JPH04190471A - Automatic generation processing system for correction picture - Google Patents

Abstract

PURPOSE:To automatically generate the correction picture without any operator's intervention by using position data on an original document for the generation of the picture for correcting data read by an optical character reader(OCR). CONSTITUTION:A reader 5 reads characters in respective items on the document 11 optically and sends them to an OCR processing part 2. At this time, the document position data 8 are used. The processing part 2 stores the sent data in a memory and displays the correction picture 10 for correcting the data on a display 9. A generation processing part 3 generates the correction picture 10 conforming with the document 11 according to the position data 8. Namely, the positions of the respective items in the picture are determined according to the position data on the respective items on the corresponding document 11. The numbers of rows and columns in the picture are smaller, so their ratio is considered. The correction picture 10 of constitution close to the document 11 is obtained on the basis of the document position data 8, so the operability of correcting operation by an operator is improved.

Description

[Detailed Description of the Invention] [Summary] This invention relates to a processing method for automatically generating a correction screen for correcting data read by OCR.

The purpose is to automatically generate a correction screen that matches the form based on location information about the form.

In an optical character reading device that reads data on a form using a reading device and displays a correction screen for correcting the reading result on a display, a generation process that generates position information for each item of the form and the correction screen. The generation processing unit is configured to generate the modified screen in accordance with the form according to the position information.

[Industrial application field]

The present invention relates to a processing method for automatically generating a modified screen.

More specifically, the present invention relates to a processing method for automatically generating a correction screen for correcting data read by OCR.

An OCR (optical character reader) reads handwritten characters written in predetermined positions on various forms.

Input as data is performed. The result of this reading is displayed on a correction screen for correction of misreading and the like. The operator corrects the misreading while looking at this correction screen.

[Conventional technology]

In order for the OCR to read the characters on the form, the positions of the characters on the form are input to the OCR.

For example, on a form, there are columns for entering data such as "customer code,""person in charge code,""year, month, and date" (these are collectively referred to as items), and each column is located at one line and one column.
The digit position is entered. OCR reads the written characters for each item according to its position information.

On the other hand, in addition to the location information about such forms, location information about the correction screen is input.

The screen is designed. If the configuration of this correction screen is similar to the ticket in question, operability for the operator will be improved. Therefore,
The three-digit line position of each item on the correction screen is input so that it is close to the position on the form.

Position information regarding the form and correction screen is input after calculating the position of each item. This work is done by one person.

[Problem to be solved by the invention]

According to the above-mentioned conventional technology, input of position information is required to design the correction screen, so there is a problem in that it takes a lot of time to design. Also.

The position information for the correction screen is similar to the position information for the form, so the information on Va is divided into 2
There is also the problem of duplicative work of having to input data multiple times.

Apart from this, since the correction screen is generally smaller than the form, the item may not fit on the correction screen with normal layout. In this case, there has conventionally been a problem that if there is even one item whose position has not been determined, the correction screen will not be displayed. For this reason, it was not possible to check even items whose display positions were fixed, and even when an attempt was made to add an item whose position had not been fixed to the correction screen, it was unclear in which part there was space available for insertion. Another problem was that the position of one item could not be changed.

SUMMARY OF THE INVENTION An object of the present invention is to provide an automatic correction screen generation processing method that automatically generates a correction screen suitable for a form based on position information about the form.

Another object of the present invention is to provide an automatic correction screen generation processing method that allows the positions of each item on the correction screen to be freely assigned.

[Means to solve the problem]

FIG. 1 is a diagram showing the basic structure of the present invention, and shows an optical character reading device according to the present invention.

In FIG. 1, 1 is a processing device, 2 is an OCR processing section, and 3
4 is a generation processing unit, and 4 is an undefined display processing unit.

5 is a reading device, 6 is reading result data, 7 is an OCR definition, 8 is form position data, 9 is a display, 10 is a correction screen, and 11 is a form.

The reading device 5 optically reads the characters written in each item on the form 11 as data and sends it to the OCR processing section 2.

For this reading, form position data 8 is used.

That is, the form position data 8 is the form 11 to be read.
This is positional information for each item, indicating the area to be read.

The OCR processing unit 2 stores the data from the reading bag 25 in the memory as reading result data 6. Also.

The OCR processing section 2 displays a modification screen 10 on the display 9 for modifying this data, that is, the result of reading.

The generation processing unit 3 follows the form position data 8.

A modified screen 10 conforming to the form 11 is generated.

[Operation] The generation processing unit 3 generates the modified screen 10 using the form position data 8 which is originally for reading the form-11. That is, the position of each item on the correction screen 10 is
It is determined according to the form position data 8 indicating the position of the corresponding item on the form 11. At this time, the ratio between the number of lines and the number of digits of the form 110 and the number of lines and the number of digits of the correction screen 10 (which generally take a smaller value) is taken into consideration.

According to the above, the correction screen 10 can be automatically generated by utilizing the form position data 8 without inputting the position information regarding the correction screen 10. Therefore, the time required to design the modified screen 10 can be extremely reduced.

Further, since the screen is generated according to the form position data 8, it is possible to obtain the modified screen 10 having a configuration similar to the form 11. Therefore, it is possible to improve the operability of the operator in correcting misreading.

〔Example〕

The optical character reading device shown in FIG. 1 will be further explained.

The processing device 1 includes a CPU (central processing unit) and a (main) memory. The OCR definition program 7 is loaded into this memory. ○CR definition body 7 consists of programs corresponding to various processes in order to perform various processes in OCR. The OCR definition body 7 on the memory and the CPU
A CR processing section 2 is configured. The OCR definition body 7 is created in advance and stored in a memory such as a DASD. The OCR definition body 7 or the OCR processing unit 2 of this embodiment is a generation processing unit 3 that generates a modified screen 10, and items that have undefined display positions and items that overlap with other items (undefined items). It has an undefined processing section 4 that displays this with a predetermined symbol (undefined symbol) in certain cases.

Based on the data read by the reading device 5, the OCR processing unit 2 generates one or more candidate characters by referring to a dictionary or the like. Then, this candidate character is stored as reading result data 6 in a memory such as a DASD. This storage is performed, for example, in order according to the ranking of candidate characters.

Further, the OCR processing unit 2 displays the reading result data 6 on the correction screen 10 by adding it.

Specifically, only the character with the first candidate character ranking is displayed in the corresponding item (field) of the correction screen 10. Further, the candidate characters are displayed on a screen other than the correction screen 10. Here, the five displayed candidate characters are one for, for example, the character at which the cursor is positioned on the correction screen 10. Note that the correction screen 10 is smaller than the entire screen because it displays necessary items in addition to the candidate characters.

The operator corrects the reading result data 6 by positioning the cursor on the character to be corrected on the correction screen 10 and selecting the correct character from among the candidate characters.

The correction screen 10 is generated by the generation processing section 3, and the two-form position data 8 is used for this purpose. An example of this form position data 8 is shown in FIG.

Further, an example of the form 11 is shown in FIG.

In FIG. 2, the form 11 includes "customer code", "
There are items such as "person in charge code". For example, in the "Customer Code", the characters rABC-0011J are written by hand or in print.

Since there are multiple items with the same name, such as "Product Code", item names such as "Product Code 1" are given in order from the top to distinguish them from each other.

The area in which the characters of each item are written (that is, the area to be read by the reading device 5) is called a field.

In particular, an area for one character is called a subfield.

All subfields and yields for the item are called the main field. In each item, the size of the subfield is common, but the size of the main field is different because the number of characters is different.

Position information for each item 5, that is, the main field (
Form position data 8) is the number of lines of the main field 1
It is made to represent the number of digits.

In FIG. 3, the form position data 8 is data in the form of an item list for the form 11.

The item list consists of the item name, attribute information, and field detailed information for each item. Of these, the main part of the form position data 8 is field detailed information. The item list is created on a memory such as a DASD by the OCR processing unit 2 using data input from a keyboard or the like.

The item name consists of a name in Japanese, which is also displayed on the form 11, and a name in alphanumeric characters, which is used as an abbreviation.

The attribute information consists of attribute 1 division and type.

"Attribute" represents the type of field. in this case.

Shows the "main" field. What is “split”?

9 is not performed because the attribute is a main field.

"r type" represents the type of one character. In this case, it indicates handwritten alphanumeric characters.

The field detailed information consists of information about the "position" of the king. The "position" information further consists of a row, a left edge, and a right edge. "Row" is form 11 of the main field of the item
Represents the row position above.

"Left end" and "right end" respectively represent the digit position at the left end and the digit position at the right end on the form 11 of the main field of the item. Specifically, the start line, start column, and end column of the main field are "rows."

Shown as "left end" and "right end".

Here, this "position" information is input from a keyboard or the like in order to be read by the reading device 5.

FIG. 4 shows the correction screen 10.

In the vertical direction, in FIG. 4(A), there are several lines between each item to make the position of each item easier to understand. Furthermore, in FIG. 4(B), the line spacing is eliminated so that as many items as possible are displayed on the correction screen 10. Note that even in the case of FIG. 4(A), items such as "product code" that are consecutive on two forms 11 are displayed without a space between two lines. That is, it conforms to the form 11.

On the other hand, when looking at the horizontal direction, Fig. 4 (A) and (B)
In both cases, each item is provided at a position corresponding to the approximate form 11.

Note that for "address", kanji characters corresponding to katakana input are displayed, and furigana characters are also added, resulting in an irregular field shape as shown in Figure 1.

Furthermore, as can be seen from the comparison between FIGS. 4(A) and 4(B) and FIG. 2, the entire horizontal direction of the corrected screen 10 corresponds to the form 11, but the details are different from the form 11. . This is due to the difference in size between the form 11 and the correction screen 10, and the fact that the size of the characters on the correction screen 10 is fixed.

FIG. 5 shows the flow of the correction screen generation process performed by the generation processing unit 3.

(2) As a pre-processing, the main fields of each item on the form 11 are arranged in ascending order of one line position, and further the main fields existing in the same line are arranged in ascending order of left end position. Specifically, they are arranged in the same order as the item numbers in FIG.

■Set parameter i (row position).

Compare this with the number of rows processed. The number of lines to be processed is 2. For example, when narrowing the line spacing as shown in FIG. 4(B), the number of lines to be processed is set to be less than the number of lines on the form 11.

When the number of rows to be processed is i - the number of rows to be processed, the processing is terminated.

■ If i<number of rows to be processed, find the variable 5tart (which is the digit position). That is.

5 tart = left end position of the field on the form 11 x horizontal size of the correction screen 10 x horizontal size of the form 11 .

■ Check to see if it is the first main field on the same line.

■ Find the variable end. That is, 7 end = start + number of digits (for number of characters) - 1.

■ Find the starting position (x, y) of the field. That is, (x, y) = (1,5 tart).

As a result, the field is displayed in the first line on the correction screen 10, starting from the digit (start). Note that the field ends at the end digit. In other words, the number of digits for one character is secured.

After this, repeat the process ① and the following.

■ In processing ■, if it is not the first field,
Compare 5tart and end+3 (number of digits of part).

■ If 5tart < end + 3, 5t
Change the value of art to end+3. This is for the purpose of correcting the main field to two characters if the interval between the main fields when displayed on the correction screen 10 is two characters or less.

Note that this is done in order to make the correction screen 10 easier to see, and the interval between two is not limited to two characters.

If 5tart < end + 3, process (2) is not performed.

As a result of the above, the main field of each item in form 11 is
It is allocated on the correction screen 10. In addition.

Allocation may be made for subfields.

The process shown in FIG. 5 moves the left end (starting position) of the main field in accordance with the horizontal ratio of the form 11 and the correction screen 10. That is.

All horizontal positional relationships on the form 11.

It is not scaled to fit within the modified screen 10.

This is due to the limitation due to the size of the characters on the correction screen 10. Therefore, the last main field in the same row number may not fit within the modification screen 10. Furthermore, if this main field is included in the correction screen 10, it may overlap with other main fields.

Therefore, for each main field, the undefined processing section 4
performs the processing shown in FIG.

■ As initialization process, variable ro@= 1, col
Let umn=1.

■ Set the parameter i and compare it with the number of main fields.

1> When the number of main fields is reached, the process ends.

Note that the processing order of this main field is 1, for example, in the order of its item number.

■ The display position of the main field is changed to correction screen 10.
Check to see if it exists inside. That is, it is checked whether each of the values of i, 5tart, and end previously obtained by the process shown in FIG. 5 falls within the corrected screen 10.

■ If it exists, check whether the main field and other main fields overlap on the correction screen]0.

If they do not overlap, process ■Repeat the following steps.

That is, since the main field is normally displayed on the modified screen IO through the process shown in FIG. 5, the fifth-order main field is processed.

■ If the main field does not exist on the correction screen 10, or if it overlaps with another main field, a predetermined symbol (hereinafter referred to as an undefined symbol) will be displayed instead of the main field. Find location. Here, first rom −1, column=1
By doing this, the search for an empty area starts from the first column of the first row of the correction screen 10.

First, row F row+ 1, column
= 1, and then compare this ROM with the horizontal size of the modified screen 10.

row > When the horizontal size of the correction screen 10 is reached, process■
Repeat the following.

■ If rom≦horizontal size of the correction screen 10, on the correction screen 10, the relevant coordinates (χ, ?) −(ro
lol.

colu+lr+) to see if there are any other main fields.

If it exists, process ■Repeat the following steps.

■ If it does not exist, check whether there is enough free space to display the undefined symbol.

If not, process 5 ■Repeat the following steps.

■ If there is, on the correction screen 10, the coordinates (x,
y) = (row, column) is the display (start) position of the undefined symbol.

After this, repeat the process ① and the following.

As described above, undefined items, ie, main fields, can be displayed as undefined symbols on the modification screen 10. Therefore.

Even if there are undefined items, it is possible to display the correction screen 10. As a result, other items can be checked on the correction screen 10, and it can also be known from the positions of other items where there is a vacant area that can be inserted. That is, for example, it is possible to add an undefined item to the modification screen 10 by slightly changing the position of another item.

〔Effect of the invention〕

As explained above, according to the present invention, in the process of generating a correction screen for correcting data read by OCR, by using form position data for reading the form, the correction screen can be created without manual intervention. Since it can be automatically created, it is possible to significantly reduce the time required to design the correction screen, and it is also possible to obtain a correction screen that matches the form in question.
゛Ease of use can be improved.

[Brief explanation of drawings]

FIG. 1 is a diagram showing the principle configuration of the present invention. FIG. 2 is a diagram showing an example of a form. FIG. 3 is a diagram showing an example of form position data. FIG. 4 is a diagram showing a correction screen. FIG. 5 shows the processing flow for generating a correction screen. Figure 6 shows the undefined display processing flow. 1 is a processing device, 2 is an OCR processing section, 3 is a generation processing section, 4
5 is an undefined display processing unit, 5 is a reading device, 6 is reading result data, 7 is an OCR definition body, 8 is form position data, 9 is a display, 10 is a correction screen, and 11 is a form. Patent applicant PFU Co., Ltd. Patent attorney Hiroshi Mori 1) (and 2 others) Figure 2 shows one row of the form position data Figure 3 shows the revised @ side Figure 4 Correction [1st page generation process flow

Claims (2)

[Claims] (1) The data on the form (11) is read by the reading device (5), and the correction screen (
10) on a display (9), the positional information (8) for each item of the form (11) and a generation processing unit (3) that generates the correction screen (10). An automatic correction screen generation processing method, characterized in that the generation processing unit (3) generates the correction screen (10) in accordance with the form (11) according to the position information (8). (2) When the generation processing unit (3) follows the location information (8), items that do not fit on the correction screen (10) and items that overlap with other items are marked with a predetermined symbol. 2. The automatic correction screen generation processing method according to claim 1, wherein the correction screen is automatically generated and displayed.