JP2001203900A - Image reader - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a color image reader which can correct the color slippage within the same horizontal scanning line and also to the edge parts of colors which are other than the white and black colors. SOLUTION: When the R, G and B outputs are different in the pixel areas of edge parts, one of R, G and B is selected as a reference color and the maximum value of output of an edge part pixel group, consisting of the edge part pixel of the reference color and the pixels existing before and after this edge part pixel, is extracted. On the basis of the extracted maximum value, the value of two colors are prepared by correction in addition to the value of reference color. Thus, the color slippage is corrected at each edge part to obtain an image, having no conspicuous color slippages and also to correct the color slippage in the horizontal scanning direction.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a color image reading apparatus which is used in a color digital copying machine, a color scanner and the like, and reads a color original image using a three-line color sensor.

[0002]

2. Description of the Related Art Conventionally, in a color image reading apparatus of this kind, when a color original image is read due to a slight shift of each reading position of a three-line color sensor or mechanical vibration, etc. R (red), G (green),
In some cases, color data different from the original color is output at an edge portion of the B (blue) image data, causing a color shift. Therefore, for the purpose of such color misregistration correction, for example, JP-A-6-334814 or JP-A-8-139949
As disclosed in Japanese Patent Application Laid-Open No. 9-98256, a technique for detecting the amount of color misregistration between each of the R, G, and B sensors from image data of a read original to correct color misregistration in the sub-scanning direction is disclosed. As shown in the official gazette, a reference chart is arranged on a portion outside the document to detect a color shift due to a reading position shift of each of the R, G, and B sensors due to a mechanical vibration or the like, and to detect a color shift in the sub-scanning direction. As disclosed in Japanese Patent Application Laid-Open No. 8-107472, a correction technique for correcting color misregistration, detects the conveyance speed of a document, calculates a deviation from a predetermined speed, and keeps the sensor reading speed constant based on the deviation. There has been proposed a technology for correcting color misregistration due to uneven reading speed by variably controlling the position of a reading cylindrical lens. In addition,
As disclosed in Japanese Patent Application Laid-Open No. 10-215381, there is a method in which a monochrome edge portion is detected from a change in sensor output, and three colors are aligned with data of a sensor disposed at the center.

[0003]

However, in the conventional color misregistration correction technique described above, for example, in the techniques disclosed in Japanese Patent Application Laid-Open Nos. 6-334814 and 9-98256, R, G, B Since the color misregistration that changes every moment in the sub-scanning direction during reading due to mechanical vibrations in the sub-scanning direction of the sensor is obtained from the original image or the reference image outside the original, it is corrected. It cannot deal with color shift within one line. In Japanese Patent Application Laid-Open No. 8-139949, each of R, G, B
Since interpolation data and a correlation function are obtained for, the processing becomes complicated, the circuit becomes complicated, and only color shift in the sub-scanning direction is corrected. In the technology disclosed in Japanese Patent Application Laid-Open No. 8-107472, the color shift is corrected based on the amount of change in the reading speed.
There is a problem that color shift in the main scanning direction due to vibration and aberration cannot be corrected. In addition, JP-A-10-2
In the technique disclosed in Japanese Patent No. 15381, the edge of a color other than black and white is not corrected because the sensor detection value does not exceed the threshold value for detecting the black and white edge portion.

The present invention has been made to solve the above-described problems, and can correct a color shift within the same main scanning line caused by a lens aberration or mechanical vibration. It is another object of the present invention to provide a color image reading apparatus capable of correcting color misregistration even for edges of colors other than white and black edges.

[0005]

According to a first aspect of the present invention, there is provided an image reading apparatus for reading an image of a document using a three-line color sensor.
Reading means having a three-line color sensor, and R (red), G (green), and B read by the reading means
An edge portion detecting means for detecting an edge portion of the (blue) image data; and a color misregistration correcting means for performing color misregistration correction processing on a pixel area of the edge portion detected by the edge portion detecting means. The means selects one of R, G, and B colors when the outputs of R, G, and B are different in the pixel region of the edge portion, uses the selected color as a reference color, and determines the edge portion pixel of the reference color and the edge portion. The maximum value of the output of the edge portion pixel group consisting of the pixels before and after the partial pixel is extracted, and correction for aligning the values of the other two colors with the value of the reference color is performed based on the extracted maximum value. .

In the above configuration, if the reading positions of the R, G, and B sensors are slightly shifted, the R, G, and B sensor outputs of the edge pixels of the image data are not equal. Some color data that does not exist appears, causing color misregistration. Therefore, when an image edge portion is detected, the maximum value of the output of the edge portion pixel group including the edge portion pixel of the reference color and the pixels before and after the edge portion pixel is extracted. The color misregistration correction is performed by aligning them. As a result, the color shift at the edge portion where the color shift is most noticeable is corrected, so that an image with less noticeable color shift can be obtained. Further, since the correction is performed using the pixel output, the color shift correction in the main scanning direction can be performed. It becomes possible.

[0007] The invention described in claim 2 is the first invention.
Wherein the color misregistration correction unit amplifies the original value and aligns the original value with the maximum value when the original value of the color to be corrected is smaller than the maximum value of the reference color. The color shift is corrected, and then the corrected value is attenuated according to the amount of amplification.

In the above arrangement, in the case of any color edge other than the white and black edges, if a color shift occurs, the original value of the color to be corrected becomes smaller than the maximum value of the reference color. However, even in such a case, the color misregistration correction can be appropriately performed.

[0009]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, an image reading apparatus according to an embodiment of the present invention will be described with reference to the drawings. FIG. 1 is a perspective view of an upward color image reading apparatus. The color reading apparatus has a platen 1 on which a document such as a book or a file is placed facing upward. Above the platen 1, a three-line color of R (red), G (green), and B (blue) is provided. An imaging camera unit (reading unit) 2 having a sensor and reading an original by scanning the sensor is provided. The imaging camera unit 2 includes a lens for automatic focusing (AF) and the like.
An illumination unit 3 for illuminating the original is provided on the upper rear side of the original table 1.
And a height measuring mirror 4 that captures the shape of the upper end portion of the document is provided behind the document table 1. In order to detect the focus position of the document, output data of the shape of the upper end portion of the document reflected on the height measuring mirror 4 is used. The AF control data (lens position) is calculated from the shape data of the upper end of the document.

When reading a document, a document such as a book or a file to be read is placed on the document table 1 and reading is started by pressing a reading start switch 5 on the document table.
Various settings and changes when reading a document can be performed by operating a setting switch on an operation unit 6 arranged on the illumination unit 3. The three-line color sensor has a G line sensor in the center, and R and B sensors arranged on both sides thereof, and is constituted by a CCD or the like.

Next, edge detection in such an original reading operation of the image reading apparatus, occurrence of color misregistration there, and a method of correcting the same will be described. First, as an example of occurrence of color misregistration, a case of detecting a boundary between white and black (edge portion) in a case where a black line or a black solid is present on a blank sheet of a read document will be described. As means for detecting an edge portion, a first derivative of each pixel output of the G sensor is constantly calculated, and when the value exceeds a certain value, an edge of the G image exists at the pixel portion. Will be. Therefore, the first derivative of the sensor output of R and B corresponding to the pixel is calculated, and if these R and B also exceed a certain value, the portion is a black line or a solid black edge. Therefore, color shift correction is performed.

FIG. 2 shows a graph for calculating the first derivative.
4 shows a typical first derivative matrix. Since there are several other primary differential matrices, it is also possible to select a primary differential matrix that matches the image to be read. Further, by combining a plurality of matrices, it is possible to detect a black line or a solid black edge with higher accuracy.

Next, the mechanism of the occurrence of color misregistration at the detected edge will be described. FIG. 3A shows the output of each of the R, G, and B sensors when the boundary between white and black in the document, that is, the edge portion is read by the three-line sensor.
The illustrated pixel position (arrow) is in the main scanning direction of the sensor.
The output of the nth pixel of the G sensor is denoted as Gn, and similarly, R
The nth pixel output of the sensor and the nth pixel output of the B sensor are Rn,
Expressed as Bn. Originally, if it is a black edge, Gn, Rn, and Bn
Are equal without color shift and should be: At this time, the color of the boundary between white and black becomes gray.

Gn = Rn = Bn (1)

However, actually, the reading positions of the R, G, and B sensors are slightly shifted due to mechanical vibration, lens aberration, and the like, and color shift occurs. When there is a color shift, FIG.
As shown in (b), the outputs of the R, G, and B sensors of the pixel n at the edge portion are not equal, and are expressed by the following equation. At this time, as the color at the boundary between white and black, some color other than gray, which is not originally there, appears.

Gn ≠ Rn ≠ Bn (Equation 2)

Next, a method of correcting a color shift will be described with reference to FIG. FIG. 4 shows a case where the outputs of the R sensor and the B sensor are corrected based on the output of the G sensor. In this method, image data Rn and Bn of an edge portion and image data shifted by less than one pixel from pixels n + 1 and n-1 adjacent to each pixel are created. That is, the distance between the n-1 pixel and the n pixel and the n pixel and the n + 1 pixel of the R sensor is m, respectively.
The equally divided interpolation data is calculated as follows.

(M-1) / m * Rn-1 + 1 / m * Rn (m-2) / m * Rn-1 + 2 / m * Rn... 2 / m * Rn-1 + (m-2) / m * Rn1 / m * Rn-1 + (m-1) / m * Rn Rn (m-1) / m * Rn + 1 / m * Rn + 1 (m-2) / m * Rn + 2 / m * Rn + 1 ·· 2 / m * Rn + (m−2) / m * Rn + 1 1 / m * Rn + (m−1) / m * Rn + 1 (Equation 3)

Similarly, interpolation data obtained by equally dividing the n-th pixel and the n-th pixel and the n-th pixel and the n + 1-th pixel of the B sensor by m are calculated as follows.

(M-1) / m * Bn-1 + 1 / m * Bn (m-2) / m * Bn-1 + 2 / m * Bn... 2 / m * Bn-1 + (m-2) / m * Bn1 / m * Bn-1 + (m-1) / m * Bn Bn (m-1) / m * Bn + 1 / m * Bn + 1 (m-2) / m * Bn + 2 / m * Bn + 1 ·· 2 / m * Bn + (m−2) / m * Bn + 1 1 / m * Bn + (m−1) / m * Bn + 1 (Equation 4)

For example, if m = 8, R interpolation data and B interpolation data are as follows.

Interpolated data of R: RH1 = 7/8 * Rn-1 + ／ * Rn RH2 = 6/8 * Rn-1 + 2/8 * Rn RH3 = 5/8 * Rn-1 + 3/8 * Rn RH4 = 4 / 8 * Rn-1 + 4/8 * Rn RH5 = 3/8 * Rn-1 + 5/8 * Rn RH6 = 2/8 * Rn-1 + 6/8 * Rn RH7 = 1/8 * Rn-1 + 7/8 * Rn RH8 = 0/8 * Rn-1 + 8/8 * Rn = Rn RH9 = 7/8 * Rn + 1/8 * Rn + 1 RH10 = 6/8 * Rn + 2/8 * Rn + 1 RH11 = 5/8 * Rn + 3 / 8 * Rn + 1 RH12 = 4/8 * Rn + 4/8 * Rn + 1 RH13 = 3/8 * Rn + 5/8 * Rn + 1 RH14 = 2/8 * Rn + 6/8 * Rn + 1 RH15 = 1/8 * Rn + 7/8 * Rn + 1 (Equation 5)

Interpolation data of B: BH1 = 7/8 * Bn-1 + 1/8 * Bn BH2 = 6/8 * Bn-1 + 2/8 * Bn BH3 = 5/8 * Bn-1 + 3/8 * Bn BH4 = 4 / 8 * Bn-1 + 4/8 * Bn BH5 = 3/8 * Bn-1 + 5/8 * Bn BH6 = 2/8 * Bn-1 + 6/8 * Bn BH7 = 1/8 * Bn-1 + 7/8 * Bn BH8 = 0/8 * Bn-1 + 8/8 * Bn = Bn BH9 = 7/8 * Bn + 1/8 * Bn + 1 BH10 = 6/8 * Bn + 2/8 * Bn + 1 BH11 = 5/8 * Bn + 3 / 8 * Bn + 1 BH12 = 4/8 * Bn + 4/8 * Bn + 1 BH13 = 3/8 * Bn + 5/8 * Bn + 1 BH14 = 2/8 * Bn + 6/8 * Bn + 1 BH15 = 1/8 * Bn + 7/8 * Bn + 1 (Equation 6)

The above-mentioned interpolation data of R {RH1,..., RH15}
RHi closest to Gn is replaced with a new Rn instead of Rn.
And Similarly, the interpolation data of B {BH1, ..., BH15}
BHi closest to Gn is set as a new Bn instead of Bn. In the example of FIG.

New Rn = RH10 (Equation 7) New Bn = BH5 (Equation 8)

As described above, by bringing the edge data Rn and Bn close to the reference Gn value, when the boundary between white and black, that is, the edge is read by using the three-line color sensor, the vibration of the machine is reduced. And a color shift caused by a lens aberration can be corrected.

In the above, the generation and correction method of the color misregistration at the white and black edge portions has been described. Further, by applying this correction method, the edge where colors other than white and black are in contact is also corrected. It can be performed. Hereinafter, the details will be described with reference to FIGS.

FIG. 5A shows the output of each of the R, G, and B sensors when the red and light blue edges in the original are read by the three-line sensor. At the border between red and light blue,
Edges are detected in all of R, G, and B. In this case, the direction of the change in the RGB sensor output is such that R decreases to the right and GB changes to the right. Also, at the black and white edges,
Although the maximum value in each sensor output group before and after the edge becomes the same value, the maximum value differs for edges of colors other than black and white. Now, assuming that the red and light blue edges are at the center of one pixel (n) of the sensor, if there is no color shift, the outputs of all the R, G, and B of that pixel are: The difference between the output immediately before and immediately after the pixel at the edge is 1 /
2 and all outputs of R, G and B should have the same value.

Rn = Rn−1 + {(Rn + 1) − (Rn−1)} / 2 (Equation 9) Gn = Gn−1 + {(Gn + 1) − (Gn−1)} / 2 (Equation 10) Bn = Bn-1 + {(Bn + 1)-(Bn-1)} / 2 (Equation 11) Rn = Gn = Bn ..... (Equation 12)

However, actually, the reading positions of the R, G, and B sensors are slightly shifted due to mechanical vibration, lens aberration, and the like, and color shift occurs. When there is a color shift, FIG.
, The outputs of the R, G, and B sensors of the pixel n at the edge do not have the same value.

The correction of color misregistration at an arbitrary color edge other than the white and black edge portions will be described with reference to FIGS. First, the maximum value is found in the output of the pixels immediately before and after the edge portion. FIG. 5 (b)
In R, the maximum value among the outputs of the pixels immediately before and immediately after the edge portion of R is Rn-1. Similarly, for G, Gn + 1, B
Is Bn + 1. Here, since the red and light blue edge portions are Rn-1 = Gn + 1 (although each pixel position having a maximum value that becomes equal is immediately before and after the edge portion), the G sensor is used as a reference. Then, for R, the above-described color shift correction method used for the black and white edge can be applied. Therefore, as shown in FIG.
Is represented by the following equation.

New Rn = RH11 (Equation 13)

On the other hand, as for B, since Bn + 1 <Gn + 1 as shown in FIG. 5B, the color misregistration correction method used for the black and white edge cannot be applied as it is. Therefore,
To apply the color shift correction method used for black and white edges,
Bn + 1 needs to be close to Gn + 1. Therefore, the data of the edge portion and the pixels before and after the edge portion are normalized with respect to the reference G. That is, Bn-1, Bn, and Bn + 1 are multiplied by the following coefficient α (amplification in claims). This is shown in FIG.

Α = Gn + 1 / Bn + 1 (Equation 14)

The normalized data of the B output thus obtained with respect to G is represented by (α * Bn-1), (α * Bn), (α * Bn +
1)

Since (α * Bn−1) :( α * Bn) :( α * Bn + 1) = Bn−1: Bn: Bn + 1 (Equation 15), Bn−1, Bn, Bn +1 and the relationship of the output change ratio is not lost. further,

Since (α * Bn + 1) = Gn + 1 (Equation 16), a color shift correction method for white and black edges can be applied.

The normalized data of B, (α * Bn-1), (α
* Bn), (α * Bn + 1), and Gn, a new α * Bn is obtained by using a color shift correction method for white and black edges.
In FIG. 7, it is as follows.

New (α * Bn) = BH12 (Equation 17)

Finally, (new α * Bn) is multiplied by α for normalization at the time of creating the interpolation data. Therefore, the actual correction value is multiplied by 1 / α (attenuation in claims). , A new Bn can be obtained. That is, the new Bn is represented by the following equation.

New Bn = BH12 / α (Equation 18)

In the above-described color misregistration correction method, since the correction is performed only by the output of the pixel, it is possible to cope with the color misregistration in either the main scanning direction or the sub scanning direction of the line sensor. Therefore, when calculating the first derivative, the direction of the line and the boundary between the colors, that is, the direction of the edge portion, are also obtained at the same time, and the direction of the color shift correction is determined from the result, and the color shift correction processing is performed in that direction. , A better image can be obtained.

FIG. 8 shows a block diagram of the control for performing the color misregistration correction processing. CCD1 that forms a line sensor
1 scans a document and generates an electric signal having image information of the document. This signal is converted into a digital signal by the A / D converter 12, and the line correction 13 and the shading correction 1 are performed.
4, image processing is performed in edge determination 15, edge direction determination 16, and color misregistration correction 17. Edge discrimination 15
, An edge is detected from a signal change, and color shift correction 1
At 7, color shift correction processing is performed.

FIG. 9 is a flowchart of the above-described edge detection and color misregistration correction processing. In this example, when an edge is detected from at least two or more sensor outputs, color shift correction is performed, and the output of G is used as a reference. First, for edge discrimination, an edge discrimination matrix is applied to one pixel output of R to check whether there is an edge in R (# 1). If there is no edge in R, it is checked whether there is an edge in G (# 2). If there is no edge, the edge discrimination matrix is moved by one pixel (# 3), and it is checked whether it is the end of the image (# 4). If not, the process returns to # 1. In # 1, if R has an edge,
It is determined whether or not G has an edge (# 5). If G has no edge, it is determined whether or not B has an edge (# 6). If none has an edge, the process proceeds to # 3. If there is an edge in G in # 2, it is checked whether there is an edge in B (# 7), and if there is no edge in B, the process proceeds to # 3.

If there is an edge in # 5, # 6, and # 7,
Since there are edges in two or more of R, G, and B, the color shift correction processing (# 8 to # 11, # 12 to # 15, # 16 to #
19) is performed. In steps # 8 to # 11, the maximum value of the G sensor output of the edge portion pixel and the maximum value of the G sensor output group around the edge portion including the G sensor outputs before and after the G sensor output are set to the maximum value of the R sensor output group at the same position as the G sensor group. After the R sensor output group is normalized by multiplying the R sensor output group by the normalization coefficient α for matching the values, the R pixels on both sides of the R pixel at the same position as the G sensor of the edge portion pixel are used as a reference. Using the output after the normalization, a plurality of R values shifted in a range of less than one pixel are calculated, an optimum R data is found from among the plurality of values, and the data is further divided by 1 / Multiply by α to obtain a new R sensor output of the edge portion pixel.

In steps # 16 to # 19, the G sensor output of the edge portion pixel and the maximum value of the G sensor output group around the edge portion composed of the G sensor outputs before and after the pixel are set to the same position as the G sensor group. By multiplying the B sensor output group by a normalization coefficient β for making the maximum value of the B sensor output group
After normalizing the B sensor output group, the B pixel at the same position as the G sensor of the edge portion pixel is used as a reference and the B pixel on both sides thereof is shifted by less than 1 pixel using the normalized output. A plurality of values of B are calculated, the optimum B data is found from the plurality of values, and the data is further divided by 1 / β
Multiply by 2 to obtain a new B sensor output of the edge portion pixel.

In steps # 12 to # 15, if there is no edge in the output of the G sensor and there are edges in both the R and B sensors, the output of the R sensor of the edge portion pixel and the R before and after the pixel are detected.
By multiplying the B sensor output group by a normalization coefficient γ for matching the maximum value of the B sensor output group at the same position as the R sensor group with the maximum value of the R sensor output group around the edge portion composed of the sensor output After normalizing the output group, B is shifted within a range of less than one pixel using the normalized output of the B pixels on both sides of the B pixel at the same position as the R sensor of the edge portion pixel. Are calculated, the optimum B data is found from among the plurality of values, and the data is further multiplied by 1 / γ to obtain a new B sensor output of the edge portion pixel.

FIG. 10 illustrates the effect of the color misregistration correction processing at the color edge from red to light blue. In the figure, the outputs of Rn-1 and Gn + 1 exceed the white threshold value, but the original data of B: Bn + 1 does not exceed the white threshold value. No correction is applied. On the other hand, in the present invention, as is apparent from the above, even if the output of B does not exceed the threshold value, the peak value of B: Bn + 1 is matched with the peak value of G: Gn + 1.
The coefficient β of β = Gn + 1 / Bn + 1 is multiplied by Bn−1 and Bn + 1, and
B'n-1 = β * Bn-1, B'n = β * Bn, B'n + 1 = β *
Bn + 1 is calculated, the peak values of Gn + 1 and Bn + 1 are combined (increment by normalization), and then a correction value candidate group of B'n: [BH
i] is calculated, and a correction value candidate: BHk closest to Gn is determined from the correction value candidate group, and finally, after correction Bn = BHk /
By resetting the normalization to β, it is possible to correct even the color edge.

The present invention is not limited to the configuration of the above embodiment, and various modifications are possible. For example, in the above-described embodiment, as an example other than the black and white edge, the change of each sensor output of RGB at the edge portion of red to light blue has been described, but the color misregistration correction processing is similarly applied to the edge of other colors. It is possible to

[0050]

As described above, according to the first aspect of the present invention, an edge portion of an image is detected from the image data read by the line sensor, and each edge is detected from the reference pixel data and the adjacent pixel data. It is a simple processing method that corrects the deviation of the reading position of the line sensor to prevent color deviation, but it can also correct the color deviation due to machine vibration and lens aberration, and also the main scanning of the line sensor. It is also possible to correct a color shift in the same line.

Further, according to the second aspect of the present invention, it is possible to correct color misregistration at edges of other colors and fine lines other than monochrome edges and fine lines.

[Brief description of the drawings]

FIG. 1 is a perspective view of an image reading apparatus according to an embodiment of the present invention.

FIG. 2 is a view showing a first-order differential matrix used for edge detection in the apparatus.

3A is a diagram illustrating outputs of R, G, and B sensors when there is no color misregistration at a monochrome edge portion, and FIG. 3B is a diagram illustrating an output when there is a color misregistration.

FIG. 4 is a diagram showing a color misregistration correction method.

5A is a diagram showing the output of each of the R, G, B sensors when there is no color misregistration at the red and light blue edges, and FIG. 5B is a diagram showing the output when there is a color misregistration;

FIG. 6 is a diagram showing a method for correcting a color shift of an R sensor.

FIG. 7 is a diagram showing a color shift correction method of a B sensor.

FIG. 8 is a diagram showing a control block for performing a color misregistration correction process.

FIG. 9 is a flowchart of a color misregistration correction process.

FIG. 10 is a view showing the effect of color misregistration correction at a color edge.

[Explanation of symbols]

 2 imaging camera section (reading means) 11 CCD (line color sensor) 15 edge determination (edge detection means) 17 color shift correction (color shift correction means)

 ────────────────────────────────────────────────── ─── Continuing on the front page (72) Inventor: Yasutaka Shintaku 88, Kuzedonojo-cho, Minami-ku, Kyoto Asahi Kosei Co., Ltd. F-term (reference) 5B047 AA01 AB04 BB03 DA01 DC11 5C055 BA06 EA05 HA36 HA37 5C072 AA01 BA19 EA05 QA17 UA05 UA18 5C077 LL19 MP07 MP08 NP01 PP32 PP39 PP43 PP47 PQ03 RR19 5C079 HB01 JA23 LA02 LA24 LA28 NA02 NA03

Claims (2)

[Claims] 1. An image reading apparatus for reading an image of a document using a three-line color sensor, comprising: reading means having the three-line color sensor; and R (red), G (green), and B read by the reading means. An edge portion detecting unit for detecting an edge portion of the (blue) image data; and a color misregistration correcting unit for performing a color misregistration correction process on a pixel region of the edge portion detected by the edge portion detecting device. The means selects one of R, G, and B colors when the outputs of R, G, and B are different in the pixel region of the edge portion, uses the selected color as a reference color, and determines the edge portion pixel of the reference color and the edge portion. Extracting the maximum value of the output of an edge pixel group consisting of pixels before and after the partial pixel, and performing correction for aligning the values of the other two colors with the value of the reference color based on the extracted maximum value. Image Reader. 2. The color misregistration correction unit according to claim 1, wherein an original value of the color to be corrected is smaller than a maximum value of the reference color.
2. The image reading apparatus according to claim 1, wherein the original value is amplified to perform color misregistration correction to the maximum value, and thereafter, the corrected value is attenuated in accordance with the amount of amplification.