JP4544691B2 - Character reader - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a character reading device that determines the possibility of erroneously recognizing a read character or character string, and performs knowledge processing or post-processing on the read character string when it is erroneously recognized. About.
[0002]
[Prior art]
Conventionally, a character reading device that reads characters recorded on a form image is known.
[0003]
In a conventional character reading apparatus, a specified area on an image is scanned, line coordinates are extracted by line cutout processing, and character coordinates within the line coordinates are detected by character cutout processing. After the detection, character recognition processing is performed on the image of the coordinates of each character.
[0004]
In a character reading apparatus, generally, character recognition is performed after cutting out lines and characters.
In order to cut out rows, the image is scanned in the horizontal direction, a histogram of the number of black pixels is created, and each row is separated at a point where the value of the histogram becomes zero.
[0005]
In order to cut out characters, the image is scanned in the vertical direction, a histogram of the number of black pixels is created, and the image is separated at each point where the value of the histogram is 0 and separated into each character.
When a character is recognized by mistake, knowledge recognition using a word matching dictionary or a process called post-processing is performed to replace misread or unread characters and improve the recognition rate.
[0006]
[Problems to be solved by the invention]
By the way, in such a conventional character reading apparatus, a line or character is cut out by creating a histogram, so there are cases where line cutting and character cutting cannot be performed correctly.
[0007]
FIG. 2 is a diagram for explaining such a conventional technique.
For example, FIG. 2A shows an example in which a plurality of characters are erased. In the figure, a rectangle indicated by a broken line indicates a correct division of one character. In this case, the character line cannot be correctly identified due to the erasing line. Therefore, all characters are mistakenly recognized as one character, and line segmentation cannot be performed correctly.
[0008]
FIG. 2B shows an example in which a plurality of characters are circled. In this case, because of the circled line, two lines of characters are recognized as one vertically long character, and correct line segmentation cannot be performed.
[0009]
FIG. 2C shows an example in which a deviation occurs in printing. In this case, there is no gap between the lines, the lines cannot be correctly identified, and the characters on the two lines are cut out as one vertically long character.
[0010]
FIG. 2D shows an example in which a line for separating digits is included in a row. In this case, a digit line that is unnecessary for character recognition processing is cut out as a character.
If there is an error in the line cutout or character cutout in this way, even if the character size or character position is clearly incorrect, the character recognition process will be carried out as it is with the detected character rectangle as each character, resulting in misreading. Or unnecessary characters are output, and the reliability of the character recognition device is lowered.
[0011]
Moreover, in order to correct such a recognition result, it is necessary for the operator to perform it manually, which increases the burden on the operator.
Therefore, it is necessary to correctly determine whether or not there is a possibility that the character has been erroneously recognized, and when the character is erroneously recognized, it is necessary to perform post-processing automatically.
[0012]
[Means for Solving the Problems]
  The present invention adopts the following configuration in order to solve the above points.
<Configuration 1>
  The character reading device according to the present invention includes an image input unit that acquires characters entered on a predetermined sheet as image data, cuts out each character from the acquired image data, and recognizes the position and size of each character from the character coordinates. And the recognition means that are adjacent to each other in the same line continuously satisfy a predetermined condition, these recognition characters are set in one block, and the position and size of the block are set for each character. Stores block recognition means for recognizing block coordinates from character coordinates, first storage means for storing the block coordinates of the recognized block, and size determination conditions for determining that the height of the block is a predetermined value or more. A second storage unit; a size determination unit that determines whether or not a height of a block obtained from the block coordinates of the recognized block satisfies the size determination condition; Control means for performing post-processing on a block determined to satisfy the size determination condition.
[0013]
<Configuration 2>
  In the character reading device having the configuration 1, the erroneous cutout determination condition is described by a script.
[0014]
<Configuration 3>
  In the character reading device having the configuration 1, the control unit performs any one of a process of replacing the character with another character, an unprocessed process, and a deleting process according to the size determination result. It is characterized by performing as follows.
[0023]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, embodiments of the present invention will be described using specific examples.
<Specific example 1>
Specific example 1 recognizes a character from image data, performs size determination by specifying size determination data adapted to the reading area as a misread determination condition for the position and size of the recognized character, and allows misreading When there is a characteristic, post-processing such as unreading is performed.
[0024]
FIG. 1 is a block diagram illustrating a configuration of the first specific example.
The character reading device of Example 1 includes an image input unit 1, a display unit 2, an input unit 3, a control unit 4, a recognition unit 5, a size determination unit 6, an image memory 21, and reading area information storage. The memory 22 includes a recognition result storage memory 23, a determination data storage memory 24, a determination result storage memory 25, and a reference coordinate storage memory 26.
[0025]
The image input unit 1 is an image input unit having a function of inputting characters and figures written on a form as image data, such as an image scanner and a FAX.
The display unit 2 has a function of displaying information to the operator, such as a display.
The input unit 3 has a function of receiving input from an operator, such as a keyboard and a mouse.
[0026]
The recognition unit 5 scans the reading area with reference to the image data stored in the image memory 21, performs line segmentation and character segmentation, specifies the position and size of each character by character coordinates, The recognition means has a function of recognizing characters and converting the recognized characters into character codes.
[0027]
FIG. 3 is an explanatory diagram of character coordinates.
As shown in FIG. 3, the character “A” is a character to be recognized, and the position and size of this character are specified by a rectangle surrounding the character “A” as indicated by a broken line in the drawing.
This rectangle is represented by the coordinates (l, t)-(r, b), where the specified position is the origin, the upper left coordinates are (l, t), and the lower right coordinates are (r, b). This coordinate becomes the character coordinate.
[0028]
The size determination unit 6 is misread determination processing means having a function of performing size determination on the position and size of characters with respect to the result recognized by the recognition unit 5 and thereby determining the possibility of misreading.
[0029]
FIG. 4 is an explanatory diagram illustrating an example of size determination data used for size determination in the first specific example.
As shown in FIG. 4, the size determination data includes a plurality of conditional expressions and a processing method when corresponding to the conditional expressions.
[0030]
Here, unreading as a processing method replaces a recognized and converted character code with a character that should not be included as a recognition result, such as “?”, And facilitates correction of the recognition result by the operator. Process.
[0031]
The unprocessed process is a process for excluding a character for which a designated condition has been set so as not to perform post-processing in advance from processes such as unreading and deleting.
Deletion is a process of deleting characters that are considered unnecessary.
The processing method is appropriately set according to the conditional expression.
[0032]
The control unit 4 has a function of controlling each block of the character reading device.
The image memory 21 is a memory for storing the image data input by the image input unit 1.
[0033]
The reading area information storage memory 22 is a memory for storing area information for performing reading processing.
FIG. 5 is an explanatory diagram of the reading area of the first specific example.
A form is shown as an example. In this form, reading areas A to F on which characters are printed are set. The size, font, and the like of the character to be recognized are different for each region, and the size determination is performed for each of the reading regions A to F. The area information is area designation information for designating the coordinates of this area and size determination data designation information for designating a size determination conditional expression applied to this area, for each of the read areas A to F. Stored in
[0034]
The recognition result storage memory 23 stores the coordinates of each region and the character coordinates and character codes recognized by the recognition unit 5.
The determination data storage memory 24 is a memory for storing size determination data as shown in FIG.
[0035]
The determination result storage memory 25 is a memory for storing the character code finally obtained as a result of the size determination and its character coordinates, and the character code stored in the determination result storage memory 25 and the character code thereof. The character coordinates are stored in the same format as the data format stored in the recognition result storage memory 23.
[0036]
The reference coordinate storage memory 26 is a memory for storing reference coordinates and character positions for the size determination unit 6 to refer to the recognition result storage memory 23. In the first specific example, the character coordinates of the character recognized by the recognition unit 5 are stored as reference coordinates. The character position is data indicating the position of the character to be referred to among the characters stored in the recognition unit 5, and is 1 when the character to be referred to is, for example, the first character.
[0037]
<Operation>
Next, the operation of Example 1 will be described.
The controller 4 reads each character by controlling each block.
[0038]
FIG. 6 is a flowchart showing the operation of the first specific example.
In step (step is indicated as “S” in the figure) 1, the image input unit 1 reads image data from a form or the like.
The image data is stored in the image memory 21, and the area information is stored in the reading area information storage memory 22.
[0039]
The control unit 4 determines whether or not area information is stored in the reading area information storage memory 22 (step 2).
Initially, a reading area A is designated as shown in FIG.
The recognizing unit 5 refers to the image data stored in the image memory 21, performs line segmentation and character segmentation on the image image in the reading area A based on the area designation information of the reading area A, Detect the coordinates of each character. Then, this character is recognized. The recognized character is converted into a character code, and the character coordinates and the converted character code are stored in the recognition result storage memory 23 (step 3).
[0040]
The size determination unit 6 performs size determination on the recognition result (step 4).
FIG. 7 is a flowchart illustrating the size determination process performed by the size determination unit 6 according to the first specific example.
In step 11, the character coordinates to be determined and their character positions are acquired from the recognition result storage memory 23, and the acquired character coordinates and character positions are stored in the reference coordinate storage memory 26. The first character position is 1. If there is no character coordinate and character position of the next character to be determined, for example, “no rectangle” information in which the character coordinate and character position are all 0 is stored in the reference coordinate storage memory 26.
[0041]
In step 12, the character coordinates stored in the reference coordinate storage memory 26 are referred to and the presence / absence of the next character coordinates to be determined is determined.
If “no rectangle” information is not stored in the reference coordinate storage memory 26, the process proceeds to step 13.
[0042]
In step 13, condition calculation is performed using the size determination data.
The size determination conditions are described by mathematical formulas for convenience. The following expressions (1) to (3) are examples of conditional expressions used for the size determination.
(B−t + 1)> Wth (1)
(R−1 + 1)> Hth (2)
t <tmin (3)
Where l, r: x coordinate of rectangle
t, b: y-coordinate of rectangle
Wth: Lower limit value of rectangle in width (x) direction (for example, 40)
Hth: lower limit value of rectangle in height (y) direction (for example, 40)
tmin: minimum value of the coordinate t (for example, 1200)
[0043]
Expression (1) indicates the misreading determination condition based on the height (y direction) of the rectangle obtained from the character coordinates, and Expression (2) indicates the misreading determination condition based on the width (x direction) of the rectangle obtained from the character coordinates. (3) uses character coordinates themselves as misreading determination conditions.
[0044]
In addition, as shown in the following expression (4), two or more conditional expressions may be combined with logical product (AND) or logical sum (OR).
(B−t + 1)> 40 AND (r−1 + 1)> 40 (4)
[0045]
This size determination data is stored in the determination data storage memory 24. The size determination data designation information stored in the reading area information storage memory 22 is referred to, and size determination data adapted to the reading area A is designated using the size determination data designation information. The character coordinates are extracted from the reference coordinate storage memory 26, and the condition determination for misreading determination is performed by applying the size determination data of the reading area A to the character coordinates.
[0046]
This apparatus stores a control program for calculating the truth of these conditional expressions.
The condition calculation is performed by first substituting the character coordinates into the first conditional expression shown in FIG. If the result of the calculation is false, that is, if the conditional expression is not satisfied, character coordinates are substituted into the next conditional expression. In this way, the character coordinates are sequentially substituted into the conditional expression, and when it becomes true, the calculation is terminated there.
[0047]
For example, as shown in FIGS. 2A and 2B, when a plurality of characters are erased, or when a plurality of characters are circled, equations (1) and (2) Will be satisfied. Further, as shown in FIGS. 2C and 2D, when printing is shifted, and when a line for separating digits is included in a row, the expression (3) is satisfied. In such a case, the calculation result of each conditional expression is true.
When the calculation result of the predetermined conditional expression becomes true, it is determined that there is a possibility of misreading, and no further conditional calculation is performed.
[0048]
On the other hand, when the calculation results for all the conditional expressions are false, the condition calculation based on the size determination data is false. At this time, it is determined that there is no possibility of misreading.
This calculation result is stored in the determination result storage memory 25.
[0049]
In step 14, first, whether the calculation result is true or false is determined.
When the calculation result is false, the process proceeds to step 15.
In step 15, the character position of the false character is acquired from the reference coordinate storage memory 26 and it is determined that there is no possibility of misreading. Therefore, the character at the character position stored in the recognition result storage memory 23 is determined. Coordinates and character codes are copied to the determination result storage memory 25 as they are.
If the result of the condition calculation is true in step 14, the process proceeds to step 16.
[0050]
In step 16, the processing method is determined.
For example, in FIG. 4, when the calculation result becomes false because the character coordinates of the recognition target character satisfy the conditional expression 2, the processing method is unprocessed.
[0051]
When the processing method is not yet processed, the process proceeds to step 15 and the character position stored in the recognition result storage memory 23 for excluding the character satisfying the preset designated condition from the processing such as unread or delete. Are directly copied to the determination result storage memory 25.
[0052]
Further, when the calculation result becomes false by satisfying conditional expression 1, the processing method becomes unread.
When the processing method is unread, the process proceeds to step 17, where the character position of the character is acquired from the reference coordinate storage memory 26, and the character coordinates of the character position stored in the recognition result storage memory 23 are determined. The character code of the character is replaced with a character that should not be included as a recognition result, such as “?”, And stored in the determination result storage memory 25. Therefore, the operator can identify at a glance the possibility of misreading by visually recognizing this character.
Further, when the processing method is deletion, unnecessary characters or symbols that are imaged with, for example, dust and are considered misread are deleted.
[0053]
After such processing for the first character is completed, the process returns to step 11 to acquire the next character coordinate and character position and execute steps 12 to 17 in the same manner.
When the “no rectangle” information is stored in the reference coordinate storage memory 26, it is determined in step 12 that the size determination has been performed for all characters recognized in the reading area A, and the process returns to step 2. .
Such processing is also performed for the reading areas B to F, and when such processing is performed for all the reading areas, the processing is completed.
[0054]
<Effect of specific example 1>
As described above, according to the first specific example, when the character to be recognized is read, the size determination data is set with respect to the coordinates of each character and the height and width obtained from the coordinates to determine the size. Therefore, it is possible to appropriately evaluate the possibility of misreading each character of the recognition result from the character coordinates.
[0055]
In addition, when the processing method is unprocessed, the character that satisfies the specified condition set not to perform the unread process in advance is unprocessed by making the character unprocessed. It can be excluded from the deletion process.
[0056]
In addition, for characters that are considered to be likely to be misread in the recognition result, the unread process is performed to replace the target character with a character that should not be included, such as “?”. Therefore, the operator can visually recognize at a glance, and the operator can easily correct the recognition result.
[0057]
Further, for example, for an obviously unnecessary character or symbol that is considered to be misread due to being imaged by dust or the like, the burden of correction work by the operator can be reduced by deleting this character.
[0058]
Furthermore, since such size determination is performed for each reading area, even in a form in which the font, character type, and size differ for each reading area, appropriate character recognition and size determination are performed for each area. Character recognition accuracy is improved.
[0059]
<Specific example 2>
Specific example 2 calculates the context of three characters based on the character coordinates, sets misreading determination conditions for the context, and performs size determination.
Similar to the first specific example, the character reading device of the second specific example includes an image input unit 1, a display unit 2, an input unit 3, a control unit 4, a recognition unit 5, a size determination unit 6, and an image memory. 21, a reading area information storage memory 22, a recognition result storage memory 23, a determination data storage memory 24, a determination result storage memory 25, and a reference coordinate storage memory 26.
[0060]
However, the reference coordinate storage memory 26 is provided with three storage areas.
FIG. 8 is an explanatory diagram of the reference coordinate storage memory 26.
The storage area b is an area for storing the character coordinates and character position of the current character to be recognized, and the storage areas a and c are the character coordinates and character position immediately before the current character, respectively, This is an area for storing character coordinates and character positions of characters.
In addition, the same code | symbol is attached | subjected about the same element as the specific example 1, and description is abbreviate | omitted.
[0061]
<Operation>
Next, the operation of the specific example 2 will be described.
In the second specific example, as in the first specific example, the flowchart of FIG. 6 is executed, and the size determination process is performed in step 4.
[0062]
FIG. 9 is a flowchart showing the size determination process of the second specific example.
In step 21, the character coordinates and character position of the first character are acquired.
The acquired character coordinates and character positions are stored in the storage area c of the reference coordinate storage memory 26, and information on “no rectangle” is stored in the storage areas a and b.
[0063]
In step 22, the character coordinates and character position of the next character are acquired.
When the next character coordinate and character position are acquired, the character coordinate and character position of the first character stored in the storage area c of the reference coordinate storage memory 26 are stored in the storage area b, and the acquired next character coordinate and The character position is stored in the storage area c.
[0064]
Thereafter, each time the character coordinates and character positions are acquired, the character coordinates and character positions stored in the storage areas b and c are stored in the storage areas a and b, respectively, and the acquired character coordinates and character positions are stored in the storage area c. To store.
[0065]
If there is no character coordinate and character position of the character whose size is to be determined next, for example, “no rectangle” information in which the character coordinate and the character position are all 0 is stored in the storage area c.
[0066]
In step 23, it is determined whether or not there is a current character to be determined.
If the information “No rectangle” is not stored in the storage area b, it is determined that there is a current character to be determined, and the process proceeds to Step 24.
[0067]
In step 24, condition calculation is performed on the character coordinates of the three characters using the size determination data.
FIG. 10 is an explanatory diagram of the second specific example.
Three character coordinates are set as shown in FIG.
An interval between characters is calculated from these three character coordinates, and size determination is performed by applying size determination data as a misread determination condition to the interval.
Expressions (5) and (6) are examples of conditional expressions as size determination data.
l-pr-1 = 0 (5)
nl-r-1 = 0 (6)
[0068]
Expression (5) is a conditional expression in which the interval between the current character and the previous character is 0, as shown in FIG. 10 (b), and Expression (6) is as shown in FIG. 10 (c). This is a conditional expression in which the interval between the current character and the next character is zero.
[0069]
It should be noted that equations (5) and (6) may be combined with logical product (AND) or logical sum (OR) in the same manner as in the first specific example.
Further, the context of the three characters is not limited to the interval between the characters, and the relationship between the sizes of the three characters may be used as a condition.
[0070]
This size determination data is stored in the determination data storage memory 24. As in the first specific example, the size determination data specification information stored in the read area information storage memory 22 is referred to and this size determination data specification information is referred to. 5 is used to specify the size determination data adapted to the reading area A shown in FIG. 5, to extract three character coordinates from the reference coordinate storage memory 26, and to apply the size determination data of the reading area A for misreading determination. The condition calculation is performed.
[0071]
The condition calculation method is the same as in the first specific example, and the calculation result is stored in the determination result storage memory 25.
However, when “no rectangle” information is stored in the storage area a or the storage area c of the reference coordinate storage memory 26, the calculation result of the conditional expression is false.
[0072]
In step 25, it is determined whether the calculation result is true or false.
When the calculation result is false, it is determined that there is no possibility of misreading because it does not correspond to any conditional expression, and the process proceeds to step 26 to acquire the character position stored in the storage area b of the reference coordinate storage memory 26. Similarly to the first specific example, the character coordinates and the character code of the character position stored in the recognition result storage memory 23 are copied to the determination result storage memory 25 as they are.
[0073]
If the calculation result is true, it is determined that there is a possibility of misreading, and the process proceeds to step 27 to determine the processing method corresponding to the conditional expression. If the processing method is unprocessed, it is stored in the storage area b. The character coordinates and character code of the current character being copied are directly copied to the determination result storage memory 25 (step 26).
[0074]
When the processing method is unread, the process proceeds to step 28, where the character position of the character is acquired from the reference coordinate storage memory 26, and the character coordinates of the character position stored in the recognition result storage memory 23 are determined. The character code of the character is replaced with a character that should not be included as a recognition result, such as “?”, And stored in the determination result storage memory 25.
When the processing method is deletion, characters or symbols considered unnecessary are deleted.
[0075]
When such processing is performed for all characters and size determination is performed for all characters (step 23), the size determination processing is terminated, and such processing is performed for all reading regions (step 23). Step 2), the process is complete.
[0076]
<Effect of specific example 2>
As described above, according to the second specific example, since the size determination is performed for the context of the three characters, not only the same effect as the first specific example is obtained but also the possibility of misreading is obtained. Therefore, it can be determined more accurately.
[0077]
<Specific example 3>
Specific example 3 is a method for performing size determination by setting a misreading determination condition to the line position, the character position, the character position from the beginning of the line, and the number of characters in the line calculated based on the character coordinates. is there.
[0078]
FIG. 11 is a block diagram illustrating a configuration of the third specific example.
The character reading device of specific example 3 includes an image input unit 1, a display unit 2, an input unit 3, a control unit 4, a recognition unit 5, a size determination unit 6, an image memory 21, and reading area information storage. The memory 22 includes a recognition result storage memory 23, a determination data storage memory 24, a determination result storage memory 25, a reference coordinate storage memory 26, and a related information storage memory 27.
[0079]
The related information storage memory 27 is a memory for storing related information such as the current character line position, character position, character position from the beginning of the line, and the number of characters in the line, as information on the current character.
[0080]
FIG. 12 is an explanatory diagram of related information of the third specific example.
The line position L, the character position I, the character position LI from the head of the line, and the number of characters LN in the line are set to values of 1 or more.
In addition, the same code | symbol is attached | subjected about the same element as the specific example 1 and the specific example 2, and description is abbreviate | omitted.
[0081]
<Operation>
Next, the operation of specific example 3 will be described.
Also in the specific example 3, as in the specific example 1, the flowchart of FIG. 6 is executed, and the size determination process is performed in step 4.
[0082]
FIG. 13 is a flowchart showing the size determination process of the third specific example.
In steps 31 to 33, the character coordinates and character positions of the first character and the next character are acquired and stored in the storage areas b and c of the reference coordinate storage memory 26, respectively, as in steps 21 to 23 of the second specific example. Proceed to 34.
[0083]
  In step 34, related information of the current character that is the determination target is set.
  That is, referring to the recognition result storage memory 23, as shown in FIG. 12, the line position L, the character position for the character to be determinedIThe character position LI from the head of the line, the number of characters LN in the line, and the like are acquired. Then, the related information is stored in the related information memory S13.
[0084]
In step 35, condition calculation is performed on the character coordinates of the current character to be determined and the related information using the size determination data.
Expressions (7) and (8) are examples of conditional expressions as size determination data.
I = 2 (7)
LN = 3 (8)
Expression (7) is a conditional expression for the second character, and Expression (8) is a conditional expression for the number of line characters of three.
[0085]
Similar to the specific examples 1 and 2, these two conditional expressions may be combined with logical product (AND) or logical sum (OR).
This size determination data is stored in the determination data storage memory 24. Like the specific examples 1 and 2, the size determination data is referred to the size determination data designation information stored in the read area information storage memory 22, and this size determination data is stored. The size determination data adapted to the reading area is specified using the designation information, three character coordinates are extracted from the reference coordinate storage memory 26, and the condition calculation for misreading determination is performed by applying the size determination data of the reading area. Do.
[0086]
The condition calculation method is the same as in the first specific example, and the calculation result is stored in the determination result storage memory 25.
However, as in the second specific example, when “no rectangle” information is stored in the storage area a or the storage area c of the reference coordinate storage memory 26, the calculation result of the conditional expression is false.
In steps 36 to 39, post-processing is performed in the same manner as in specific examples 1 and 2.
[0087]
When such processing is performed for all characters and size determination is performed for all characters (step 33), the size determination processing is terminated and such processing is performed for all reading areas (step 33). Step 2), the process is complete.
[0088]
<Effect of specific example 3>
  As described above, according to the specific example 3, not only the positional relationship between the current character and the preceding and following characters, but also the line position L and the character position as related information of the current character.ISince the size determination is performed with respect to the character position LI from the beginning of the line and the number of characters LN of the line, the effects of specific examples 1 and 2 can be obtained, and the processing conditions can be set for a specific line and character. More accurate processing can be applied to misreading and unnecessary characters due to errors in line cutout processing and character cutout processing.
[0089]
<Specific Example 4>
In the fourth specific example, line coordinates of three characters are calculated based on the character coordinates, a misreading determination condition is set to the line coordinates, and size determination is performed.
[0090]
As in the third specific example, the character reading device of the fourth specific example includes an image input unit 1, a display unit 2, an input unit 3, a control unit 4, a recognition unit 5, a size determination unit 6, and an image memory. 21, a reading area information storage memory 22, a recognition result storage memory 23, a determination data storage memory 24, a determination result storage memory 25, a reference coordinate storage memory 26, and a related information storage memory 27. Has been.
In addition, about the same element as the specific examples 1-3, the same code | symbol is attached | subjected and description is abbreviate | omitted.
[0091]
<Operation>
Next, the operation of the specific example 4 will be described.
Also in the specific example 4, as in the specific example 1, the flowchart of FIG.
[0092]
FIG. 14 is a flowchart illustrating the size determination process of the fourth specific example.
In step 41, a rectangular area in the first row is created.
FIG. 15 is an explanatory diagram showing a method of creating line coordinates according to the fourth specific example.
As shown in FIG. 15, rectangular areas {circle around (1)}, {circle around (2)}, and {circle around (3)} indicated by broken lines each indicate a rectangular area surrounding one character.
[0093]
The rectangular areas (1) to (3) are represented by the following character coordinates, for example.
Character coordinates of rectangular area (1): (pl, pt)-(pr, pb)
Character coordinates of rectangular area (2): (l, t)-(r, b)
Character coordinates of rectangular area (3): (nl, nt)-(nr, nb)
[0094]
In order to create the line coordinates of the first line, the minimum rectangular area (4) is set so as to include all the characters. The character coordinates of the rectangular areas {circle around (1)} to {circle around (3)} are extracted from the recognition result storage memory 23, and line coordinates (nl, pt) − (nr, b) are created.
[0095]
Further, the character position of the last character of this line is taken as the character position of this line, the line coordinates (nl, pt)-(nr, b) of the created first line and the character position are stored in the reference coordinate storage memory 26 storage area. In the storage area a and b, information “no rectangle” is stored.
[0096]
In step 42, as in step 41, a rectangular area of the next row is created.
When the rectangular area of the next line is created, the line coordinates of the first line stored in the storage area c of the reference coordinate storage memory 26 and the character position of the last character of the line are stored and created in the storage area b. The next line coordinate and the character position of the last character of the line are stored in the storage area c.
[0097]
Thereafter, every time line coordinates are created, the data stored in the storage areas b and c are stored in the storage areas a and b, respectively, and the created character coordinates and character positions are stored in the storage area c.
[0098]
If there is no line coordinate of the line to be subjected to size determination next, for example, “no rectangle” information in which the line coordinates and the character position of the last character of the line are all 0 is stored in the storage area c.
[0099]
In step 43, the presence / absence of the current line to be determined is determined.
If the information “No rectangle” is not stored in the storage area b, it is determined that there is a current line to be determined, and the process proceeds to Step 44.
[0100]
  In step 44, the relevant information of the current line to be determined is set.
  That is, the last character of the current line is acquired from the storage area b of the reference coordinate storage memory 26, the recognition result storage memory 23 is referred to for the character at the character position, and the line position L, character positionIThe character position LI from the head of the line, the number of characters LN in the line, and the like are acquired. Then, the related information is stored in the related information memory S13.
[0101]
In step 45, condition calculation is performed on the current line and related information using size determination data.
As for the conditional expression, the positional relationship of the current line, the size of the current line, etc. can be set.
Further, as in the first to third examples, the two conditional expressions may be combined with logical product (AND) or logical sum (OR).
[0102]
The size determination data is stored in the determination data storage memory 24. Like the specific examples 1 to 3, the size determination data specification information stored in the read area information storage memory 22 is referred to, and the size determination data specification is performed. The size determination data adapted to the reading area is specified using the information, the three character coordinates are extracted from the reference coordinate storage memory 26, and the condition calculation for misreading determination is performed by applying the size determination data of the reading area. .
[0103]
The size determination data includes a conditional expression similar to the specific example 1 (FIG. 4) and a processing method corresponding to the conditional expression.
The condition calculation method is the same as in the first specific example, and the calculation result is stored in the determination result storage memory 25.
[0104]
However, as in the second specific example, when “no rectangle” information is stored in the storage area a or the storage area c of the reference coordinate storage memory 26, the calculation result of the conditional expression is false.
[0105]
In step 46, the calculation result is determined. If the calculation result is false, the process proceeds to step 47.
In step 47, the character position I of the current line, that is, the final character position of the current line and the number of characters LN of that line are obtained from the related information storage memory 27, and the final character position is determined from the start character position (I-LN + 1) of this line. The character coordinates and character codes up to I are copied from the recognition result storage memory 23 to the determination result storage memory 25 as they are.
[0106]
If the calculation result is true, it is determined that there is a possibility of misreading, and the process proceeds to step 48 to determine the processing method corresponding to the conditional expression.
When the processing method is unprocessed, the process proceeds to step 47, and the same processing as when the determination result is false is performed.
If the processing method is unread, the process proceeds to step 49.
[0107]
In step 49, the character position I of the current line, that is, the last character position and the number of characters LN of that line are obtained from the related information storage memory 27, and the start character position (I-LN + 1) of this line to the last character position I is acquired. The character coordinates are copied from the recognition result storage memory 23 to the determination result storage memory 25, and the character code is replaced with a character that should not be included as a recognition result, such as “?”, And stored in the determination result storage memory 25. .
When the processing method is deletion, delete the character.
[0108]
When such processing is performed for all rows and size determination is performed for all rows (step 43), the size determination processing is terminated and such processing is performed for all reading areas (step 43). Step 2), the process is complete.
[0109]
<Effect of specific example 4>
As described above, according to the fourth specific example, the positional relationship before and after the current line is calculated, and the size determination is performed with respect to this positional relationship. And post-processing can be performed.
[0110]
<Specific example 5>
In the specific example 5, when characters having the same position condition continue in the same line, these characters are grouped into blocks, and the size is determined in units of blocks.
[0111]
The related information storage memory 27 of the fifth specific example stores conditional expressions for grouping into blocks based on the preceding and following character intervals.
For example, the character coordinates of character positions i and i + 1 are (L (i), T (i))-(R (i), B (i)) and character coordinates (L (i + 1), T (i +), respectively. 1)) − (R (i + 1), B (i + 1)), the interval D is calculated by the following equation (9).
D = L (i + 1) -R (i) -1 (9)
[0112]
Expressions (10) to (15) are expressions indicating conditions for creating a block based on the interval D.
D = Dthl (10)
D ≠ Dthl (11)
D <Dthl (12)
D ≦ Dthl (13)
D> Dthl (14)
D ≧ Dthl (15)
Where Dthl: predetermined value
[0113]
These expressions (10) to (15) are stored in the related information storage memory 27.
The determination data storage memory 24 of the specific example 5 stores size determination data applied to this block.
[0114]
FIG. 16 is an explanatory diagram illustrating an example of the size determination data of the fifth specific example.
The reference coordinate storage memory 26 of the specific example 5 has three storage areas a to c as in the specific example 2.
In addition, about the same element as the specific examples 1-4, the same code | symbol is attached | subjected and description is abbreviate | omitted.
[0115]
<Operation>
Next, the operation of Example 5 will be described.
In the specific example 5, as in the specific example 1, the flowchart of FIG. 6 is executed, and the size determination process is performed in step 4.
[0116]
FIG. 17 is a flowchart showing the size determination process of the fifth specific example.
In step 51, referring to the character in the area acquired from the recognition result storage memory 23 from the beginning, the interval D is calculated, and the conditional expressions (10) to (15) are evaluated, and any of the conditions is the same. When corresponding continuously in a line, the minimum rectangular area including these characters is taken as one block.
[0117]
FIG. 18 is an explanatory diagram of this block.
As shown in FIG. 18, when the characters P, Q, and R are arranged on the same line, the interval D between the characters P and Q is expressed by the equation (9) as described above.
[0118]
For example, when the interval D between the characters P and Q corresponds to any one of the equations (10) to (15), the characters P and Q are combined into one block. A region indicated by a broken line in FIG. 18 indicates one block created in this way.
[0119]
When characters are printed in the areas (1), (2), (3) as shown in FIG. 15, the area (4) indicated by the solid line is the smallest rectangular area, and this is one block. Become.
This block is specified by block coordinates (L (i), T (i))-(R (i + 1), B (i + 1)).
[0120]
First, information “no rectangle” is stored in the storage areas a and b of the reference coordinate storage memory 26, and the block coordinates of the first block are stored in the storage area c together with the last character position of the block.
[0121]
In step 52, the next block is created in the same manner as in step 51.
The reference information stored in the storage areas a, b, and c of the reference coordinate storage memory 26 is moved one by one, the reference coordinates of the next block are acquired from the recognition result storage memory 23, and the reference coordinates of this block are stored. The area c is stored together with the last character position of the block. If there is no next line, information of “no rectangle” is stored.
[0122]
In step 53, it is determined whether there is a block whose size should be determined.
When there is a block whose size is to be determined, the process proceeds to step 54.
In step 54, related information of the current block that is the determination target is set.
In order to set this related information, the character position of the last character of the current block is obtained from the character position of the current block stored in the reference coordinate storage memory 26, and the recognition result storage memory 23 is stored for the character at that character position. The line position L, the character position I, the character position LI from the head of the line, the number of characters LN and the number of block characters BN of the line are acquired and stored in the related information storage memory 27. Note that L, I, LI, LN, and BN have values of 1 or more.
[0123]
In step 55, the block immediately before the current block stored in the areas a, b, and c of the reference coordinate storage memory 26, the current block, the block coordinates of the next block, and the related information storage memory 27 are stored. The size determination data stored in the determination data storage memory 24 according to the area information of the area stored in the read area information storage memory 22 and the size determination data conditions Calculates the truth of an expression.
The conditional expressions may be the same conditional expressions as in the first to fourth examples, or may be compounded by logical product (AND) or logical sum (OR).
[0124]
The size determination data includes a conditional expression similar to the specific example 1 (FIG. 4) and a processing method corresponding to the conditional expression.
The condition calculation method is the same as in the first specific example, and the calculation result is stored in the determination result storage memory 25.
[0125]
However, as in the second specific example, when “no rectangle” information is stored in the storage area a or the storage area c of the reference coordinate storage memory 26, the calculation result of the conditional expression is false.
[0126]
In step 56, the true / false of the calculation result is determined. If the calculation result is false, the process proceeds to step 57.
In step 57, the character position I of the current block, that is, the final character position of the current block and the number of characters BN of the block are obtained from the related information storage memory 27, and the final character position is determined from the start character position (I-BN + 1) of this block. The character coordinates and character codes up to I are copied from the recognition result storage memory 23 to the determination result storage memory 25 as they are.
[0127]
If the calculation result is true, it is determined that there is a possibility of misreading, and the process proceeds to step 58 to determine the processing method corresponding to the conditional expression.
When the processing method is not yet processed, the process proceeds to step 57 and the same processing as when the determination result is false is performed.
If the processing method is unread, the process proceeds to step 59.
[0128]
In step 59, the character position I of the current block, that is, the final character position and the number of characters BN of the block are obtained from the related information storage memory 27, and from the start character position (I-BN + 1) to the final character position I of this block. The character coordinates are copied from the recognition result storage memory 23 to the determination result storage memory 25, and the character code is replaced with a character that should not be included as a recognition result, such as “?”, And stored in the determination result storage memory 25. To do.
When the processing method is deletion, delete the character.
[0129]
When such a process is performed for all the blocks and the size determination for all the blocks is completed (step 53), the process returns to step 2 to recognize all the region information (step 3) and the size determination process (step 4). ) Is performed (step 2), all processing is terminated.
[0130]
<Effect of Specific Example 5>
As described above, according to the specific example 5, when characters with the same condition are consecutive in the same line, these characters are grouped into blocks, and size determination is performed on the blocks. Post-processing can be performed by discriminating errors in line cut-out processing and character cut-out processing.
[0131]
<Specific Example 6>
In specific example 6, the size determination data is described as script data.
[0132]
FIG. 19 is a block diagram illustrating the configuration of the sixth specific example.
The character reading device of specific example 6 includes an image input unit 1, a display unit 2, an input unit 3, a control unit 4, a recognition unit 5, a size determination unit 6, a script data analysis unit 7, an image memory 21, a reading area information storage memory 22, a recognition result storage memory 23, a determination data storage memory 24, a determination result storage memory 25, a reference coordinate storage memory 26, a related information storage memory 27, and a script data storage memory 28.
[0133]
The script data storage memory 28 is a memory for storing size determination data described in a script, and the script is described in text.
The script data analysis unit 7 refers to the size determination data stored in the script data storage memory 28, performs a syntax analysis, and has a function of converting into internal size determination data that can be used by the size determination unit 6. It is.
In addition, about the same element as the specific examples 1-5, the same code | symbol is attached | subjected and description is abbreviate | omitted.
[0134]
<Operation>
Next, the operation of the specific example 6 will be described.
In the second specific example, as in the first specific example, the flowchart of FIG. 6 is executed, and the size determination process is performed in step 4.
[0135]
FIG. 20 is a flowchart showing the size determination process of the sixth specific example.
In step 61, the size determination data described in the script is analyzed.
In order to analyze the size determination data, a script is acquired from the script data storage memory 28 according to the information of the area stored in the reading area information storage memory 22.
[0136]
The script data analysis unit 7 parses the script, converts it into internal size determination data that can be used by the size determination unit 6, and stores the converted size determination data in the determination data storage memory 24.
[0137]
Expression (16) is an expression showing an example of the size determination data described by this script.
Processing unit, (condition 1) processing 1 | (condition 2) processing 2 | ... | (condition n) processing n ... (16)
[0138]
Processing units such as character units and line units are described as processing units, and conditional expressions (10) to (15) are described as conditions 1 to n, for example. And the process 1-n corresponding to the conditions 1-n is enumerated.
[0139]
In steps 62 to 66, the related information of the current line is set as in steps 52 to 55 of the specific example 5, and the condition calculation is performed using the size determination data stored in the determination data storage memory 24.
In steps 67 to 70, post-processing similar to that in the specific example 5 is performed.
[0140]
When such a process is performed for all the blocks and the size determination for all the blocks is completed (step 64), the process returns to step 2 to recognize all the region information (step 3) and the size determination process (step 4). ) Is performed (step 2), all processing is terminated.
[0141]
<Effect of Specific Example 6>
As described above, according to the specific example 6, by describing the size determination data in a script, the same effects as the specific examples 1 to 5 can be obtained, and the conditional expression can be easily defined. For this reason, it is possible to easily cope with a change due to an error in the size determination data.
[Brief description of the drawings]
FIG. 1 is a block diagram showing a configuration of a specific example 1;
FIG. 2 is a conventional explanatory diagram.
FIG. 3 is an explanatory diagram of character coordinates of a specific example 1;
FIG. 4 is an explanatory diagram showing an example of size determination data of specific example 1;
FIG. 5 is an explanatory diagram of a reading area in specific example 1;
FIG. 6 is a flowchart showing the operation of the first specific example.
FIG. 7 is a flowchart showing a size determination process of specific example 1;
FIG. 8 is an explanatory diagram of a reference coordinate storage memory of a specific example 2;
FIG. 9 is a flowchart illustrating size determination processing according to the second specific example.
10 is an explanatory diagram of specific example 2. FIG.
FIG. 11 is a block diagram showing a configuration of a specific example 3;
FIG. 12 is an explanatory diagram of related information of specific example 3;
FIG. 13 is a flowchart illustrating a size determination process according to a specific example 3;
FIG. 14 is a flowchart illustrating a size determination process according to a specific example 4;
FIG. 15 is an explanatory diagram illustrating a method for creating line coordinates according to a fourth specific example;
FIG. 16 is an explanatory diagram showing an example of size determination data in specific example 5;
FIG. 17 is a flowchart illustrating size determination processing according to a specific example 5;
FIG. 18 is an explanatory diagram of a block of specific example 5;
FIG. 19 is an explanatory diagram showing a configuration of a specific example 6;
FIG. 20 is a flowchart illustrating size determination processing according to a specific example 6;
[Explanation of symbols]
1 Image input section
4 Control unit
5 recognition part
6 Size determination part
22 Reading area information storage memory
23 Recognition result storage memory
24 judgment data storage memory
25 Judgment result storage memory
26 Reference coordinate storage memory

Claims (3)

Image input means for acquiring characters entered on a predetermined sheet as image data;
Recognizing means for cutting out each character from the acquired image data and recognizing the position and size of each character from the character coordinates;
When the interval between adjacent characters in the same line satisfies a predetermined condition continuously , these recognized characters are set in one block, and the position and size of the block are determined from the character coordinates of each character to the block coordinates. Block recognition means that recognizes as
First storage means for storing the block coordinates of the recognized block;
Second storage means that stores determines size determination condition that the height of the block is equal to or greater than a predetermined value,
Size determination means for determining whether or not the height of the block obtained from the block coordinates of the recognized block satisfies the size determination condition;
A character reading apparatus comprising: control means for performing post-processing on a block determined to satisfy the size determination condition. The erroneous clipping determination condition, character reading apparatus according to claim 1, characterized in that that will be described by the script. The control means performs any one of a process of replacing the character with another character, an unprocessed process, and a deleting process as the post-process according to the size determination result. The character reader according to 1 .

Images