JPH1040312A - Device for preparing slip picture - Google Patents

Abstract

PROBLEM TO BE SOLVED: To automatically prepare an image of a slip not filled in yet from a filled slip. SOLUTION: An image of a plurality of filled slips in the same format are inputted (100), those input images are positioned (101), and the AND image is prepared (102). A rectangle surrounding black run linking components is extracted from each image (103), the rectangle of the input image is made to be correlated with the rectangle of the AND image, and an unnecessary image area in the AND image is specified by comparison between the mutual corresponding rectangles, and deleted (104). The AND image after the unnecessary image is deleted is outputted as an unfilled slip image (105).

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to image processing, and more particularly to a technique for automatically creating an image of an unfilled form from a completed form.

[0002]

2. Description of the Related Art There are many forms in an office, and sometimes it is necessary to create a form in the same format as a completed form. However, there was no means to automatically create an unfilled form from a completed form, so if those forms were not digitized, the food processor would have to create a form in the same format again. was there. If the format of the form is complicated, the amount of work is very large. In order to avoid such inconveniences, it is desired to realize a means for automatically creating an unfilled form from a completed form.

[0003] The following prior arts are known for image registration related to the present invention. (A) At a representative point on an image, pattern matching is performed by shifting the image little by little, and a position shift is detected by using a function of similarity as a weighting coefficient (Japanese Patent Application Laid-Open No. Sho 63-3).
388). (B) A method of calculating a shift amount at which the integrated projections in the X direction and the Y direction best match, and using the calculated shift amount as an image shift amount between input images (Japanese Patent Laid-Open No. 2-54495). (C) A method of detecting and correcting an image shift by extracting a mouse or a predetermined mark by shape recognition (Japanese Patent Laid-Open No. Hei 4-261259). (D) A method of searching for the best matching position while moving the template image in the window area and performing image positioning using the position as a true corresponding point (Japanese Patent Laid-Open No. 4-336677).

[0004]

SUMMARY OF THE INVENTION An object of the present invention is to provide an apparatus for automatically creating an image of an unfilled form from a completed form.

[0005]

According to the first to twelfth aspects of the present invention, there is provided a form image creating apparatus for creating an image of an unfilled form from a completed form.

According to the first aspect of the present invention, there is provided a form image creating apparatus for inputting a plurality of filled-in form images in the same format, aligning the entered plurality of filled-out form images, Means for creating an AND image of the plurality of completed form images after the combination. This AND
Most of the images are images common to a plurality of filled-out form images, that is, images of fixed information which are pre-printed on the form, and in some cases, may be used as is as an image of an unfilled form as it is. It is possible. But this A
The ND image often includes some unnecessary images that do not originally exist in the blank form. The form image creation device according to the invention of each of claims 2 to 7 is configured to convert an AND image
By deleting such unnecessary images, a more accurate blank form image is created.

In other words, the form image creating apparatus according to the second aspect of the present invention is capable of performing
There is further provided a means for specifying an unnecessary portion of the AND image to be deleted by comparing the images, and deleting the unnecessary portion from the AND image.

According to a third aspect of the present invention, there is provided a form image creating apparatus comprising: a plurality of filled form images after alignment;
Means for extracting a rectangle surrounding the black run connected component from the image;
A rectangle extracted from the filled-out form image after the alignment is associated with a rectangle extracted from the AND image, and an unnecessary portion of the AND image to be deleted is specified from a relationship between the corresponding rectangles. From the AND image.

A vote image creating apparatus according to the invention of claim 4 is
Means for extracting a rectangle surrounding the black run connected component from the plurality of filled-out form images and the AND image after the alignment, and a rectangle extracted from the filled-out form image after the alignment, and a positional relationship included therein. The rectangles extracted from the AND image are associated with each other.
Means for identifying an unnecessary portion of the D image to be deleted and deleting the unnecessary portion from the AND image.

According to a fifth aspect of the present invention, there is provided a form image creating apparatus.
Means for extracting a rectangle surrounding the black run connected component from the plurality of filled-out form images and the AND image after the alignment, and a rectangle extracted from the filled-out form image after the alignment and the AND
Means for associating with the rectangles extracted from the D image, identifying the unnecessary portion of the AND image to be deleted by comparing the sizes of the corresponding rectangles, and deleting the unnecessary portion from the AND image. Have more.

According to a sixth aspect of the present invention, there is provided a form image creating apparatus, comprising: a plurality of filled form images after alignment;
Means for extracting a rectangle surrounding the black run connected component from the image;
The rectangle extracted from the filled-in form image after the alignment is associated with the rectangle extracted from the AND image, and the number of black pixels of each rectangle extracted from the filled-in form image after the alignment and the corresponding number of black pixels Means for identifying an unnecessary portion of the AND image to be deleted by comparing the number of rectangular black pixels extracted from the AND image and deleting the unnecessary portion from the AND image.

According to a seventh aspect of the present invention, there is provided a form image creating apparatus, comprising: a plurality of filled form images after alignment;
Means for extracting a rectangle surrounding the black run connected component from the image;
The rectangle extracted from the filled-in form image after the alignment is associated with the rectangle extracted from the AND image, and extracted from the AND image corresponding to each rectangle extracted from the filled-in form image after the alignment. Depending on the number of rectangles
Means for identifying an unnecessary portion of the AND image to be deleted and deleting the unnecessary portion from the AND image.

In general, characters and the like (images that must not be deleted) to be printed on a form in advance are arranged at positions near the frame lines. Paying attention to this, the form image apparatus according to the invention of claim 11 further has means for extracting a line segment corresponding to a ruled line or a frame line on the filled-in form image after the alignment, and deletes unnecessary portions. The means for adjusting adjusts a threshold value for determining whether to delete a portion related to the rectangle extracted from the filled-out form image after the alignment, from the extracted line segment, in accordance with the distance.

In addition, a large rectangle is often extracted from an image that should not exist in a blank form. Focusing on this, in the form image creating apparatus according to the twelfth aspect, the means for deleting the unnecessary part is related to the rectangle according to the size of the rectangle extracted from the filled-in form image after the alignment. Adjust the threshold for determining whether to delete the part.

According to the form image creating apparatus according to the second to seventh aspects of the present invention, an AND image from which unnecessary images have been deleted can be obtained, which has a sufficient quality as an image of an unfilled form. . However, if there is a "blurred" or "missing" of a character or the like in the filled-out form image serving as the base of the AND image, it also appears in the AND image.

There are no such "faints" or "chips"
In order to obtain a higher quality blank form image, the form image creating apparatus according to the invention of claim 8 uses the AN after deleting unnecessary portions.
The image processing apparatus further includes means for performing image correction on the D image with reference to the filled-out form image after the alignment. According to the ninth aspect of the present invention, the image correcting means replaces the image of the part which has not been deleted in the AND image with the OR image of the corresponding part in the plurality of filled-out form images after the alignment. Perform image correction. According to the tenth aspect,
The image correction means performs the image correction by replacing the image of the portion that has not been deleted in the AND image with an image obtained by averaging the images of the corresponding portions of the plurality of filled-out form images after the alignment.

[0017]

Embodiments of the present invention will be described below with reference to the drawings.

[Embodiment 1] The apparatus configuration of this embodiment can be represented as shown in FIG. FIG. 2 shows a processing flow of the present embodiment. 3 and 4 are diagrams for explaining the processing contents. Hereinafter, the present embodiment will be described with reference to these drawings as appropriate. In the processing flow of FIG. 2, the AND image buffer 107 is shown separately for easy understanding, but these are the same. This is the same in other embodiments described later.

The image input unit 100 shown in FIG. 1 uses a plurality of image input devices, such as a scanner and a facsimile, to input a plurality of filled-out forms in the same format and stores them in an input image buffer (step 110 in FIG. 2). . In the image registration unit 101, for example, by using a method such as the above-described conventional technology, the input images are aligned with respect to one image in the input image as a reference, and input images other than the reference input image are determined. , And converts the converted image and the reference input image into the input image buffer 106 (step 1 in FIG. 2).
11). Hereinafter, unless otherwise specified, the input image is:
This is the input image after the alignment processing.

The AND image creating unit 102 performs an AND operation on a plurality of input images in the input image buffer 106 for each pixel, and obtains an obtained result image (AN
D image) is stored in the AND image buffer 107 (FIG. 2).
Step 112). Specifically, each pixel on the AND image is a black pixel if the corresponding pixels on all input images are black pixels, and a white pixel if the corresponding pixels on at least one input image are white pixels. . Next, in the rectangle extracting unit 103, for all input images and their AND images,
The connected components of the black run are extracted, and the coordinates of the rectangle surrounding the connected components (for example, the coordinates of two opposing corners) are stored in the rectangular coordinate buffer 108 (step 113 in FIG. 2).

FIG. 3 schematically shows an example of the above-mentioned creation of an AND image and extraction of a rectangle. In FIG. 3, 1
Reference numerals 30 and 131 denote input images of the completed form (after position adjustment). Here, the number of input images is two, but this is merely an example, and it goes without saying that three or more input images may be used. 1
An AND image 32 is obtained by ANDing the input images 130 and 131. 140, 141 and 14
2 are input images 130 and 131 and AND image 1 respectively
32, a rectangular region extracted therefrom is superimposed as a rectangular frame.

As can be understood from the example of the AND image 132, most of the entry information other than the fixed image such as the character or the like previously printed on the form has been removed from the AND image at this stage. This AND image can be used as an image of a blank form. However, it is often unavoidable that a certain amount of unnecessary image remains at the stage of AND composition, and it is practical to remove the unnecessary image.

In FIG. 1, the unnecessary image deleting unit 103
A process of referring to the rectangular coordinates stored in the rectangular coordinate buffer 108 and deleting an image of a rectangular area determined to be unnecessary on the AND image in the AND image buffer 107 (making all pixels in the rectangular area white pixels) I do. Step 1 of FIG.
Processing loops 14 to 121 are parts for performing the unnecessary image deletion processing.

That is, first, the input image buffer 106
One input image is selected from the plurality of input images (after the alignment) stored in the input image (step 114), and one rectangle extracted from the input image is selected as a rectangle A (step 115), and the variable S is set. Clear to zero (Step 11
6). Then, one rectangle extracted from the AND image is selected as a rectangle B (step 117). Rectangle A and rectangle B
Are compared (stored in the rectangle coordinate buffer 108), and it is checked whether the rectangle A includes the rectangle B (step 118). If not included, the process returns to step 117, and another rectangle of the AND image is selected again as rectangle B. If the rectangle A includes the rectangle B, the area of the rectangle B is calculated, the area is added to the value of the variable S, and the result is set to the value of the variable S (step 119), and the process returns to step 117 and returns to step 117. Select rectangle B.

Thus, a rectangle A on the input image
Are found and the total area is obtained as a variable S, the absolute value of the difference between the S value and the area of the rectangle A is calculated, and if it is not smaller than a certain threshold Th, It is determined that the area on the AND image corresponding to the rectangle A should be deleted (step 120), the image in that area is deleted (step 121), and the process returns to step 117 to select another rectangle A, but deletes it. If it is determined that it should not be performed, the process proceeds to step 117 without proceeding to step 121. While a rectangle to be selected as rectangle A remains, the loop from step 115 to step 121 is repeated.

In this way, the unnecessary image of the AND image is deleted by paying attention to a certain input image. When all the rectangles extracted from the input image of interest are selected as the rectangle A and processed, the process exits the processing loop, returns to step 114, selects the next input image, and returns to step 115.
From step to step 121 are repeated. When the processing relating to the last input image is completed, an image in which unnecessary images have been deleted from the AND image (the intended blank image of the form) is obtained in the AND image buffer 107. This image is output to the outside by the form image output unit 105 (step 122).

FIG. 4 is an explanatory diagram of determination and deletion of an unnecessary area of an AND image related to a certain input image. In FIG. 4, reference numeral 150 denotes a certain input image in which a rectangular area is superimposed as a rectangular frame, and 151 denotes an AND image in which the rectangular area is superimposed as a rectangular frame. The rectangle 152 (rectangle A) of the input image includes the rectangles 153, 154, 156 (rectangle B) of the AND image, and the area b1 of the rectangle 152 and the total area b2 of the rectangles 153, 154, 155 (the value of the variable S) Since the absolute value of the difference is not smaller than the threshold Th, the image in the area of the AND image corresponding to the rectangle 152 is determined to be unnecessary (determined to be the entry information of the form), and is deleted. On the other hand, since the rectangle of the AND image included in the rectangle 160 of the input image is only the rectangle 161 and the absolute value of the difference between the areas a1 and a2 is smaller than the threshold Th, the image of the area of the AND image corresponding to the rectangle 160 Is not determined to be unnecessary and is therefore not deleted. By performing the same processing between two or more input images and the AND image, an AND image from which entry information has been deleted, that is, an image 163 of a blank form can be obtained.

In general, as the number of input images used for the unnecessary image deletion processing increases, the possibility of missing entry information from the AND image decreases, and the accuracy of the obtained blank form image improves. However, if the number exceeds a certain number, the accuracy does not improve so much in spite of the increase in the processing amount. Therefore, an appropriate number may be appropriately determined in consideration of the expected processing accuracy and the processing amount.

[Embodiment 2] The apparatus configuration of this embodiment can be represented as shown in FIG. FIG. 5 shows a processing flow of the present embodiment.

Image input unit 100, image alignment unit 1
01, the processing by the AND image creation unit 102, the rectangle extraction unit 103, and the form image output unit 105 are exactly the same as in the first embodiment. The processing content of the unnecessary image deletion unit 104 is partially different from that of the first embodiment.

Referring to the processing flow of FIG. 5, input of a plurality of filled-out form images, creation of AND images,
Steps 210 to 213 which are processing parts up to rectangle extraction
And the form image output step 222 are the same as steps 110 to 113 and step 122 of the above embodiment.

The loop of steps 214 to 221 which is a processing part for deleting unnecessary images of an AND image is basically the same as the loop of steps 114 to 121 of the first embodiment. Step 21 for
Only the contents of 9, 220 differ from the first embodiment. That is, in the present embodiment, the sum of the number of black pixels in the rectangle B of the AND image included in the rectangle A of the input image is determined as a variable S (step 219). The absolute value of the difference from the total number of black pixels in the included rectangle B is compared with a threshold Th, and if the absolute value of the difference is smaller than Th, it is determined that the rectangle A area should not be deleted, and Th
If so, it is determined that it should be deleted (step 220).

FIG. 6 is an explanatory diagram of such an unnecessary image deletion determination and deletion process. In FIG. 6, reference numeral 250 denotes an input image in which a rectangular area is superimposed as a rectangular frame, and reference numeral 251 denotes an AND image in which the rectangular area is superimposed as a rectangular frame. The rectangles 253, 2 of the AND image are included in the rectangle 252 (rectangle A) of the input image.
54, 256 (rectangle B), and the absolute value of the difference between the number b1 of black pixels in the rectangle 252 and the total number b2 of black pixels in the rectangles 253, 254, 255 (value of the variable S) is smaller than the threshold Th. Since there is no image, the image in the area of the AND image corresponding to the rectangle 252 is determined to be unnecessary (it is determined to be the entry information of the form), and is deleted. On the other hand, the rectangle of the AND image included in the rectangle 260 of the input image is only the rectangle 261, and the absolute value of the difference between the numbers of black pixels a1 and a2 inside each is smaller than the threshold Th, so that the AND image corresponding to the rectangle 260 The image in the area is not determined to be unnecessary and is not deleted. By performing the same processing between two or more input images and the AND image, an AND image from which entry information has been deleted, that is, an image 263 of a blank form can be obtained.

As described above, the present embodiment focuses on the number of black pixels in a rectangle, so that the processing amount is increased as compared with the first embodiment using a rectangular area, but an improvement in the detection accuracy of unnecessary images can be expected.

[Embodiment 3] The apparatus configuration of this embodiment can be represented as shown in FIG. FIG. 7 shows a processing flow of the present embodiment.

Image input unit 100, image alignment unit 1
01, the processing by the AND image creation unit 102, the rectangle extraction unit 103, and the form image output unit 105 are exactly the same as in the first embodiment. The processing content of the unnecessary image deletion unit 104 is partially different from that of the first embodiment.

Referring to the processing flow of FIG. 7, input of a plurality of filled-out form images, creation of an AND image,
Steps 310 to 313 which are processing parts up to rectangle extraction
And the form image output step 322 are the same as steps 110 to 113 and step 122 of the above embodiment.

The loop of steps 314 to 321 which is a processing part for deleting unnecessary images of an AND image is basically the same as the loop of steps 114 to 121 of the first embodiment, Step 31 for
Only the contents of 9,320 are different from those of the first embodiment. That is, in the present embodiment, the variable S is incremented by 1 each time a rectangle B of the AND image included in the rectangle A of the input image is found (step 319), and the number of rectangles of the AND image included in the rectangle A (corresponding to (The number of rectangles) is obtained as a variable S value. Then, the obtained number of corresponding rectangles is compared with a threshold Th, and if the number of corresponding rectangles is smaller than Th, it is determined that the area of the rectangle A should not be deleted, and if it is not less than Th, it is determined that the area should be deleted (step 320). ). If it is determined that the image should be deleted, the image in the area of the AND image corresponding to the rectangle A is deleted (step 321).

FIG. 8 is an explanatory diagram of such an unnecessary image judging and deleting process. In FIG. 8, reference numeral 350 denotes a certain input image in which a rectangular region is superimposed as a rectangular frame, and reference numeral 351 denotes an AND image in which the rectangular region is superimposed as a rectangular frame. Rectangle 3 of input image
52 (rectangle A) includes three rectangles 353, 3 of the AND image.
54, 356 (rectangle B) are included, and the number of corresponding rectangles is three, which is equal to or greater than the threshold value Th. Therefore, it is determined that an image in the area of the AND image corresponding to the rectangle 352 is unnecessary (in the form entry information). Judge that there is) and delete it. On the other hand, the rectangle of the AND image included in the rectangle 360 of the input image is the rectangle 3
Since only 61 and the corresponding rectangle number are smaller than Th, the rectangle 36
The image in the area of the AND image corresponding to 0 is not determined to be unnecessary and is not deleted. By performing the same processing between two or more input images and the AND image, an AND image from which entry information has been deleted, that is, an image 363 of a blank form is obtained.

It is to be noted that the unnecessary image determination method of the present embodiment and the unnecessary image determination method of the first embodiment are selectively used depending on the quality of the image to be processed, or the determination conditions of both methods are applied simultaneously. This makes it possible to delete unnecessary images with higher accuracy.

[Embodiment 4] The apparatus configuration of this embodiment can be represented as shown in FIG. In FIG. 9, the image input unit 4
00, image alignment unit 401, AND image creation unit 4
02, the rectangle extracting unit 403, the unnecessary image deleting unit 404, the input image buffer 406, the AND image buffer 407, and the rectangular coordinate buffer 408 are the image input unit 100 of FIG. Positioning unit 101, AND image creation unit 102, rectangle extraction unit 10
3, unnecessary image deletion unit 104, input image buffer 106,
AND image buffer 107 and rectangular coordinate buffer 108
And the function is the same. The method of determining an unnecessary image in the unnecessary image deletion unit 104 may be the same method as in the first, second, or third embodiment, or may be a combination of these methods.

The difference of this embodiment from the above embodiments in the apparatus configuration is that image shaping is performed on the AND image (in the AND image buffer 107) after the unnecessary image has been deleted by the unnecessary image deletion unit 404. And an OR image / multi-valued image buffer 410 related to the image shaping, and the form image output unit 40
5 is an OR image / multi-value image buffer 4 for the final form image
10 and output to the outside.

FIG. 10 shows a processing flow of this embodiment. In FIG. 10, an image of a completed form is input (step 420), and an AND image from which unnecessary images have been deleted is created (step 421). The processing up to this point is the same as the processing up to the point before the image output in the first, second or third embodiment. Subsequent steps 422 to 424 are part of the form image shaping process.

First, as a preparation work, at step 422, O
Initialize the R image / multi-valued image buffer 410. Then, in step 423, the input image buffer 406
, One input image A is selected from a plurality of input images (aligned images), and in step 424, the image in the OR image / multi-valued image buffer 410 and the input image A are ORed.
The combined image is again converted into an OR image / multi-valued image buffer 410.
To save. Similar OR synthesis is performed for all input images. When this is completed, in step 425, A
An OR image of an area corresponding to a rectangle remaining without deletion of the ND image is extracted from the OR image / multi-value image buffer 410 and output to the outside. That is, substantially, AND
An OR image of a plurality of input images is created for an area corresponding to a rectangle remaining without being deleted from the image.

Therefore, in step 424, A
OR synthesis of the input image is performed only on the rectangular area remaining without deleting the ND image, and a form image in which all pixels outside the rectangular area are white pixels is directly created in the OR image / multi-valued image buffer 410. In this case, in this case, it is sufficient to simply output the form image created in the OR image / multi-valued image buffer 410 without being aware of the area in step 425.

FIG. 11 is an explanatory diagram of the form image shaping described above. In FIG. 11, reference numeral 430 denotes an AND image, in which the rectangular area remaining without being deleted (not determined as an unnecessary image part) is overlapped as a rectangular frame. 431 and 432 are input images (1) and (2), respectively.
2 shows the rectangular area superimposed as a rectangular frame. The rectangle 435 of the AND image is the image area of the character “sentence”, but the image is disturbed (blurred or missing). Such “fading” or the like may occur not only when “fading” or the like exists in the character itself on the form, but also due to an error in reading the form, an error in positioning the input image, or the like. . 436 and 437 are rectangles 435
Are the images of the areas of the input images (1) and (2) corresponding to. In these images 436 and 437, “blurring” or “missing” is seen. In the OR image 438, such “blurring” or the like is restored. And this OR
Since the image 438 is employed as an output image instead of the image of the rectangle 435, it is possible to output the blank form image 439 in which “blurring” or the like has been restored.

[Embodiment 5] The apparatus configuration of this embodiment can be represented as shown in FIG. FIG.
Shows the processing flow of this embodiment.

This embodiment is the same as Embodiment 4 except for the processing contents of the form image shaping unit 409. 12, the step 520 of inputting an image and the step 521 of creating an AND image with unnecessary image deleted are the same as steps 420 and 421 of the fourth embodiment, respectively. The contents of steps 522 to 524 for shaping the form image are different.

That is, as a preparatory operation, the OR image / multi-value image buffer 410 is initialized in step 522. Then, in step 523, one input image A is selected from a plurality of input images (aligned images) in the input image buffer 406, and in step 524,
The pixel value in the OR image / multi-value image buffer 410 corresponding to the black pixel of the input image A is incremented by one. The same operation is performed for all input images, and a multi-valued image in which the number of black pixels at each pixel position of a plurality of input images is set to a pixel value is ORed.
It is created in the image / multi-value image buffer 410. When this is completed, in step 525, a region of the multi-valued image corresponding to the rectangle remaining without being deleted from the AND image (not determined to be an unnecessary image portion) is extracted, and the pixel value of the region is set to a threshold value. The image is binarized by Th (typically, if the pixel value is a majority of the number of input images, the pixel is a black pixel; otherwise, the pixel is a white pixel) and output as an output image. Naturally, the pixels other than the rectangular area are white pixels.

That is, in the region corresponding to the rectangle remaining without being deleted from the AND image, the pixel position that is a black pixel in the input image of Th number or more is a black pixel,
The other pixels are white pixels, and the other regions create output images in which all pixels are white pixels. Therefore, in step 524, the white / black determination of the pixel and the increment of the pixel value may be performed only for the rectangular area remaining without being deleted from the AND image. In this case, step 52 is performed.
In 5, the multivalued image created in the OR image / multivalued image buffer 410 may be simply binarized and output as a form image without regard to the area. As is apparent from the above description, the image shaping method according to the present embodiment uses AND
It can also be said that a method of replacing an image of a rectangular region that is not deleted in the image with an image obtained by averaging the images of the regions corresponding to the plurality of input images.

FIG. 13 is an explanatory diagram of the above-described form image shaping. In FIG. 13, reference numeral 530 denotes a rectangular area that remains on the AND image without being deleted (not determined as an unnecessary image part) as a rectangular frame. Reference numerals 531, 532, and 533 denote input images (1), (2), and (3), respectively, in which rectangular regions are superimposed as rectangular frames. The rectangle 535 of the AND image is an image area of the character “sentence”.
Chipping) is recognized. Reference numerals 536, 537, and 538 are images of the areas of the input images (1), (2), and (3) corresponding to the rectangle 535. The number of black pixels is counted for each pixel position of these images 536, 537, and 538 to create a multivalued image and binarize it (specifically, black if the pixel value is 2 or more, white if less than 2). Output image 539
Create This output image 539 is composed of images 536 and 53
Since the image has the characteristics of 7,539 averaged images, "faint", "missing", "dirt", and the like seen in those images are restored. Then, the same restoration is performed for the other rectangular areas.
0 is obtained.

In the present embodiment, since the number of black pixels is counted and the binarization process is performed, the processing amount is increased as compared with the fourth embodiment, but a blank form image can be restored with higher accuracy.

[Embodiment 6] FIG.
It can be shown as follows. FIG. 15 shows a processing flow of this embodiment.

In FIG. 14, an image input unit 600, an image alignment unit 601, an AND image creation unit 602, a rectangle extraction unit 603, an unnecessary image deletion unit 604, a form image output unit 605, an input image buffer 606, and an AND image buffer 607 and a rectangular coordinate buffer 608 are parts corresponding to the parts of the same name shown in FIG. 1, and their functions are the same except for the unnecessary image deletion unit 604. Line segment extraction unit 60
Reference numeral 9 denotes a newly added portion, which detects a long black run from the input image and extracts a black run longer than a certain threshold as a line segment. The line segment extracted here is, specifically, a ruled line or a frame line on a form. The unnecessary image deletion unit 604 basically has the same function as the unnecessary image deletion unit 104 in the first embodiment, but sets a threshold value for determining whether to delete a rectangular area by using a line segment extraction unit 609. This embodiment is different from the first embodiment in that the distance is changed according to the distance between the line segment extracted from the rectangle and the rectangle. This difference will be further described in the processing flow of FIG.

When the processing flow of FIG. 15 is compared with the processing flow of FIG. 2 relating to the first embodiment, it is apparent that the processing contents are basically the same. That is, it is obvious that steps 620 to 623 correspond to steps 110 to 113 in FIG. 2 and that the processing loop of steps 624 to 632 corresponds to the processing loop of steps 114 to 121 in FIG. However, unlike the processing flow of FIG. 2, in the processing loop of steps 624 to 632, the step 625 of line segment detection is added, and not only the variable S is initialized in step 627 but also the variable e The “distance from the line segment” is substituted. In step 631, the fixed value Th is set as a variable as a threshold value to be compared with the absolute value of the difference between the total area S of the rectangle B included in the rectangle A and the area of the rectangle A. The value divided by e (distance from the line segment) is used. This difference will be specifically described with reference to FIG.

In FIG. 16, reference numeral 650 denotes an input image in which a rectangular area is superimposed as a rectangular frame, 653 denotes a frame on the input form, and the frame line is a line segment extraction unit 609. Is extracted from the input image. 6
Reference numeral 51 denotes an AND image in which the rectangular area is superimposed as a rectangular frame.

When the rectangle 654 of the input image is selected as the rectangle A, the rectangle A is the rectangle 656, 6 of the AND image.
57,658 (rectangle B), and the total area is b2
(Value of the variable S), assuming that the area of the rectangle 654 is b1, in the first embodiment, if | b1-b2 | is equal to or larger than the fixed threshold, the area of the rectangle 654 is determined to be an unnecessary image. But,
In this embodiment, the distance from the line segment of the rectangle A is detected (this is set as a variable e in step 627). The distance detected here is the shortest distance in the row direction (wider one of the vertical and horizontal directions). That is, in the case of the rectangle 654, the distance e1 in the row direction of the frame line 655 is detected. Then, a value obtained by dividing the fixed value Th by the distance e1 is used as a determination threshold. That is, | b1−b2 | ≧ Th
If / e1, the rectangle 654 is determined to be an unnecessary image.

When the rectangle 660 of the input image is selected as the rectangle A, the rectangle of the AND image included in the rectangle is only the rectangle 662, and their areas are a1 and a2. Is d1. If | a1−a2 | <Th / d1, the area of the rectangle 660 is determined to be an unnecessary image.

The same processing is performed by ANDing two or more input images.
By deleting the image in the area determined to be an unnecessary image of the AND image, the AND image from which the entry information is deleted, that is, the blank image 665 of the form is obtained. Output by the unit 605 (step 633).

In general, the characters and the like (images that should not be deleted) of the items printed in advance on the blank form are often arranged near the frame lines. The letters “1”, “2”, and “3” in the form handled in FIG. 16 are such examples (probably item numbers). On the other hand, the entry frame of information to be arbitrarily entered often has a certain margin in the line direction, and the distance between the entry character and the frame line in the line direction tends to be larger than the character of the item to be entered in advance. There is. Focusing on this point, in the present embodiment, the accuracy of the unnecessary image deletion is improved by changing the threshold value of the unnecessary image determination so as to be in inverse proportion to the distance between the character and the entry frame. In the present embodiment, the threshold value is simply set to be inversely proportional to the distance. However, the threshold value may be changed according to the distance by using another appropriate distance function.

[Embodiment 7] The device configuration of this embodiment can be represented as shown in FIG. Further, the processing flow can be represented in substantially the same manner as the processing flow of the sixth embodiment shown in FIG. 15, and only the differences between the processing contents of the sixth embodiment and the present embodiment will be described with reference to FIG. explain.

In the present embodiment, in the loop of steps 628 to 630, the rectangle B of the AND image included in the rectangle A of the input image is found in the same manner as in the sixth embodiment. As in the second embodiment, the total number of black pixels in the rectangle B is obtained as a variable S value. Then, in step 631, the absolute value of the difference between the S value and the number of black pixels in the rectangle A and the fixed value Th are set to the distance e from the frame line of the rectangle A.
It is determined whether or not the area of the rectangle A is an unnecessary image by comparing with the value Th / e divided by the following equation. That is, this embodiment employs the same unnecessary image determination method as that of the second embodiment,
The determination threshold is controlled in the same manner as in the sixth embodiment.

[Embodiment 8] The apparatus configuration of this embodiment can be represented as shown in FIG. Further, the processing flow can be represented in substantially the same manner as the processing flow of the sixth embodiment shown in FIG. 15, and only the differences between the processing contents of the sixth embodiment and the present embodiment will be described with reference to FIG. explain.

In the present embodiment, in the loop of steps 628 to 630, the rectangle B of the AND image included in the rectangle A of the input image is found in the same manner as in the sixth embodiment. As in the third embodiment, the number of rectangles B is obtained as a variable S value. Then, step 631
, The value Th / e obtained by dividing the fixed value Th by the distance e from the frame line of the rectangle A is used as a determination threshold, and a comparison with the S value is performed to determine whether the area of the rectangle A is an unnecessary image. Determine whether or not. That is, the present embodiment employs the same unnecessary image determination method as in the third embodiment, and controls the determination threshold in the same manner as in the sixth embodiment.

In the fourth embodiment or the fifth embodiment, it is obvious that the determination threshold value may be controlled in the same manner as in the present embodiment.

[Embodiment 9] Images other than characters on an unfilled form are often extracted as large rectangles. Focusing on this, the present embodiment changes the threshold value for the unnecessary image determination according to the size of the rectangle. Specifically, the threshold is adjusted so that a large rectangle such as a rectangle surrounding the entire form is not easily deleted.

The device configuration of this embodiment can be represented as shown in FIG. 1 as in the first embodiment. However, the content of the unnecessary image determination by the unnecessary image deletion unit 104 is slightly different from the case of the first embodiment. This difference will be described with reference to the processing flow of the present embodiment shown in FIG.

When the processing flow of FIG. 17 is compared with the processing flow of FIG. 2 relating to the first embodiment, it is clear that the processing contents are basically the same. That is, it is apparent that steps 920 to 923 correspond to steps 110 to 113 in FIG. 2, and the processing loop of steps 924 to 931 corresponds to the processing loop of steps 114 to 121 in FIG. However, unlike the processing flow of FIG. 2, in the processing loop of steps 924 to 931, step 62
7, not only the variable S is initialized, but also the “area of the rectangle A” is substituted for the variable q, and in step 930, the difference between the total area S of the rectangle B included in the rectangle A and the area of the rectangle A is calculated. As the threshold value to be compared with the absolute value, a value obtained by dividing the fixed value Th by a variable q (the area of the rectangle A) is used. This difference will be specifically described with reference to FIG.

In FIG. 18, reference numeral 950 denotes an input image in which a rectangular area is superimposed as a rectangular frame, and 951 denotes an AND image in which the rectangular area is superimposed as a rectangular frame. When a rectangle 954 of a number printed in advance on an unfilled form is selected as the rectangle A, this rectangle 954 includes only the rectangle 955 (rectangle B) of the AND image, and their areas are a1 and a2. | A1
If −a2 | <Th / a1, it is determined that the area of the rectangle 954 is not an unnecessary image. On the other hand, when the relatively large rectangle 956 of the arbitrarily entered character is selected as the rectangle A, the threshold value for the unnecessary image determination is a fixed value Th of the rectangle 9
The value divided by the area b1 of 56 is used, which is b1> a
Since it is 1, the value is smaller than the determination threshold value used for the rectangle 954. Then, when the rectangle 957 includes rectangles 957, 958, and 959 of the AND image, and their total area is b2, if | b1−b2 | ≧ Th / b1,
The area of the rectangle 956 is determined to be an unnecessary image and is deleted.

The same processing is performed by ANDing two or more input images.
By deleting the image of the area determined to be an unnecessary image of the AND image, the AND image from which the entry information is deleted, that is, the blank image 960 of the form is obtained and output to the outside. (Step 932).

In the present embodiment, the threshold value for the unnecessary image determination is simply inversely proportional to the area of the rectangle A. However, if the size of the image to be deleted is known to some extent, another suitable function ( If a function that minimizes the determination threshold when the area of the rectangle A corresponds to the size of the deletion target is used, unnecessary image determination with higher accuracy can be performed.

[Embodiment 10] The apparatus configuration of this embodiment can be represented as shown in FIG. Further, since the processing flow can be represented in substantially the same manner as the processing flow of the ninth embodiment shown in FIG. 17, only the differences between the processing contents of the ninth embodiment and the present embodiment will be described with reference to FIG. explain.

In the present embodiment, in the loop of steps 927 to 929, the rectangle B of the AND image included in the rectangle A of the input image is found as in the ninth embodiment. As in the second embodiment, the total number of black pixels in the rectangle B is obtained as a variable S value. Then, in step 930, the absolute value of the difference between the S value and the number of black pixels in the rectangle A is compared with the value Th / q obtained by dividing the fixed value Th by the area q of the rectangle A, and the rectangle A is determined. It is determined whether the area is an unnecessary image. That is, the present embodiment employs the same unnecessary image determination method as that of the second embodiment, and controls the determination threshold value in the same manner as that of the ninth embodiment.

[Embodiment 11] The apparatus configuration of this embodiment can be represented as shown in FIG. Further, since the processing flow can be represented in substantially the same manner as the processing flow of the ninth embodiment shown in FIG. 17, only the differences between the processing contents of the ninth embodiment and the present embodiment will be described with reference to FIG. explain.

In the present embodiment, in the loop of steps 927 to 929, the rectangle B of the AND image included in the rectangle A of the input image is found in the same manner as in the ninth embodiment.
, The number of rectangles B is determined as the variable S value in the same manner as in the third embodiment. Then, in step 930, it is determined whether or not the area of the rectangle A is an unnecessary image by comparing and determining the S value and a value Th / q obtained by dividing the fixed value Th by the area q of the rectangle A. That is, the present embodiment employs the same unnecessary image determination method as in the third embodiment, and controls the determination threshold value in the same manner as in the ninth embodiment.

In the fourth embodiment or the fifth embodiment, it is obvious that the determination threshold value may be controlled in the same manner as in the present embodiment.

[0077]

As is apparent from the above detailed description, according to the invention of each of the first to twelfth aspects, if there is a filled-out form, the image of the form in an unfilled state is created based on the form. Since it can be created automatically, there is no need for uneconomic work such as re-creating a form using a word processor when a blank form is required. According to the invention described in each of claims 2 to 7, the invention described in claim 11, or the invention described in claim 12, it is possible to automatically create a high-quality blank form image that does not include an unnecessary image. it can. According to the eighth, ninth, or tenth aspect of the invention, it is possible to automatically create a higher-quality non-filled form image without “blurring” or “missing” of characters and the like.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an example of a device configuration of the present invention.

FIG. 2 is a flowchart illustrating a processing flow according to the first exemplary embodiment of the present invention.

FIG. 3 is a diagram illustrating an example of rectangle extraction of an input image and an AND image.

FIG. 4 is a diagram for explaining an unnecessary image determination according to the first embodiment of the present invention.

FIG. 5 is a flowchart illustrating a processing flow according to a second embodiment of the present invention.

FIG. 6 is a diagram for explaining an unnecessary image determination in Embodiment 2 of the present invention.

FIG. 7 is a flowchart illustrating a processing flow according to a third embodiment of the present invention.

FIG. 8 is a diagram for explaining an unnecessary image determination in Embodiment 3 of the present invention.

FIG. 9 is a block diagram showing another example of the device configuration of the present invention.

FIG. 10 is a flowchart illustrating a processing flow according to a fourth embodiment of the present invention.

FIG. 11 is a diagram for explaining an unnecessary image determination in Embodiment 4 of the present invention.

FIG. 12 is a flowchart illustrating a processing flow according to a fifth embodiment of the present invention.

FIG. 13 is a diagram for explaining an unnecessary image determination in Embodiment 5 of the present invention.

FIG. 14 is a block diagram showing another example of the device configuration of the present invention.

FIG. 15 is a flowchart illustrating a processing flow according to a sixth embodiment of the present invention.

FIG. 16 is a diagram for explaining an unnecessary image determination in Embodiment 6 of the present invention.

FIG. 17 is a flowchart illustrating a processing flow according to a ninth embodiment of the present invention.

FIG. 18 is a diagram for explaining an unnecessary image determination in Embodiment 9 of the present invention.

[Explanation of symbols]

 100, 400, 600 Image input unit 101, 401, 601 Image alignment unit 102, 402, 602 AND image creation unit 103, 403, 603 Rectangle extraction unit 104, 404, 604 Unnecessary image deletion unit 105, 405, 605 Form Image output unit 106, 406, 606 Input image buffer 107, 407, 607 AND image buffer 108, 408, 608 Rectangular coordinate buffer 409 Form image shaping unit 410 OR image / multi-valued image buffer 609 Line segment extraction unit

Claims (12)

[Claims] 1. A form image creating apparatus for creating an image of an unfilled form from a completed form, comprising: means for inputting a plurality of filled-out form images in the same format; A form image creating apparatus comprising: means for aligning a completed form image; and means for creating an AND image of a plurality of filled form images after the alignment. 2. A form image creating apparatus for creating an image of an unfilled form from a completed form, comprising: means for inputting a plurality of filled-out form images in the same format; Means for aligning the completed form image, means for creating an AND image of the plurality of completed form images after the alignment,
A form image creating apparatus comprising: means for identifying an unnecessary portion of the AND image to be deleted by comparing the D images and deleting the unnecessary portion from the AND image. 3. A form image creating apparatus for creating an image of an unfilled form from a completed form, comprising: means for inputting a plurality of filled-out form images in the same format; Means for aligning the completed form images, means for creating an AND image of the plurality of completed form images after the alignment, and surrounding the black run connected component by the plurality of completed form images and the AND images after the alignment. Means for extracting a rectangle, and associating the rectangle extracted from the filled-in form image after the alignment with the rectangle extracted from the AND image, and removing an unnecessary portion of the AND image to be deleted from the relationship between the corresponding rectangles. And a means for deleting unnecessary portions from the AND image. 4. A form image creating apparatus for creating an image of an unfilled form from a filled-out form, comprising: means for inputting a plurality of filled-out form images in the same format; Means for aligning the completed form images, means for creating an AND image of the plurality of completed form images after the alignment, and surrounding the black run connected component by the plurality of completed form images and the AND images after the alignment. A means for extracting a rectangle is associated with a rectangle extracted from a filled-out form image after alignment, and a rectangle extracted from an AND image having a positional relationship included in the extracted form image. Identify unwanted parts of the image that need to be removed,
Means for deleting the unnecessary part from the AND image. 5. A form image creating apparatus for creating an image of an unfilled form from a completed form, comprising: means for inputting a plurality of filled-out form images in the same format; Means for aligning the completed form images, means for creating an AND image of the plurality of completed form images after the alignment, and surrounding the black run connected component by the plurality of completed form images and the AND images after the alignment. Means for extracting a rectangle, and associating the rectangle extracted from the filled-in form image after the alignment with the rectangle extracted from the AND image, and comparing the sizes of the corresponding rectangles to each other, A form image creating apparatus comprising: means for specifying an unnecessary portion to be deleted and deleting the unnecessary portion from the AND image. 6. A form image creating apparatus for creating an image of an unfilled form from a completed form, comprising: means for inputting a plurality of filled-out form images of the same format; Means for aligning the completed form images, means for creating an AND image of the plurality of completed form images after the alignment, and surrounding the black run connected component by the plurality of completed form images and the AND images after the alignment. Means for extracting a rectangle, and associating the rectangle extracted from the filled-in form image after the alignment with the rectangle extracted from the AND image, and determining each rectangle extracted from the filled-in form image after the alignment. By comparing the number of black pixels with the number of rectangular black pixels extracted from the corresponding AND image, an unnecessary portion of the AND image to be deleted is specified, and the unnecessary portion is identified by AN.
A form image creating apparatus comprising: means for deleting from a D image. 7. A form image creating apparatus for creating an image of an unfilled form from a filled-out form, comprising: means for inputting a plurality of filled-out form images in the same format; Means for aligning the completed form images, means for creating an AND image of the plurality of completed form images after the alignment, and surrounding the black run connected component by the plurality of completed form images and the AND images after the alignment. Means for extracting a rectangle, and associating the rectangle extracted from the filled-in form image after the alignment with the rectangle extracted from the AND image, to each rectangle extracted from the filled-in form image after the alignment. A form image creating apparatus comprising: means for identifying an unnecessary portion of the AND image to be deleted based on the number of rectangles extracted from the corresponding AND image, and deleting the unnecessary portion from the AND image. 8. An AND image from which an unnecessary portion has been deleted,
The form image creating apparatus according to any one of claims 2 to 7, further comprising means for performing image correction by referring to the filled form image after the alignment. 9. The form image creating apparatus according to claim 8, wherein the image correcting means replaces an image of a part which has not been deleted in the AND image with a corresponding part in the plurality of filled form images after the alignment. A form image creating apparatus for performing image correction by replacing with an OR image. 10. The form image creating apparatus according to claim 8, wherein the image correcting means converts an image of a part of the AND image that has not been deleted into a corresponding part of a plurality of filled form images after alignment. A form image creating apparatus for performing image correction by replacing the image with an averaged image. 11. A means for extracting a line segment corresponding to a ruled line or a frame line on a filled-in form image after alignment, and a means for deleting an unnecessary portion is extracted from the filled-in form image after alignment. 8. The form image according to claim 5, 6 or 7, wherein a threshold value for judging whether or not a portion related to the rectangle should be deleted is adjusted according to a distance of the extracted rectangle from the extracted line segment. Creating device. 12. The form image creating apparatus according to claim 5, wherein the means for deleting an unnecessary portion is provided in accordance with the size of the rectangle extracted from the filled-in form image after alignment. A form image creating apparatus for adjusting a threshold value for determining whether to delete a portion related to a form.