JPH0713984A - Image processor - Google Patents

Abstract

PURPOSE:To provide an image processor which is capable of recognizing marks from the images obtained from general scanner and facsimile, etc., and the images obtained through a network, and easily canceling the marks. CONSTITUTION:Single-color input images obtained through an image scanner, a facsimile and a network are inputted in an area detection means 1 and areas where marks are possible to exist are segmented. In a line segment detection means 2, a long line which exists in the segmented image and is an ineffective information, is detected. A mark discrimination means 3 outputs a signal where the mark is ineffective when the long line is detected in the line segment detection means 2. When the ineffective information is not detected in the line segment detection means 2, whether the marks are imparted to the segmented areas or not, is discriminated and the discrimination result is outputted.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention extracts a specific shape from an image read by an image scanner, a facsimile or the like or an image of a single color obtained through a network, and recognizes information checked in the specific shape. The present invention relates to an image processing device.

[0002]

2. Description of the Related Art Conventionally, as a known technique for reading a mark, for example, there is a technique used for a mark sheet. A frame on a document such as a mark sheet to which a mark is given is drawn in a color different from that of a writing instrument to be checked, such as a pencil or a ballpoint pen. Since this color uses so-called dropout color that is not read by an image scanner, there is no frame for marking in the image of the read document. The position of the mark can be known by reading the information such as the format number and timing mark written on the document, or by inputting the position information in advance, so the absolute coordinates are used based on this information. Whether or not the mark is checked can be determined from the number of pixels at the mark position.

However, such an original can be read only at a specific position, and the mark frame is not removed in an image obtained by a general scanner, a facsimile or the like, or an image obtained through a network. There was a drawback that I could not. Further, with respect to the document creation, since the dropout color that cannot be identified by the scanner is used, the document cannot be easily created.

On the other hand, when a mark is erroneously made on the above-mentioned original, it is necessary to erase the mark. When a pencil is used as a writing tool, it can be easily erased with an eraser, but when a ballpoint pen or the like is used, it is difficult to erase. Also,
When performing a large amount of marking processing, it is necessary to make erasing particularly easy. for that reason,
For example, Japanese Patent Application Laid-Open No. 2-8978 describes that a mark entry field for cancellation is provided. However, if a cancel mark entry field is separately provided, a space for the manuscript is required, and there is a problem that the manuscript surface cannot be effectively used. In addition, there is a problem that other mark entry fields cannot be freely arranged depending on the cancellation mark entry field.

[0005]

SUMMARY OF THE INVENTION The present invention has been made in view of the above circumstances, and can recognize a mark from an image obtained by a general scanner, a facsimile or the like, or an image obtained through a network, and at the same time, the mark can be recognized. It is an object of the present invention to provide an image processing device capable of easily canceling.

[0006]

According to the present invention, in an image processing apparatus for recognizing a mark described in an image, a region extracting means for extracting a mark describing region from the image, and an area extracting means for extracting the mark in the image. Line segment detection means for detecting a line segment in the mark description area, and a line segment detected by the line segment detection means while detecting the mark described in the mark description area extracted by the area extraction means Is provided with mark identifying means for invalidating the mark when the length of the mark is greater than or equal to a predetermined threshold value.

[0007]

According to the present invention, the area in which the mark is written is extracted by the area extracting means and the mark is detected by the mark identifying means, so that it can be applied to an image of a single color obtained through a general image scanner, facsimile, or network. The mark description frame and the mark can be extracted and recognized. This allows the user to freely design and use the mark writing area.

Further, by detecting the line segment in the mark writing area by the line segment detecting means, for example, when erasing is instructed by a double line or the like, the double line indicating the erasing is detected. When the erasing instruction is detected, the mark is invalidated by the mark identifying means, so that it is determined that there is no added mark, and the mark is canceled.

[0009]

1 is a block diagram showing an embodiment of an image processing apparatus of the present invention. In the figure, 1 is a region detecting means, 2 is a line segment detecting means, and 3 is a mark identifying means. The image input from the image scanner and the input image of a single color obtained through the facsimile or the network are input to the area detecting means 1. The area detection unit 1 cuts out an area where a mark may exist from the input image. The region to be cut out can be a region wider than the mark entry frame or the like. The cut out image is input to the line segment detecting means 2 and the mark identifying means 3. The line segment detecting means 2 detects a long line existing in the cut image. This long line was added to cancel the mark, for example,
Double lines, etc. The mark identifying means 3 is the line segment detecting means 2
If a long line is detected at, the mark is output as an invalid signal. Alternatively, a signal without a mark can be output. If a long line is not detected by the line segment detecting means 2, it is discriminated whether or not a mark is added to the area cut out by the area detecting means 1,
Output the identification result.

FIG. 2 is a flow chart for explaining the operation of an embodiment of the image processing apparatus of the present invention. When an image for mark recognition is input to the image processing apparatus, S11
In, the area detecting means 1 cuts out an area where a mark may exist. In step S12, invalid information, such as a double line added for cancellation, is detected from the cut image area. If there is invalid information, the mark is invalid. If the invalid information does not exist, in S13, the information printed in advance on the original such as the frame for adding the mark is removed. Then, in S14, the presence or absence of the mark is identified. As an example of the identification method, the number of black or white pixels in the clipped image area may be counted and compared with a preset threshold value to detect the presence or absence of a mark. For example, if the marked portion is a black pixel, the number of black pixels is counted, and if the count result is greater than or equal to the threshold value, it is determined that there is a mark, and if it is less than the threshold value, it is determined that there is no mark. Of course, other identification methods may be used.

The above-described image processing such as detection of invalid information, removal of a frame, and identification of a mark can be realized by a method using a morphology (inter-image logical operation) processing in which a pixel count is added. . By using this method, software and hardware processing algorithms can be made common, macros can be formed, and the processing speed can be increased, so that even a normal CPU can perform software processing at high speed. Also, it is easy to convert to hardware,
Further speedup can be implemented. Hardware can be standardized and the processing speed can be increased. Further, since the morphology processing can extract the features all at once, the shapes can be extracted and recognized at high speed by combining these features. Further, there is little influence on image distortion and noise, and highly accurate identification is possible.

FIG. 3 is an explanatory diagram of an example of a check box for marking. In the following description, a frame or dot that indicates the position of the mark to be printed on the document in advance is called a check box. FIG. 3 shows an example of check boxes and states of various markings.
These checkboxes and checks for markings can consist of only a single color, eg black.

FIG. 3A shows a case where a rectangle is used as indicated by a check box. Checking for marking is as shown in ~,
Various things such as ○, ×, /, \, and filling can be performed. FIG. 3B shows a case where dots are used as shown as a check box.
Checks for marking, as shown in
Similar to the case of FIG. 3 (A), it can be performed by using a ◯ mark, an X mark, a check mark, a ● mark, or the like. Further, in FIG. 3C, a combination of dots and hook-shaped symbols is used as indicated by a check box. The check for marking can be performed in the same manner as in the case of FIG. In these examples, the mark “OFF” state is the case of only the check box, and the mark “ON” state is the case of the marking example.

In addition, in order to cancel the check after checking the check box due to writing failure, etc., please refer to FIG.
As shown in (D) and (E), the check box is drawn by drawing a plurality of long vertical lines. In this example, cancellation is instructed by drawing a double line vertically.

It is of course possible to use check boxes other than the example shown in FIG. 3, and the checks are only general ones, and are not limited to those shown in the example. . Further, the long line for canceling can also be constructed so as to be drawn horizontally, and, of course, only one line may be provided, and three or more lines may be provided.

Hereinafter, each process will be described in detail using the first specific example. In the first specific example, a case where the rectangle shown in FIG. 3A is used as a check box will be described as an example. First, a process of detecting invalid information of a check box will be described. As shown in FIG. 3D, when the check box is disabled, a plurality of long vertical lines are drawn on the check box. Therefore, if invalidation is instructed,
There is a long vertical line in the cut out area. That is, the invalid area can be identified by extracting only the long vertical line.

FIG. 4 is an explanatory diagram of a process of extracting invalid information in the first specific example. FIG. 4 shows the processing result of the image for each processing step in the case where the vertical double line that invalidates the mark is described and in the case where the check mark is added as a normal check. In addition, the processing contents of each processing step are shown on the right.

First, FIG. 4A shows a clipped image. The image is cut out with some margin in consideration of the distortion of the input / output system. However,
A long vertical line indicating invalidity is assumed to be drawn from the top to the bottom of the clipped image.

As shown in FIG. 4B, vertical lines are extracted from the image thus cut out, and components other than the vertical lines are removed. FIG. 5 is an explanatory diagram of an example of vertical line detection. To detect a vertical line, the operation of "ANDing the original image with the original image shifted one dot below" is repeated a plurality of times, and then "the image with the processed image shifted one dot above and the processed image By repeating the operation of "ORing.""The same number of times, only vertical lines with a certain length or more are extracted.
For example, if the image shown in FIG.
An image shown in FIG. 5B is obtained by ANDing the image shown in FIG. 5A with the original image shown in FIG. 5A. can get.
Further, the image shown in FIG.
The image shown in FIG. 5C is obtained by creating an image in which the image shown in FIG. 5B is shifted down by one dot and taking the AND with the image shown in FIG. 5B. Next, FIG.
An image shown in FIG. 5D is obtained by creating an image in which the image shown in FIG. 5C is shifted by one dot and taking the OR with the image shown in FIG. 5C. Furthermore, FIG.
An image shown in FIG. 5E is obtained by creating an image in which the image shown in FIG. 5D is shifted by one dot and taking the OR with the image shown in FIG. 2 like this
By performing the AND processing twice and the OR processing twice, it is possible to obtain an image in which only a vertically continuous portion of 3 dots or more remains.

Although an example of vertical line extraction has been described here, AND processing is performed while swelling the image to the left and right, and processing such as swelling the image and performing OR with the original image is performed to faithfully reproduce the original image. It is also possible to extract a simple line. All of these processes are performed on a face-by-face basis, but BitBlt (Bit Bundary bl
Since it can be realized only by sequential transfer and calculation of each bit by means of an occ transfer, etc., high speed processing is possible.

The horizontal line can also be extracted by the same method. For example, the operation of ANDing an image obtained by shifting the original image to the right by one dot and the original image is repeated a plurality of times, and then the processed image is made 1 time. By repeating the operation of taking the OR of the image shifted to the left of the dot and the processed image the same number of times, only the horizontal line having a certain length or more can be extracted. This is effective when a plurality of horizontal lines are used as invalid information for cancellation.

Returning to FIG. 4, the lines of the check box itself and the check lines also remain in the vertical lines thus extracted. Therefore, next, it is necessary to identify the vertical line of the frame and the vertical line indicating invalidity. In order to extract the invalid information from the vertical lines, first, the continuous components of the image from the top to the bottom may be extracted. First, by extracting consecutive black pixels from the uppermost part, the image shown in FIG. 4C is obtained. In this state, in the case of a normal check having a vertical line extending from the upper part like a check mark, a line other than invalid information still remains. Furthermore, by extracting continuous black pixels from the lowermost part, only the vertical line entered as invalid information can be extracted as shown in FIG.

FIG. 6 is an explanatory diagram of an example of the continuous component extraction processing. As a process for extracting a component in which black pixels are continuous from the uppermost part or the lowermost part, first, adjacent A lines are enlarged horizontally by one dot to create a line C. This processing is performed by, for example, an image moved by 1 dot to the left and an image moved by 1 dot to the right and O.
It can be realized by taking R. The created line C is ANDed with the line B one line below,
Write the processing result to the position. Such a process between two lines can be realized by sequentially repeating the process from the top to the bottom. That is, the top line is 1
AN of the image enlarged to the left and right of the dot and the image of the second line
D processed and written to the second line, written 2
An image obtained by enlarging the image of the line on the right and left of one dot and the image of the third line is ANDed and written on the third line. In this way, the process is repeated up to the bottom line. When the processing is completed, only the vertical line in which black pixels are continuous from the top is extracted. Further, by performing this processing from the bottom to the top, it is possible to extract only lines of invalid information that are continuous from the top to the bottom of the image.

When a horizontal line is used as the invalid information, the same processing is performed from the leftmost edge to the rightmost edge and from the rightmost edge to the leftmost edge, so that black pixels continue from the leftmost edge to the rightmost edge. The horizontal line can be extracted. Of course, when vertical lines and horizontal lines are used in combination, vertical lines that are continuous from the top to the bottom and horizontal lines that are continuous from the leftmost to the rightmost are separately performed by performing both processes. Obtainable.

When the invalid information is detected as described above, the entry field of the mark becomes invalid regardless of whether or not there is a check. When invalid information is not detected, it is identified whether or not the check box is checked. Therefore, first, the check box information is removed. In this specific example, the check box has a rectangular shape, and thus is composed of vertical lines and horizontal lines. Further, as a typical check method, check methods such as those in FIGS. 3A to 3 can be considered. Since there is an oblique line in the checks up to, the vertical line and horizontal line removal processing is effective. However, in the simple vertical line and horizontal line removal processing, all the checks are removed, resulting in erroneous recognition. However, the width of the line is different when compared with the check box frame. Therefore, by performing vertical line and horizontal line removal processing for a limited line width,
Can be identified as OFF and ~ as ON.

The horizontal line removal algorithm with the above line width limited will be described. FIG. 7 is an explanatory diagram of an example of a check box information removal process in the first specific example. In FIG. 7, the case of FIG. 3A is shown as an example. FIG. 7 (A) shows the clipped original image. This image is the original image. First, t from the original image
An image composed of pixels that are continuous for h dots or more is extracted. The extracted image is as shown in FIG. Here, th dots are two dots in the upper and lower parts. Specifically, it can be obtained by ANDing an image obtained by shifting the original image by 2 dots upward and an image obtained by shifting the original image by 2 dots downward. Next, the extracted image, that is, the image of FIG. 7B is removed from the original image, and then the image of FIG.
Get the image shown in. Specifically, it is obtained by exclusive OR processing of the original image and the image of FIG. Further, horizontal lines are extracted from the obtained image to obtain the image shown in FIG. This horizontal line extraction method can be performed by applying the method described above with reference to FIG. 5 to horizontal line extraction. By removing the horizontal lines thus extracted from the original image, the image shown in FIG. 7E is obtained.

The horizontal lines are removed by the above processing. By performing the same process in the vertical direction on the image from which the horizontal lines have been removed, only the graphic in the frame remains. Either the horizontal line or the vertical line may be removed first. Alternatively, removal of horizontal lines and removal of vertical lines are performed separately, and AND processing of the processed images is performed, whereby the figure in the frame can be obtained.

Regarding the image processed in this way,
The number of remaining pixels is counted, and when the number of pixels is equal to or larger than the specified value, the check box is determined to be ON, and when the value is less than the specified value, the check box is determined to be OFF. Of course, it is also possible to use a method other than the method of using the count value to determine whether or not there is a check.

The second specific example will be described below. Second
In the specific example, a case will be described as an example in which a check box in which dots and hook-shaped symbols are combined as shown in FIG. 3C is used. Note that the same processing can be performed in the case of the check box using dots shown in FIG.

First, the process of detecting the invalid information of the check box will be described, but the process is the same as in the first specific example. FIG. 8 is an explanatory diagram of a process of extracting invalid information in the second specific example. In FIG. 8, for each processing step, a vertical double line that invalidates the mark is described, a ○ check is added as a normal check, and a check mark is added. The processing result of the image of is shown. In addition, the processing contents of each processing step are shown on the right.

FIG. 8A shows the cut out image. When cropping the image, considering the distortion of the input / output system,
It is cut out with some margin. However, it is assumed that a long vertical line indicating invalidity is written from the top to the bottom of the clipped image.

As shown in FIG. 8B, vertical lines are extracted from the image thus cut out, and components other than the vertical lines are removed. As the method of detecting the vertical line, the method described in FIG. 5 above can be used. In the extracted vertical line,
The line of the check box itself and the line of the check remain. Therefore, next, it is necessary to identify the vertical line of the frame and the vertical line indicating invalidity. In order to extract the invalid information from the vertical lines, first, the continuous components of the image from the top to the bottom may be extracted. First, by extracting consecutive black pixels from the uppermost part, the image shown in FIG. 8C is obtained. In this state, in the case of a normal check having a vertical line extending from the upper part like a check mark, a line other than invalid information still remains. Further, by extracting continuous black pixels from the lowermost part, only vertical lines filled in as invalid information can be extracted as shown in FIG. For the black pixels continuous from the uppermost part and the black pixels continuous from the lowermost part, the method described in FIG. 6 can be used.

When the invalid information is detected as described above, the entry field of the mark becomes invalid regardless of the check. When invalid information is not detected, it is identified whether or not the check box is checked. Therefore, first, the check box information is removed. In this specific example, since the check box has a shape in which dots and hook-shaped symbols are combined, vertical lines and horizontal lines are present in some of the components. Further, as a typical check method, check methods such as those in FIGS. 3C to 3 can be considered. Since there is a diagonal line in the check of ,,, it is effective to remove the vertical line and the horizontal line.
However, in the simple vertical line and horizontal line removal processing, all the checks of may be removed, which may cause erroneous recognition. Therefore, by performing vertical line and horizontal line removal processing for a limited line width,
Can be identified as OFF and ~ as ON.

The horizontal line removal algorithm with the line width limited is similar to that of the first specific example in the second specific example. FIG. 9 is an explanatory diagram of an example of the check box information removal processing according to the second specific example. In FIG. 9, the case of FIG. 3C is shown as an example.
FIG. 9A shows the original image that has been cut out. This image is the original image. First, an image composed of vertically continuous pixels of th dot or more is extracted from the original image. The extracted image is as shown in FIG. Here th
The dots are upper and lower two dots. Specifically, it can be obtained by ANDing an image obtained by shifting the original image by 2 dots upward and an image obtained by shifting the original image by 2 dots downward. Next, the extracted image from the original image, that is,
The image in FIG. 9 (B) is removed to obtain the image shown in FIG. 9 (C). Specifically, it is obtained by exclusive OR processing of the original image and the image of FIG. 9 (B). Further, horizontal lines are extracted from the obtained image to obtain the image shown in FIG. This horizontal line extraction method can be performed by applying the method described above with reference to FIG. 5 to horizontal line extraction. By removing the horizontal lines thus extracted from the original image, the image shown in FIG. 9E is obtained.

The horizontal lines are removed by the above processing. By performing the same process in the vertical direction on the image from which the horizontal lines have been removed, only the graphic in the frame remains. Either the horizontal line or the vertical line may be removed first. Alternatively, removal of horizontal lines and removal of vertical lines are performed separately, and AND processing of the processed images is performed, whereby the figure in the frame can be obtained.

Regarding the image processed in this way,
The number of remaining pixels is counted, and when the number of pixels is equal to or larger than the specified value, the check box is determined to be ON, and when the value is less than the specified value, the check box is determined to be OFF. In FIG. 9E, the check mark and the filled portion may be missing, but when the presence or absence of the check is finally determined by counting the number of pixels, the change in shape has no effect.

[0037]

As is apparent from the above description, according to the present invention, a check box with a specific shape is provided for an image read by an image scanner, a facsimile, or an image of only a single color obtained through a network. Is recognized, ON or OFF of each check box, and
The effect is that invalidity can be identified.

[Brief description of drawings]

FIG. 1 is a block diagram showing an embodiment of an image processing apparatus of the present invention.

FIG. 2 is a flowchart for explaining the operation of an embodiment of the image processing apparatus of the present invention.

FIG. 3 is an explanatory diagram of an example of a check box for marking.

FIG. 4 is an explanatory diagram of a process of extracting invalid information in the first specific example.

FIG. 5 is an explanatory diagram of an example of vertical line detection.

FIG. 6 is an explanatory diagram of an example of continuous component extraction processing.

FIG. 7 is an explanatory diagram of an example of a check box information removal process in the first specific example.

FIG. 8 is an explanatory diagram of a process of extracting invalid information in the second specific example.

FIG. 9 is an explanatory diagram of an example of check box information removal processing according to a second specific example.

[Explanation of symbols]

1 area detecting means, 2 line segment detecting means, 3 mark identifying means.

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Hiroshi Iida 3-2-1 Sakado, Takatsu-ku, Kawasaki-shi, Kanagawa KSP R & D Business Park Building Fuji Xerox Co., Ltd.

Claims (1)

[Claims] 1. In an image processing apparatus for recognizing a mark written in an image, a region extracting means for extracting a mark writing region from the image, and a line in the mark writing region extracted by the region extracting means. A line segment detecting means for detecting a segment, and a threshold value for detecting the mark written in the area in which the mark is extracted by the area extracting means and the length of the line segment detected by the line segment detecting means is predetermined. In the above case, the image processing apparatus is provided with a mark identifying means for invalidating the mark.