing means and recognizing a cell surrounded by the ruled line; Item column character recognizing means for recognizing a character, cell attribute determining criterion storing means for storing a criterion for determining a cell attribute, and the item field character recognizing means based on the criterion of the cell attribute criterion storing means. A cell attribute determining unit for determining a cell attribute other than the item column from the item column character recognized in the above, and a character recognizing unit for recognizing a character in each cell according to the cell attribute determined by the cell attribute determining unit. A table reader having an automatic cell attribute determination function.