JPH03141482A - Character segmenting device - Google Patents

Abstract

PURPOSE:To improve precision by scanning inside a mask scanning area by using a scanning line mask only when a character line area whose length in a line direction exceeds a threshold is detected, and detecting a scanning path in which the number of black bits becomes minimum, and segmenting the character pattern of one unit on the basis of this position. CONSTITUTION:In the case the character line area whose length in the line direction exceeds the prescribed threshold is detected by a block detecting means 12, it is considered to be a contact character area where adjoining character patterns overlap each other or are in contact with each other, and the mask scanning area is set for the contact character area by a scanning area setting means 14, and a mask scanning means 16 scans it by using plural scanning line masks of different paths. Thus, the scanning path is which the number of the black bits on the path becomes minimum is detected from among the scanning paths of the scanning line masks, and a character segmenting means 18 divides the contact character area by a dividing line set on the basis of the position of the minimum scanning path, and segments the character pattern of every divided contact character area as the character pattern of one unit. Thus, the character pattern of one unit can be segmented precisely.

Description

DETAILED DESCRIPTION OF THE INVENTION (Field of Industrial Application) The present invention relates to a character cutting device for cutting out a character/graphic pattern in units of - from an original image pattern of a character/graphic character marked "W&" on a form.

(Prior Art) Conventionally, in character recognition processing, a pattern (data) for one character has been separated from an original image pattern (data) of a character figure stored on a line-by-line basis.

In the conventional technology widely used for this separation, the original image pattern for one line is scanned in the column direction perpendicular to the character line direction, and black bits representing the character line portion are accumulated for each main scanning position. Create a histogram of black bits for the image pattern. Then, the cumulative number of black bits is compared with a predetermined specific value S for each main scanning position, and the main scanning position where the cumulative number of black bits is S or more and a value smaller than S as the character start and end positions in the line direction. Detects the main scanning position where the character starts and ends, and separates (cuts out) the pattern of the area from the start of the character to the end of the character as a pattern of one character.

However, in this prior art, when adjacent characters overlap or touch each other, the distribution areas of the black bits representing these characters overlap and become one, making it impossible to correctly cut out a pattern for the characters.

As a conventional technique for solving such problems, Document I:
There is a technique disclosed in Japanese Patent Publication No. 62-46039.

This prior art is a technique for cutting out a pattern for one character from the original image pattern of a plurality of characters written in a band-shaped character frame of a constant width. A change in the first character distance in the column direction from the frame line to the character line part and a change in the second character distance in the column direction from the other frame line of the character frame to the character line part are detected, and changes in these character distances are detected. Detects the change point where the change is more than a predetermined value, and detects the change point at least in the character line direction between the first change point detected on one frame side and the second change point detected on the other frame side. A pair of change points within a predetermined distance of one length are detected as one set, and a character cutting position is determined in accordance with the position of the pair of change points. As a result, even when adjacent characters overlap, it is possible to accurately separate these touching characters and cut out a pattern for one character.

(Problem to be Solved by the Invention) However, in the prior art of the literature engineer described above, in order to detect a change in the first character distance, the original image pattern is scanned over the entire surface of the character frame, and the change in the second character distance is roughly detected. In order to detect the character frame, it is necessary to scan the original image pattern over the entire surface of the character frame, and there is a problem in that the processing speed becomes slow because these two scans are performed.

The purpose of this invention is to solve the above-mentioned conventional problems by providing a character cutting device M that can cut out a pattern for one character from an original image pattern with a small amount of scanning when characters overlap or touch each other. It is about providing.

(Means for Solving the Problem) In order to achieve this object, the character cutting device of the present invention cuts out one unit of character figure pattern from the original image pattern of character figure. block detection means for detecting a character line area of one row of the original image pattern based on a histogram of cumulative black and nt counts in the column direction obtained by projecting the original image pattern in a column direction perpendicular to the first row direction; scanning area setting means for determining the character line area to be a contact character area and setting the position of a mask scanning area for the contact character area when the length of the character line area in the row direction exceeds a predetermined threshold; a mask scanning means for scanning an original image pattern within a mask scanning area using a plurality of scanning line masks, and detecting a minimum scanning path in which the number of black bits on the scanning path is minimized from among the scanning paths of the scanning line mask; , character cutting means that divides the contact character area by a dividing line set based on the position of the minimum scanning path, and cuts out the character figure pattern of each divided contact character area as one unit character figure pattern. A character cutting device characterized by:

(Function) According to the character cutting device H having such a configuration, when a character line area whose length in the line direction exceeds a predetermined threshold is detected, adjacent character figures overlap each other in the character line area. is considered to be the touching character area (hereinafter,
These character figures that overlap or touch each other are called touching character figures). Then, a dividing line is set to divide this touching character area into each character figure of the touching character figures. The character/graphic pattern of the contact character area divided by the dividing line is cut out from the original image data as one unit, for example, a pattern for a - character.

In setting the dividing line, a mask scanning area is set for the contact character area, and the inside of this mask scanning area is scanned using a plurality of scanning line masks having different scanning paths. and,
A scanning path with the minimum number of black bits on that path is detected from among the scanning paths of the scanning line mask. Since the number of black bits on this minimum scanning path is the minimum, it can be considered that the character lines of the touching character figures have the least amount of overlap or no overlap in the minimum scanning path and the area near this path.

Therefore, by setting a dividing line at any suitable position within the area that includes the minimum scanning path and the area near this path (for example, by setting the minimum scanning path as the dividing line), it is possible to accurately cut out one unit of character/figure pattern. .

(Embodiments) Hereinafter, embodiments of the present invention will be described with reference to the drawings. It should be noted that the drawings are merely shown schematically to the extent that this invention can be understood, and therefore the shapes of each component and the arrangement values 1
and dimensions are not limited to the illustrated example.

FIG. 1 is a functional block diagram for explaining the configuration of an embodiment of the present invention.

As shown in FIG. 1, the character cutting device 10 of this embodiment produces a histogram of the cumulative number of black bits in the column direction, which is obtained by projecting the original image pattern for one row in the column direction perpendicular to the row direction. Based on this, a block detection means 12 detects a character line area of the original image pattern for one line, and when the length of the character vJ area area in the row direction exceeds a predetermined threshold, the character line area is regarded as a contact character area. Scanning area setting means 14 sets the position of the mask scanning area in the contact character area, and scans the original image pattern in the mask scanning area using a plurality of scanning line masks each having a different scanning path, and sets the scanning path of the scanning line mask. The mask scanning means 16 detects the minimum scanning path with the minimum number of black bits on the scanning path from among the scanning paths, and the contact character area is divided by a dividing line set based on the position of the minimum scanning path, and the contact character area is divided into the divided contact characters. Character cutting means 1 for cutting out character and figure patterns for each region as one unit of character and figure patterns
8, and a mask memory 20 that roughly stores a scanning line mask.

This example will be explained in more detail below.

In this embodiment, as the original image pattern, the photoelectric conversion unit 22
The image pattern of the elephant is used.

The photoelectric conversion unit 22 optically scans the text line writing area of the form one line at a time, photoelectrically converts the optical signal from the text line writing area, and converts it into a quantized image pattern, for example, converts the text line portion into black bits. The 0 image memory 24, which outputs a black and white binary digital signal in which the character background is expressed with white bits, stores the image pattern for one line from the photoelectric converter 22 as an original image pattern.

FIG. 2 shows an example of the original image pattern stored in the image memory 24, and in the figure, 26 indicates the original image pattern.

As shown in FIG. 2, an X-Y coordinate system is set on the image memory 24 with an arbitrary suitable position as the origin ○, and the data of the original image pattern at the pixel position represented by this coordinate system is set. It is structured so that it can be read freely. In the example shown in FIG. 2, since the character lines are written horizontally, the X-axis direction is hypothetically the row direction along the character lines, and the Y-axis direction is the column direction perpendicular to the row direction.

When the original image pattern is stored in the image memory 24, the block detection unit 12 projects one row of the original image pattern 26 in the column direction to create a histogram of the cumulative number of black bits in the column direction.

FIG. 3 is a diagram showing an example of a histogram created by scanning the image pattern 26P8 in FIG. 2, in which the vertical axis is the cumulative number of black hits, and the horizontal axis is the It is.

In order to create a histogram, the block detection means 12
The original image data 26 in the image memory 24 is scanned with the column direction as the main scanning direction. Then, for each sub-scanning position

Next, the block detection means 12 examines the histogram and detects character line areas in the row direction of the original image data 26.

Fig. 4 is a diagram showing the distribution state of character line areas detected by examining the histogram in Fig. 3. In Fig. 1, the horizontal axis is the X axis corresponding to the , the position of the detected character line area is represented by a hatched rectangular area. In FIG.
=1.2.3). Also, in Figure 4,
The starting end position and end value M of the character line area Bn are shown with the symbol ×3□ and Xgn attached, and for the sake of rough understanding, the starting end position of FIG. 4 in FIGS. il Xs
n and end position X. The positions H corresponding to are shown with the symbols XS++, XE, and 1. In the following description, the detected character line area Bn, start position XSn, and end value MX will be described.

simply character line area B, start position 11Xs and end position XE
Also expressed as

In order to detect the character line area B, the print detection means 12 detects the character line area B so that the cumulative number of black bits in the column direction is set to a predetermined threshold THL+
(For example, THLl=1) The sub-scanning position It
Detect and THL 1 of these positions x8 and XEFffi
Find the length X in the row direction of the area [8] where the above cumulative black bit numbers are continuous. Then, the length ΔX is compared with a predetermined threshold value THL2 (for example, THL2=2), and a region B having a length of X greater than or equal to the threshold value THL2 is detected as a character line region. The positions Wxs and XE at this time are the starting and ending positions of the character line area B in the row direction. Threshold THL
By setting + and THL2 arbitrarily and suitably, it is possible to avoid detecting a noise area in the original image data 26 as a character line area.

When the character line area B is detected, the scanning area setting means 14 compares the length X of the character line area B inputted from the block detection means 12 with a predetermined threshold value T)IL3. Threshold T
HL3 is for determining how many units of character/figure patterns the character line area tftB includes. In this embodiment, a threshold value THL3 of one arbitrary suitable value is set, and the threshold value T
A character line area B having a length ΔX of HL3 or less includes one unit of character/figure patterns, and a character line area B having a length ΔX exceeding a threshold T)lL3 includes overlapping or contacting character figures (contacting For example, the numbers in Figure 2
It is assumed that this is a touching character area that includes two units of patterns, ``1'' and ``2''.

When the length ΔX is less than or equal to the threshold T)IL3, the scanning area setting means 14 sets the starting end position X of the character line area B of the length ΔX and x6, and information P indicating that the character line area B is not divided. Output.

Further, the scanning area setting means 14 has a length ΔX as a threshold i1 THL
When the length exceeds 3, the character line area B with the length ΔX is a contact character area that includes contact character figures, so the position of the mask scanning area is set to obtain a dividing line to separate these character figures unit by unit. Then, the start position xs and end value M X E of the touched character area B, and the start end position XM8 and end value WIXM in the row direction of the mask scanning area. and information Q indicating that the touched character area B is divided.

In this embodiment, the minimum cumulative number lFr of black bits is detected from among the cumulative number of black bits in the column direction of the entire area or local area of the touching character area B, and the position in the row direction where this minimum cumulative number of black bits is obtained is determined. The position XMS and X ut' of the mask scanning area so as to include Xff, i□ and the area near the position
Set jr. The settings of XMS and XME may be set so that, for example, X s S X MS<X ME≦Xε.

FIG. 5 is a diagram showing an example of setting the mask scanning area.
A histogram of the cumulative number of black bits in the column direction similar to the figure is shown.

In the example shown in FIG. 5, a local area is set at an arbitrary suitable position, for example, the center of the contact character area B, and the contact character area B is set to have a length of ΔX/2 in the line direction, for example.
Xc (ΔX74)≦X≦Xe+(Δx/4)
Let the node gyrus of be the local region. However, Xc represents the center position of the contact character area B in the row direction.

Further, the length of the mask scanning area 3o in the row direction is, for example, X/4.
For example, in contact character area B (
7)X... -(ΔX/8)≦x≦x,,.

The area within the range of +(ΔX/8) is defined as a mask scanning area of 3°.

The mask scanning means 16 receives information P from the scanning area setting means 14.
Or input Q and perform the operation according to this information.

When the information P is input, the mask scanning means 16 outputs the information P input from the scanning area setting means 14 and the positions XS and X of the character line area B.

Further, when information Q is input, the mask scanning means 16 selects the position xa of the contact character area B input from the scanning area setting means 14.
and XE, and the positions XM8 and XMl of the mask scanning area:
Based on this, an operation for processing using a scanning line mask is started. This operation will be explained with reference to FIGS. 6 to 8.

FIG. 6 is a diagram for explaining the position detection of the touched character area in the column direction, and shows the same original image pattern as FIG. 2.

First, the mask scanning means 16 detects the starting end position Y and the ending end position Yt of the contact character area B in the column direction.

In order to detect these positions Ys and YE, the mask scanning means 16 projects the original image pattern 26 in the contact character area B (area where xs≦X≦X) in the row direction and calculates the cumulative number of black bits in the row direction. Create a histogram of In this creation, original image data 26 of the contact character area [8] is scanned with the line direction as the main scanning direction, and 1j! SC2
(See FIG. 6) The above black bit numbers are accumulated to obtain a histogram consisting of the cumulative black bit numbers for each ItY. Then, this histogram is examined in the same manner as in the case of detecting the character line area B in the row direction by the block detection means 12, and the starting end position Y of the character line area in the column direction on the side of the contact character figure is determined.

and the terminal position NY.

As shown in Fig. 6, the points (Xs, Ys), (XE
, YS ), (Xs , Ye ) and (XE, Y
The rectangular frame connecting E) becomes the circumscribing frame G of the contact character figure.

Then, the mask scanning means 16 detects 1tYs and Y
6, the starting position of the mask scanning area 3o in the column direction!
Set Yms and YMI-. These positions VMS and Y■ may be set so that, for example, Y□≦Y8 and YE≦YME within the storage area of the image pattern 26 for one line, but in the following explanation, VMS = Ys and Let YME=Yi.

After setting the positions Y Ml and Y Mi of the mask scanning area 30, the mask scanning means 16 then scans the mask scanning area 30.
Mask scanning is performed in the area where XMII≦X≦XME and YM8≦Y≦Ywe.

FIG. 7 is a diagram showing an example of the flow of mask scanning processing by the mask scanning means.

The mask scanning means 16 scans the mask scanning area 30 at a position Y
After setting MS, Y ME, the corresponding mask area 3
Start mask scanning for the contact character area B where 0 is set (5TART).

The mask scanning means 16 that has started the mask scanning calculates a minimum value register (
MIN register) to an initial value of 1000, for example! Set and M
Initialize the IN register (Sl), and scan No. to specify the number (No) assigned to each scan line mask.
The scanning number counter is initialized by setting the counter to an initial value, for example, 11Fr (S2).

Next, the mask scanning means 16 reads the scanning line mask of the number specified by the stored value of the scanning number counter from the mask memory 20, and counts the number of black bits on the scanning path SC3 of the scanning line mask within the mask scanning area 30. (cumulative),
The count result is stored in the 81ACK register (S3).

FIGS. 8(A) to 8(F) are diagrams showing an example in which six types of scanning line masks are prepared as one set of scanning line masks.
A), (B), (C), (D), (E) and (F) are each scan N.

1.2.3.4.5 and 6 are shown as scan line masks. In these figures, line segments ■, ■, ■, ■, and ■ are straight lines X=×□, Y=YMs+△Y/2, and X=XM, respectively.
E, Y = YM8+△Y/4 and Y=YM8+(Δ
Y・3/4), where ΔY is the position YM
8 and YME in the column direction), as shown in FIG. By any suitable combination of 61! A scanning path SC3 is formed.

Next to S3, the mask scanning means 16 checks whether the contents of the BLACK register are ○ or not, that is, the scanning path SC3 where there is no black hit on the scanning path SC3! Determine whether or not it has been detected (
S4).

If the contents of the register 5LAG: are not O in S4, the mask scanning means 16 performs the processing in S5.

The mask scanning area 30 in FIG. 6 is scanned by the scanning number in FIG. 8(A).
When scanning is performed using a scanning line mask of 1, the scanning path SC
The cumulative number of black bits on 3 is 2, so 2 is stored in the register 81AC, and the process of S5 is performed after S4.

In S5, the contents (stored values) of the BLACK register and the MIN register are compared.

If the contents of the BLAC register are smaller than the contents of the MIN register in S5, the contents of the MIN register are changed to BLACK in order to detect the minimum number of cumulative black bits on the scanning path SCa.
Rewrite the contents of the register, and along with this, the minimum scanning path S0
In order to detect 3%, the contents of the 5CAN register are rewritten to the contents of the scan No. register at this time (S6), and then scan N
The scan number is updated by adding 1 to the value stored in the o register (S7).

In scan No. 1, the contents of the BLACK register are 2.
Since the content of the MIN register is 1000, 86 and S7 are performed after 85.

Further, if the contents of the BLACK register are equal to or larger than the contents of the MIN register in S5, the process of 87 is performed without performing 86 after S5.

Next to S7, the mask scanning means 16 determines whether or not mask scanning has been completed for all of the scanning line masks (for example, the six types of scanning line masks shown in FIG. 8) stored in the mask memory 20 (S8 ).

If there are remaining scanning line masks that have not been subjected to mask scanning in S8, the process returns to S3.

Furthermore, if mask scanning has been completed for all of the scanning line masks in S8, the scanning path SC3 of the scanning line mask with the number specified by the value stored in the 5CAN register is the minimum scanning path, so the position of the minimum scanning path is , the position X in the row direction of the touched character area B, ,X, and the information Q should be outputted (S9).
Next, the mask scanning process is finished, and when the next touching character area SIJ< is detected and the position Y Ml Y Ml of the mask scanning area 30 is set for the next touching character area SIJ, the process waits to start the process from Sl ( END).

Further, if the contents of the BLACK register are 0 in S4, the mask scanning means 16 determines that the scanning path SC3 of the scanning line mask specified by the contents (stored value) of the scanning No. register at this time is the minimum scanning path. Then, the content of the 5CAN register is rewritten to the content of the scan NO register at this time (SIO), and then the process of S9 is performed.

The scan number of the mask scanning area 30 in FIG. 6 is M8 (CC).
When scanning using a scanning line mask of 3, the scanning path SC
The cumulative number of hits on a becomes 0, and since the content of the 8LACK register is 07: in S4, the process of 810 is performed after S4. In this case, in S10, the position of the minimum scanning path SC3 is determined, for example, the end point D 1 (X
ME, Ys), D4 (X, S, Y ME) and bending point D2 (XME, Y.

+3・△Y/4), D3 (XMS, Y3+3・△y
/4) position is output.

When the character cutting means 18 receives the information P and the row direction positions 1fXs and Xg of the character line area B from the mask scanning means 16, the character cutting means 18 converts the pattern of the position X5 and XE circumference of the image data 26 for one line into one unit of character figure. Cut out and output as a pattern.

Further, when the character cutting means 18 inputs the information Q, the position of the minimum scanning path 1tD1 to D4, and the positions XS and XE of the contact character area B from the mask scanning means 16, The touching character area B is divided based on ,

As shown in FIG. 6, for example, the point (XME, o), Dl
, D2, D3, D4 and (XMS, Y-a-) @ may be used as dividing lines to divide the contact character area B.

The present invention is not limited to the embodiments described above, and therefore the numerical conditions, ranges, shapes, positional relationships, operations, and flows of input and output signals of each component can be changed as desired.

For example, in the above-mentioned embodiment, six masks were prepared as scanning line masks, but by increasing the number of masks, characters can be cut out accurately without missing a unit of character figure pattern or with minimizing the missing part. You will be able to do this.

FIGS. 9A to 9H are diagrams showing an example in which a 5tis scanning line mask is prepared. In these figures, the line segments ■, ■,
■ and ■ are straight lines Y=YMs+ΔY/8, Y=Y, respectively
M! ++ΔY −3/8, Y=YM8+ΔY・5/8 and Y=VMS+ΔY・7/8, showing the scanning path on the 9th
As shown in the figure, scanning path ■, scanning path ■, ■, ■
Alternatively, eight types of scanning paths 5C38 are formed by combining and combining (1) and scanning path (2) at arbitrary suitable positions and lengths.

The positions of the scanning paths ■, ■, ■, ■ in the row direction in one mask of these eight types of scanning lines change in units of ΔY/8, and therefore, contact is made accurately according to the change in units of ΔY/8. It is possible to divide the character area and cut out one unit of character/graphic pattern.

The shape of the scanning path of the scanning line mask can be changed arbitrarily and suitably, and in addition to the above-mentioned scanning line mask, it can be used to separate two consecutive characters (or contacting character figures) , the scanning line mask that can be separated most efficiently is determined in advance through statistical processing, and the mask is stored in mask memo 1.
It may also be registered and used in .

Further, in the above embodiment, an example was explained in which one unit of character/figure pattern is cut out from the original image pattern stored in one line in the image memory, but the original image pattern is stored in units of multiple lines,
The original image pattern for one line is cut out, or the original image pattern for one line is scanned using format information such as line position, and one unit of character/figure pattern is cut out in the same manner as described above. Good too.

Further, the mask scanning area is not limited to the area including the position II x ff1in and its neighboring area, but can be set as desired. For example, the area at the center of the touched character area may be set as the mask scanning area.

Furthermore, the cutout positions of the - unit character/figure pattern in the column and row directions are not limited to those of the above-mentioned embodiments, but can be changed as desired.

Furthermore, an image background that has been subjected to preprocessing such as background processing may be used as the original image pattern.

(Effects of the Invention) As is clear from the above description, the character segmentation device of the present invention projects the original image pattern for one line in the column direction and examines the histogram obtained. The position in the direction % is detected, and the length in the line direction of the character line area is determined from this position in the line direction.

Then, the character line area whose length in the line direction exceeds a predetermined threshold (
(Contact character area) 7 & Only when detected, scan the inside of the mask scanning area using a scanning line mask and detect the minimum scanning path that minimizes the cumulative number of black bits on that path for the scanning path of the scanning line mask. , the contact character area is divided based on the position of this minimum scanning path, and - unit character figure patterns are cut out.

Since one unit of character/figure pattern is cut out based on the position of the minimum scanning path, even if the letters I#l+I overlap or touch each other and are written on a form by hand, there will be no loss of - unit of the character/figure pattern. Cutting can be performed with high precision without chipping or with very little chipping.

In addition, only when a character line area (contact character area) whose length in the line direction exceeds a predetermined threshold is detected, the mask scanning area is scanned using a scanning line mask, which is better than conventional character extraction processing. The amount of scanning of the original image pattern can be reduced, and therefore characters can be cut out at high speed.

Therefore, if the present invention is applied to a character recognition device, a high-speed character recognition device can be realized.

[Brief explanation of the drawing]

FIG. 1 is a functional block diagram for explaining the configuration of an embodiment of the present invention. FIG. 2 is a diagram showing an example of an original image pattern. FIG. 3 is a diagram showing an example of a histogram of the cumulative number of black bits in the column direction. 4 is a diagram showing an example of the distribution state of the character line area, FIG. 5 is a diagram showing an example of setting the mask scanning area in the embodiment, and FIG. 6 is a diagram showing the position of the contact character area in the column direction in the embodiment. Figure 7 is a diagram showing an example of the flow of mask scanning processing in the embodiment; Figures 8 (A) to (F) are diagrams showing an example in which six types of scanning line masks are prepared; , FIGS. 9A to 9H are diagrams showing an example in which eight types of scanning line masks are prepared. DESCRIPTION OF SYMBOLS 10... Character cutting device, 12... Block detection means 14... Scanning area setting means + 6-... Mask scanning means, 26... Original image pattern 30... Mask scanning area. Fig. 2 Example of a histogram Fig. 3 ×1 HXsi Xt2 An example of Fig. 7 u An example of preparing 6 types of scanning line masks Fig. 8 30 An example of preparing 8fi type scanning line masks Fig. 9 0 An example of preparing 8 types of scanning line masks Fig. 9

Claims (2)

[Claims] (1) In a character cutting device that cuts out one unit of character/figure pattern from the original image pattern of character/figures, the cumulative black in the column direction obtained by projecting the original image pattern of one line in the column direction perpendicular to the row direction. Based on the bit number histogram, the above 1
block detection means for detecting a character line area of an original image pattern for a line; when the length of the character line area in the line direction exceeds a predetermined threshold, the character line area is regarded as a contact character area; scanning area setting means for setting the position of a mask scanning area; and scanning an original image pattern within the mask scanning area using a plurality of scanning line masks each having a different scanning path, mask scanning means for detecting a minimum scanning path in which the number of black bits on the scanning path is minimized; dividing the touching character area by dividing lines set based on the position of the minimum scanning path; 1. A character cutting device comprising: a character cutting means for cutting out each character figure pattern as one unit character figure pattern. (2) The scanning area setting means detects the minimum cumulative number of black bits from among the cumulative number of black bits in the column direction of the contact character area, and detects the position in the row direction where the minimum cumulative number of black bits is obtained; 2. The character cutting device according to claim 1, further comprising means for setting a mask scanning area that includes an area near the position.