JP3268084B2 - Form reading system - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a form reading system for reading an irregular form using, for example, a text reader type OCR (optical character reading device).
The present invention relates to a form reading system using OCR that analyzes a form layout according to LC (layout information).

[0002]

2. Description of the Related Art A method of recognizing characters on a form by OCR can be roughly divided into the following three methods. (1) A method in which an FC (format control information) designating the coordinates of the reading area, the number of digits to be read, the character type, and the like is transmitted to the OCR in advance, and the OCR recognizes characters according to the FC.

(2) The LC (layout information) of the character area on the form is transmitted to the OCR in advance, and the L
A method in which the OCR determines a reading area according to C and recognizes a character.

(3) O from the scanned form image
A method in which a CR automatically recognizes a character area to recognize a character. The OCR by the method (1) is called a form OCR,
Used to process OCR-formatted forms.

On the other hand, the OCR by the method (3)
Is referred to as a text reader, and is used to process forms that are not converted to OCR. Since the absolute coordinates of the reading area are not specified, the recognition result is somewhat ambiguous.

The OCR according to the method (2) is also called a text reader. However, since character recognition is performed using LC, the reliability of the recognition process is higher than that of the OCR according to the method (3).

[0007] In the conventional form reading system using OCR according to the method (2), an existing form such as a salary transfer request form that has not been converted to OCR is read using LC as follows.

The text reader type OCR performs a layout analysis on a form image captured by a scanner from layout information on the form included in the LC, and calculates a character area to be read.

Then, the character area to be read calculated by the layout analysis is recognized, and the recognition result data is output to the host controller. The application on the host controller specifies the type of the read item from the read result data and the LC and displays the recognition result data on the correction screen or outputs the edited data to the output file.

[0010]

In a conventional form reading system using a text reader type OCR, an existing form can be easily read in accordance with the LC, but layout analysis is erroneously performed due to dust on the form and blurring of ruled lines. In such a case, there is a problem that the correspondence between the recognition result data and the read item does not work well, and the correction screen and the output data are broken.

A conventional text reader type OCR
In the form reading system using, the recognition result data in which a layout analysis error has occurred has to be either discarded or processed by key entry while referring to the actual form.

The present invention has been made in view of the above circumstances, and as a result of causing a text reader type OCR to recognize an existing form such as a salary transfer request form that has not been converted into an OCR, a layout analysis error has occurred. It is an object of the present invention to provide a form reading system suitable for confirming and correcting the recognition result data in the case where the recognition result data is obtained.

[0013]

To achieve the above object, according to the solution to ## form reading system of the present invention, the layout information storage
Read form image according to layout information obtained from
Analyze the layout of the
A character recognition device that determines a character area and performs character recognition, and recognition result data output from the character recognition device and the number of tables, the number of items in the table, and the items of the layout information
Compare with the inner row number information, and if they do not match, the layer
Error detection means for outputting an error analysis error, and the layout analysis error output from the error detection means .
Display the form image corresponding to the location where the error occurred, and
Correction means for checking and correcting errors .

[0014]

According to the above-described structure, for a non-OCR form, the character recognition means analyzes the layout of the form image of the form based on the layout information designating the layout of the form to be read. performing character recognition of the read target that is determined by the layout analysis. Further, the error detection means, for detecting a layout analysis error in the character recognition means and a recognition result and the layout information output from the character recognition means. When the error detecting means detects the layout analysis error, a form image corresponding to the error occurrence location is displayed.

[0015] Thus, the operator, with the recognition result of the layout analysis error has occurred, it is possible to a data correction with reference to the display of document images.

[0016]

An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing the configuration of the form reading system according to the embodiment. This system includes an OCR 1, an LC storage unit 2, and an OCR control device 3, as shown in FIG. OCR1 is a text reader type O
It is a CR and is composed of an image scanner and a recognition device. This recognition apparatus analyzes the layout of a form image read by an image scanner, and determines a character area to be cut out and recognized according to the analysis result. The character recognition result of each character area is transferred to the OCR control device 3 as recognition result data. The recognition device also has a function of transferring the form image received from the image scanner to the OCR control device 3 as it is.

The OCR control device 3 controls the operation of the OCR 1, and is constituted by, for example, a personal computer. The OCR control device 3 includes a layout analysis error detection unit 4, a data correction unit 5, and a data output unit 8 as a mechanism for controlling the OCR 1. The data correction unit 5
Comprises an error correction unit 6 having an image display control unit 7.

The layout analysis error detecting section 4 has an OCR
1 based on the recognition result data transferred to the OCR control device 3, an error is detected in the layout analysis processing by the OCR 1. This error detection is performed by comparing various in-form information between the OCR1 recognition result data and the LC used by the OCR1. When detecting the layout analysis error, the error detection unit 4 notifies the error correction unit 6 of the data correction unit 5 of the detection.

The data correction unit 5 displays the recognition result data on the screen, and corrects the recognition result data according to a key input from the user. The error correction unit 6 is for correcting the error of the form in which the layout analysis error has occurred, responds to the error notification from the layout analysis error detection unit 4, controls the image display control unit 7, and Display the form image corresponding to the error location on the screen. The display control unit 7 has a function of controlling the change of the magnification of the displayed form image, the change of the display target position on the form, and the change of the display position on the screen according to the keyboard operation by the operator.

The operator confirms the location of the error and corrects the recognition result data by key input while looking at the form image displayed on the screen. The corrected clean data is sent to the data output unit 8. Then, it is stored as read data of the form.

Next, the operation of the embodiment will be described. OCR
1 performs layout analysis and character recognition of a form in accordance with the LC of the LC storage unit 2, but here, first, the LC
Will be described with reference to FIGS.

FIG. 2 is a conceptual diagram for explaining the LC according to the embodiment, and FIG. 3 is table information on a form included in the LC according to the embodiment, item information in each table, and information in each item. It is a conceptual diagram for explaining line information.

FIG. 2A shows an image of a form output from the OCR 1 that has not been converted to an OCR format.
An entry of information 2 is performed. At this time, an area in which an item having the same attribute exists is specified on the screen using a pointing device such as a mouse. FIG. 2B shows an example of the designation. Using this LC, the OC
R1 recognizes the form. An example of this recognition is shown in FIG.

The LC includes information on the number of tables on the form, information on the number of items in each table, information on the number of rows in each item, and the like.
Here, the table on the form, the items in the table, and the rows in the items are 10a to 10c (table on the form), 11a to 11b (items in the table), and 12a to 12b shown in FIG.
(Line in the item).

Next, an outline of a processing procedure of the OCR control device 3 for the recognition result data transmitted from the OCR 1 will be described with reference to FIGS. FIG. 5 is a flowchart for explaining a processing procedure of the OCR control device 3 according to the embodiment, and FIG. 6 is a flowchart for explaining a processing procedure of the error detection unit according to the embodiment.

First, the error detection unit 4 performs a layout analysis error check on the recognition result data of the OCR 1 and the corresponding LC in the LC storage unit 2 (step A1 in FIG. 5).
In this layout analysis error check, first, it is checked whether or not the number of tables in the form matches the recognition result data with the information on the number of tables in the LC (step B in FIG. 6).
1). Next, it is checked whether the number of items in the table matches the recognition result data and the number of items in the table of the LC2 (step B2 in FIG. 6). Furthermore, it is checked whether the number of lines in the item matches the recognition result data with the number of lines in item of LC2 (step B in FIG. 6).
3). These check results are comprehensively determined, and as a result, the presence / absence of a layout analysis error and the location of the error occurrence are detected.

As a result of the above check, when a layout analysis error is detected (Y in step A2 of FIG. 5), the fact is notified to the error correction unit 6 of the data correction unit 5, and a layout analysis error has occurred. Modify the form (Fig. 5
Step A4).

First, the error correction unit 6 causes the image display control unit 7 to display a form image corresponding to the location where the layout analysis error has occurred on the screen. In addition, the display control unit 7
Based on key input data from a keyboard or the like by an operator, control is performed such as changing the magnification of the displayed form image, changing the display target position on the form image, and changing the display position on the screen.

The concept of this operation will be described with reference to FIG. FIG. 4 is a conceptual diagram illustrating control of screen display according to the embodiment. FIG. 4A is a form image read by the OCR1 image scanner. FIG. 4B shows a state in which a form image corresponding to the location where the layout analysis error is detected is displayed on the screen. Here, FIG. 4C shows the result of the operator performing a magnification changing operation via the keyboard. FIG. 4D illustrates a change in the display target position on the form image. Further, the display position can be changed to an arbitrary position on the screen by operation.

Thus, the operator can deal with the detected location of the layout analysis error while referring to the form image. The error detection unit 4
If the result of the check is that there is no layout analysis error (N in step A2 in FIG. 5), the data correction unit 5 performs any processing such as confirmation and correction as needed (step A3 in FIG. 5). It is possible.

The recognition result data confirmed and corrected by the data correction unit 5 is output to the data output unit 8 (FIG. 4).
Step A5). This makes it possible to easily correct the recognition result data of the form in which the layout analysis error has occurred.

Next, with reference to FIGS. 7 to 9, a specific procedure of an error detection process by the layout analysis error detection unit 4 and an example of display of a location where a layout analysis error has occurred will be described.

FIG. 7 is a flowchart showing an example of a specific processing procedure of the error detection unit according to the embodiment. FIG. 8 is a conceptual diagram for explaining recognition result data transmitted from the OCR according to the embodiment. FIG. 9 is a conceptual diagram illustrating an example of display of a layout analysis error occurrence location according to the embodiment.

First, the number of tables on the form is obtained from the recognition result data (step C1 in FIG. 7). The number of tables is obtained from the form data of the recognition result data as shown in FIG.
The recognition result data is a value obtained by the layout analysis processing of the OCR 1 and includes form data, in-table data,
It contains item line data, item data and character data. The form data contains the number of tables on the form and the address of the start position of the form. The data in the table contains the total number of items in each table and the address of the start position of the table. The number and the address of the start position of the item line, the item data include the number of lines in the item and the address of the start position of the item, and the character data include the recognition result data and the address of the start position of the line.

Then, the number of tables on the obtained form and the LC
(Step C2 in FIG. 7). If the comparison results do not match (N in step C3 in FIG. 7), the address of the start position of the form image is set as a layout analysis error location (step C11 in FIG. 7).
The error detection processing ends.

As a result, for example, an area as shown at 14a in FIG. 9 is displayed on the screen as a layout analysis error. If the comparison results match (Y in step C3 in FIG. 7), the number of items in each table is obtained from the recognition result data (step C4 in FIG. 7). The number of items in the table is obtained from the data in the table of the recognition result data as shown in FIG. Then, the obtained number of in-table items is compared with the in-table item number information of the LC (step C5 in FIG. 7). This is performed by determining that the comparison result matches when the total number of items in the table is an integral multiple of the number of items in the table of LC. If the comparison results do not match (N in step C6 in FIG. 7), the number of items in item line units is checked from the number of items in the item line data of the recognition result data as shown in FIG. The start position of the item row that does not match the information on the number of items in the table is set as a layout analysis error location (step C11 in FIG. 7), and the error detection process ends. Here, the item row refers to 13 shown in FIG.

As a result, for example, an area as shown at 14b in FIG. 9 is displayed on the screen as a layout analysis error. If the comparison results match (Y in step C6 in FIG. 7), the number of lines in each item is obtained from the recognition result data (step C7 in FIG. 7). The number of lines in this item is shown in FIG.
Is obtained from the item data of the recognition result data as shown in FIG. Then, the obtained number of lines is compared with the information on the number of lines in the LC item (step C8 in FIG. 7). If the comparison results do not match (N in step C9 in FIG. 7), the item start position is set as a layout analysis error location (step C11 in FIG. 7), and the error detection process ends.

As a result, an area such as that shown at 14c in FIG. 9 is displayed on the screen as a layout analysis error. If the comparison results match for all inspection items (Y in step C9 in FIG. 7), it is determined that there is no layout analysis error (step C10 in FIG. 7), and the error detection processing ends.

As a method for comparing the number of items in the table, a method may be employed in which an inspection is performed for each item row from the beginning, instead of checking from the total number of items in the table. Instead of ending the error detection processing when a layout analysis error is detected, a method may be used in which the number of tables, the number of items in a table, and the number of rows in an item are checked and the presence or absence of a layout analysis error is determined.

[0040]

According to the present invention, even if a layout analysis error occurs as a result of recognizing an existing form such as a payroll transfer request that has not been converted into an OCR with the text reader type OCR, the form image displayed as a layout analysis error is referred to. This eliminates the need for discarding recognition result data and performing key entry with reference to actual forms as in the past, making it possible to make simple corrections and achieve paperless operation. Make it possible.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of a form reading system according to an embodiment of the present invention.

FIG. 2 is a conceptual diagram for explaining LC (layout information) according to the embodiment.

FIG. 3 is a conceptual diagram for explaining table layout information, item number information in each table, and row number information in each item included in the LC according to the embodiment.

FIG. 4 is an exemplary conceptual diagram for explaining screen display control according to the embodiment.

FIG. 5 is a flowchart for explaining a processing procedure of the OCR control device according to the embodiment.

FIG. 6 is a flowchart conceptually showing a processing procedure of an error detection unit according to the embodiment.

FIG. 7 is an exemplary flowchart illustrating an example of a specific processing procedure of an error detection unit according to the embodiment.

FIG. 8 is a conceptual diagram for explaining recognition result data transmitted from the OCR according to the embodiment.

FIG. 9 is an exemplary conceptual diagram explaining an example of display of a layout analysis error occurrence location according to the embodiment.

[Explanation of symbols]

1 ... OCR, 2 ... LC (layout information) storage unit, 3 ...
OCR control device, 4 ... layout analysis error detection unit, 5
... Data correction section, 6 ... Error correction section, 7 ... Image display control section, 8 ... Data output section, 9 ... Form, 10a-10c
... Table on the form, 11a-11b ... Items in the table, 12a-
12b ... line in item, 13 ... item line, 14a-14c ...
Screen display target area.

Claims (1)

(57) [Claims] 1. A ray obtained from a layout information storage unit.
The layout of the read form image according to the
A character recognition device that performs character recognition by determining a character area to be recognized from the analysis result, a recognition result data output from the character recognition device, a table number information of the layout information, and a table. Item number information and terms
Compare with the number of rows in the eye and if they do not match,
Error detection means for outputting an out-analysis error, and the layout analysis output from the error detection means
Display the form image corresponding to the error occurrence location of the error
And a correcting means for performing error confirmation and correction .