JP2812704B2 - Character extraction device - Google Patents

Description

Description: BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a character extracting device for extracting characters from a character line with ruled lines.

(Prior Art) In recent years, in order to increase the processing efficiency of a large amount of documents, there is an increasing demand for enabling information written in the documents to be mechanically input to a computer. The character recognition device used for this mechanical input separates (cuts out) an image of one character from a printed or handwritten character group image in character units, and performs character recognition of the separated image of one character. .

Generally used documents are often provided with ruled lines in addition to characters, but for character recognition, it is necessary to cut out an image of one character so as not to include the added ruled lines.

Examples of a device and a method for cutting out so as not to include a ruled line include an optical character reading device disclosed in JP-A-60-160487 and a character disclosed in JP-A-62-217385. There is a separation method.

In the former conventional apparatus, a row direction and a column direction are virtually set on an image memory storing image data from the reading unit (the column direction is a direction perpendicular to the row direction. Hereinafter, virtually set). The row direction and the column direction are referred to as a predetermined row direction and a predetermined column direction), and the image data of the document is scanned in the row direction.
In this scanning, the number of black pixels is accumulated in the row direction to create a peripheral distribution in the row direction, and the area of the peripheral distribution exceeding the predetermined number of black pixels is defined as an underline portion and the area of the peripheral distribution less than the predetermined number of black pixels. The position of the edge of the character portion in the column direction is detected as the start position and the end position of the character in the column direction. Then, only the character portion is scanned except for the underline portion, and the number of black pixels is accumulated in the column direction by this scanning to create a peripheral distribution in the column direction. Detect start and end positions.

In this conventional device, in order to accurately detect the start and end positions of characters in the line direction when the ruled line is added along the line direction, the true line direction of the image is made to coincide with the predetermined line direction. Image data such as manuscript
At the time of creating the marginal distribution, the overlapping of characters and ruled lines in a predetermined line direction is reduced.

In the latter conventional method, a plurality of local regions are set on an image including a character line with a straight line such as an underline,
The images of these local regions are symbolized into predetermined direction elements by pattern matching, and the direction elements are further classified. Then, a line segment having a predetermined condition (for example, length or existing area) is selected from line segments composed of direction line elements (for example, horizontal straight line components) having the same direction as a line segment unnecessary for character recognition such as an underline. Is detected, and the detected line segment is removed from the image as an unnecessary line segment. After obtaining an image from which unnecessary line segments have been removed in this way, this image is scanned to create a peripheral distribution in the row and column directions, and the starting position of the character in a predetermined row and column direction is determined from the created peripheral distribution. The end position was detected.

(Problems to be Solved by the Invention) However, in the former conventional apparatus described above, if the predetermined row direction set in the image memory is deviated from the true row direction of the image, the predetermined The overlap of characters and ruled lines in the direction increases, so that black pixels of the character and black lines of the ruled line coexist on the same main scanning line. It is difficult to detect well.

Further, in the former conventional device described above, even if the predetermined line direction is deviated from the true line direction, the character start and end positions in the predetermined line direction are all the same within the same character line. An incomplete character that is practically unrecognizable due to a missing end is cut out. The degree of the lack of this character becomes more severe near the end of the line.

Moreover, if the character and the ruled line are in contact with each other in addition to the mismatch between the predetermined line direction and the true line direction, the accuracy of detecting the position of the character and the ruled line is further deteriorated, and the degree of missing of the character is further increased.

In the latter conventional method described above, the read image is divided into a plurality of local areas, the read image of each local area is compared with a predetermined plurality of basic patterns by matching, and the local area is oriented based on the comparison result. Convert to line elements. Such a local matching process is complicated, and therefore, the device configuration of the device adopting the conventional method is complicated, which causes a problem that the size of the device is increased and the cost of the device is increased.

SUMMARY OF THE INVENTION An object of the present invention is to solve the above-described conventional problems, so that even when the predetermined row direction and the true row direction do not match, the predetermined row direction and column direction are determined based on the change in the number of accumulated black pixels. It is an object of the present invention to provide a character cutout device that can accurately detect the start position and the end position of a character so that the character is not lost.

(Means for Solving the Problems) In order to achieve this object, a character extracting device according to the present invention optically reads an information medium in which a character line with a ruled line is described and outputs image data of the information medium. And an image data storage unit for storing image data from the reading unit, and setting a character part area in a character line area with a ruled line so as not to include a ruled line, and storing image data in the character part area of the image data storage part. The scanning is performed to detect the cumulative number of black pixels in a predetermined column direction for each scanning line, and based on a change in the cumulative number of black pixels in the column direction, a character start position and an end position in a predetermined row direction are detected for each character. Based on the character start and end positions in the line direction, the first ruled line detection area includes a ruled line at a position adjacent to the character start position in the predetermined line direction and opposite to the character end position in the predetermined line direction. Set for each character A ruled line detection area setting unit that sets a second ruled line detection area for each character so as to include a ruled line at a position adjacent to a character end position in a predetermined line direction and opposite to a character start position in the predetermined line direction. And scanning the image data in the first and second ruled line detection areas of the image data storage unit to detect the cumulative number of black pixels in a predetermined row direction for each scanning line, and based on the change in the cumulative number of black pixels in the row direction. The ruled line detection processing in the first and second ruled line detection areas is performed for each character, and based on the ruled line detection processing result, the character start and end positions in a predetermined column direction are positioned closer to the character part area than the ruled line. As described above, the operation of the character ruled line separation unit, which is set for each character, the reading unit, the ruled line detection area setting unit, and the character ruled line separation unit is controlled, and the character start position and end position in the predetermined row direction and the predetermined column direction are controlled. Character start and end positions Characterized in that the made and a controller for outputting on an individual character in the ruled line with a character row region.

(Operation) According to the character extracting device having such a configuration, the reading unit optically reads the information medium in which the character line with the ruled line is described, outputs image data of the information medium, and the image data storage unit And the image data from the reading unit.

The ruled line detection area setting unit sets the character part area in the character line area with a ruled line so as not to include a ruled line, and scans the image data in the character part area of the image data storage unit to accumulate the data in a predetermined column direction. The number of black pixels is detected for each scanning line. Then, based on the detected change in the cumulative number of black pixels in the column direction, the character start and end positions in the predetermined row direction are detected for each character.

Further, the ruled line detection area setting unit sets the first ruled line detection area at a position adjacent to the character start end position in the predetermined line direction and a character end position in the predetermined line direction based on the character start position and the end position in the line direction. A ruled line is set for each character so as to include a ruled line on the opposite side, and the ruled line detection area is a position adjacent to the character end position in the predetermined line direction and opposite to the start end position in the predetermined line direction. Is set for each character to include.

The character ruled line separation unit scans the image data in the first and second ruled line detection areas of the image data storage unit, detects the cumulative number of black pixels in a predetermined row direction for each scan line, A ruled line detection process in the first and second ruled line detection areas is performed based on a change in the number of pixels. Further, based on the result of the ruled line detection processing, the character start and end positions in a predetermined column direction are set for each character so as to be located closer to the character portion area than the ruled line.

The control unit controls the operation of the reading unit, the ruled line detection area setting unit, and the character ruled line separating unit, and determines the start and end positions of the character in the predetermined row direction and the start and end positions in the predetermined column direction with a ruled line. Output for each character in the character line area.

As described above, the process of detecting the start and end positions of the character is simplified by detecting the start and end positions of the character in the predetermined row direction based on the change in the cumulative number of black pixels in the predetermined column direction in the character portion area. it can. Moreover, by setting the character portion area to include only the character portion of one character line so as not to include the ruled line, the black pixels constituting the ruled line are not counted as the cumulative number of black pixels in a predetermined column direction. Or almost only a small number can be counted, and as a result, the detection accuracy of the start and end positions of the moss can be improved.

Further, the start and end positions of the characters in the predetermined column direction are detected based on the change in the cumulative number of black pixels in the predetermined row direction in the first and second ruled line detection areas, so that these start and end positions are detected. Detection processing can be simplified.

In addition, since the first ruled line detection area is set to include a ruled line at a position adjacent to the character start position in the predetermined line direction and opposite to the character start position in the predetermined line direction,
It is possible to prevent black pixels constituting a character from being counted or hardly counted as the cumulative number of black pixels in the predetermined row direction in the first ruled line detection area. As a result, the position of the ruled line in the first ruled line detection area can be detected with high accuracy. Similarly, the second ruled line detection area is set to include a ruled line at a position adjacent to the character end position in the predetermined line direction and opposite to the character start position in the predetermined line direction. The constituent black pixels can be prevented from being counted or almost not counted as the cumulative number of black pixels in a predetermined row direction in the second ruled line detection area. As a result, the position of the ruled line in the second ruled line detection area can be detected with high accuracy.

Therefore, since the positions of the ruled lines in the first and second ruled line detection areas can be detected with high precision, characters can be cut out without causing any missing characters or with less missing characters than before.

Hereinafter, embodiments of the present invention will be described with reference to the drawings. It should be noted that the drawings are only schematically shown to the extent that the present invention can be understood, and therefore the configuration of each component, input / output signals, the flow of input / output signals, the connection relation of signal lines, and the flow of operation are illustrated in the drawings. It is not limited to.

FIG. 1 is a functional block diagram schematically showing a configuration of an embodiment of the present invention.

The character extracting apparatus according to this embodiment includes a reading unit 10, an image data storage unit 12, a ruled line detection area setting unit 14, a character ruled line separating unit 16
And a control unit 18.

The reading unit 10 optically reads the information medium on which the character line with the ruled line is described, and outputs image data of the information medium. The image data storage unit 12 stores the image data from the reading unit 10.

The ruled line detection area setting unit 14 sets the character part area in the character line area with a ruled line so as not to include a ruled line, scans the image data in the character part area of the image data storage unit 12, and scans the image data in a predetermined column direction. The cumulative number of black pixels is detected for each scanning line, and based on the change in the cumulative number of black pixels in the column direction, the character start and end positions in a predetermined row direction are detected for each character.

Then, the ruled line detection area setting unit 14 sets the first and second ruled line detection areas based on the character start and end positions in the predetermined line direction. The ruled line detection area setting unit 14 sets the first ruled line detection area to a position adjacent to the character start position in the predetermined line direction and one line so as to include a ruled line at a position opposite to the character start position in the predetermined line direction. The second ruled line detection area is set for each character so as to include a ruled line at a position adjacent to the character end position in the predetermined line direction and opposite to the character start position in the predetermined line direction. Set.

Further, the character ruled line separating section 16 scans the image data in the first and second ruled line detection areas of the image data storage section 12, detects the cumulative number of black pixels in a predetermined row direction for each scanning line, and A ruled line detection process in the first and second ruled line detection areas is performed based on the change in the number of black pixels. Further, based on the result of the ruled line detection processing, a character start position and a character end position in a predetermined column direction are set for each character so as to be located closer to the character portion area than the ruled line.

The control unit 18 includes the reading unit 10 and the ruled line detection area setting unit 14 described above.
In addition to controlling the operation of the character ruled line separating unit 16, the character start position and end position in the predetermined line direction and the character start position and end position in the predetermined column direction are output for each character in the character line region with ruled lines. I do.

Hereinafter, this embodiment will be described in more detail. In the following description, the predetermined character start position and end position in the line direction are referred to as the line direction start position and the line direction end position, and the predetermined character start position and end position in the column direction are referred to as the column direction start position and column direction. Called the end position.

(Reading Unit) The reading unit 10 of this embodiment converts the reflected light P from a recording medium such as a manuscript into a black and white binary quantized electric signal (image data), not shown, and converts this image data. It comprises a photoelectric conversion unit for outputting in pixel units, and a scanning mechanism for relatively moving the photoelectric conversion unit and the recording medium for scanning the recording medium.

(Image Data Storage Unit) The image data storage unit 12 of this embodiment is configured using an image memory that stores the image data from the reading unit 10 in a scanning order, and stores one line or more of the image data.

An XY coordinate system is set on the image memory. For example, assuming a horizontally written document, the X-axis direction is a predetermined row direction and the Y-axis direction is a predetermined column direction.

The true row direction of the image may coincide with the X-axis direction or may slightly deviate from the X-axis direction.

It is most preferable that the true line direction of the image coincides with the X-axis direction in order to cut out the characters without causing the loss of characters. However, even if the true line direction is slightly deviated from the X-axis direction, Characters can be cut out with less chipping than before or without chipping.

(Ruling Line Detection Area Setting Unit) The ruled line detection area setting unit 14 of this embodiment sets a character area and detects the cumulative number of black pixels (column cumulative pixel number) in a predetermined column direction in the character area. Marginal distribution generator 20
And a first peripheral distribution storage unit 22 for storing the detected column accumulated black pixels, a black block detection unit 24 for detecting a row direction start position and a row direction start position, and a first and a second ruled line detection area are set. And a detection area position determining unit 26 that performs the detection.

The first marginal distribution creating unit 20 includes one edge position Y _T1 and the other edge in a predetermined column direction of a ruled character line region including one character line and a ruled line added to the character line. Position Y _B1
Is input from the control unit 18. Then, for example, in accordance with the following equation (1) and (2), sets an edge position Y _T2 and the other end edge positions Y _B2 one character region in a given column.

Y _T2 = Y _T1 + α (1) Y _B2 = Y _B1 −α (2) α is a predetermined distance between the character and the ruled line and a predetermined inclination of the true line direction with respect to the predetermined line direction. The constant value arbitrarily and suitably set, for example, α = 16, assuming that it varies within the range. Note that α is a value proportional to the separation distance between the positions Y _T1 and Y _B1 ,
For example, α = (Y _T1 −Y _B1 +1) / 8 may be set.

Then, the first marginal distribution creating unit 30 scans the image data in the character portion area of the image data storage unit 12 based on the positions _YT2 and _YB2 , and, for each position X in a predetermined row direction, a predetermined The number of column accumulated pixels is detected by accumulating black pixels on the scanning lines in the column direction, and the detected column accumulated pixel number is stored in the first peripheral distribution storage unit 22.

Further, the black block detection unit 24 sequentially reads the column accumulated pixel number detected for each position X from the smaller (or larger) position X, and compares the read column accumulated pixel number with a predetermined threshold. . Investigated changes in column cumulative number of pixels by the comparison, the row direction starting end position X _L (one row direction end position X number of columns cumulative pixels above the threshold value when a column cumulative number of pixels is changed from less than a threshold value above the threshold Position X _R ), and when the column cumulative pixel number changes from the threshold value to the threshold value, the position X of the column cumulative pixel number not less than the threshold value is defined as the row direction end position X _R (or the row direction start position X _L ). To detect. Single character between this way detected position X _L and X _R are considered to be present. Hereinafter referred to the region between the position X _L and X _R borders with character line area as the character region.

For each individual character of the black block detecting section 42 row area, to detect the position X _L and X _R.

The detection area position determining unit 26, an adjacent Shikatsu row direction end position X _R in the first ruled-line detection region row direction starting end position X _L is set so as to include a border position on the opposite side, thus the first ruled line a detection region outside the character region adjacent to the row direction starting end position X _L side of character area is set.

In order to improve the detection accuracy of the ruled line in the first ruled line detection area, the width of the first ruled line detection area in a predetermined row direction is set to be within the distance between adjacent characters, and the black pixels constituting the character are in the first ruled line detection area. It is good to reduce the number included in.

Similarly detection area position determining unit 26 sets to include a border position opposite to the row direction end position X _R on adjacent Shikatsu row direction starting end position X _L, thus character region and the second ruled line detection area outside adjacent to the row end position X _R side of character area of the set by. In order to improve the detection accuracy of the ruled line in the second ruled line detection area, the width of the second ruled line detection area in a predetermined line direction is set to a width within the distance between adjacent characters, and the black pixels constituting the character are defined by the second ruled line. It is preferable to reduce the number included in the detection area.

The detection area position determination unit 26 sets the first and second ruled line detection areas for each individual character in the character line.

For example, the first ruled line detection area is set according to the following equations (3) and (4), and the second ruled line detection area is set according to the following equations (5) and (6).

However, sequentially counted n-th character coordinate X of a predetermined row direction to progressively larger direction (hereinafter, simply referred to as n-th character) the position X _L and X _L (n) and the position X _R of X _R (n), n
X _LL (n) represents one edge position of the first ruled line detection area of the second character in the predetermined line direction, and X represents the other edge position.
_LR (n), one edge position in a predetermined line direction of the second ruled line detection area of the n-th character is X _RL (n), the other edge position is X _RR (n), and one character line is within one line. _Is the total number of characters detected in, n is a natural number in the range of 1 ≦ n ≦ n _LS , β
Is a constant that is arbitrarily and suitably set. If the resolution of the optical reading of the reading unit 10 is 16 pixels / mm, for example, β
= 32.

X _LR (n) = _XL (n) −1 (3) • When n = 1 or _XL (n) -1-β> X _R (n−1) +1, X _LL (n) = X _{L (n) -1-β ·} n ≧ 2 and _{X L (n) -1-β} ≦ X R (n-1) +1 when _{X LL (n) = X R} (n-1) +1 ...... ( 4) X _RL (n) = X _R (n) +1 (5) n = n _LS or X _R (n) + 1 + β < _XL (n + 1) -1
_{X RR (n) = X R} (n) + 1 + β · n ≦ n LS -1 and _{X R (n) + 1 +} β ≧ X L (n + 1) when the -
_XRR (n) = _XL (n + 1) -1 when (1) (character ruled line separating unit) The character ruled line separating unit 16 determines the cumulative number of black pixels in a predetermined row direction in the first ruled line detection area. (Hereinafter, referred to as a first row cumulative pixel number), and a second peripheral distribution for detecting a cumulative black pixel number in a predetermined row direction (hereinafter, referred to as a second row cumulative pixel number) in the second ruled line detection area. A creating unit 28, a second peripheral distribution storage unit 30 that stores the detected first and second row accumulated pixel numbers, and a ruled line detection process in the first ruled line detection area based on a change in the first row accumulated pixel number. A ruled line detection unit 32 for performing a ruled line detection process in the second ruled line detection area based on the change in the cumulative number of pixels in the second line, and a starting position and a column direction in the column direction based on the ruled line detection processing result of the ruled line detection unit 32 And a ruled line separation position determining unit 34 for setting the end position to the character portion area side of the ruled line.

In this embodiment, the second peripheral distribution creating unit 28 scans the image data in the first ruled line detection area of the image data storage unit 12 and, for each position Y in a predetermined column direction, in a predetermined row direction. The number of black pixels on the scanning line is accumulated to detect the first row accumulated pixel number, and the detected first row accumulated pixel number is stored in the second peripheral distribution storage unit 30. Similarly, the second marginal distribution creating unit 28 stores the second marginal accumulated pixel number in the detected second marginal distribution storage unit 30.

The ruled line detection unit 32, the edge positions of one side of the first ruled line detection region and Y _T1 at a predetermined column direction, the first ruled lines detected from the position Y _LS1 character line margin of the first ruled line detection area Y is sequentially reduced to the edge position Y _T1 on one side of the area, and the first row accumulated pixel number is read out (where Y _LS1 > Y _T1 and Y
Y so that ruled lines and margins are included between _LS1 and Y _T1
Set _LS1 . It is sufficient to set Y _LS1 = Y _T2 . ), And compare the read first row accumulated pixel count with a predetermined threshold value. By this comparison, a change in the row accumulated pixel number is examined.

When the cumulative number of pixels in the first row changes from less than the threshold to greater than or equal to the threshold, the position Y of the cumulative number of pixels in the first row that is greater than or equal to the threshold is detected as the first ruled line detection processing result _YLT . However, it is assumed that the position Y when the first row accumulated pixel number first changes is the first ruled line detection processing result _YLT .

When finishing the first line reads all the cumulative number of pixels from the first row accumulated without detecting a change in the number of pixel positions Y _LS1 to Y _T1 is the first ruled line detection result Y _LT and edge position Y _T1.

Furthermore ruled line detection unit 32, the other side of the first ruled line detection area in a given column direction edge positions and Y _B1, from the position Y _LS2 character line margin of the first border region of the first ruled line area Y is sequentially increased to the edge position Y _B1 on the other side, and the accumulated pixel number of the first row is read out (however, Y _LS2 <Y _B1 and Y _LS2 to Y _B1
Set _YLS2 to include ruled lines and margins before _B1 . For example, Y _LS2 = Y _B2 may be set. ), And examine the change in the cumulative number of pixels in the first row.

If the first row cumulative pixel number changes from less than the threshold value to the threshold value or more, the position Y of the first row cumulative pixel number that is equal to or greater than the threshold value is detected as the second ruled line detection processing result _YLB . However, the position Y when the first row cumulative number of pixels is first changed is assumed to be the second ruled line detection processing result Y _LB.

When all the first row accumulated pixel numbers from positions Y _LS2 to Y _{B1 have} been read out without detecting a change in the first row accumulated pixel number, the second line detection processing result Y _{LB is set} to the edge position Y _B1 . I do.

The ruled line detection unit 32, the second ruled line detection area of the edge positions of one side in a given column direction and Y _T1, the second ruled lines detected from the position Y _RS1 character line margin of the region second border detection Y is sequentially reduced to the edge position Y _T1 on one side of the area, and the second row accumulated pixel number is read out (where Y _RS1 > Y _T1 and Y
Y so that ruled lines and margins are included between _RS1 and Y _T1
Set _RS1 . For example, Y _RS1 = Y _T2 may be set. ), Examine the change in the second row cumulative number of pixels, the third detection processing result Y _RT corresponding to the result of investigating changes obtained in the same manner as in the first ruled line detection region described above. Further, the ruled line detection unit 3 sets the edge position on the other side of the second ruled line detection area to Y _B1, and sets the edge on the other side of the second ruled line area from the position Y _RS2 of the character line margin in the second ruled line area. Y is sequentially increased to the position Y _B1 and the cumulative number of pixels in the second row is read out (provided that Y _RS2 <Y _B1 and Y
To include ruled lines and blank between the _RS2 to Y _B1 to set the Y _RS2. For example, Y _RS2 = Y _B2 may be set. ), A change in the cumulative number of pixels in the second row is checked, and a fourth detection processing result Y _RB corresponding to the result of checking the change is obtained in the same manner as in the above-described first ruled line detection area.

The appearance of the change in the first and second row cumulative pixel numbers can be regarded as a ruled line being detected, and the detection processing result Y _LT , Y _LB , Y _RT or Y Y when these row cumulative pixel numbers have changed.
_RB can be regarded as indicating the edge position of the ruled line on the character portion area side. When the detection processing results Y _LT = Y _B1 , Y _LB = Y _B1 , Y _RT = Y _T1 or Y _RB = Y _B1 , the first ruled line detection area in the range of Y _T1 ≦ Y ≦ Y _LS1 , Y _LS2 ≦ First ruled line detection area in the range of Y ≦ Y _B1 , Y
Second ruled line detection area in the range of _T1 ≤ Y ≤ Y _RS1 , or Y _RS2 ≤ Y
It can be considered that no ruled line was detected in the second ruled line detection area in the range of ≦ Y _B1 .

Borders separating position determining section 34, for example, in accordance with the following judgment formula (7) and (8), sets the column direction starting end position Y _T and the column end position Y _B in the character region side than the position of the border.

• When Y _LT ≥ Y _RT Y _T = Y _LT · When Y _LT <Y _RT Y _T = Y _RT ... (7) • When Y _LB ≥ Y _RB Y _B = Y _RB · Y _LB <Y _RB In the case of Y _B = Y _LB (8) (Control Unit) The control unit 18 determines the edge position in the predetermined column direction of the character line area with ruled lines according to the format of the recording medium set in the reading unit 10. Y _T1 , Y _B1 and the edge position in the predetermined row direction
Load X _S, and X _E format information storage unit (not shown).
The read positions Y _T1 and Y _B1 and the positions X _S and X _E are output to the ruled line detection area setting unit. Character line region with borders an area between the position Y _T1 to Y _B1 with and position X _S from X _E
It is an area between.

In addition, not only those previously stored in the format information storage unit as the positions Y _T1 and Y _B1 and the positions X _S and X _E but also a part or the entire image data of the recording medium is scanned by a conventionally known line detecting unit. Position Y _T1 , Y _B1 and position X _S ,
It may be used X _E.

The position X _L and X _R of character control unit 18 which borders the detection area setting unit 14 detects a position X _LL and X _LR Border Detection area,
The positions X _RL and X _RR are input, and these positions are stored for each individual character in the ruled character line area.

Then, the control unit 18 outputs the position X _LL , X _LR and the position X _RL , X _RR of the ruled line detection area to the character ruled line separating unit 16 for each character.

Character line separation unit 16 determines the position of the character with respect to one character
When the detection of Y _T and Y _B is completed, the control unit 18 returns to the character ruled line separation unit 16.
Enter and save the position Y _T , Y _B from the saved position Y _T , Y
_B and the positions X _L and X _{R of} the characters from which the positions Y _T and Y _B are obtained are stored as character cutout information. Control unit with this
18 inputs the positions Y _T and Y _B and outputs the positions X _LL and X _LR and the positions X _RL and X _RR of the ruled line detection area for the next character to the character ruled line separation unit 16.

The control unit 18 determines that the character ruled line separation unit 16 has one character Y _T , Y _B
Is detected, the characters Y _T , Y _B , _XL ,
May output the X _R, or letters Y _T of the separating portion 16 all or any suitable number of character lines in the region of detection already character When finished detection of Y _B Y _T, Y _B , X _L and X _R may be output for each character.

Description of Operation Next, in order to deepen the understanding of the present invention, the operation of this embodiment will be described using a more specific example of operation.
In the following description, a character extracting operation for one character line will be described. However, it is apparent that the present invention can be applied to character extracting for a plurality of character lines.

When a processing start signal is input to the control unit 18 via an input unit (not shown), the control unit 18 outputs a reading start signal to the reading unit 10 and outputs the edge position of the character line area with ruled lines from the format information storage unit. Enter Y _T1 , Y _B1 , X _S , X _E and enter these positions Y _T1 , Y _B1 ,
X _S, and outputs the X _E in the first row peripheral distribution generator 20. In preparation for the start of the operation of the ruled line detection area setting section 14, the first line peripheral distribution creating section 20 determines a character area from the input edge positions Y _T1 and Y _B1 and the equations (1) and (2). to the edge position Y _T2 and the one end edge positions, for example, X _S and the other an edge position of the side for example X _E in a given row of a character area and sets the Y _B2 in the row direction setting.

When a reading start signal is input, the reading unit 10 starts optical reading within a predetermined reading range of the recording medium, and outputs image data in pixel units. Image data is output with white pixels constituting the ground of the recording medium being, for example, "0" and black pixels constituting the character being, for example, "1". The image data storage unit 12 stores the image data from the reading unit 10 in a scanning order.

When reading of the entire surface of the predetermined reading range is completed and storage of image data of the entire surface of the reading range is completed, the ruled line detection area setting unit 14 starts operating. Hereinafter, the ruled line detection area setting section
The operation 14 will be described with reference to FIGS. 2 and 3.

FIGS. 2A and 2B are diagrams showing an example of the distribution of the image data and the column cumulative pixel number. FIG. 2A shows image data of one character line included in a predetermined reading range 36 of the recording medium. In FIG. 2A, the X axis and the Y axis are in the XY coordinate system set on the image memory. Represents a coordinate axis. Fig. 2 (B)
Shows the distribution of the number of column accumulated pixels detected for the character row shown in FIG. 2A, where the horizontal axis represents the position X in a predetermined row direction and the vertical axis represents the column accumulated black pixel number at the position X. Represents FIG. 3 is a diagram showing an example of the operation flow of the ruled line detection area setting unit. In FIG. 2A, a region surrounded by a dashed line and denoted by reference numeral 40 is an example of a character line region with a ruled line.

When the ruled line detection area setting section 14 starts operation, the first marginal distribution creating section 20 detects the column accumulated pixel number of the character section area 38 (the area surrounded by the two-dot chain line in FIG. 2A). Start.

The first row peripheral distribution generator 20, the main scanning position X in the character portion area and X _S (S1), between the scans the image data in a predetermined row direction from Y _T2 at position X to Y _B2 By counting the number of black pixels, the column accumulated pixel number HL (X) at position X
Is detected (S2). And the detected column cumulative pixel number HL
(X) is stored in the first peripheral distribution storage unit 22 as HL (X) of the position X.

 The column accumulated pixel number HL (X) can be expressed by the following equation (9).

Here, P (X, Y) represents image data, and P (X, Y)
Y) = 1 represents a black pixel and P (X, Y) = 0 represents a white pixel.

Next, the first marginal distribution creating unit 20 determines whether or not the column accumulated pixel number HL (X) at all positions X in the character portion area 38 has been detected (S3). Position X = X _E unless the next position X
In step S4, 1 is added to the position X in order to detect the column accumulated pixel number in step S4. Position X = assumes that detects a column cumulative pixel number HL (X) in the X all positions if X _E finishes detecting column cumulative number of pixels.

When the detection of the column accumulated pixel number is completed, the black block detection unit 24
Starts detection of the line end position X _L and end position X _R , and sets the number of detected characters n and the read position X of HL (X) to the initial value 1
And the position X _S of the previous position X _S columns in -1 accumulated pixel number HL (X _S -1) is initialized to an initial value X _S virtually
HL and _{(X S -1) = 0 (} S5). Next, the column accumulated pixel number HL (X) at the position X is read from the first peripheral distribution storage unit 22, and the read HL (X) is compared with a predetermined threshold value THL (eg, THL = 1). It is determined whether or not it is a black block element (S6). HL (X) satisfying HL (X) ≧ THL is regarded as a black block element constituting a character, and HL (X) <THL
HL (X) is regarded as a white block element constituting the ground of the recording medium. By setting the threshold value THL to any suitable value, the HL that accumulates noise in the image data
The opportunity to detect (X) as a black block element can be reduced.

When a black block element is detected in S6, the black block detection unit 24 reads the column accumulated pixel number HL (X-1) of the position X-1 immediately before the position X from the first peripheral distribution storage unit 22, The read HL (X-1) is compared with the threshold value THL, and HL (X-
It is determined whether 1) is a white block element (S7).
If the result of the determination in S7 is a white block element, it indicates that a white block element has been detected at position X-1 and a black block element has been detected at position X, and therefore the start of the n-th character in the predetermined line direction The position X is stored as the position _XL (n) (S8), and the operation is performed in S9. If the determination result in S7 is a black block element, S9 is performed without performing S8.

When the white block element is detected in S6, the black block detection unit 24 reads the column accumulated pixel number HL (X-1) of the position X-1 immediately before the position X from the first peripheral distribution storage unit 22. HL (X-1) is compared with the threshold value THL, and HL (X
It is determined whether -1) is a black block element (S1).
0). If the result of the determination in S10 is a black block element, it indicates that a black block element has been detected at position X-1 and a white block element has been detected at position X. Therefore, the end position of the n-th character in the predetermined line direction The position X is stored as X _R (n) (S11), and 1 is added to n to detect the positions X _L (n) and X _R (n) of the next character (S12), and then S9 is performed. . If the determination result in S10 is a white block element, S9 is performed without performing S11 and S12.

In S9, the black block detection unit 24 determines whether or not all the column accumulated pixel numbers HL (X) in the character portion area 38 have been read.
S9 in X = X _E unless all columns accumulated pixel number HL
Since the reading of (X) has not been completed, the HL at the next position X
To read (X), add 1 to X (S13), then S6
Return to Since read all columns accumulated pixel number HL (X) if X = X _E in S9, the black block detection unit 24 is the row direction starting end position X _L and the end position for all the characters in the character area 38 to end the detection of X _R.

When the detection of the position X _L and X _R is terminated, the detection area position determining unit 26 starts the setting of the first and second ruled line detection area. The detection area position determination unit 26 informs the X direction in the character area 38 that the size of the X becomes gradually larger.
_L (n) and the end position X _R (n) is read from the black block detection unit 24, the read position X _L (n), from the X _R (n) Equation (3) - and (6), n th The edge position X _LL (n), X _LR (n) of the first ruled line detection area of the character and the edge position X _RL (n), X _RR (n) of the second ruled line detection area are set to detect the ruled line. The area is set (S14), and these positions X _L (n) and X _R
(N) and position X _LL (n), X _LR (n), X _RL (n), X
_RR (n) is output to the control unit 18. The control unit 18 is 1
The start and end positions in the row direction of the nth to _nLSth characters and the edge positions of the first and second ruled line detection areas are input and saved for each character.

When the detection area position determination unit 26 outputs the line direction start and end positions of the n _LS- th character and the edge positions of the first and second ruled line detection areas, the ruled line detection area setting unit 14 ends the operation. .

When the ruled line detection area setting unit 14 ends the operation, the control unit 18
Outputs the start and end positions in the row direction and the edge positions of the first and second ruled line detection areas to the character ruled line separation unit 16 for each character from the first character to the _nLSth character. The character ruled line separation unit 16 inputs the starting position and the ending position of one character in the line direction and the edge positions of the first and second ruled line detection areas, and based on these positions, detects the character column direction of one character. The start position and the end position are detected and output to the control unit 18. When the control unit 18 inputs the start position and end position of one character in the column direction from the character ruled line separation unit 16, the control unit 18 determines the input start position and end position of the input column direction and the start position and end position of the detected character in the row direction. In addition to outputting the cut-out information, the start and end positions of the next character in the line direction and the end positions of the first and second ruled line detection areas are determined by the character ruled line separating unit 16.
Output to

The control unit 18 by entering the column direction starting end position and the end position of the n _LS th character from the character ruled line separating section 16, the n _LS-th character string direction start and end positions in the row direction beginning and end positions and the output of After that, the character extraction for one character line ends.

Hereinafter, with reference to FIG. 4 and FIG.
An example of the operation will be described.

FIG. 4 (A) is a diagram mainly showing image data of one character, and FIGS. 4 (B) and (C) are the first row cumulative pixel count and the second row cumulative of the image data shown in FIG. 4 (A). It is a figure showing distribution of the number of pixels. FIGS. 5A and 5B are diagrams showing an example of the operation flow of the character ruled line separation unit.

Character ruled line separating section 16, to the row direction starting end position and the end position X _L (n) and X _R (n) from the control unit 18, end edge positions X _LL of the first ruled line detection area (n), X _LR (n ) And the edge positions X _RL (n) and X _RR (n) of the second ruled line detection area, the
Th starts detection in the column direction starting end position and the end position Y _T and Y _B character, first the second peripheral distribution generator 28 is the n-th character of the first ruled line detection region 42 in line cumulative pixel BLH ( Y) and detection of the row cumulative pixel number BRH (Y) in the second ruled line detection area 44 are started. In FIG. 4, the first ruled line detection area 42 is indicated by a three-dot chain line, and the second detection area 44 is indicated by a four-dot chain line.

Second marginal distribution creating unit 28 that has started detecting the row cumulative pixel number
Sets the scanning position Y in the predetermined column direction to Y _T1 (* S1), and then scans the image data in the predetermined row direction at the position Y to determine the position from X _LL (n) to X _LR (n). The number of black pixels between them is counted to detect the first row accumulated pixel number BLH (Y), and the detected BLH (Y) is stored in the second peripheral distribution storage unit 30 as the first row accumulated pixel number at the position Y. . At the same time, the image data is scanned at the position Y and X _RL (n) is
_The number of black pixels up to _RR (n) is counted, the second row cumulative pixel number BRH (Y) is detected, and the detected BRH (Y) is used as the second row cumulative pixel number at the position Y to obtain the second peripheral distribution. It is stored in the storage unit 30 (* S2).

The row accumulated pixel numbers BLH (Y) and BRH (Y) can be expressed by the following equations (10) and (11).

Next, the second margin distribution creating unit 28 sets the ruled line detection areas 42 and 44
It is determined whether or not the row accumulated pixel number has been detected at all positions Y in (* S3). If the position Y is not equal to Y _B1 , one is added to the position Y (* S4) in order to detect the number of accumulated pixels at the next position Y (* S4), and then the process returns to * S2. Position Y = X _E a long Invite column cumulative number of pixels in all positions Y BLH (Y), BRH
Assuming that (Y) has been detected, the detection of the row cumulative pixel number ends.

Line the accumulated pixel number of the detection is completed ruled line detecting unit 32 starts the detection of the column direction starting end position Y _T and the column end position Y _B, BLH
Reading position Y (Y) is set to Y _T2 as Y _LS1 (* S5). Then the Compared position Y column cumulative number of pixels BLH a (Y) from the second peripheral distribution storage unit 30 reads, the read BLH a (Y) with a predetermined threshold value THL _L BLH (Y) is whether black block elements Is determined (* S6).

The value of THL _L can be any suitable value. In this example, THL _L is set by, for example, the following equation (12).

When detecting a black block element in * S6, the ruled line detection unit 32 detects the ruled line, and stores the detection position Y of the black block element as a ruled line detection processing result _YLT (* S7), and then * S10 Do. * If it does not detect the black block elements in S6, judges read Taka whether all BLH a (Y) from Y _LS1 to Y _T1 (* S8). If Y = Y _T1 in * S7, all the BLH (Y) have been read out without detecting the ruled line, and the position Y = Y _T1 is stored as the ruled line detection processing result Y _LT (* S7), and then * Perform S10. * Y = Y _{T1 at} S7
Otherwise, position Y to read BLH (Y) at next position Y
1 (* S9), and then returns to * S6.

The ruled line detection unit 32 sets the read position Y of BLH (Y) at * S10.
Set to Y _B2 as Y _LS2 . Then read position Y column cumulative number of pixels BLH a (Y) from the second peripheral distribution storage unit 30, the reading BLH a (Y) is compared with a threshold value THL _L BLH (Y) is determined whether the black block elements (* S11).

When detecting a black block element in * S11, the ruled line detection unit 32 detects a ruled line, and stores the detection position Y of the black block element as a ruled line detection processing result _YLB (* S12).
Next, * S15 is performed. * If it does not detect the black block in S11, it determines read Taka whether all BLH a (Y) from Y _LS2 to Y _B1 (* S13). If Y = Y _B1 in * S13, reading of all BLH (Y) is completed without detecting a ruled line, and the position Y = Y _B1 is stored as a ruled line detection processing result _YLB (* S12). * Perform S15. * Y in S13
If not = Y _B1 , 1 is added to the position Y to read BLH (Y) at the next position Y (* S14), and then the process returns to * S11.

The ruled line detection unit 32 sets the read position Y of BRL (Y) in * S15.
Set to Y _T2 Y as Y _RS1 . Then read position Y column cumulative number of pixels BRL a (Y) from the second peripheral distribution storage unit 30, the BRL the read BRH a (Y) compared with a threshold value THL _R (Y) is whether black block elements Judge (* S16).

The value of THL _R can be any suitable value. In this example, THL _R is set by the following equation (13).

When detecting a black block element in * S16, the ruled line detection unit 32 stores the detected position Y of the black block element as a ruled line detection processing result Y _RB (* S17), and then performs * S20. If no black block element is detected in * S17, it is determined whether or not all BRH (Y) from Y _RS1 to Y _T1 have been read (* S18). * If Y = Y _T1 in S18, all the BRH (Y) have been read out without detecting the ruled line, and the position Y = Y _T1 is stored as the ruled line detection processing result Y _RT (*
S17), then perform * S20. Unless Y = Y _{T1 in} * S18, 1 is subtracted from the position Y to read the BRH (Y) at the next position Y (* S19), and then the process returns to * S16.

The ruled line detection unit 32 sets the read position Y of BRY (Y) at * S20.
Set to Y _B2 as Y _RS2 . Then read position Y column cumulative number of pixels BRH a (Y) from the second peripheral distribution storage unit 30, the BRH the read BRH a (Y) compared with a threshold value THL _R (Y) is whether black block elements It is determined (* S21).

When the ruled line detection unit 32 detects a black block element in * S21, the detection position Y of the black block element is stored as a ruled line detection processing result Y _RB (* S22), and the ruled line detection processing of the n-th character is performed. The detected Y _LT , Y _LB , Y _RT, and Y _RB are output to the ruled line separation position determination unit 34, and the ruled line detection process is completed. * If it does not detect the black block elements in S21, determines read Taka whether all BRH a (Y) from Y _RS2 to Y _B1 (* S23). * If Y = Y _B1 in S23, all the BRH (Y) have been read out without detecting the ruled line, and the position Y = Y _B1 is stored as the ruled line detection processing result Y _RB (*
S22), the n-th character in the ruled line detection processing result as the detected Y _LT, Y _LB, ends the output was ruled line detection process with respect to Y _RT and Y _RB borders separating position determining unit 34. * Unless Y = Y _{B1 in} S23, 1 is set to the position Y where BRH (Y) of the next position Y should be read.
Addition (* S24), and then return to * S21.

Upon receiving the ruled line detection processing result from the ruled line detection unit 32, the ruled line separation position determination unit 34, in accordance with equations (7) and (8), starts the column direction start position Y _T and column direction end position Y _T of the n-th character. detecting a _B (* S25), controls the detected Y _T and Y _B section 1
Output to 8.

For example, from the image data shown in FIG. 4, Y _RB is detected as Y _LT as the column direction starting end position Y _T character and column end position, thus the true row direction of the image deviates from a predetermined row direction Even when the character 46 and the ruled line 48 are in contact with each other, the character 46 and the ruled line 48 can be separated and cut out.

The present invention is not limited to the above-described embodiment. Therefore, the operation flow of each component, input / output signals, operation timing, connection relationship, and configuration can be arbitrarily and suitably changed.

For example, in the above-described embodiment, a description has been given of an example in which the character start and end positions in a predetermined line direction and the character start and end positions in a predetermined column direction are detected using image data of a horizontally written document as an example. Obviously, can be applied to vertically written documents.

FIG. 6 is a diagram showing an example of image data of a vertically written document. For example, as shown in the figure, if an XY coordinate system is set on the image memory and the X axis is set to a predetermined row direction, the character starting position in the predetermined row and column direction characters and The end position can be detected.

(Effects of the Invention) As is clear from the above description, according to the character extracting device of the present invention, the character start position and the end position in the predetermined line direction are determined by the cumulative number of black pixels in the character portion region in the predetermined column direction. , The detection process of the start and end positions can be simplified. In addition, since the character portion area is set so as to include only the character portion of one character line so as not to include the ruled line, the black pixels constituting the ruled line are not counted as the cumulative number of black pixels in the predetermined column direction. Alternatively, it is possible to count almost only a small amount, and as a result, it is possible to improve the detection accuracy of these start and end positions.

Further, the detection of the character start position and the end position in the predetermined column direction is performed based on the change of the cumulative black pixel number in the predetermined row direction in the first and second ruled line detection areas. Since the processing is started from the processing, the detection processing of these start and end positions can be simplified.

Moreover, since the first ruled line detection area is set to include a ruled line at a position adjacent to the character start position in the predetermined line direction and opposite to the character end position in the predetermined line direction,
It is possible to prevent black pixels constituting a character from being counted or hardly counted as the cumulative number of black pixels in the predetermined row direction in the first ruled line detection area. As a result, the position of the ruled line in the first ruled line detection area can be detected with high accuracy. Similarly, the second ruled line detection area is set to include a ruled line at a position adjacent to the character end position in the predetermined line direction and opposite to the character start position in the predetermined line direction. The constituent black pixels can be prevented from being counted or hardly counted as the cumulative number of black pixels in a predetermined row direction in the second ruled line detection area. As a result, the position of the ruled line in the second ruled line detection area can be detected with high accuracy.

Therefore, since the position of the ruled line in the first and second ruled line detection areas can be detected with high accuracy, even if the predetermined line direction is slightly shifted from the true line direction of the image, and / or the ruled line is in contact with the character. However, a character can be cut out at a character start position and an end position in a predetermined column direction so that the character is not chipped or the character is cut less than before.

Further, since the detection of the ruled lines in the first and second ruled line detection areas is performed based on the change in the number of accumulated black pixels, the detection of the number of accumulated black pixels and the threshold processing are performed. Can be done with

Therefore, according to the clipping device of the present invention, a device capable of accurately detecting the start and end positions of a character in a predetermined row and column direction by simple processing even when the predetermined row direction is deviated from the true row direction. Can be provided. In addition, since the processing is simple, the structure of the apparatus can be reduced in size and simplified, so that a low-cost character extracting apparatus can be obtained.

If the present invention is applied to a character recognition device, for example, when characters or ruled lines are printed on the document with a slight inclination, even if the reading unit main-scans the document in a direction deviating from the true line direction, the character Can be extracted with high accuracy, and therefore, accurate character recognition can be performed.

[Brief description of the drawings]

FIG. 1 is a functional block diagram showing an example of the configuration of an embodiment of the present invention. FIGS. 2 (A) and (B) are diagrams showing the distribution of image data and the number of accumulated black pixels in a column. FIG. 4A shows an example of one character image data, and FIGS. 4B and 4C show row accumulation in the first ruled line detection area. FIGS. 5A and 5B show an example of the distribution of the number of black pixels and an example of the distribution of the cumulative number of black pixels in the line in the second ruled line detection area. FIGS. FIG. 6 is a diagram showing an example of image data of a vertically written document. 10 reading unit 12 image data storage unit 14 ruled line detection area setting unit 16 character ruled line separation unit 18 control unit 20 first peripheral distribution creating unit 22 first peripheral distribution storage Unit 24 Black block detection unit 26 Detection area position determination unit 28 Second peripheral distribution creation unit 30 Second peripheral distribution storage unit 32 Ruled line detection unit 34 Ruled line separation position determination unit

Continued on the front page (58) Fields surveyed (Int.Cl. ⁶ , DB name) G06K 9/20 G06K 9/34

Claims (1)

(57) [Claims] A reading section for optically reading an information medium on which a character line with a ruled line is described and outputting image data of the information medium; an image data storage section for storing image data from the reading section; The area is set in the character line area with a ruled line so as not to include a ruled line, and the image data in the character part area of the image data storage unit is scanned, and the cumulative number of black pixels in a predetermined column direction is determined for each scanning line. Detecting, based on the change in the cumulative number of black pixels in the column direction, detects the character start position and end position in the predetermined line direction for each character, based on the character start position and end position in the line direction,
The first ruled line detection area is set for each character so as to include a ruled line at a position adjacent to the character start position in the predetermined line direction and opposite to the character end position in the predetermined line direction. A ruled line detection area setting unit that sets, for each character, a ruled line adjacent to the character end position in the predetermined line direction and opposite to the character start position in the predetermined line direction; and Scanning the image data in the first and second ruled line detection areas of the section to detect the cumulative number of black pixels in a predetermined row direction for each scanning line, and based on the change in the cumulative number of black pixels in the row direction, A character rule separation unit configured to perform a rule detection process in the second rule detection region for each character, and to set a character start position and an end position in the predetermined column direction for each character based on a result of the rule detection process; Character extraction device characterized by comprising .