JP3732254B2 - Format information generation method and format information generation apparatus - Google Patents

Description

[0001]
[Industrial application fields]
The present invention optically reads a character on a form having a reading field (reading area) in which letters, numbers, symbols, marks, and the like (in this specification, these are collectively referred to simply as “characters”) are written. The present invention relates to a format information generation method for generating format information, which is read information used for reference in reading an optical character reader (hereinafter referred to as OCR), and the format information generator.
[0002]
[Prior art]
In the conventional OCR, a form image of a certain format (format) is captured by an image scanner or the like, and the image data of the form image is stored in an image memory (image storage device). Then, using the format information specifying the reading field of the form (that is, the format information specifying the reading position and character type of the form), the reading field is cut out by the character cutting means from the image data stored in the image memory. The character on the form is recognized by extracting the character features in the extracted reading field by the character recognition means.
That is, in order to identify the format of the form, a form ID represented by numbers or the like is entered or printed at a predetermined position (for example, 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, the reading field is cut out from the form image, and the characters in the cut out reading field are recognized as a dictionary. It has come to recognize using.
[0003]
[Problems to be solved by the invention]
However, the format information used in the conventional OCR is created by the operator manually measuring the position and size of each reading field on the form with a ruler or the like. The created format information is stored (registered) in an OCR storage device by an operator using a personal computer or the like. Therefore, there are problems such as the following (a) to (c), and it is difficult to solve them.
(A) When there are a plurality of forms with the same form ID, the operator selects one form judged to be appropriate from the plurality of forms in consideration of work efficiency, and the position of the reading field based on the form. Field information is created by measuring the size and size with a ruler. As described above, since the operator manually creates the field information, the work efficiency is not good.
(B) The operator creates format information from one form in consideration of work efficiency. However, the position of the reading field on the form varies depending on the printing error and the cutting error. When the printing error or cutting error of these forms is large, the error of the created format information also becomes large.
(C) As described above, field information is usually created from a single form in consideration of work efficiency, and it is difficult to detect the mistake even if a different field is selected as a reading position by mistake. .
In the present invention, as the problem that the conventional technology has, since the operator creates the format information manually, the work efficiency is not good, and the format information is created from a single form. There is an error in the format information created due to cutting errors, etc., or the error is large, and even if a different field is selected as the reading position by mistake, it is difficult to detect abnormal formatting. It is an object of the present invention to provide a solved format information generation method and format information generation apparatus.
[0004]
[Means for Solving the Problems]
In order to solve the above-described problems, the first invention provides a form ID for form identification and format information for reference used when reading the character on the form having a reading field in which the character is entered by OCR. In the format information generation method to be generated, an image acquisition process for acquiring images of a plurality of forms having the same format with the same form ID and outputting a plurality of form images, a format information creation process, and a comparison The selection process is executed. Here, in the format information creation process, the position of the line segment and / or the position of the character included in each form image is detected, and the format generation rule corresponding to the form ID is applied to the detection result. Determine the reading field position. Then, when all the reading fields can be detected for one form image, it is output as OK format information, and when there is one or more reading fields that cannot be detected, it is output as NG format information for each form. In the comparison and selection process, the OK format information is compared with each other,Difference in information indicating the position and size of the same reading field (ie,Difference between those OK format information)One of the most average OK format information is selected by majority vote from those having a predetermined margin value (margin value) or less, and this is output as format information of the form ID.
[0005]
In the second invention, the comparison and selection process of the first invention has the following processing contents. That is, in the comparison and selection process of the second invention, when there is one OK format information output by the format information creation process, the OK format information is selected and output unconditionally. On the other hand, when there are a plurality of OK format information output by the format information creation process, if there is one or more sets of OK format information that can be regarded as the same format information not exceeding a predetermined margin value, the most among them is One piece of average OK format information is selected and output by majority decision. If there is no set of OK format information, a process of generating a format information generation error (ie, format information generation NG) is performed.
According to a third aspect, in the format information generation method according to the first or second aspect, the position and size of the reading field is a rectangular shape surrounding the reading item based on XY coordinates with the origin 0 at the upper left of the form. The upper left position p (x, y), the width w in the X-axis direction, and the height h in the Y-axis direction are used. The format information includes the form ID, the number of read fields, and p (x, y), w, and h information for each read field.
According to a fourth aspect, in the format information generation method of the second aspect, when the result of the comparison and selection process is a format information generation error, the result is displayed on the screen of the display means.
[0006]
According to a fifth aspect of the invention, in the format information generating apparatus for generating the format ID for reference used when reading the form ID having the form ID for the form identification and the form on the form having the reading field filled with the character by OCR, Image acquisition means for acquiring images of a plurality of forms having the same format with the same form ID and outputting a plurality of form images, and format information creating means are provided. Here, the format information creating means detects the position of the line segment and / or the position of the character included in each form image, and applies the format generation rule corresponding to the form ID to the detection result. The reading field position is determined and output as OK format information when all the reading fields can be detected for one form image, and as NG format information when there is one or more reading fields that cannot be detected for each form. To do.
In the fifth aspect of the invention, furthermore, all the OK format information output from the format information creating means is compared,The difference in information indicating the position and size of the same reading field (that is, the difference between the OK format information)Comparing means for detecting OK format information that can be regarded as the same format information equal to or less than a predetermined margin value and OK format information that is regarded as different OK format information, and OK format information that can be regarded as the same format information detected by the comparing means And counting means for selecting the maximum OK format information of the count value of the counting means and outputting this as format information of the form ID is provided.
[0007]
[Action]
According to the first and third inventions, form images of a plurality of forms with the same form ID are acquired by the image acquisition process. The position of the line segment and / or the position of the character in each form image acquired by the image acquisition process is detected by the format information creation process. That is, in this format information creation process, for example, the detection result and the format generation rule are collated (matched), and the detection result that matches (matches) the format generation rule is searched for each reading field, Determine the position of the reading field. Then, in the format information creation process, when all the read fields can be detected for one form image, it is output as OK format information, and when there is one or more read fields that cannot be detected, it is output as NG format information.
In the comparison and selection process, a plurality of OK format information output from the format information creation process is compared, and one OK format is selected by majority vote from among those whose difference between the OK format information is a predetermined margin value or less. This is output as the format information of the form ID.
[0008]
According to the second and third inventions, one OK format information is selected from the OK format information output by the format information creation process by the comparison and selection process, for example, as follows. That is, when there is one OK format information output by the format information creation process, the OK format information is selected and output as it is. On the other hand, when there are a plurality of OK format information output by the format information creation process, if there are one or more sets of OK format information that can be regarded as the same format information, one OK format information is determined by majority vote from among them. Select and output. At this time, if the set of OK format information does not exist, a format information generation error occurs.
According to the fourth invention, when the result of the comparison and selection process of the second invention is a format information generation error, the result is displayed on the screen of the display means. Then, the operator can select optimal OK format information based on the information on the screen, for example.
[0009]
According to the fifth aspect, the image acquisition unit acquires the form images of a plurality of input forms and sends them to the format information creation unit. The format information creating means detects the position of the line segment and / or the position of the character included in each form image. Then, for example, the detection result and the format generation rule are matched, the detection result that matches the format generation rule is searched for each reading field, the position of the reading field is determined, and OK format information or NG format is determined. Output information for each form.
All OK format information output from the format information creating means is compared by the comparing means, and OK format information that can be regarded as the same format information and OK format information that is regarded as different OK format information are detected. The number of OK format information that can be regarded as the same format information is counted by the counting means, the OK format information having the maximum count value (count value) is selected by the selecting means, and is output as the format information of the form ID.
[0010]
【Example】
In the embodiment of the present invention, the format information generation device reads a plurality (z) of the forms 11 to 1z with the same form ID, matches the contents of the forms 11 to 1z with the format generation rule, A method of creating format information corresponding to each of the forms 11 to 1z and selecting and outputting one optimum format information from the plurality of format information will be specifically described. Therefore, FIG. 2 shows an example of one form 11 out of a plurality (z) with the same form ID to be read.
In the form 11 of FIG. 2, a form ID 20 is entered or printed at a predetermined position, for example, at the upper right, in order to identify the format. The form ID 20 is provided at a common position of the forms 11 to 1z having the same format, and is usually expressed by a number unique to the form of 2 to 5 digits (a 4-digit number in FIG. 2). Further, the form 11 is provided with a plurality (n) of reading fields (items) FIELD in which characters to be read are entered (in FIG. 2, four reading fields FIELD 1 to FIELD 4 are provided. ). Of these four reading fields FIELD 1 to FIELD 4, for example, the upper reading field FIELD 1 is the billing month field, the lower reading field FIELD 2 is the customer number field, the lower left reading field FIELD 3 is the logistics route field, The lower right reading field FIELD 4 is a grand total field.
In the present embodiment, these four reading fields FIELD 1 to FIELD 4 are optically read, their positions and sizes are detected, a plurality of format information corresponding to the respective forms 11 to 1z are created, and the plurality of formats are detected. Select the most appropriate piece of information.
[0011]
FIG. 3 is a diagram for explaining the expression format of the positions of the reading fields FIELD 1 to FIELD 4 shown in FIG.
The positions and sizes of the respective reading fields FIELD 1 to FIELD 4 in FIG. 2 are determined based on the XY coordinates with the origin 0 as the upper left of each of the forms 11 to 1z, and the rectangular upper left position p1 (x1, y1), X It is expressed using the width w in the axial direction and the height h in the Y-axis direction.
FIG. 1 is a schematic functional block diagram of a format information generating apparatus showing an embodiment of the present invention.
This format information generation apparatus reads a plurality (z) of forms 11 to 1z with the same form ID 20 as shown in FIG. 2, and a plurality of forms 11 to 1z corresponding to the forms 11 to 1z based on the read result. Is a device that selects and outputs one of the most suitable format information, and has a photoelectric conversion means 31 that reads each of a plurality of input forms 11 to 1z, on the output side The image storage means 32 and the control unit 40 are connected to each other. The photoelectric conversion means 31 is configured by an image scanner or the like, optically reads each of the input forms 11 to 1z, acquires the form images IMAGE 1 to IMAGE z, and includes the forms 11 to 1z in the respective forms 11 to 1z. It has a function as an image acquisition means for recognizing the form ID 20. The image storage means 32 is a memory for storing a plurality of form images IMAGE 1 to IMAGE z read by the photoelectric conversion means 31 in the form of image data.
[0012]
The control unit 40 functions to control the entire format information generating device, create a plurality of format information based on the image data stored in the image storage means 32, and select and output the optimum one of the format information And one or a plurality of personal computers. In the control unit 40, a format information creating unit 41, a comparison / selection unit 42, an input / output control unit 43, and the like are provided.
The format information creating means 41 creates and outputs OK format information COD1 to CODm or NG format information ALFNG based on each image data stored in the image storage means 32. That is, in the format information creating means 41, when all the reading fields FIELD can be detected for one form image IMAGE, the format information is output as OK format information COD1 to CODm, and the reading fields FIELD that could not be detected are output. When there is one or more, it has a function of outputting the format information as NG format information ALFNG, and has an image analysis means 41a, a character recognition means 41b, and a collation search means 41c.
[0013]
Based on each image data stored in the image storage means 32, the image analysis means 41a analyzes the position of the line segment and the character contained in each of the form images IMAGE 1 to IMAGE z and outputs the image analysis result. The character recognition means 41b and the collation search means 41c are connected to the output side. The character recognizing unit 41b has a function of extracting a character at the position from the character position of the image analysis result analyzed by the image analyzing unit 41a and recognizing the character using a recognition dictionary or the like. A means 41c is connected. A format generation rule storage unit 51 provided outside the control unit 40 is connected to the collation search unit 41c. The format generation rule storage means 51 is a memory for storing a format generation rule created in advance corresponding to the form ID 20 of the forms 11 to 1z to be read.
[0014]
The collation search unit 41c matches the image generation result analyzed by the image analysis unit 41a with the format generation rule corresponding to the form ID 20 read from the format generation rule storage unit 51, and matches the format generation rule. The image analysis result is searched for each reading field FIELD, the position of the reading field FIELD is determined, and the format information (OK format information COD1 to CODm or NG format information ALFNG) is created, and its output side The format information storage means 52 and the comparison / selection means 42 are connected. The format information storage means 52 is a memory for storing the format information (OK format information COD1 to CODm or NG format information ALFNG) output from the collation search means 41c.
The comparison / selection means 42 compares all the OK format information COD1 to CODm stored in the format information storage means 52, selects the optimum one of the OK format information COD, and uses this to select the format of the form ID 20 It has a function of outputting as information, and has a comparison means 42a, a counting means 42b, and a selection means 42c.
[0015]
The comparison unit 42a compares all the OK format information COD1 to CODm stored in the format information storage unit 52, and the OK format information COD that can be regarded as the same format information not more than a predetermined margin value read from the margin storage unit 53. The OK format information COD regarded as different OK format information is detected, and the detection result is stored in the comparison result storage means 54. The counting means 42b is connected to the output side. The counting means 42b has a function of counting the number of OK format information COD that can be regarded as the same format information stored in the comparison result storage means 54, and a selection means 42c is connected to the output side thereof. The selecting unit 42c has a function of selecting the OK format information COD having the maximum count value of the counting unit 42b and outputting it as the format information of the form ID 20, and registering format information outside the control unit 40 on the output side. Means 55 are connected. The format information registration means 55 stores (registers) the OK format information COD of the form ID 20 output from the selection means 42c together with the form ID 20, and is constituted by a storage device such as a magnetic drum or a magnetic disk.
[0016]
Further, an input / output control unit 43 is provided in the control unit 40, and an input operation unit 56 and a display unit 57 outside the control unit 40 are connected to the input / output control unit 43. The input operation unit 56 includes a keyboard and a mouse, and has a function of inputting data to the control unit 40 and instructing work contents. The display means 57 is composed of a display such as a CRT or a liquid crystal, and has a function of displaying output data of the control unit 40 on the screen. Input / output with respect to the control unit 40 of the input operation means 56 and the display means 57 is controlled by the input / output control means 43.
FIG. 4 is an overall flowchart of format information generation processing in the format information generation method of the present embodiment using the format information generation apparatus as described above.
[0017]
In the format information generation process of FIG. 4, in the image acquisition process of step S <b> 1, a plurality of forms 11-1 z with the same form ID 20 are read by the photoelectric conversion means 31, and the plurality of form images IMAGE 1-1. IMAGE z is stored in the image storage means 32 in the form of image data. Next, in the format information creation process in step S2, the format information creation means 41 reads the image data stored in the image storage means 32, and the position of the line segment included in each of these form images IMAGE 1 to IMAGE z. And the character position are analyzed, the analysis result is matched with the format generation rule corresponding to the form ID 20 stored in the format generation rule storage means 51, and the analysis result matching the format generation rule is matched with each read field. Each field 1 to FIELD n is searched to determine the positions of the reading fields FIELD 1 to FIELD n, and format information (OK format information COD 1 to CODm or NG format information ALFNG) is output. This format information is stored in the format information storage means 52 in step S3.
Thereafter, in the comparison / selection process in step S4, the comparison / selection means 42 reads the OK format information COD1 to CODm stored in the format information storage means 52 and compares them, and the difference between the OK format information COD1 to CODm.(Ie, the difference in information indicating the position and size of the same reading field)However, the average one OK format information COD is selected from the ones below the predetermined margin value stored in the margin storage means 53 by majority decision, and this is output as the format information of the form ID. The output format information is registered in the format information registration means 55 in step S5.
[0018]
Next, specific examples of the image acquisition process (1), the format information creation process (2), and the comparison / selection process (3) will be described with reference to FIGS.
(1) Image acquisition processing
When the format information generation process is started by a command from the input operation unit 56, in step S1 of FIG. 4, the photoelectric conversion unit 31 has a plurality of forms 11 to 11 having the same form ID 20 as shown in FIG. 1z form images IMAGE1 to IMAGEz are photoelectrically converted (for example, blank pages are converted to "0" for bit units, line segments (ruled lines) and character portions are converted to "1") as image data The image is stored in the image storage unit 32. In step S1, the form ID 20 described at the upper right of the forms 11 to 1z in FIG. 2 is also recognized.
FIG. 5 is a diagram showing an example of the form image IMAGE 1 read in step S1 of FIG.
The form image IMAGE 1 is represented by, for example, XY coordinates with the upper left as the origin (0, 0). Numbers 0 to 24 on the X axis and numbers 0 to 34 on the Y axis are given for convenience of explanation. Further, rectangular broken lines representing the billing month field (reading field FIELD 1) and the distribution route field (reading field FIELD 3) are given for convenience of explanation and do not appear on the form image IMAGE1. It is assumed that, for example, four reading fields FIELD 1 to FIELD 4 to be read are determined in advance by the format generation rule. For convenience of explanation, coordinate positions p (x, y) are attached to the four corners of each rectangular reading field FIELD 1 to FIELD 4, but these do not appear on the form image IMAGE 1.
[0019]
The image data of the form image IMAGE 1 represented by such XY coordinates is stored in the image storage means 32.
Note that in step S1 in FIG. 4, if the form ID 20 cannot be recognized due to light printing or the like, the following measures can be taken.
That is, when the form ID 20 cannot be recognized, the control unit 40 displays a form ID code input screen on the screen of the display means 57, and the form image IMAGE 1 of the form (for example, 11) for which the form ID 20 could not be recognized. May be displayed, and the operator may input the form ID 20 using the keyboard of the input operation means 56 while viewing the form image IMAGE 1 on the screen.
[0020]
(2) Format information creation processing
In step S2 of FIG. 4, after image analysis processing is performed by the image analysis means 41a, for example, as described in the specification and drawings of Japanese Patent Application No. 7-142122 filed earlier by the applicant of the present application. Then, the collation search process is performed by the collation search means 41c to create the format information.
First, the image analysis unit 41a reads the image data stored in the image storage unit 32, and detects the positions of all line segments and characters included in each form image IMAGE1 to IMAGEz by a projection method or the like ( The image analysis result is sent to the collation search means 41c. Further, from the image analysis result analyzed by the image analysis unit 41a, the character recognition unit 41b detects (cuts out) the character block at the character position as necessary, and recognizes the detected character block using the recognition dictionary. Is recognized, and the character recognition result is sent to the collation search means 41c.
That is, in the image analysis process, a line segment parallel to the X axis is detected by the projection process in the X axis direction with respect to the form image IMAGE 1 in FIG. 5, and further, parallel to the Y axis by the projection process in the Y axis direction. Detects line segments. From the detected line segment parallel to the X axis and line segment parallel to the Y axis, the coordinates of the intersection can be obtained.
[0021]
For example, the customer number field (FIELD 2) in FIG. 5 includes a line segment in the Y-axis direction of x = 7, a line segment in the Y-axis direction of x = 23, and a line segment in the X-axis direction of y = 6. , A rectangular region surrounded by a line segment in the X-axis direction of y = 8. The field information of this customer number field (FIELD 2) is shown in FIG. 3 because the coordinate position of the upper left starting point is p (7, 6) and the lower right end point is p (23, 8). Thus, it is represented by the coordinate position p (7, 8) of the starting point, the width w = 16 in the X-axis direction, and the height h = 2 in the Y-axis direction.
To detect the position of a character in a character block that is not surrounded by a rectangular frame such as the billing month field (FIELD 1), the projection result in the X-axis direction is compared with the projection result in the Y-axis direction. The position of the frame surrounding each character in the billing month field (FIELD 1) consisting of character blocks, that is, the position of each character can be detected. Therefore, a character recognition result is obtained by cutting out the billing month field (FIELD 1) made up of character blocks by the character recognition means 41b and recognizing each character in the character block using a recognition dictionary or the like.
FIG. 6 is a diagram illustrating an example of the format generation rule stored in the format generation rule storage unit 51.
In this format generation rule, the number of read fields FIELD corresponding to the form ID 20 of the forms 11 to 1z to be read, the name of the read field FIELD, the characteristics of the read fields FIELD 1 to FIELD 4, and the like are defined. The characteristics of each of the reading fields FIELD 1 to FIELD 4 are defined as contents suitable for each of the reading fields FIELD 1 to FIELD 4 such as the size (dimension) of the field frame and the character block, the existing position, and the search method. .
[0022]
Next, in the collation search means 41c that performs collation search processing, the line segments and characters detected by the image analysis means 41a and the character recognition means 41b and the form ID 20 read from the format generation rule storage means 51 in FIG. Such a format generation rule is matched, and line segments and characters that match the format generation rule are searched. Then, the collation search means 41c sequentially determines the positions on the XY coordinate axes of the respective reading fields FIELD1 to FIELD4, and sequentially stores each field information as the position information in the format information storage means 52. If the position of the reading field FIELD cannot be determined (detected) (NG), the NG information is also stored in the format information storage means 52. When the field information of each of the read fields FIELD 1 to FIELD 4 is stored in the format information storage means 52, the format information including the field information is created.
When all the reading fields FIELD 1 to FIELD 4 can be detected for one form image IMAGE, the format information is OK format information COD. When there is one or more reading fields FIELD that could not be detected, the format information Is NG format information ALFNG. An example of this format information is shown in FIG.
The format information in FIG. 7 includes information such as the form ID 20 = 1234 corresponding to the forms 11 to 1z to be read, the number of reading fields FIELD = 4, and the field positions and attributes of the reading fields FIELD1 to FIELD4. ing. When there is a reading field FIELD that could not be detected, for example, an NG mark is added to the reading field FIELD column.
[0023]
(3) Comparison selection process
In step S4 of FIG. 4, the comparison / selection means 42 reads out a plurality of OK format information COD1 to CODm with the same form ID 20, compares the OK format information COD1 to CODm, and averages them among them. One format information is selected by majority vote. At this time, when there are a plurality of OK format information CODs with the same form ID 20 and there is no set of OK format information COD regarded as the same format information, the format information is set as NG format information ALFNG. .
Hereinafter, a method for selecting format information by majority decision will be described in detail with reference to FIGS.
FIG. 8 is a diagram showing OK format information COD11 to CODmn stored in the format information storage means 52. As shown in FIG.
For example, a plurality of form images IMAGE 1 to IMAGE z are acquired for the same form ID 20 by the photoelectric conversion means 31, and format information is created by the format information creation means 41 for these form images IMAGE 1 to IMAGE z. As a result, the format information of m images of the form images IMAGE 1 to IMAGE m becomes OK format information COD, and for each item of the reading fields FIELD 1 to FIELD n, as shown in FIG. It is assumed that the format information of CODmn has been created.
[0024]
In FIG. 8, the OK format information generated as a result of executing the format information creation process for the form image IMAGE k is CODk, and the format information item corresponding to the reading field FIELD j that is the item of CODk is the OK format information CODkj. The OK format information CODkj is expressed by the coordinates of the form image IMAGE k and the reading field FIELD j.That is, OK format information CODkj is a form image. IMAGE Read field of OK format information CODk corresponding to k FIELD j, and as shown in FIG. FIELD It consists of information indicating the position of j (for example, the upper left position p1 (x1, y1)) and the size (the width w in the X direction and the height h in the Y direction).
First, the comparison means 42a reads all the OK format information COD11 to CODmn for each form image IMAGE1 to IMAGEm from the format information storage means 52, and compares all the OK format information COD11 to CODmn. The comparison result is shown in FIG.For example, the same reading field of OK format information COD1 and COD2 FIELD Are calculated by calculating w1-w2 for the width w in the X direction, h1-h2 for the height h in the Y direction, and x1-x2, y1 for the upper left position P (x, y). -Y2.
In FIG. 9, the result of comparing the OK format information CODk corresponding to the form image IMAGE k and the OK format information CODj corresponding to the form image IMAGE j is described as CMPkj. As a result of comparison, “◯” indicates that the format information is the same, and “x” indicates that the format information is different. Then, all the OK formats of the OK format information COD1 to CODm are compared with each other, and the number of ◯ is counted by the counting means 42b. In FIG. 9, the number of o in the OK format information CODk is written as Ck. As a result, the format information with the most ○ is selected by the selection means 42 c as the format information of the form ID 20 and registered in the format information registration means 55.
In the majority test as described above, for example, the maximum value of the number of format information (number of forms) m for the same form ID 20 is set to 5, and the format information after OK format information COD6 is ignored without being subject to the majority test. May be. In this way, the majority test process can be simplified.
[0025]
A method for calculating the comparison result shown in FIG. 9 will be described below.
The comparison result CMPki is obtained by comparing the margin values of the OK format information CODk and the OK format information CODi stored in the margin storage means 53. If it is the same in the case of ◯ and the count number Ck is the maximum among C1, C2,..., Ck,..., Cm, the OK format information CODk is representative.
For example, when the comparison result CMP21 is obtained by comparing the OK format information COD1 and the OK format information COD2 in FIG.
(COD21-COD11): Margin1
(COD22-COD12): Margin2
(COD2j-COD1j): Marginj
(COD2n-COD1n): Marginn
・
・
・
As shown in FIG. 5, for each OK format information COD11, COD21,..., The difference between the two is taken, and it is determined whether or not the difference is less than or equal to a predetermined margin value (Margin1, Margin2,...). If all the read fields FIELD j are equal to or less than the margin value (Marginj), the OK format information COD1 can be regarded as the same as COD2.
[0026]
FIG. 10 is a flowchart showing the comparison / selection processing of the OK format information CODk and CODi as described above.
In this flowchart, the count number c of the counting means 42b is set to 0 in step S11, and 1 is set to k and i in step S12. In step S13, the format information CODk is read from the margin storage means 53, and in step S14, the format information CODi is read, and the process proceeds to step S15. In step S15, the comparison unit 42a takes the difference α between the OK format information CODk and CODi, and determines whether or not the difference α is equal to or less than a predetermined margin value. This margin value is preset according to the reading field FIELD (for example, 0.5 mm to 1.0 mm) and is stored in the margin storage means 53.
If the difference α is equal to or less than the predetermined margin value, the count number c is incremented by 1 in step S16, and i is incremented by 1 in step S17. If the difference α is larger than the margin value, the process proceeds directly to step S17. In step S18, it is determined whether or not up to m pieces of OK format information CODi have been processed. If not, the process returns to step S14 to repeat the above processing.
When the comparison process of m pieces of OK format information CODi is performed, the count number c is stored in step S19. Thereafter, the count number c is set to 0 in step S20, +1 is added to k in step S21, and the process proceeds to step S22. In step S22, it is determined whether or not up to m pieces of OK format information CODk have been compared. If not, the process returns to step S13 to repeat the above processing. When the comparison process is performed for up to m pieces of OK format information CODk in step S22, the selection means 42c selects the format information CODk from the count number c in step S23, and the comparison selection process ends.
[0027]
The following two cases (i) and (ii) are the conditions for OK in the majority test as described above.
(I) If there is one OK format information COD for one form ID 20, the OK format information COD is selected unconditionally, and the majority test result is OK.
(Ii) When a plurality of OK format information COD exists for one form ID 20
If there is one or more sets of OK format information COD that are regarded as the same format information, the majority test result is OK. On the other hand, if it does not exist, the majority test result is NG. This NG results in a format creation error when the difference between the OK format information for each of the forms 11 to 1z is larger than a preset margin value. In this case, for example, the NG is displayed on the screen of the display means 57 to notify the operator. The operator may select an appropriate form and create format information from the selected form.
[0028]
As described above, this embodiment has the following advantages (a) to (c).
(A) Since the format information can be automatically created using the format information generating apparatus of FIG. 1, the burden on the operator can be reduced. Further, highly accurate format information can be obtained, and the creation work efficiency can be remarkably improved.
(B) Format information is created based on a plurality of forms 11 to 1z, and an optimum one is selected from the format information. Therefore, a printing format, cutting error, etc. in the form are absorbed and the format is highly accurate. Information can be obtained.
(C) Since format information is created from a plurality of forms 11 to 1z and an optimum one is selected from the format information, an abnormality in the creation of format information can be detected easily and accurately.
[0029]
The present invention is not limited to the illustrated embodiment, and various modifications can be made. Examples of such modifications include the following (i) to (iii).
(I) In the format information generation process of FIG. 4, the format information selected by the comparison / selection process of step S4 is registered as it is in step S5, but before this registration, the selected format information is displayed on the display means 57. If the information is displayed on the screen and the operator confirms whether or not it has been correctly created, the format information can be registered correctly.
(Ii) In the format information generation process of FIG. 4, the contents of the forms 11 to 1z, the format generation rules, and the format information created by the format information creation means 41 are not limited to those shown in the figure, but various types It can be arbitrarily changed to fit the form.
(iii) In the format information generating apparatus of FIG. 1, the character recognition unit 41b performs character recognition from the image analysis result, but the character recognition unit 41b may be omitted depending on the read contents of the forms 11 to 1z. In addition, the forms 11 to 1z, such as adding another functional block for improving the function in the control unit 40, or configuring the plurality of storage units 32 and 51 to 55 by one or a plurality of common storage units. The device configuration can be arbitrarily changed according to the type and content of the device.
[0030]
【The invention's effect】
As described above in detail, according to the first, second, and third inventions, the image acquisition process, the format information creation process, and the comparison / selection process are executed to create the format information. A plurality of format information is automatically created from a plurality of forms using a format generation rule, and optimum format information can be selected and output from the plurality of format information. Therefore, not only the burden on the operator can be reduced, but also the format information creation accuracy and creation work efficiency can be significantly improved. Furthermore, since one format information for the form ID is created based on a plurality of forms with the same form ID, printing errors and cutting errors in the form can be absorbed, and highly accurate format information can be obtained. In addition, after creating a plurality of format information, the format information is compared with each other, and the most average format information is selected and used as the format information of the form ID. Can be detected.
According to the fourth invention, when the result of the comparison selection process is a format information generation error, the result is displayed on the screen of the display means. For example, the operator looks at the format information generation error, It is also possible to create format information by selecting an appropriate form.
According to the fifth aspect, since the format information creating means is provided, a plurality of format information can be created easily and accurately using the format generation rule. As the plurality of created format information, the most appropriate format information is selected by the comparison unit, the counting unit, and the selection unit, so that the format information with high accuracy can be created automatically.
[Brief description of the drawings]
FIG. 1 is a schematic functional block diagram of a format information generating apparatus showing an embodiment of the present invention.
FIG. 2 is a diagram illustrating an example of a form input to the format information generation device in FIG. 1;
FIG. 3 is a diagram for explaining a representation format of a position of a reading field FIELD in FIG. 2;
4 is an overall flowchart of format information generation processing using the format information generation apparatus of FIG. 1. FIG.
FIG. 5 is a diagram illustrating an example of a form image IMAGE1 in FIG.
6 is a diagram showing an example of a format generation rule in FIG. 1. FIG.
7 is a diagram illustrating an example of the format information in FIG. 1. FIG.
8 is an explanatory diagram of OK format information COD11 to CODmn in FIG. 4;
FIG. 9 is a diagram showing a comparison result of FIG.
10 is a flowchart of a comparison / selection process of OK format information CODk and CODi in FIG. 9;
11 is a diagram illustrating an example of the majority test of FIG.
FIG. 12 is a diagram illustrating an example of the majority test of FIG.
[Explanation of symbols]
11-1z form
20 Form ID
31 Photoelectric conversion means
32 Image storage means
40 Control unit
41 Format information creation means
41a Image analysis means
41b Character recognition means
41c collation search means
42 Comparison selection means
42a comparison means
42b Counting means
42c selection means
43 I / O control means
51 Format generation rule storage means
52 Format information storage means
53 Margin storage means
54 Comparison result storage means
55 Format information registration means
56 Input operation means
57 Display means
S1 Image acquisition processing
S2 format information creation process
S3 format information storage processing
S4 Comparison selection process
S5 Format information registration process

Claims (1)

In a format information generating apparatus that generates a format ID for reference used when optically reading the text on a form having a form ID for identifying the form and a reading field in which characters are entered.
Image acquisition means for acquiring images of a plurality of forms in the same format each having the same form ID and outputting a plurality of form images;
The position of a line segment and / or the position of a character included in each form image is detected, and the detection result is provided corresponding to the form ID and the position of the reading field of the form is defined. The read field position is determined by applying the format generation rule, and when all the read fields can be detected for one form image, the OK format information is obtained. When there is one or more read fields that cannot be detected, the NG format information is obtained. Format information creating means for outputting each form, and
The read field positions of all the OK format information output from the format information creating means are compared, and the difference between the information indicating the position and size of the same read field can be regarded as the same format information with a predetermined margin value or less. A comparison means for detecting format information and OK format information regarded as different OK format information;
Counting means for counting the number of OK format information that can be regarded as the same format information detected by the comparison means;
Selecting the maximum OK format information of the count value of the counting means, and outputting the selected format ID format information as the form ID;
A format information generating apparatus comprising:

Images