JPH096869A - Method and device for format information generation - Google Patents

Abstract

PURPOSE: To decrease the capacity of a memory which stores reference format information for document reading. CONSTITUTION: Image data on a document 20 which is given a new document ID are stored in an image memory 22 through a photoelectric converting means 21. A new format information generating means 31 generates format information on the document 20. A margin setting means 32 sets margin ranges by read fields according to a format generation rule. A retrieval and registration means 33 uses the margin ranges to retrieve the same format information as the format information on the document 20 among pieces of format information stored in a format information memory 40. The document ID of the document 20 is stored and registered in the format information memory 40 together with the group name given to the format information extracted by the retrieval.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a form mark (hereinafter referred to as a form ID) for reading necessary items described in a form by an optical character reader (hereinafter referred to as OCR). The present invention relates to a format information generation method and a format information generation device for generating the above format information.

[0002]

2. Description of the Related Art OCR is used to read necessary items described in a form. In the conventional OCR, an image of a form is captured by an image scanner or the like, and the image data of the form image is stored in an image memory. Then, using the format information designating the reading field of the form, the reading field is cut out from the image data stored in the image memory by the character cutting means, and the character feature in the cut reading field is extracted by the character recognition means. By doing so, the characters on the form are recognized. That is, in order to identify the format of the form, the form ID represented by numbers or the like is written or printed at a predetermined position (for example, the upper right).
Therefore, in the OCR, first, the form ID is recognized, the format information prepared in advance corresponding to the form ID is referred to, and the reading field is cut out from the form image,
The characters in the cut-out read field are recognized using a recognition dictionary.

[0003]

However, the conventional method of reading necessary items in the form using the OCR has the following problems. Format information is for each form I
It is necessary to register every D unit. If there are many types of form IDs or the number of types of form IDs increases, the amount of information increases, so the capacity of the format information memory must be increased. In addition, if the capacity of the format information memory is large, it will take time to retrieve the necessary format information when the necessary portion of the form is read by the OCR.

[0004]

In order to solve the above-mentioned problems, the first invention is to read necessary items by OCR for a form whose format is fixed for each form ID. In the format information generating method for generating and registering the unique format information for reading and referring to the form corresponding to, the following processing is performed. That is, in the format information generating method of the first invention, the new format information generating process for generating the new format information for the unregistered form in which the format information is not registered, and the previously generated format generating rule are used. A margin setting process for setting a margin range is performed on the new format information based on the above. Then, what is considered to be the same as the new format information within the margin range is extracted from the registered format information, and the same group name as the extracted registered format information is used as the form of the unregistered form. If the registered format information that can be regarded as being the same is not registered, the search registration processing is performed in which the new format information, the form ID thereof, and the new group name are registered in association with each other. There is. According to a second aspect of the present invention, the new format information generation process according to the first aspect takes in image data of an unregistered form in which the format information is not registered and processes the image data based on a format generation rule created in advance. Then, the new format information is generated.

In the third invention, in order to read necessary items by OCR for a form whose format is defined for each form distinguished by the form ID, read reference is made to the form corresponding to each form ID. In the format information generation device that respectively generates and registers the unique format information for
The following photoelectric conversion means, image memory, new format information generation means, margin setting means, format information memory, and search / registration means are provided. The photoelectric conversion means acquires an image of the target form and converts it into image data. The image memory stores the image data. The new format information generating means has a function of recognizing the form ID and generating new format information based on the image data stored in the image memory. The margin setting means has a function of setting a margin range for the new format information based on a format generation rule created in advance. The format information memory includes a plurality of the format information, a group name attached to each of the format information, and a form ID to which the format information is applied.
The group is stored in association with the group. Then, the search / registration means extracts, from the format information stored in the format information memory, format information that is within the margin range and can be regarded as the same as the new format information, and attaches it to the extracted format information. When the registered group name and the form ID of the unregistered form are stored and registered in the format information memory and the format information that can be regarded as the same does not exist, the new format information, the form ID, and a new group name And has the function of storing and registering in the format information memory in association with each other.

In a fourth invention, the margin setting means in the third invention uses a margin value stored in association with each read field in the format information, and respectively sets the margin range for each read field. The configuration is set. The fifth invention is
Each read field of the format information in the third or fourth invention is a type 1 read field represented by a positional relationship in which the end of the target form is the origin, and an arbitrary line segment drawn on the target form. And the second type reading field represented by the positional relationship with the point as the origin. The margin setting means uses a first margin value corresponding to the first type reading field and a second margin value corresponding to the second type reading field and different from the first margin value. The margin range for each read field is set.

[0007]

According to the first aspect of the invention, since the format information generating method is configured as described above, the new format information for the unregistered form is generated by the new format information generating process. A margin range is set in the margin setting process for the new format information. Then, by the search registration process, the format information that is within the margin range and can be regarded as the same as the new format information is extracted from the registered format information, and the extracted registered format information is added to the form ID of the unregistered form. The same group name is added and registered. If there is no registered format information that can be regarded as the same, the new format information, the form ID thereof, and the new group name are registered in association with each other. That is, if there is registered format information that can be regarded as the same as the new format information, the new format information is not stored in the format information memory. According to the second invention, in the new format information generation processing in the first invention, the image data of the unregistered form in which the format information is not registered is fetched and the image data is processed based on the format generation rule. New format information is generated. That is, the format information can be automatically generated by preparing the format generation rule in advance.

According to the third invention, the image of the target form is acquired and converted into image data by the photoelectric conversion means. The image memory stores image data. The new format information generation means generates new format information based on the image data stored in the image memory, and the margin setting means sets the margin range. On the other hand, the format information memory stores a plurality of format information, a group name assigned to each format information, and a form ID group to which the format information is applied, in association with each other. Then, by the search / registration means, the format information stored in the format information memory is extracted as format information that is within the margin range and can be regarded as the same as the new format information. The form ID of an unregistered form is stored in the format information memory and registered. If there is no format information that can be regarded as the same, the new format information, its form ID, and the new group name are stored and registered in the format information memory.

According to the fourth aspect of the present invention, the margin range is stored by the margin setting means in association with each read field with respect to the format information generated by the new format information generating means in the third aspect. Are set respectively. That is, the margin range corresponding to each reading field is set. According to the fifth invention, the format information generated by the new format information generating means in the third or fourth invention is the first type reading field represented by a positional relationship with the end of the target form as the origin. And a second-type reading field represented by a positional relationship with an arbitrary point on a line segment drawn on the target form as an origin. Then, the margin setting means uses the first margin value corresponding to the first type reading field and the second margin value corresponding to the second type reading field and different from the first margin value. The margin range for the read field is set respectively. That is, the margin ranges for the first type read field and the second type read field are set independently of each other. Therefore, the above problem can be solved.

[0010]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS FIGS. 2 (1) and 2 (2) are diagrams for explaining the format information of the form shown in FIG. 1. FIG. 2 (1) shows an example of the format information, and FIG. Indicates the field position. Moreover, FIG. 3 (1), (2)
FIG. 6 is a diagram illustrating a form of a form. These Figure 2
The gist of the embodiment of the present invention will be described with reference to (1), (2) and FIGS. 3 (1), (2). The format information for reading characters is registered in correspondence with each form ID. As shown in FIG. 2A, the number of read fields to be cut out for reading and the field positions of those read fields. It is composed of information and attribute information of a character or a group of characters to be read. In the field position information in the format information, the coordinates P (x ₁ , y ₁ ) of the upper left position of each read field are designated. Coordinate P
2 (2), the origin O of the coordinates is designated as the upper left end portion of the form 10, for example (x ₁ , y ₁ ). Further, in order to indicate the size of each reading field, a width w in the x direction and a height h in the y direction of the reading field are designated. Each attribute has a character type of a character or a character group to be read in the reading field, the number of characters, a character interval Pt, and the like, and is stored in the format information memory together with field position information.

FIGS. 3A and 3B show two different forms 15 and 16 as examples. Each form 1
The form IDs 15 ₁ and 16 ₁ of 5 and 16 are respectively written in predetermined positions in the upper right of the form. Each form 15,1
6, the customer number, the product number, the unit price, the quantity, the total amount, the total amount, the distribution route, the shipping date, etc. are described in each field according to the format. Although the formats of the two forms 15 and 16 are different, for example, each customer number 1
5 ₂ , 16 ₂ and distribution route 15 ₃ , 16 ₃ and shipping date 1
5 ₄ and 16 ₄ are written at almost the same positions on the form. When the customer numbers 15 ₂ and 16 ₂ , the distribution routes 15 ₃ and 16 ₃ and the shipping dates 15 ₄ and 16 ₄ are read by the OCR, the read fields to be cut out are the same, and each form 15,
The 16 pieces of format information are the same. In this embodiment, even if the form IDs are different as described above, if there are forms with the same format information, the number of format information to be registered is reduced.

FIG. 1 is a functional block diagram of a format information generating apparatus showing an embodiment of the present invention. This format information generating apparatus includes a photoelectric conversion unit 21 such as an image scanner that captures an image of a form 20, and a photoelectric conversion unit 2.
The image memory 22 is connected to the output side of 1. The photoelectric conversion means 21 and the image memory 22 are O
It may be combined with CR. The output side of the image memory 22 is a control means 3 composed of a personal computer or the like.
Connected to 0. A format information memory 40 is connected to the output side of the control means 30. The format information memory 40 stores a plurality of format information generated by the control means 30, and has group names A, B, C ... Assigned to each format information and each group name A, B ,. The form ID group to which the format information with C ... is applied is stored in association with each other. The control means 30 uses the new format information generation means 3
1, margin setting means 32, search registration means 33,
The display control means 34 is provided. The new format information generating means 31 generates format information of the target form from the image data, and the image memory 2
Connected to 2.

FIG. 4 is a block diagram showing the configuration of the new format information generating means shown in FIG. The new format information generation means 31 includes an image analysis means 31a connected to the image memory 22. Image analysis means 3
Collation search means 31b is connected to the output side of 1a. The collation search means 31b is connected to the format generation rule storage means 31c which stores the format generation rules. The output side of the collation search means 31b is connected to the margin setting means 32. The search registration means 33 is connected to the output side of the margin setting means 32. The search / registration means 33 is connected to the format information memory 40. The margin setting means 32 has a function of setting a margin range for position information of each read field of the format information generated by the new format information generating means 31. The search / registration unit 33 has a function of searching the format information stored in the format information memory 40 and storing the format information of the target form 20 in the format information memory 40 to register it. The display control means 34 controls display on the display 35 connected to the display control means 34, and is connected to the image memory 22 and the format information memory 40.

Next, the format information generating method of this embodiment using the format information generating apparatus of FIG. 1 will be described. FIG. 5 is a diagram showing the form 20 of FIG. The form ID 20 ₁ is described in the form 20. Form ID
The number 20 ₁ is written in the upper right of the form according to the rules.
Formatting information for this form 20 is not yet registered form, the format information is one that is not registered in the format information memory 40 in correspondence with the form ID 20 _1. Here, the information to be read by the OCR is recorded in the field of the customer number 20 ₂ and the distribution route 20.
_The information described in the field of ₃ and the information described in the field of the shipping date 20 ₄ .

FIG. 6 is a flow chart showing a format information generation process using the format information generation device of FIG. 1, and FIG. 7 is a diagram showing a new format information generation process in FIG. In the image capture process of step ST10 in FIG. 6, the surface of the form 20 to which the new form ID is attached is imaged on the photoelectric conversion means 21 such as an image scanner by an optical system. The photoelectric conversion unit 21 acquires an image of the form 20 and photoelectrically converts it to generate image data. The image data is stored in the image memory 22. In the new format information generation process of step ST20, the new format information generation means 31 recognizes the form ID of the form 20 and generates new format information for the form 20. For example, Japanese Patent Application No. 7-14212 filed earlier by the applicant of the present application.
The format production rules are applied as described in No. 2 specification and drawings. The new format information generation process of step ST20 is performed by steps ST21 to ST2 of FIG.
In step ST21 to ST25, the following (1) image analysis and (2) collation search are performed.

(1) Image analysis In step ST21, based on the image data stored in the image memory 22, the image analysis means 31a detects (analyzes) the positions of all the line segments and the characters included in the form image. Then, the image analysis result is sent to the matching search means 31b. FIG. 8 is a projection view of a form image for explaining the method of calculating the coordinates of the position of the line segment and the position of the character. In FIG. 8, HX and HY are projections of the form image in the X-axis direction and the Y-axis direction (sum of black points).
And the outside (background) of the form image is formed by only black dots. Then, the projection process in the X-axis direction obtains the projection HX in the X-axis direction, and by comparing the projection HX with the X-axis direction frame detection slice SX, the line segment parallel to the X-axis can be detected. The Y-axis coordinate value of each line segment is 6,
It becomes 8, 9, ... Similarly, a projection process HY in the Y-axis direction is obtained by the projection process in the Y-axis direction, and a line segment parallel to the Y-axis can be detected by comparing the projection HY with the Y-axis direction frame detection slice SY. . The X-axis coordinate values of each line segment are 1, 7, 9, .... The values of the X-axis direction frame detection slice SX and the Y-axis direction frame detection slice SY can be changed.

From the detected line segment parallel to the X-axis and the line segment parallel to the Y-axis, the coordinates of their intersections can be obtained. For example, the physical distribution route field 20 ₃ is Y for x = 1.
A line segment in the axial direction, a line segment in the Y-axis direction at x = 13, and y = 2
This is a rectangular area surrounded by a line segment in the X-axis direction of 7 and a line segment in the X-axis direction of y = 32. In the field position information of the physical distribution route field 20 ₃ , the coordinate position of the starting point at the upper left with respect to the origin of the end of the form 20 is p (1,27), and the coordinate position of the ending point at the lower right is p (13,32). ), The coordinate position p (1,27) of the starting point and the width w = 1 in the X direction
2 and the coordinate value of the height h = 5 in the Y direction. To detect the position of a character in a character block that is not enclosed in a rectangular frame like the billing month field 20 ₅ ,
For example, the following method may be used. That is, the billing month field 20 ₅ consisting of a character block is obtained by comparing a portion having a low height and a wide width in the projection HX in the X-axis direction with a portion having a low height and a wide width in the projection HY in the Y-axis direction. It is possible to detect the position of the frame surrounding each character inside, that is, the position of each character. Therefore, cut out billing month field 20 ₅ consisting of a text block, by performing the recognition of each character of the character block using the recognition dictionaries, a character recognition result is obtained.

The image analysis means 31a detects (cuts out) a character block at a character position from the image analysis result as needed, and detects the character block from the detected character block using a recognition dictionary (not shown). To recognize. Below the physical distribution route field 20 ₃ , a shipping date field 20 ₄ for writing the shipping date is provided. Shipping date field 20 ₄ is composed of a character block. The image analysis means 31a uses the shipping date field 20.
Each character in the character block in ₄ is recognized. This character recognition recognition result is also sent to the collation search means 31b. It also to be form ID 20 ₁ also simultaneously recognized. An example of the image analysis result obtained by the image analysis means 31a is shown in FIG.

(2) Collation search In step ST22, the collation search means 31b detects the line segment and the character detected by the image analysis means 31a and the format generation rule corresponding to the form ID 20 read from the format generation rule storage means 31c. To search for line segments and characters that match the format generation rule. Then, the collation search means 31b determines the position on the XY coordinate axis of the read field (for example, the customer number field 20 ₂ ) of the field number i = 1, and the field position information which is the information of the position is determined in step ST.
At 23, it is stored in a storage means (not shown).

FIG. 10 is a diagram showing an example of the format generation rule. The format productions, form ID20 number reading field corresponding to ₁ n of the form 20 as a reading target, reading the field names, and characteristics of each read field is defined. As characteristics of each reading field, contents suitable for each reading field such as attribute information of described items, size (dimension) of the field frame or character block, existing position, and search method are defined. For example, for a customer number field 20 _2, the size of the field frame (width w = 8, the height h = 2), the location of the approximately (above the maximum frame, the character block customer number right) provisions Has been done. Such a format generation rule and the image analysis result obtained by the image analysis means 31a are matched by the matching search means 31b to search for an image analysis result that matches the format generation rule, and the customer number field 20 ₂ Create field information.

At step ST24, it is determined whether or not the positions of all the read fields 20 _{2 to} 20 ₄ with the read field number n = 3 have been determined. If not, the field number i is incremented by +1 ( Increment), and the processing of steps ST22 and ST23 is performed on the next reading field (for example, the physical distribution route field 20 ₃ ) in the same manner as above. Process is completed for the shipping date field 20 _4, in step ST24, the retrieval processing for all the read field number of fields n = 3 is determined to have ended, and proceeds to the next step ST25. In step ST25, the position information of all the read fields is displayed on the display, for example, and the operator confirms whether the format information for the form 20 is properly created. If it is not correct, the new format information is completed by correcting it.

FIG. 11 is a diagram showing the format information of the form 20 generated by the new format information generating means of FIG. Based on the format generation rule, format information for the form 20 as shown in FIG. 11 is generated. Here, each read field in the format information is divided into a first type read field and a second type read field, depending on how the position is designated. In the customer number field 20 ₂ and the physical distribution route field 20 ₃ which are first type reading fields, the upper left position P (x ₁ , y ₁ ) of each field frame is designated with the upper left end of the form 20 as the origin O. Shipping date field 20 which is the second type field
_{The position of 4} is designated with the point on the line segment drawn on the form 20 as the origin O ^* . The origin O ^{* in} this case is represented by a point (1, 27) on the frame surrounding the physical distribution route field 20 ₃ . The positional information from the origin O ^* constitutes the positional information of the shipping date field 20 ₄ . For example, as the shipping date field 20 _4, the frame described portions is small, the two reading fields are employed. Cutting errors and printing errors may be considered in the form. Generally, the printing error is sufficiently smaller than the cutting error. Therefore, as in the present embodiment, the first-type read field and the second-type read field are managed separately, which improves the character recognition accuracy.

FIG. 12 is a diagram showing the origin of the shipping date field. If the position information of the shipping date field 20 ₄ is specified from the origin O at the upper left end of the form 20, there is a cutting error in reality, and when the items described in the shipping date field 20 ₄ are recognized as characters, the OCR is the physical distribution. Route field 20 ₃
There is a possibility that character recognition will be performed across. Since characters are entered in the dispatch date field 20 ₄ is small, so that the accuracy of character recognition is lowered. Therefore, the upper left end of the physical distribution route field 20 ₃ is the inner frame origin O ^* (x _o ^* , y _o ^* ).
The position information of the shipping date field 20 ₄ is designated as. That is, the point P (x ₁ ^* , y _{1) in the} upper left part of the shipping date field 20 ₄ viewed from the inner frame origin O ^* (x _o ^* , y _o ^* )
^*), And the date of shipment field 20 ₄ of height h ^* and the width w ^* is specified. As a result, the position of the shipping date field 20 ₄ becomes clear, it increases the character recognition accuracy by OCR.
After the format information is generated, the process proceeds to step ST30 in FIG. In the margin setting process of step ST30, the margin setting means 32 refers to the format generation rule and sets the margin range for each field of the format information. For example, the customer number field 20 ₂ and the distribution route field 20 _{3 of} the first type reading field
For, the margin range of ± 1.0 mm on the form is set. Shipment date field 20 of type 2 reading field
_{For 4} , set a margin range of ± 0.5 mm.

Next, the search registration process of step ST40 is performed. The search registration process of step ST40 is composed of steps ST41 to ST45. Step ST
At 41, the search and registration means 33 reads the first format information stored in the format information memory 40 together with the group name A attached to the format information. In step ST42, the search registration means 33
The format information of the form 20 generated in step ST30 is compared with the field position information of the format information read from the format information memory 40. FIG.
FIG. 7 is a diagram showing a relationship between a margin range set in step ST30 of FIG. 6 and a field position. For example, with respect to the reading field F of the customer number field 20 ₂ indicated by diagonal lines, the margin range is set by the margin M ₁ set outside and the margin M ₂ set inside. Form 2 for height h and width w of reading field F
The height and width of the margin M ₁ at 0 are h + 1,
w + 1. The height and width of the margin M ₂ in the form 20 are h−1 and w−1, respectively. Step ST42
In the comparison, if the field position is between the margin M ₁ and the margin M ₂ , it is determined as the same field position information. The search / registration means 33 performs comparison for each field, and if the difference between all the field position information is within the margin range set in step ST30, the format information is considered to be the same. Then, in step ST43, the form ID of the form 20 is added with the group name A of the read format information and stored in the format information memory 40.

When it is determined that the format information of the group name A read from the format information memory 40 cannot be regarded as the same as the format information of the form 20 as a result of the comparison at step ST42, the format information memory 40 at step ST43. It is checked whether all the format information stored in the has been compared. When the number of format information for which comparison has been completed has not reached the number k of format information stored in the format information memory 40, the process returns to step ST41. That is,
The comparison is repeated in the order of the format information of the group names B, C, ... And the format information which can be regarded as the same as the format information of the form 20 is searched. If the target format information is not extracted by repeating the comparison, a new group name is added to the created format information and the form ID of the form 20 in step ST45, and these are added to the format information memory. 40 and registered.

FIG. 14 is a diagram for explaining registration in the format information memory shown in FIG. In the format information memory 40, each format information and group name A, B,
C, ... Are registered correspondingly. And OCR
The form IDs having the same format information that is referred to when they are read are registered under the same group name. After the search registration process of step ST40, step S
At T50, confirmation is done. Display control means 34
Reads out the format information corresponding to the form ID of the form 20, superimposes it on the image data stored in the image memory 22, and displays the image on the display 35. The operator confirms that the read format information matches the form 20 and ends the processing.

As described above, in this embodiment, the photoelectric conversion means 21, the image memory 22, the new format information generation means 31, the margin setting means 32, the search registration means 33, and the format information memory 40 are provided. I have it. Then, the format information of the form 20 having the new form ID is generated by the new format information generation means 31, and the margin setting means 32 and the search registration means 33 can identify the format information of the form 20 from the registered format information. Information is extracted and its form ID2
0 ₁ is registered in the format information memory 40 with the same group name as the format information that can be regarded as the same. Therefore, when there is format information that can be regarded as the same, it is not necessary to newly register the format information, and the capacity of the format information memory 40 can be reduced. This also shortens the search time when reading by OCR. Further, since each reading field is divided into the first type reading field and the second type reading field to specify the position information and set the margin range, the character recognition accuracy by OCR is sufficiently ensured.

The present invention is not limited to the above embodiment, and various modifications can be made. For example, there are the following modifications. (1) The new format information generation means 31 refers to the format generation rule to convert the form 2 from the image data.
Although the format information of 0 is generated, it is also possible to measure the form 20 with a ruler to obtain the field position information and define the attribute information with a keyboard or the like. (2) The form of the form 20 or the fields read by OCR are not limited to those shown in FIG. (3) In the above embodiment, an example of reading a character by OCR is shown, but it is effective not only for a character but also for a description that a specific symbol is selected by a symbol or a circle.

[0029]

As described in detail above, according to the first aspect of the present invention, new format information for an unregistered form is generated by the new format information generation process, and the margin range set by the margin setting process is used. Then, the format information that can be regarded as the same as the new format information in the search registration processing is extracted from the registered format information. Then, the group name attached to the format information that can be regarded as the same is attached to the form ID of the target form and registered. Therefore, it is not necessary to register the format information in the format information memory for each form ID of the unregistered form, and the capacity of the format information memory can be reduced.
Further, this makes it possible to shorten the search time for the format information when reading the description items on the form by the OCR. According to the second invention, the new format information generation process in the first invention takes in the image data of the unregistered form in which the format information is not registered, and the image data is generated based on the format generation rule created in advance. Since the new format information is generated by processing, the new format information can be automatically generated. Therefore, for example, the troublesomeness of measuring each reading field of the form with a ruler or the like is reduced.

According to the third invention, a new format information generating means, a margin setting means for setting a margin, a format information memory for storing format information by a group name, and the same format as the format information of an unregistered form. The format information generating device is provided with a search / registration means for searching information. Therefore, it is not necessary to register the format information in the format information memory for each form ID of the unregistered form, and the capacity of the format information memory can be reduced. Further, this makes it possible to shorten the search time for the format information when reading the description items on the form by the OCR. According to the fourth invention, the margin setting means in the third invention is configured to set the margin range for each reading field by using the margin value stored in association with each reading field in the format information. Therefore, the margin range suitable for each reading field is set, and the accuracy of character recognition can be improved. According to the fifth invention, each reading field in the third or fourth invention is composed of a first type reading field and a second type reading field, and the margin setting means corresponds to the first type reading field. The margin range for each reading field is set using the first margin value and the second margin value corresponding to the second type reading field and different from the first margin value. Therefore, for example, even if there is a cutting error in the form, the character recognition accuracy by OCR does not decrease.

[Brief description of drawings]

FIG. 1 is a functional block diagram of a format information generation device showing an embodiment of the present invention.

FIG. 2 is a diagram illustrating format information of the form in FIG.

FIG. 3 is a diagram illustrating a form of a form.

4 is a configuration block diagram of a new format information generating unit in FIG.

5 is a diagram showing a form 20 in FIG. 1. FIG.

6 is a flowchart showing a format information generation process using the format information generation device of FIG.

7 is a diagram showing a new format information generation process in FIG.

FIG. 8 is a projection diagram of a form image for explaining a method of calculating coordinates of a position of a line segment and a position of a character.

FIG. 9 is a diagram showing an example of an image analysis result obtained by the image analysis means 31a.

FIG. 10 is a diagram showing an example of a format generation rule.

11 is a diagram showing format information of the form 20 generated by the new format information generating means of FIG.

FIG. 12 is a diagram showing the origin of a shipping date field.

13 is a diagram showing a relationship between a margin range set in step ST30 of FIG. 6 and a field position.

FIG. 14 is a diagram illustrating registration in the format information memory in FIG. 1.

[Explanation of symbols]

 20 forms 21 photoelectric conversion means 22 image memory 31 new format information generation means 32 margin setting means 33 search registration means 34 display control means 40 format information memory ST10 image import processing ST20 new format information generation processing ST30 margin setting processing ST40 search registration processing

Claims (5)

[Claims] 1. A unique format for reading reference to a form corresponding to each form ID in order to read necessary items with an optical character reading device for the form having a fixed format for each form ID. In a format information generation method for generating and registering information, new format information generation processing for generating the new format information for an unregistered form in which the format information is not registered, and a format generation rule created in advance Margin setting processing for setting a margin range for the new format information based on the above, and what is considered to be the same as the new format information within the margin range is extracted from the registered format information, and the extracted The same group name as the registered format information has the form ID of the unregistered form. When there is no registered format information that can be regarded as the same, the search registration processing for registering the new format information, the form ID thereof, and the new group name in association with each other is performed. Characteristic format information generation method. 2. The new format information generation process,
2. The image data of an unregistered form in which the format information is not registered is fetched, the image data is processed based on a format generation rule created in advance, and the new format information is generated. Format information generation method. 3. A read reference is made to a form corresponding to each form ID in order for the optical character reader to read necessary items for the form having a fixed format for each form distinguished by the form ID. In a format information generation device for respectively generating and registering unique format information for use in a device, photoelectric conversion means for acquiring an image of a target form and converting it into image data, an image memory for storing the image data, the image memory A new format information generating means for recognizing the form ID and generating new format information based on the image data stored in, and a margin range for the new format information based on a format generation rule created in advance. Margin setting means to set, a plurality of the format information, and Among the format information stored in the format information memory, a group name assigned to each format information, a format information memory that stores a form ID group to which the format information is applied in association with each other, Format information that is considered to be the same as the new format information within the margin range is extracted, and the group name attached to the extracted format information and the form ID of the unregistered form are stored in the format information memory. If there is no format information that is registered and can be regarded as the same, a search / registration unit that stores the new format information, the form ID thereof, and the new group name in the format information memory in association with each other is registered. A format information generation device characterized by the above. 4. The margin setting means is configured to set the margin range for each reading field by using a margin value stored in association with each reading field in the format information. The format information generation device according to claim 3. 5. Each reading field in the format information includes a first-type reading field represented by a positional relationship with an end of the target form as an origin, and an arbitrary point on a line segment drawn on the target form. And a second type reading field represented by a positional relationship with the origin as the origin. The margin setting means corresponds to the first type margin value corresponding to the first type reading field and the second type reading field. 5. The format information generating apparatus according to claim 3, wherein a margin range for each of the read fields is set by using a second margin value different from the first margin value.