JP2003244437A - Image processor - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an image processor for eliminating the base texture for the unit not only of an entire document image or specific area but also of a line and for easily reading character information by an OCR even from the document image having the high gray level value of a background. <P>SOLUTION: In a system having a scanner part 101 for digitizing and capturing character information written on the surface of paper, an image signal processing part 103 for performing an image processing operation and an image storage part 102 for storing image information and the processing result, the system has a means 203 for finding a gray level histogram in a subscanning direction, a means 204 for finding the gray level histogram of maximum appearance frequency for each line in the subscanning direction and a means 205 for converting the gray level value of the image on the basis of the gray level histogram information of the maximum appearance frequency for each line in the subscanning direction, the gray level value of the background is efficiently specified and only the gray level value of that background is eliminated. <P>COPYRIGHT: (C)2003,JPO

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image processing apparatus such as a copying machine, a facsimile, a printer, etc., and more particularly to an image processing apparatus for removing a background of a document image.

[0002]

2. Description of the Related Art Conventionally, documents have been digitized,
Text information written on paper is often saved as image information, but since image information such as bitmaps has a large capacity and document editing is difficult,
It has been desired to convert it to a character code. Therefore, the OCR (Opti that reads the character image and converts it to the character code
cal Character Reader: optical reader) is used. At this time, in order to correctly convert the character code into the OCR, it is necessary to correctly separate the character area and the background area by the binarization processing performed in the OCR.

However, an image having a high background density value, such as a newspaper, may not be correctly separated into a character and a background pixel by the binarization process performed in the OCR.
As a result, the OCR recognition rate decreases. Therefore, before applying the OCR, it is necessary to perform background removal as a first-stage process in order to improve the OCR recognition rate for an image in which the background density value is particularly high.

In this respect, as a conventional apparatus for removing the background, the background removal processing method and apparatus in the image processing of Japanese Patent Application No. 6-48571 and the background removal method of the image storage device of Japanese Patent Application No. 3-7573 are available. Yes, and Japanese Patent Application No. 2-145
An image processing device of No. 100 and the like are known. These conventional apparatuses use the density histogram of the vertical axis: frequency and the horizontal axis: density value as a feature amount to remove the background.

Next, FIG. 9 shows the configuration of a general document image background removal processing apparatus. A scanner unit 901 that digitizes and reads character information written on paper, an image storage unit 902 that stores image information captured by the scanner unit and image information after image processing, and binarizes an input multivalued digital image. The image signal processing unit 903 is configured to operate.

Arrows in FIG. 9 indicate the flow of the image information. First, the multi-valued image read by the scanner unit 901 is temporarily stored in the image storage unit 902. next,
An image signal processing unit 903 is applied to the captured multivalued image.
The background density value is detected by and the background removal processing is performed. As a result of this background removal process, the image is stored again in the image storage unit 902. Further, after storing the image after the background removal processing in the image storage unit 902, the OCR 904 sets the image signal processing unit 903.
Character recognition processing is performed by using the image processed in step 1 as an input.
At this time, since binarization is performed with a simple threshold value in the OCR 904,
An image in which the background area has a low density value while the density value of the character area is hardly changed and the background is removed is sent to the OCR 904 as a target image for character recognition.

[0007]

However, according to such a conventional apparatus, it is possible to scan the information of the entire image once to remove the background of the entire document image or only a specific area, but in line units. There is a problem that it is impossible to remove the background.

For this reason, when the background density is not uniform, the local information is not used, so that it depends on the entire density information, and the background removal in the local area is performed. Is difficult.

Further, when reading character information from a document image having a high background density value, the difference between the density values of the character area and the background area becomes small, and there is a problem that the image is not clear and it is difficult to read.

Therefore, according to the present invention, the vertical axis is the density value of the maximum frequency in the unit of one line in the sub-scanning, and the horizontal axis is the histogram of the number of lines in the sub-scanning as the trace amount of the background removal, and the density is determined for each sub-scanning line An object of the present invention is to provide an image processing device that removes the background in accordance with an extremely small amount by using a conversion method.

[0011]

In order to solve the above-mentioned problems, the invention as claimed in claim 1 is a scanner unit for digitizing and capturing character information written on a paper, and image signal processing for performing image processing operation. In a sub-scanning direction, and a unit for obtaining a density histogram in the sub-scanning direction.
Efficiently having means for obtaining a density histogram with the maximum appearance frequency for each line and means for converting the density value of the image based on the density histogram information with the maximum appearance frequency for each line in the sub-scanning direction. The feature is that the density value of the background is specified and only the density value of the background is removed.

According to a second aspect of the invention, in the image processing apparatus according to the first aspect, the appearance for each line obtained by the means for obtaining the density histogram having the maximum appearance frequency for each line in the sub-scanning direction. The density histogram of the maximum frequency is further smoothed, and the background density value reflects the density value of the input image and is uniform in the processed image after the background removal of the entire image. It is characterized in that

Further, the invention according to claim 3 is the same as claim 1
Alternatively, the image processing apparatus according to the second aspect further includes a unit for normalizing the density value before the unit for obtaining the density histogram having the maximum appearance frequency for each line in the sub-scanning direction. An image having a high value is characterized in that the pixels having an intermediate density value are made high, the background density value is easily identified for the subsequent processing, and the background is removed more efficiently.

On the other hand, according to the invention of claim 4, in the image processing apparatus according to any one of claims 1 to 3, the density histogram information having the maximum appearance frequency for each line in the sub-scanning direction is used. In addition to the means for converting the density value of the image, there is further provided means for changing the means for converting using the density information, and the output image has little change in the density value for the character pixels and has a background for the background pixels. It is characterized in that the effect of removal is high.

[0015]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The embodiments of the image processing apparatus according to the present invention will now be described in detail with reference to the attached document.

[Embodiment 1] This embodiment relates to claim 1. FIG. 1 shows the flow of the background removal processing of claim 1 performed by the image signal processing unit 903 of FIG. In the processing order of FIG. 1, a density histogram is created 103 for the input multi-valued document image 101, and a density histogram with the maximum appearance frequency for each line is created 104 based on the density histogram information. Based on the histogram information for each line, the background removal 105 is performed on the pixels distributed in that line by the density conversion formula, and the document image 102 with the background removed is output to the image storage unit 902.

First, with respect to the multi-valued document image 101, a density value for one line is searched in the sub-scanning direction, and a density histogram creation 103 process is performed to create a density histogram for one line. As a result, the density distribution information for one line in the sub-scanning direction is stored. The histogram is used when searching for a density value having the highest appearance frequency in the sub-scanning direction.

Next, based on the information of the density histogram,
A density histogram having the maximum appearance frequency for each sub-scanning line is created 104. At this time, the frequency of all density values is searched based on the information of the histogram created in the density histogram creation 103, and the density value information of the maximum frequency is searched for in the density histogram in which the appearance frequency of one sub-scanning line is the maximum. Save as a value. After the creation, the density histogram creation 103 process is performed again in the next sub-scanning direction, and the density histogram 104 having the maximum appearance frequency for each line is created 104 to maximize the appearance frequency for each line in all sub-scanning directions. Create a density histogram.

When the density histogram having the maximum appearance frequency for each line is created in all sub-scanning directions, the background removal 105 process is performed by the density conversion formula to remove the background. Since the image is searched in all sub-scanning directions at the time of this processing, the minimum density value in the entire image is searched, and this value and the density value of the pixel of interest of the input image are used in the conversion formula to convert. To do. For pixels belonging to one sub-scanning direction, a density value having the maximum appearance frequency of the line is searched for and obtained, and
The minimum density value of the entire image is used in the conversion formula to perform density conversion. In this process, the density histogram with the highest appearance frequency for each line in all sub-scanning directions is searched, the density value is converted for the entire image, and the background is removed. Since the density value of the pixel in the character area is high before the processing, the density value does not change so much, but the density value of the pixel in the background area decreases due to the conversion formula.

FIG. 2 shows a document image before background removal,
The example of the image which performed the background removal is shown. FIG. 2 is an image obtained by cutting out a part of a newspaper read by a scanner. In FIG. 2A, the newspaper originally has a high density value in the background. On the other hand, in FIG. 2B, the background density value is effectively converted to a low value, and the character area does not change much.

FIG. 3 shows a density histogram of the images of FIGS. 2 (a) and 2 (b). As you can see from both histograms,
In (a), the pixels that are mostly distributed around the middle density value 100 as a background are shifted to around 25 density values in (b). In addition, for pixels with a density value of 150 or more in (a),
In (b), there is almost no change in the density value.

[Embodiment 2] This embodiment relates to claim 2. FIG. 4 shows the flow of the background removal processing of claim 2 performed by the image signal processing unit 903 of FIG. In the processing order of FIG. 4, the density histogram is created 103 for the input multi-valued document image 101, and the density histogram having the maximum appearance frequency for each line is created 104 based on the density histogram information. . Where 1
The histogram created in the creation 104 of the density histogram having the maximum appearance frequency for each line is smoothed by the smoothing 201 of the density histogram having the maximum appearance frequency for each line. Based on the smoothed histogram information for each line, the background distribution removal 105 is performed on the pixels distributed in that line by the density conversion formula, and the document image 102 from which the background is removed is output to the image storage unit 902.

First, for the multi-valued document image 101, a density value for one line is searched in the sub-scanning direction, and a density histogram creation 103 process is performed to create a density histogram for one line. As a result, the density distribution information for one line in the sub-scanning direction is stored. This histogram is used when searching for the density value having the highest appearance frequency in the sub-scanning direction.

Next, based on the information of the density histogram, the density histogram 104 having the maximum appearance frequency for each line is created 104. Based on the information of the histogram created in 103 of creating the density histogram, the frequency of all density values is searched, and the density value information of the maximum frequency is stored as the value of the density histogram in which the appearance frequency of one sub-scanning line is the maximum. To do. After the creation, the density histogram creation 103 process is performed again in the next sub-scanning direction, and the density histogram creation 104 having the maximum appearance frequency for each line is performed to determine the appearance frequency for each line in all sub-scanning directions. Create the maximum density histogram.

The density histogram having the maximum appearance frequency for each line is created 104 in all sub-scanning directions to create the density histogram having the maximum appearance frequency for each line, and then the appearance frequency for each line is generated. Performs the smoothing 201 process of the maximum density histogram. Since the density histogram with the maximum appearance frequency for each line is a function that uses the maximum frequency of pixels of the same density in one-dimensional sub-scan as a variable, smoothing is performed by taking the first-order derivative of this one-dimensional function. .

As shown in FIG. 5, the frequency value a of interest is smoothed by taking an average value in the vicinity. As shown in FIG. 5, in the case of smoothing in the vicinity of a, if the preceding and following frequency values are b and c, respectively, the frequency amount of interest a is (b +
c) The average value of 2 is replaced with the frequency amount a of interest, and the smoothing process of the maximum density value of the appearance frequency is performed.

After the density histogram having the maximum appearance frequency for each line is created in all sub-scanning directions, the background removal 105 process is performed by the density conversion formula to remove the background. Since the image is searched in all sub-scanning directions at the time of this processing, the minimum density value in the entire image is searched, and this value and the density value of the pixel of interest of the input image are used in the conversion formula. Convert. Furthermore, for pixels belonging to one sub-scanning direction,
Find the density value with the highest appearance frequency of the line,
Using this value and the minimum density value of the entire image in the conversion formula,
Perform density conversion. In this process, a density histogram with the highest appearance frequency for each line in all sub-scanning directions is searched, density value conversion is performed on the entire image, and the background is removed. Since the density value of the pixel in the character area is low before the processing, the density value does not change much, but the density value of the pixel in the background area becomes low due to the conversion formula.

FIG. 6 shows a result image without smoothing and with smoothing. Without the smoothing process (a), the density conversion formula may vary greatly from line to line depending on the density value of the maximum frequency, so the value of the conversion formula may change greatly even in the vicinity, and the image may look unnatural at the center of the image. Edges may occur. On the other hand, in the smoothing processing (b), since the average value in the vicinity of the line is taken, the density conversion formulas for the lines are not significantly different, and an unnatural edge is less likely to occur in the sub-scanning direction.

[Embodiment 3] This embodiment relates to claim 3. FIG. 7 shows the flow of the background removal processing of claim 3 performed by the image signal processing unit 903 of FIG. In the processing order of FIG. 7, the density value normalization 701 is performed on the input multi-valued document image 101, the minimum density value of the input image is normalized to 0, and then the density histogram is created.
3 is performed, and based on the density histogram information, a density histogram having the highest appearance frequency for each line is created 10
Do 4. Based on the histogram information for each line,
For the pixels distributed on that line, the background removal 105 is performed by the density conversion formula, and the background is removed from the document image 102.
Is output to the image storage unit 902.

The input image has a high background density value of the background,
In addition, since the density value of the character area is even lower, the density histogram has a small difference between the density value of the background area and the density value of the character area, and the overall width is narrow. Normalization 701 of the density value is performed on the histogram in which the density range is narrow. The normalization formula is as follows: Normalized density value = {maximum gradation value of input image− (density value before normalization × maximum gradation value of input image) / maximum density value of input image} To increase the difference between the density values of the character area and the background area.

After that, the density histogram creation 103 according to the first aspect, the density histogram 104 having the highest appearance frequency for each line, and the background removal 105 by the density conversion formula are performed to perform the same background removal. To process.

[Embodiment 4] This embodiment relates to claim 4. FIG. 8 shows the flow of the background removal processing of claim 4 performed by the image signal processing unit 903 of FIG. In the processing order of FIG. 8, a density histogram is created 103 for the input multi-valued document image 101, and based on the density histogram information, a density histogram 104 having the maximum appearance frequency for each line is created 104. , Based on the histogram information for each line, the background conversion is performed by replacing the density conversion formula with a or b according to the density value of the pixel of interest for the pixels distributed on that line, and the document with the background removed. The image 102 is output to the image storage unit 902.

First, the multi-valued document image 101 is searched for a density value for one line in the sub-scanning direction, and the density histogram creation 103 process is performed to create a density histogram for one line. As a result, the density distribution information for one line in the sub-scanning direction is saved. It should be noted that this histogram is used when searching for a density value having the highest appearance frequency in the sub-scanning direction.

Next, based on the information of the density histogram, the density histogram 104 having the maximum appearance frequency for each line is created 104. Based on the information of the histogram created in 103 of creating the density histogram, the frequency of all the density values is searched, and the density value information of the maximum frequency is used as the value of the density histogram in which the appearance frequency of one sub-scanning line is the maximum. save. Once created, the density histogram creation 103 processing is performed again in the next sub-scanning direction, and the density histogram creation 104 having the maximum appearance frequency for each line is created 104, and the appearance frequency for each line in all sub-scanning directions is performed. Creates the maximum density histogram.

When a density histogram having the maximum appearance frequency for each line in all sub-scanning directions can be created, the density conversion formula is changed for each density value of the pixel of interest, and the background removal 801 by the density value conversion formula, 802 processing is performed to remove the background. Since the document image is more distributed in the background area, the density value with the highest appearance frequency for each line is assumed to be the background density value of that line, and this value is used as a threshold value, and the density value is high due to the height relationship. In this case, the density value conversion equation a for the character area is processed separately, and when the density value is low, the density value conversion equation b for the background area is processed separately.

Since the image is searched in all sub-scanning directions at the time of this processing, the minimum density value of the entire image is searched, and this value and the density value of the target pixel of the input image are converted into a conversion formula. Used to convert. For pixels belonging to one sub-scanning direction, the density value with the highest appearance frequency of the line is searched for and obtained, and the minimum density value in the entire image is used in the conversion formula to perform density conversion. In this process, the density histogram with the highest appearance frequency for each line in all sub-scanning directions is searched, the density value is converted for the entire image, and the background is removed.

The pixels in the character area have a high density value before processing, and the density conversion is performed by the density conversion equation a for the character area, so the density value does not change much, but the pixels in the background area are Since the density value is converted by the density conversion equation b for the density value, the density value becomes lower than the density value before the conversion.

[0038]

According to the invention described in claim 1, since the density value of the background of the input image can be specified by using the density histogram having the maximum appearance frequency for each line,
Based on this information, the background density value can be converted to a low value, and the background can be removed.

According to the second aspect of the present invention, since the difference in the neighboring area is smoothed by smoothing the density histogram in which the appearance frequency of each line is the maximum, the output image has the density values of the neighboring areas. The difference is small, and it is difficult for an unnatural edge to occur. Further, the density conversion formula lowers the density value of the background and removes the background.

According to the third aspect of the present invention, first, by normalizing the minimum density value of the input image to the minimum density value of the gradation, the minimum density value approaches the origin, so that the background density value is With a high image, the background density value can be efficiently set low, and the background of the output image is removed.

According to the fourth aspect of the present invention, the background density value is lowered by changing the density conversion formula using the density histogram having the maximum appearance frequency for each line and the density value information of the input image. Since the density of the character changes little and the density can be converted, it is possible to output an image in which the background of the output image is removed and the density of the character is the same as that of the input image.

[Brief description of drawings]

FIG. 1 is a diagram showing a flow of background removal processing according to an embodiment of the present invention.

FIG. 2 is a diagram showing an example of a document image (a) before background removal and a document image after background removal.

FIG. 3 is a density histogram of the images of FIGS. 2 (a) and 2 (b).

FIG. 4 is a diagram showing a flow of background removal processing for smoothing a density histogram according to the embodiment of the present invention.

FIG. 5 is a diagram showing a part of a density histogram having the maximum appearance frequency for each line.

FIG. 6 is a diagram showing comparison of presence / absence of smoothing processing.

FIG. 7 is a diagram showing a flow of background removal processing for normalizing density values according to the embodiment of the present invention.

FIG. 8 is a diagram showing a flow of background removal processing in which the density conversion formula is changed according to the embodiment of the present invention.

FIG. 9 is a configuration diagram of a general background removal processing device.

[Explanation of symbols]

101 Multi-valued document image 102 Document image with the background removed 103 Creating a density histogram 104 Creating a density histogram with the highest appearance frequency for each line 105 Background removal by a density conversion formula 201 The appearance frequency for each line is the maximum Smoothing of histogram of density 701 Normalization of density value 801 Background removal by density conversion equation a 802 Background removal by density conversion equation b 901 Scanner unit 902 Image storage unit 903 Image signal processing unit 904 OCR (optical reading device)

Claims (4)

[Claims] 1. A system comprising a scanner unit for digitizing and capturing character information written on paper, an image signal processing unit for performing image processing calculation, and an image storage unit for storing image information and processing results. A means for obtaining a density histogram in the scanning direction, a means for obtaining a density histogram having the maximum appearance frequency for each line in the sub-scanning direction, and a density histogram having the maximum appearance frequency for each line in the sub-scanning direction An image processing apparatus comprising: means for converting a density value of an image based on information. 2. A means for performing smoothing on the density histogram having the maximum appearance frequency for each line obtained by the means for obtaining the density histogram having the maximum appearance frequency for each line in the sub-scanning direction, The image processing apparatus according to claim 1, further comprising: 3. The method according to claim 1, further comprising a unit for normalizing the density value before the unit for obtaining the density histogram having the maximum appearance frequency for each line in the sub-scanning direction. 2. The image processing device according to item 2. 4. A means for converting a density value of an image based on density histogram information having the maximum appearance frequency for each line in the sub-scanning direction, and a means for changing a conversion means using the density information. 4. The method according to claim 1, further comprising:
The image processing device according to item.