JPH10143607A - Image processor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To appropriately separate a character part and a background part even for images for which a part of characters is thinned by comparing line widths corresponding to respective connection components with a threshold value, judging whether the connection component is the character part or the background part and outputting a result. SOLUTION: A connection component detection part 30 detects the connection component of the same color as character information from inputted source image data and the respective connection components detected by the connection component detection part 30 are stored in a connection component storage part 40. For the respective connection components outputted from the connection component storage part 40, a line width parameter is detected by a line width detection part 50. The line width parameter values of the respective connection components outputted to a character/background judgement part 60 are compared with the threshold value internally possessed by the character/background judgement part 60 and whether the respective connection components are a character data part or the background part is judged.

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 for separating a character portion from a background portion of an image, and more particularly to noise and noise contained in an original image such as a document captured as a binary image. It is suitable for application to an apparatus for removing a tint block or the like.

[0002]

2. Description of the Related Art In an optical character reading apparatus or a facsimile, a document or the like is optically read by using a CCD sensor or the like, and an original image image of a document or the like quantized into black and white binary values is obtained. In the case of a document or the like, the original image image includes image data (hereinafter, referred to as character data) of characters and line drawings (hereinafter, collectively referred to as characters) and image data of a character background portion (hereinafter, referred to as background). Data). Generally, background data often includes unnecessary noise that hinders image processing, and for image processing, a process of removing noise included in background data is performed in advance in many cases.

[0003] Further, as seen in a headline article of a newspaper or the like, some documents have a line pattern such as a vertical line, a horizontal line, or an oblique line in a character background portion. In this case, the background data includes image data of a line pattern (hereinafter, referred to as pattern data). If the image processing of the original image data is performed while such pattern data is included, initial image processing can be performed. Inconveniences such as no image processing and poor image processing efficiency occur.

For example, when character recognition processing is performed using an optical character reader, it is very difficult to perform character extraction processing without removing pattern data. Generally, processing for removing pattern data is performed. After that, character extraction processing is performed.

As a method of processing such pattern data or image data of a character background portion such as noise, a method using a black run length or contraction (degeneration) is known (for example, reference 1 pp. 54). -56).

Literature 1, "Toshiyuki Sakai, Seminar on Fundamental Information Science, Corona Co., 1986" The method of using the black run length is, for example, to scan the original image data in the vertical direction and the horizontal direction, and to set the black run length to a predetermined value. By removing black runs smaller than the threshold value, unnecessary elements such as background line patterns and noise lines that are thinner than character lines are removed. When the character color is white, similar processing is performed according to the white run length.

In the method using erosion, for example, a 4-neighbor erosion is repeated in the entire original image until there is no background line pattern, and then expansion processing is performed to restore characters.

[0008]

However, in any of the above methods, accurate background removal is not possible unless the background line pattern, noise line, or the like is thinner than the character line.

FIG. 2A is an example of an image with a line pattern in which a black background pattern is added to a character (hereinafter, referred to as a black character) image in which a character line is composed of black pixels. b)
Is an example of an image with a line pattern in which a white background pattern is added to a character (hereinafter, referred to as a white character) image in which a character line is composed of white pixels. This shows a case where the thickness is almost the same.

[0010] For example, when the background data, which is a black vertical line pattern, is removed from the original image data shown in FIG. 2A using the black run length, as shown in FIG. The horizontal line disappears. On the other hand, even when shrinkage is used, FIG.
As shown in (1), the horizontal line of the "character" disappears, and it is impossible to restore the horizontal line of the lost character even if the expansion processing is performed thereafter.

In the case of an original image in which characters are printed in a typeface similar to the Mincho style as shown in FIG. 2A, the horizontal lines of the characters become extremely thin. However, there is a problem that character removal processing and character recognition processing, which are subsequent processing, cannot be performed accurately.

Therefore, there is a demand for an image processing apparatus that can appropriately separate a character portion from a background portion even in an image in which a portion of a character is thin.

[0013]

In order to solve this problem, an image processing apparatus according to the present invention comprises: (1) a connected component detecting means for detecting a connected component having the same color as a character line color from original image data; (2) a line width detecting means for detecting a line width of each of the connected components; and (3) a line width corresponding to each of the connected components is compared with a threshold to determine whether each of the connected components is a character portion. Or a character background determining unit that determines whether the character is a background part and outputs a determination result.

[0014]

BEST MODE FOR CARRYING OUT THE INVENTION

(A) First Embodiment Hereinafter, a first embodiment of an image processing apparatus according to the present invention will be described in detail with reference to the drawings.

(A-1) Configuration of First Embodiment FIG. 1 is a functional block diagram of an image processing apparatus according to this embodiment. 1, the image processing apparatus of this embodiment includes an image recording medium 10, an image storage unit 20, a connected component detection unit 30, a connected component storage unit 40, and a line width detection unit 50.
, A character background determination unit 60, a background removal unit 70, and an output image storage unit 90.

The image recording medium 10 stores an image of an original image such as a plurality of documents including characters and line drawings as a binary image.

The image storage unit 20 inputs the binary image to be processed from the image recording medium 10 as original image data and stores it (for example, “0” for white and “1” for black). And outputs it to the connected component detection unit 30 and the background removal unit 70 described later.

The connected component detecting section 30 can handle the character color of the character line, whether black or white, and character color information indicating the character color of the original image at that time and the image storage section 2.
The original image data output from 0 is input, and a connected component having the same color as the character color information is detected from the original image data.

More specifically, the connected component detection unit 30 receives the original image data from the image storage unit 20 and scans the entire original image data. And the same color as the character color information,
A series of pixels having a 4-connected relationship with each other are detected from the original image data as connected components, and output to the connected component storage unit 40 and the background removal unit 70. For example, a pixel having the same color as the target pixel is selected from four neighborhoods of the target pixel (four pixels before, after, right and left of the target pixel), and the same processing is performed on the selected pixel, thereby forming the connected component. To detect.

The connected component storage 40 stores the connected component detector 3
Storing one or more connected components detected by 0;
This is output to the line width detection unit 50.

The line width detecting section 50 includes a connected component storing section 40
, A value (hereinafter, referred to as a line width parameter) corresponding to the line width of each connected component is detected, and the line width parameter is output to the character background determination unit 60.

More specifically, the line width detecting section 50 has an observation window of two rows and two columns consisting of a0 to a3 as shown in FIG. All the pixels constituting the image data for each connected component) are displayed in the observation window a0.
Then, the 2 × 2 observation window is raster-scanned so as to be viewed through the observation windows a0 to a3.

Then, a comparison is made between the 2 × 2 observation window and the pixels on the memory of the connected component storage unit 40 seen from the 2 × 2 observation window, and as a result, the connection observed from the 2 × 2 observation window is obtained. When all the pixels on the memory of the component storage unit 40 have the same color as the character color indicated by the character color information (the character color is assumed to be black in this specific description), the value of the number of black points Q is increased by one. To be counted. When the pixel on the memory of the connected component storage unit 40 viewed from the observation window a0 has the same color as the character color information, the value of the number A of black points is counted up by one.

However, the initial values of the number of black spots Q and the number of black spots A are 0. When scanning image data, 2 × 2
When a part of the observation window is located outside the memory of the connected component storage unit 40, the pixel seen from the observation window is different from the character color.

In the connected component, a value obtained by dividing the number of black points A by a value obtained by subtracting the number of four black points Q from the number of black points A is defined as a line width parameter W as shown in the equation (1).

Line width parameter: W = A / (A−Q) (1) The character background determination unit 60 determines each of the line width parameter values of each connected component detected by the line width detection unit 50 by a predetermined value. By comparing with a threshold value, it is determined whether the connected component is a character data portion or a background data portion.
That is, the character background determination unit 60 stores a threshold value for the line width parameter value of the connected component in advance, and when the line width parameter value of the connected component is equal to or larger than the threshold value, the connected component When the line width parameter value is less than the threshold value, the connected component is determined to be a background data portion.

The background removing unit 70 receives the judgment result output from the character background judging unit 60, and if the connected component output from the connected component detecting unit 30 is a background data part, The connected component portion is removed from the output original image data.

More specifically, the background removing unit 70 determines the original image data from the image storage unit 20, the connected component detecting unit 30 determines the position information of the connected component in the original image data, and the character / background determining unit 60 determines the determination result. If the connected component is a background data portion, the connected component portion in the original image data output from the image storage unit 20 is black-and-white inverted and output to the output image storage unit 90, and the connected component is In the case of a character data portion, the connected component portion of the original image data output from the image storage unit 20 is output to the output image storage unit 90 as it is.

The output image storage section 90 stores the image data output from the background removal section 70 from which the background data portion has been removed.

(A-2) Operation of First Embodiment The image processing apparatus of the first embodiment having the above configuration is
The following operation is performed.

A binary image such as a plurality of documents having characters is stored in the image recording medium 10. The binary image to be processed output from the image recording medium 10 is stored by the image storage unit 20 and output to the connected component detection unit 30 and the background removal unit 70. The connected component detection unit 30 detects a connected component having the same color as the character color information from the input original image data. Each connected component detected by the connected component detection unit 30 is stored in the connected component storage unit 40. For each connected component output from the connected component storage unit 40, the line width detection unit 50 uses the line width parameter W
Is detected.

For example, FIG. 2A, which is an image composed of a black character image and a background image including a black line pattern, is used as an original image.
When input to the image processing apparatus of this embodiment, the value of the line width parameter W of each connected component detected by the line width detection unit 50 is as shown in FIGS. 5A and 5B. .
According to FIG. 5A, the number of connected components in the background data portion of the image shown in FIG. 2A is 56, and the value of the line width parameter W of the connected components is 1.0 to 2.0. Value within the range. On the other hand, according to FIG. 5B, the number of connected components in the character data portion of the image shown in FIG. 2A is 4, and the value of the line width parameter W of the connected components is 3.0 to 3 .
The value is in the range of 6.

The line width parameter value of each connected component output to the character background determination unit 60 is compared with a threshold value internally stored in the character background determination unit 60, and each connected component is determined to be a character data portion or It is determined whether it is a background data part.

For example, when the image shown in FIG. 2A is input to this embodiment, the range of the line width parameter value of the character data portion is 3.0 to 3.6 as described above, and The line width parameter value range of the data part is 1.0
Since the threshold value is set to 2.5 in the character background determination unit 60, for example, 2.5 is set as the threshold value.
It is possible to correctly determine whether each connected component is a character data portion or a background data portion.

When the connected component output from the connected component detecting unit 30 is a background data portion, the background removing unit 70 inverts the connected component portion in the original image data output from the image storage unit 20 into black and white and outputs an output image. Output to the storage unit 90,
When the connected component output from the connected component detection unit 30 is a character data portion, the connected component portion of the original image data output from the image storage unit 20 is used as it is in the output image storage unit 90.
Output to

The image data from which the background data portion has been removed is stored in the output image storage section 90.

FIG. 6A shows an image of the image obtained by removing the background data portion from the black character image of FIG. 2A including the background data portion and extracting only the target characters according to the first embodiment. It is an example.

FIG. 6B shows an image obtained by removing the background data portion from the white character image of FIG. 2B including the background data portion and extracting only the target character according to the first embodiment. This is an example.

(A-3) Effects of the First Embodiment According to the image processing apparatus of the first embodiment, concatenation of the same color as the character color information from the original image data (character image data and background data) By detecting the component, calculating the line width parameter value for each connected component, and comparing it with the threshold value, it is determined whether the connected component is a character data portion or a background data portion. Occurs.

In the original image, even when a part of a character is thin, a line width parameter value having a certain constant value is obtained, and the line width parameter value and the background line pattern, noise line, etc. Can be easily distinguished, and only unnecessary lines such as line patterns and noise lines can be removed without losing thin character lines.

As a result, when performing character recognition or the like, character extraction processing and character recognition processing are performed accurately. In particular, it is effective to apply the image processing apparatus of this embodiment to character recognition of a headline article of a newspaper in which character data and background data are mixed.

(B) Second Embodiment Next, an image processing apparatus according to a second embodiment of the present invention will be described with reference to the drawings.

The second embodiment has a function of automatically determining the color of a character line of an input image in addition to the function of the image processing apparatus in the first embodiment.

(B-1) Configuration of the Second Embodiment FIG. 7 is a functional block diagram of an image processing apparatus according to the second embodiment. 7, in the second embodiment, the image recording medium 10, the image storage unit 20, the connected component detection unit 30, the connected component storage unit 40, and the line width detection unit 50
, A character background determination unit 60, a background removal unit 70, a character color determination unit 82, and an output image storage unit 90.

The difference between the second embodiment and the first embodiment is that, in the first embodiment, character color information is supplied from outside to the connected component detection unit 30 and the line width detection unit 50. However, in the second embodiment, the color of the character line in the original image is determined by the character color determination unit 82, and the determination result is used as character color information as the connected component detection unit 30 and the line width detection unit 5.
0.

Therefore, the image recording medium 10, the image storage unit 20, the connected component detection unit 30, and the connected component storage unit 40
The functions of the line width detection unit 50, the character background determination unit 60, the background removal unit 70, and the output image storage unit 90 have been described in detail in the first embodiment. In the embodiment, those descriptions are omitted.

The character color determination unit 82 receives the original image data output from the image storage unit 20, determines the color of the character line in the original image data, and outputs the determination result as character color information. The character color determination unit 82 can be divided into three components according to the functions of the character color determination unit 82. Hereinafter, each unit constituting the character color determination unit 82 will be described in detail.

FIG. 8 is a functional block diagram of the character color judgment section 82. The character color determination unit 82 includes a black lump calculation unit 84,
It comprises an image contracting section 86 and a determining section 88.
Hereinafter, the above components will be described in detail.

The image shrinking section 86 receives the original image data output from the image storage section 20 as input, holds a memory having an area of the same size as the original image data, and shrinks the original image data by about 8 times. Is created and the contracted image data is output.

The black lump ratio calculating unit 84 receives the original image data output from the image storage unit 20 and the contracted image data output from the image contracting unit 86 as inputs, and outputs the original image data and the contracted image data respectively. Is used to calculate the black lump ratio.

Specifically, as shown in FIG. 9, the black lump calculation unit 84 includes an observation window of three rows and three columns, b1 to b9, and calculates all pixels constituting the original image data and the contracted image data. Observation window b which is the center window of the observation window of the above 3 × 3 mask
5 and 3 × 3 masks b1 to b9 so that the 3 × 3 mask
The observation window of the mask is raster-scanned. At this time, observation window b
When the pixel of No. 5 is black, counting is performed so that the value A2 of the number of black points is increased by one. When all the pixels visible from the observation window of the 3 × 3 mask are black, the counting is performed so that the black block number B is increased by one. However, the initial values of the number of black points A2 and the number of black blocks B are 0.
And

Then, as shown in the equation (2), the black lump ratio KG in the original image data is defined by the ratio of the number of black lump BG of the original image data to the number of black dots A2G. Further, as shown in equation (3), the number of black blocks BS and the number of black points A in the contracted image data are obtained.
The black lump ratio K in the contracted image data is determined by the ratio with 2S.
Define S. However, when the number of black dots A2G of the original image data is 0, the black lump ratio KG of the original image data is 0, and when the number of black dots A2S of the contracted image data is 0, the black lump ratio KS of the contracted image data is 0. And

Black lump ratio of original image data: KG = BG / A2G (2) Black lump ratio of contracted image data: KS = BS / A2S (3) The judgment unit 88 outputs from the black lump ratio calculation unit 84 By comparing the black block rate KG of the original image data and the black block rate KS of the contracted image data, the character color of the original image data is determined, and the determination result is output as character color information.
That is, when the value of the black block rate KG of the original image data is smaller than the value of the black block rate KS of the contracted image data,
The character of the original image data is determined to be black, and when the value of the black lump ratio KG of the original image data is equal to or greater than the value of the black lump ratio KS of the contracted image data, the character of the original image data is determined to be white. It is output to the connected component detection unit 30 and the line width detection unit 50 as character color information.

(B-2) Operation of the Second Embodiment As described above, the difference between the second embodiment and the first embodiment is that in the first embodiment, the difference is given externally. In the second embodiment, the character color information is generated by the character color determination unit 82 included in the image processing apparatus according to the second embodiment.

Therefore, in the second embodiment,
Only the operation of the character color determination unit 82 will be described in detail.

From the original image data output from the image storage unit 20, the contracted image data that has been contracted by about 8 is converted to the image contracted unit 8.
6 is created. The black lump ratio calculating unit 84 counts the number of black points A2G of the original image data and the number of black lump BG of the original image data from the original image data output from the image storage unit 22, and calculates the black lump ratio KG of the original image data. , The number of black points A2G and the number of black blocks BG. Similarly, from the contracted image data output from the image contracting unit 86, the number of black points A2S of the contracted image data and the number of black blocks BS of the contracted image data are counted, and the black block ratio KS of the contracted image data is calculated. The black block rate KG of the original image data and the black block rate KS of the contracted image data output from the black block rate calculation section 84 are determined by the determination section 8.
Then, the colors of the characters constituting the original image data are determined based on the magnitude relation, and the determination result is output as character color information.

For example, when each of FIG. 2 (a) and FIG. 2 (b) is input to the image processing apparatus of the second embodiment as an original image, each of the black lump calculating unit 84 and the determining unit 88 Is performed as follows.

First, when FIG. 2A is used as the original image data, the black lump calculator 84 calculates the number of black points A of the original image data.
2G is counted as 7577, black lump count BG is counted as 1339,
As a result, the black lump ratio KG of the original image data is calculated to be 0.17. The number of black points A2S of the contracted image data is 1339.
Is counted as 530, and the black chunk ratio KS of the contracted image data is calculated as 0.39. Therefore, since the black lump ratio KG of the original image data is smaller than the black lump ratio KS of the contracted image data, the determination unit 88 determines that the original image data is a black character, and outputs the determination result as character color information.

On the other hand, when the original image data is shown in FIG. 2B, the black lump calculating section 84 calculates the number of black points A of the original image data.
2G is counted as 40645, and the number of black lumps BG is counted as 33588. Eventually, the black lumps ratio KG of the original image data is calculated as 0.82. The number of black points A2S of the contracted image data is 335.
88 and the number of black lumps BS are counted as 26097, and the lumps ratio KS of the contracted image data is calculated as 0.77. Therefore,
The black block rate KG of the original image data is the black block rate K of the contracted image data.
Since it is larger than S, the determination unit 88 determines that the original image data is a white character, and outputs the determination result as character color information.

The character color information output from the character color determination unit 82 is input to the connected component detection unit 30 and the line width detection unit 50. Note that the subsequent operation of the second embodiment is not different from the operation of the first embodiment, and thus the description thereof is omitted.

FIG. 6A illustrates the character color of the image shown in FIG. 2A (the character color is black) according to the second embodiment, and is similar to that of the first embodiment. 2A is an example of an image in which the background data portion is removed from the image of FIG. 2A including the background data portion and only the intended characters are extracted. FIG. 6 (b) is the same as FIG.
As in the case of FIG. 2B, the second embodiment makes it possible to use FIG.
This is an example of an image in which the character color of the image is determined (the character color is white), the background data portion is removed, and only the target character is extracted.

(B-3) Effects of the Second Embodiment The image processing apparatus of the second embodiment also detects connected components of the same color as the character color from the original image data, and sets the line width for each connected component. By calculating the parameter value and comparing it with a threshold value, it is determined whether the connected component is a character data portion or a background data portion. A line width parameter value having the calculated value is obtained, and an unnecessary background element such as a background line pattern or a noise line is obtained according to the value of the line width parameter.
Character lines can be easily distinguished, and unnecessary background elements such as line patterns and noise can be removed without losing thin character line portions.

In addition to this, according to the image processing apparatus of the second embodiment, the black block ratio of the original image data and the black block ratio of the contracted image data are calculated, and these two black block ratios are compared. In this way, since it has a function of automatically determining the character color of the original image, even when it is unknown whether a black character image or a white character image is input, it is possible to appropriately separate the character portion from the background portion. It can be carried out.

Therefore, it is effective to apply the image processing apparatus of the second embodiment to character recognition of a newspaper headline article or the like in which whether it is described as a black character image or a white character image is unified.

(C) Other Embodiments In the first and second embodiments, the threshold value required for determining the character data portion and the background data portion by the line width parameter is determined by the character background determination. Although it is supposed to be stored in advance in the section, for example, a method of automatically setting an appropriate threshold value by estimating the character size from information such as the width and height of the original image may be used. good. Further, the user may be able to arbitrarily set the setting from outside.

In the first and second embodiments, when the connected component is the background data portion in the background removing unit,
The connected component is output after being inverted in black and white, but is not limited to this. The output image of the output image storage unit is initialized with a color different from the character color, and the character background determination unit determines the background data part. A method of removing the background data portion by not outputting the connected component from the background removal unit may be used.

Further, in the second embodiment, when the character color of the original image is white, the character color of the final output image (the image stored in the output image storage unit) is white, and the color of the character background portion is black. However, after the character color is determined by the character color determination unit, the entire original image data may be inverted between black and white only when the character color of the original image is white, and the processing after the connected component detection unit may be continued as black characters.

Furthermore, in the second embodiment, when the image is contracted in the image contracting section, the image is contracted by using near-eight contraction. However, the present invention is not limited to this method. . An appropriate contraction method according to the resolution of the image may be used for the contraction of the image. For example, 4-neighbor contraction may be used.

In the second embodiment, the character color determination unit calculates the black lump ratio of the original image data and the black lump ratio of the contracted image data, and compares these two black lump ratios. Either the black character image or the white character image was determined, but the black character ratio of the original image data and the black character ratio of the expanded image data were calculated, and by comparing these two black character ratios, the black character ratio was determined. Either the image or the white character image may be determined. In this case, the criterion for determining the character color based on the magnitude of the black lump ratio of the original image data and the black lump ratio of the expanded image data is opposite to that of the embodiment.

Further, in the second embodiment, the black chunk calculating section defines the black chunk as a set of nine black pixels filling a 3 × 3 mask, but the present invention is not limited to this. Absent. For example, a black block may be defined as a group of black pixels having an appropriate size and shape according to the resolution of an image. Alternatively, the character color may be determined using a white lump. Also in this case, the criterion of the character color based on the size of the white lump ratio of the original image data and the white lump ratio of the contracted image data is opposite to that of the embodiment.

Further, in the first and second embodiments,
Although the image recording medium is used as the input device, the present invention is not limited to this. For example, an image reading device such as a so-called scanner or facsimile may be used as the input device.

Further, in the first and second embodiments, the line width detecting unit detects the line width parameter from the number of black points and the number of four black points. Of course, other possible methods may be used.

Furthermore, in the second embodiment, the character color determination unit calculates the black lump ratio for each of the original image data and the contracted image data of the original image data, and compares the ratios to obtain the original image data. Although the color of the character line of the data is determined, another method that can determine the color of the character line of the original image data may be used.

[0074]

As described above, according to the present invention, a connected component having the same color as a character color is detected from original image data, a line width is calculated for each connected component, and the line width is compared with a threshold value. By doing so, it is determined whether the connected component is a character data portion or a background data portion. Therefore, even in an image in which a part of a character is thin in the original image, the character portion and the background portion can be appropriately separated. .

[Brief description of the drawings]

FIG. 1 is a functional block diagram of a first embodiment.

FIG. 2 is an example of an original image.

FIG. 3 is an image example after a background data portion is removed.

FIG. 4 is a diagram showing an observation window for detecting a black spot.

FIG. 5 shows connected components and corresponding line width parameter values.

FIG. 6 is an image after removing a background data portion.

FIG. 7 is a functional block diagram of the second embodiment.

FIG. 8 is a functional block diagram of a character color determination unit 82;

FIG. 9 is a diagram showing an observation window for detecting a black lump.

[Explanation of symbols]

Reference numeral 30 denotes a connected component detection unit, 50 denotes a line width detection unit, and 60 denotes a character background determination unit.

Claims (4)

[Claims] 1. A connected component detecting means for detecting a connected component having the same color as a character line color from original image data; a line width detecting means for detecting a line width of each of the connected components; Image processing characterized by having a character background determining means for determining whether each of the connected components is a character part or a background part by comparing the determined line width with a threshold value and outputting a determination result apparatus. 2. The image processing apparatus according to claim 1, further comprising a background removing unit that removes only the connected component in the original image data when the connected component is a background according to the determination result of the character background determining unit. An image processing apparatus according to claim 1. 3. The image processing apparatus according to claim 1, further comprising a character color determination unit configured to determine a color of a character line of the original image data.
Or the image processing apparatus according to 2. 4. The character color determining means creates contracted image data from the original image data, and, for each of the original image data and the contracted image data, a pixel having the same color as the character color among all pixels constituting the image data. Is counted as the number of character color pixels, and for each of all pixels constituting the image data, the number of pixels of interest in which all surrounding pixels around the pixel of interest are the same color as the character color is defined as the number of character color blocks. Counting, calculating the value obtained by dividing the number of character color blocks by the number of character color pixels, as a character color block rate, and calculating the color constituting the character by the character color block rate of the original image data and the character color block rate of the contracted image data The image processing apparatus according to claim 3, wherein: