JPH0821059B2 - Unnecessary data removal method for character contour generation - Google Patents

Description

DETAILED DESCRIPTION [Table of Contents] Outline Industrial field of application Conventional technology Problems to be solved by the invention Means for solving the problems Action Example Effect of the invention [Outline] Necessary for performing character recognition A character contour generation method that efficiently erases unnecessary image data in a character frame that surrounds a character when generating a character contour, and aims to improve the character recognition rate and the processing efficiency. From the circumscribed quadrilateral circumscribing the vector to be created, a first character frame with a predetermined size enlarged and a second character frame with a predetermined size further enlarged from the first character frame are created. A character contour generation method for generating a contour table as character contour information from image data in a frame, which is a contour table for creating a contour table from image data existing in a second character frame. Bull making method,
After selecting image data with continuous areas from the contour table and assigning labels with different values to the image data having independent areas, the maximum and minimum coordinate values of the area of the image data for each label value are calculated in the contour table. Labeling processing means for creating a label area table obtained from the image data, image data in which one pixel area is continuous with reference to the label area table, image data having an area of N × N pixels or less, and a first character frame And a cutout processing unit that selects image data having a large area existing outside the area occupied by the area and deletes the information of the selected image data from the contour table.

[Industrial applications]

The present invention relates to an unnecessary data removal method at the time of generating a character contour that efficiently removes unnecessary image data existing in a character frame surrounding the character when generating a character contour necessary for character recognition.

In order to recognize a character written in a drawing or the like where a character entry position is arbitrary, a character frame surrounding the character is provided as a preprocessing. This is because the area of one character is determined, the character contour is generated from the image data in this character area, and the character is recognized.

By the way, in this character frame, image data due to background stains such as drawings or noise generated by a reading device, or part of other characters may intrude as image data. If it exists in the character frame, when the character contour is generated from the image data in the character frame, a character contour different from the read character is generated, which may cause a character recognition error. It is necessary to delete the data efficiently to improve the character recognition rate and the processing efficiency.

[Conventional technology]

 FIG. 9 is a diagram illustrating a conventional technique.

As shown in FIG. 9 (a), the character frame is a vector 51 obtained by vectorizing the character image data 50.
A circumscribed quadrilateral 52 circumscribing to is possible. However, if only the image data in the character frame consisting of the circumscribed quadrangle 52 is taken out, as can be seen from FIG. 9 (a), a part of the character image data 50 outside the character frame is deleted. I will end up.

Therefore, a range in which the circumscribed quadrangle 52 of the vector 51 is expanded by an appropriate amount is set as a character recognition target area. For example, a virtual character frame as shown at 53 in FIG. 9 (b), which is expanded outward by a calculated by a = (character image line thickness + α) × 1/2, is generated. Here, α is experimentally determined. Then, a contour table showing the character contour is created from the image data within the range of the virtual character frame 53 determined in this way. At this time, as shown in the shaded portion of FIG. In the character frame 53, image data based on background stains on the drawing or noise generated by a reading device or the like, or part of image data of adjacent characters may enter.

When creating the contour table, for example, as shown in FIG. 9C, it is assumed that the virtual character frame 53 has coordinates 1 to 16 in the X direction and 1 to 13 in the Y direction. The image data in the inside are scanned in the Y direction, and the coordinate value of the point changed from the white pixel to the black pixel and the coordinate value of the point changed from the black pixel to the white pixel are obtained. A contour table as shown in is requested.

That is, for example, when the X coordinate is 1, there is no pixel change point when scanned in the Y direction, and when the X coordinate is 2, when there is a point where the white pixel of the image data changes to a black pixel when scanned in the Y direction. The Y coordinate value is 8, and the Y coordinate value of the point where the black pixel changes to the white pixel is 9. Similarly, when the X coordinate is 5, when scanning in the Y direction, the Y coordinate value of the point of changing the white pixel to the black pixel of the image data is 4, and the Y coordinate value of the point of changing the black pixel to the white pixel. Is 6, the Y coordinate value of the point of changing the white pixel to the black pixel of the next image data is 8, and the Y coordinate value of the point of changing the black pixel to the white pixel is 9.

Further, the image data in the virtual character frame 53 is scanned in the X direction, and the coordinate value of the point where the white pixel is changed to the black pixel and the coordinate value of the point where the black pixel is changed to the white pixel are obtained. The contour table as shown in FIG.

That is, for example, when the Y coordinate is 3, the X coordinate value of the point of changing the white pixel to the black pixel of the image data when scanning in the X direction is 14, and the X coordinate of the point of changing the black pixel to the white pixel. The value is 15.

Similarly, when the Y coordinate is 4, the X coordinate value of the point at which the white pixel changes to the black pixel of the image data when scanning in the X direction is 5, and the X coordinate value of the point at which the black pixel changes to the white pixel. Is 7, the X coordinate value of the point at which the white pixel changes to the black pixel again is 8, and the X coordinate value of the point at which the black pixel changes to the white pixel is 12.

Then, the X coordinate value of the point of changing the white pixel to the black pixel of the image data is 14, and the X coordinate value of the point of changing the black pixel to the white pixel is 16.

In this way, when the contour table is created in a state where image data due to noise and image data of other characters are mixed in addition to the image data of the recognition target character of FIG. 9C, when the contour table is created, FIG. Unnecessary coordinate values are mixed in for the recognition target character contour, such as the coordinate values shown in and of and the coordinate values shown in and of FIG. 9 (e). Therefore, if the character contour is generated and the character recognition is performed based on the contour table, the character recognition cannot be performed or the character recognition is erroneous.

For this reason, as a means for removing noise, a mask processing device is added to a reading device for reading a drawing or the like to remove image data based on noise.

 FIG. 10 is a diagram for explaining noise removal.

The mask processing device removes, for example, image data of 3 × 3 pixels or less, but does not remove image data of a size larger than this. That is, the size of the frame in FIG. 10 (a) is 3
Assuming that the number of pixels is × 3 pixels, the image data of a size that falls within this frame is erased as indicated by the diagonal lines, but the image data is outside this frame as indicated by the hatched portions in FIGS. 10 (b) and (c). If it appears, it is impossible to remove the slender image data such as one pixel in width and four pixels in length which cannot be actually written as shown in FIG. 10 (c).

[Problems to be Solved by the Invention]

As described above, conventionally, when slender image data based on noise or image data in which a plurality of pixels are gathered is mixed in a virtual character frame, and image data of another character is mixed, the coordinate value of unnecessary image data is changed. Since the characters are mixed in the contour table, when the character recognition is performed based on such a character contour table, there is a problem that the character recognition rate decreases.

In view of such a problem, the present invention aims to efficiently remove unnecessary coordinate values mixed in the character contour table, improve the character recognition rate, and improve the processing efficiency.

[Means for solving the problem]

 FIG. 1 is a diagram for explaining the principle of the present invention.

The contour table creating means 31 stores binary image data,
The coordinate value of the vector obtained from the character image data is input, and the contour table creating means 31 creates a circumscribed quadrilateral circumscribing the character vector based on the binary image data and the coordinates of the vector, and the circumscribed quadrilateral is predetermined from the circumscribed quadrilateral. Create a first character frame that is enlarged by the size of, and then create a second character frame that is enlarged by a predetermined size from this first character frame. From the image data in this second character frame, , A contour table as shown in FIG. 9 (d) or FIG. 9 (e) is created.

From the contour table created by the contour table creating means 31, the labeling processing means 32 selects image data in which regions are continuous, assigns different value labels to the image data having independent regions, and then labels each image data. A label area table is created by obtaining the maximum and minimum coordinate values of the area of the image data for each value from the contour table.

The cut-out processing means 33 refers to the label area table created by the labeling processing means 32, and the area of the image data for each label value is elongated image data in which one pixel area is continuous and N × N pixels. Image data having the following areas and image data having a large area existing outside the area occupied by the first character frame are selected,
Information on the selected image data is deleted from the contour table based on the label value given to the selected image data.

[Action]

With the above-described configuration, the contour table creating means 31 creates the first and second character frames each having a predetermined size enlarged from the circumscribed quadrangle circumscribing the character vector, and the second character frame is created. From the image data in the character frame, one type of contour table out of the two types of contour tables conventionally created is created.

Then, the labeling processing means 32 selects image data having independent areas from the contour table, assigns a label value to each, and then sets the maximum and minimum coordinate values of the image data area for each label value. And create a label area table.

From the label area table, the cut-out processing means 33 outputs the area occupied by each image data to the outside of the area occupied by the first character frame, that is, a continuous elongated one pixel, N × N pixels or less. Image data with large area is selected.

Then, the selected image data is determined as unnecessary image data and deleted from the contour table, so that only the image data necessary for generating the character contour can be left in the contour table.

Therefore, if character recognition is performed using this contour table,
Since there is no unnecessary image data that causes character recognition error, the character recognition rate is improved and the processing efficiency is improved because the processing is easy.

〔Example〕

FIG. 2 is a block diagram illustrating an embodiment of the present invention.
FIG. 3 is a flow chart for explaining the operation of FIG. 2, FIG. 4 is a diagram for explaining a virtual character frame, FIG. 5 is a diagram showing an example of a contour table, and FIG. 6 is an example of a label table. FIG. 7 is a diagram showing an example of the label area table, and FIG. 8 is a diagram showing an example of the label selection table.

FIG. 9 (a) via the terminal A to the contour table creating section 34.
The binary image data shown in the character image data 50 and the coordinate values of the vector shown in the vector 51 of FIG. 9 (a) of the character image data are input. The contour table creation unit 34 creates a vector circumscribed quadrilateral of the character as shown in FIG. 3 based on the binary image data and the coordinate values of the vector.

Then, as shown in FIG. 3, the size and orientation of the input character image data are adjusted. That is, the size and orientation of the character image are adjusted to erect up to a predetermined size, for example. Then, in the same manner as described with reference to FIG. 9B, the virtual character frame 53 is created as shown in FIG. Then, from the virtual character frame 53, for example, a virtual character frame 54 further expanded outward by 2a is created.

By creating the virtual character frame 54, the image data of the adjacent character shown by (c) in FIG. 9 largely invades the virtual character frame 54 as shown in FIG.

Then, the contour table creating unit 34 uses the virtual character frame 54.
A contour table as shown in FIG. 5 is created from the image data existing inside the contour table in the same manner as the contour table shown in FIG.
In this case, of course, a contour table as shown in FIG. 9 (d) may be created.

In the present embodiment, description will be made regarding the case where a contour table having Y on the horizontal axis is created as shown in FIG.

The labeling unit 35 creates a label table as shown in FIG. 6 from the contour table shown in FIG. 5 stored in the contour table memory 37 and stores it in the label table memory 38.

That is, as shown in FIG. 3, the initial value of the label value is set to 1, and then, in the contour table of FIG. 5, the coordinate values in the Y direction are sequentially incremented from the minimum value to +1 as shown in FIG.
Extract the coordinate values. In this example, X coordinate values 20 and 21 are extracted at the coordinate value 9 in the Y direction. At this time, the X coordinate values 20 and 21 are used as comparison values, and the label value 1 is given to the extracted coordinate value 9 in the Y direction as shown in FIG.

Next, as shown in FIG. 3, it is checked whether or not the X coordinate value of the Y coordinate value ± 1 has a coordinate value that overlaps with the comparison value. That is, when the coordinate value in the Y direction is 10 or 8, the X coordinate value is 20 to 21.
Check if there is any overlap. That is, it is checked whether there are the same numbers as the comparison values 20 to 21. In the case of FIG. 5, there is no overlap in the coordinate value 8 in the Y direction, but in the coordinate value 10 in the Y direction, there are 20 and 22 in the X coordinate values, and the comparison value 20
Since this is the coordinate value that overlaps with and 21, this coordinate value in the Y direction
The area of X coordinate value 20 to 22 in 10 is determined to be the image data of the area continuous with the image data of X coordinate value 20 to 21 in the Y direction coordinate value 9.

That is, 20 and 21 of the X coordinate value at the coordinate value 9 in the Y direction
And the black pixels whose X coordinate values are 20 and 21 in the Y-direction coordinate value 10 are adjacent to each other in the Y-direction coordinate values 9 and 10, and are therefore determined to be the same image data.

Subsequently, the labeling processing unit 35, in FIG.
Some of the X coordinate values of the Y coordinate value ± 1 had coordinate values that overlapped with the comparison value.
Examine whether there are two or more excluding the already extracted area.

That is, with respect to the image data area extracted in FIG. 3, it is checked whether or not the area of the adjacent image data is divided into two or more branched areas except the already extracted area.

If there is no branched region, move to Fig. 3,
If there is a branched region, the labeling unit 35 stores the Y coordinate value extracted in FIG. 3 and the comparison value in the internal stack, and as shown in FIG. 3, it is detected in FIG. It is checked whether the adjacent image data is on the -1 side on the Y coordinate.

On the -1 side, as shown in FIG. 3, 1 is subtracted from the Y coordinate value, and as shown in FIG. 3, the X coordinate value on the Y coordinate value has a coordinate value that overlaps the comparison value. Check if there is. That is, it is checked whether there is any overlap with 20 to 21, and if there is any overlap, the label value 1 is given, but if it is the +1 side, 1 is added to the Y coordinate value as shown in FIG. As shown in FIG. 3, an X coordinate value on the Y coordinate value having a coordinate value overlapping with the comparison value is extracted. That is, in the case of FIG. 5, the Y coordinate value 10 may overlap the comparison values 20 to 21 of the Y coordinate value 9. Therefore, the X coordinate values 20 to 22 of the Y coordinate value 10 are set as the comparison values and As shown in Fig. 6, coordinate values in the Y direction
At 10, a label value of 1 is given.

Subsequently, the labeling processor 35 returns to the routine of FIG. 3 and checks whether or not the X coordinate value of the Y coordinate value ± 1 has a coordinate value that overlaps the comparison value. That is, the coordinate value in the Y direction 11
Or, in the X coordinate value in 9, it is checked whether there is any overlap with the comparison value 20 to 22.

Then, if there is an overlap, it is checked whether there are two or more excluding the already extracted area. In the case of FIG. 5, Y coordinate value 9
In the above, since the X coordinate values 20 to 21 are already extracted areas, except for this, in the Y coordinate value 11, the X coordinate values 21 to 22
Therefore, the label value 1 is given to this area in FIG.

This operation is repeated, and for the area shown in FIG.
As shown in FIG. 6, if all label values are given, the third
In the figure, adjacent image data is no longer detected.

Here, the labeling unit 35, as shown in FIG.
Check if there is data in the stack. If there is data in the stack, restore the Y coordinate value and the comparison value, and return to the routine of Fig. 3. If there is no data in the stack, label all contour data. Check whether That is, in FIG.
Check if the labels have been labeled for each.

In this case, since the unlabeled contour data still remains in FIG. 4, the labeling processing unit 35 assigns the label value to the unlabeled contour data having the smallest Y coordinate value. +1 to the label value, then
That is, the label value is set to 2, and the process returns to the routine of FIG.

In FIG. 5, labels up to the Y coordinate value 9 are already attached, and at the Y coordinate value 10, X coordinate values 11 and 13 which are not labeled are extracted. Therefore, the labeling processor 35
Assigns a label value of 2 as shown in FIG. 6 using the X coordinate values 11 to 13 as comparison values, and executes the processing of FIG.

That is, in the Y coordinate value 11, since the X coordinate values 11 to 16 are coordinate values that overlap the comparison values 11 to 13 of the Y coordinate value 10, the labeling processing unit 35 gives the label value 2. Then, as detected by the routines of FIG.
Since the coordinate values 14 and 18 are coordinate values that overlap the comparison values 11 to 16, it is determined that the area is a branched area, and the X coordinate values 11 and 13 of the Y coordinate value 10 are already extracted and are excluded. Values 14 and 18
A label value of 2 is given to the label as shown in FIG.

Further, since the X coordinate values 17 and 19 of the Y coordinate value 11 are coordinate values that overlap the comparison values 14 to 18 of the Y coordinate value 10, the label value 2 is similarly given.

In the Y coordinate value 12, the X coordinate values 12 to 14 are coordinate values that overlap the comparison values 11 to 16 of the Y coordinate value 11, and in the Y coordinate value 13, the X coordinate values 13 to 15 are the comparison values of the Y coordinate value 12. The coordinate value overlaps 12 to 14, and the Y coordinate value is 14 and the X coordinate value is 14 to 17.
Are coordinate values that overlap the comparison values 13 to 15 of the Y coordinate value 13.

Furthermore, in the Y coordinate value 15, the X coordinate values 16 to 18 are coordinate values that overlap the comparison values 14 to 17 of the Y coordinate value 14, and the Y coordinate value 16
In, the X coordinate value 17-18 is the comparison value 16-18 of the Y coordinate value 15.
Is a coordinate value that overlaps
16 to 18 are coordinate values that overlap the comparison values 17 to 18 of the Y coordinate value 16.

In the Y coordinate value 18, the X coordinate values 13 to 17 are shown in FIG.
As detected by the routine shown in (1), the coordinate value overlaps with the comparison values 12 to 14 of the Y coordinate value 17, and also overlaps with the X coordinate values 16 to 18 of the Y coordinate value 17, and the branched region It is determined that there is.

In the Y coordinate value 17, the X coordinate values 16 to 18 are excluded because they are already extracted regions, and the X coordinate values 12 to 14 overlap the comparison values 13 to 17 of the Y coordinate value 18, and the Y coordinate value 17 is compared. Values 12-14 are Y
Since the coordinate value is a coordinate value that overlaps the X coordinate values 12 to 13 of the coordinate value 16, the labeling processing unit 35 assigns the label value 2 to this adjacent area as shown in FIG.

Similarly, the labeling unit 35 extracts the unlabeled X coordinate values 8 and 12 in the Y coordinate value 14 of FIG. 5, and sets the X coordinate values 8 to 12 as the comparison value to obtain the third value. The process shown in the figure is executed, but the comparison value 8 is added to the Y coordinate value 13 and the Y coordinate value 15.
As there are no coordinate values that overlap with ~ 12, as shown in Fig. 6, Y
At coordinate value 14, the unlabeled X coordinate values 8 and 12
A label value of 3 is assigned to.

After labeling all contour data, the labeling unit 35 stores the label table as shown in FIG. 6 in the label table memory 38, and then refers to this label table, as shown in FIG. Virtual character frame 54 shown in FIG.
At, the maximum and minimum coordinate values of the contour data having the same label value are obtained, and these are stored in the label area table memory 39 as a label area table.

That is, the label value of the contour data shown in FIG. 4 is 1, but the maximum value of the Y-direction coordinate value of this contour data, that is, the coordinate value of the point changing from the black pixel to the white pixel is the same as that shown in FIG. As shown in FIG. 7, the value is 18 and the minimum value of the X-direction coordinate value, that is, the coordinate value of the point at which the white pixel changes to the black pixel is 20. Further, the minimum value of the Y-direction coordinate value, that is, the coordinate value of the point that changes from the white pixel to the black pixel is 9, and the maximum value of the X-direction coordinate value, that is, the coordinate of the point that changes from the black pixel to the white pixel. value is
28.

Next, the maximum value of the Y-direction coordinate value of the contour data of label value 2, that is, the coordinate value of the point changing from the black pixel to the white pixel is 19, and the minimum value of the X-direction coordinate value, that is, from the white pixel. The coordinate value of the point that changes to a black pixel is 11. Further, the minimum value of the Y-direction coordinate value, that is, the coordinate value of the point that changes from the white pixel to the black pixel is 10, and the maximum value of the X-direction coordinate value, that is, the coordinate of the point that changes from the black pixel to the white pixel. The value is 19.

Further, the maximum value of the Y-direction coordinate value of the contour data of label 3, that is, the coordinate value of the point where the black pixel changes to the white pixel is 15
The minimum value of the X-direction coordinate value, that is, the coordinate value of the point where the white pixel changes to the black pixel is 8. Further, the minimum value of the Y-direction coordinate value, that is, the coordinate value of the point that changes from the white pixel to the black pixel is 14, and the maximum value of the X-direction coordinate value, that is, the coordinate of the point that changes from the black pixel to the white pixel. The value is 12.

The cutout processing unit 36 uses the label area table memory 39 to
The label area table as shown in FIG. 7 is read out and referred to, and as shown in FIG. 3, it is checked whether there is image data in which one pixel area is continuous, that is, image data having an elongated area having a width of one pixel.

When an area of contour data having a label value of 3 is created from the label area table shown in FIG. 7, an area having the size shown in FIG. 4 is created, and the image data has an elongated area having a width of 1 pixel. I understand.

Here, when there is image data having an elongated area having a width of one pixel, the clipping processing unit 36 creates a label selection table as shown in FIG. 8 as shown in FIG. C0H is added and registered as a code indicating that the image data is unnecessary image data based on noise.

Subsequently, the cutout processing unit 36, as shown in FIG.
It is checked whether there is image data having a region of N pixels or less, for example, a region of 3 × 3 pixels or less.

When the contour data area of label value 2 is created from the label area table shown in FIG. 7, the area of the size shown in FIG. 4 is created, and when the contour data area of label value 1 is created, FIG. An area of the size shown in is created.

Since the area of and is 3 × 3 pixels or more,
The cut-out processing unit 36 does not perform the processing shown in FIG. 3 and shifts to the processing shown in FIG. That is, it is checked whether there is intrusion image data from adjacent characters or the like.

This is compared with the virtual character frame 53 shown in FIG. 4, and it is checked whether or not there is larger contour data in the area outside the virtual character frame 53.

As described above, when the contour data area having the label value 2 is created, there is no area outside the virtual character frame 53.

However, when the contour data area having the label value 1 is created, the area is large outside the virtual character frame 53. This is because the X coordinate value of the virtual character frame 53 is 22 and the X coordinate value of the virtual character frame 54 is
Since it is 28, it can be judged from the ratio (22-20) / (28-22).

Therefore, as shown in FIG. 8, the cut-out processing unit 36 assigns, for example, the reference numeral 0 to the contour data of the label value 2, and to the contour data of the label value 1 as unnecessary image data based on the intrusion, for example, the reference numeral 80H. Is added and registered in the label selection table, and this label selection table is stored in the label selection table memory 30.

The cutout processing unit 36 then determines the label selection table memory 30.
By referring to the label selection table as shown in FIG. 8 stored in FIG. 8, it is recognized that the image data of the label value assigned the code of C0H and 80H is unnecessary, and the image data of the label value assigned the code of 0 is recognized. As shown in FIG. 3, a contour table in which the horizontal axis is in the Y direction is created by using only the above. That is, the contour table except for the contour table of FIG. 5 is created.

Then, a contour table in which the horizontal axis is in the X direction is created from the contour table in which the horizontal axis is in the Y direction. Then, as the cut-out character contour information, the two newly created contour tables are transmitted via the terminal B.

〔The invention's effect〕

As described above, according to the present invention, unnecessary image data in a character frame created for recognizing a character can be efficiently removed, so that the character recognition rate and the processing efficiency can be improved.

[Brief description of drawings]

1 is a diagram for explaining the principle of the present invention, FIG. 2 is a block diagram for explaining an embodiment of the present invention, FIG. 3 is a flow chart for explaining the operation of FIG. 2, and FIG. 4 is a virtual character frame. FIG. 5 is a diagram showing an example of a contour table, FIG. 6 is a diagram showing an example of a label table, FIG. 7 is a diagram showing an example of a label area table, and FIG. 8 is a label selection table. FIG. 9 is a diagram illustrating an example, FIG. 9 is a diagram illustrating a conventional technique, and FIG. 10 is a diagram illustrating noise removal. In the figure, 31 is a contour table creating means, 32 is a labeling processing means,
33 is a cutout processing means, 34 is a contour table creation unit, 35 is a labeling processing unit, 36 is a cutout processing unit, 37 is a contour table memory, 38 is a label table memory, 39 is a label area table memory, and 40 is a label selection table. Memory, 50 is character image data, 51 is vector, 52 is circumscribed quadrilateral, 5
3, 54 are virtual character frames.

Claims (1)

[Claims] 1. In order to recognize a character contour image, a first character frame in which a predetermined size is enlarged from a circumscribed quadrangle circumscribing a vector obtained from the character image, and a predetermined character further than the first character frame are provided. And a second character frame in which the size of the second character frame is enlarged, and a contour table as character contour information is generated from the image data in the second character frame. From the image data existing in the character frame, the contour table creating means (31) for creating the contour table, and the image data in which regions are continuous from the contour table created by the contour table creating means (31) are selected, and each is independent. After assigning different value labels to the image data having the specified area, the label area table is created by obtaining the maximum and minimum coordinate values of the area of the image data for each label value from the contour table. By referring to the labeling processing means (32) and the label area table created by the labeling processing means (32), image data in which one pixel area is continuous and image data having an area of N × N pixels or less And the image data having a large area existing outside the area occupied by the first character frame, and selected from the contour table based on the label value given to the selected image data. A unnecessary data removing method at the time of generating a character contour, which is provided with a cutting processing means (33) for erasing information of image data.