JP4828060B2 - Image reading apparatus, correction apparatus, and correction method - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to an image reading apparatus that reads an original while relatively moving the original and reading means, and in particular, an image reading apparatus that acquires image data by simultaneously reading an original with an image sensor having a plurality of sensor lines such as three primary colors. The present invention belongs to a technique for detecting and correcting, for example, color blur and color misregistration caused by inconsistencies between sensor lines of a read image resulting from document surface fluctuations accompanying document transport.
[0002]
[Prior art]
As a single image sensor, there is a so-called three-line color sensor having three sensor lines and three primary color filters mounted on the photoelectric conversion element surface of each sensor line. This 3-line color sensor can not only form a compact color scanner (color image reading device) compared to the case where three image sensors are used, but also so-called light source switching that reads a color document while sequentially turning on the three primary color light sources. Compared to the method, it has the advantage of being able to be made faster.
[0003]
There are two typical types of scanners using such a three-line color sensor. One is a so-called original fixing type scanner in which an original is fixedly placed on a glass surface and read while moving the reading optical system, and the second is a reading optical system while the original is conveyed and moved by a conveying means. This is a sheet-through scanner that reads a fixed image.
[0004]
A sheet-through type scanner that feeds and conveys originals one by one and scans them as they are will improve stability due to undulations and fluttering of the original, or the impact of the pressing point of the conveyance roller used in the conveyance system separating from the original. Variation occurs on the missing image.
[0005]
As the sheet-through type, in addition to the sheet-through scanner configured integrally with the conveyance portion, an auto document feeder (hereinafter referred to as ADF) that separates and sends one by one because a large number of documents are placed on the document-fixing type scanner. There are types that are placed and read. Recently, in order to meet the need to read both sides of a document, a configuration in which a reading unit is included in the ADF and both sides are read by the scanner and the ADF has appeared.
[0006]
In addition, a fixed document type scanner that reads a document with a fixed document tends to be considered to be able to read the document stably without unstable operation. In a scanner with a high reading resolution, even in the original fixed type, the mirror carriage that forms the reading line and the optical carriage that integrally forms the optical system are moved or vibrated in the sub-scanning direction, which is the moving direction. Also, the fluctuation in the reading width direction appears in the image as the same behavior as the wave of the original.
[0007]
[Problems to be solved by the invention]
As an example for solving the color blur or color misregistration that occurs when an image reading apparatus such as an ADF with a reading unit or a scanner is configured using such a three-line color sensor, for example, disclosed in JP-A-7-143281. There is a method of using a prism. It is configured such that light from the same reading line reaches each sensor line position of the 3-line color sensor by color separation.
[0008]
However, the adjustment of this arrival position requires extremely high accuracy, and the yield of the production process is low, and the image acquisition cannot be performed due to the positional deviation caused by vibration after shipment. There is. In view of such problems, the present invention acquires features from the image data of the three primary color lines acquired in the conventional configuration without using a particularly complicated and special configuration, and solves color shift and color blur on the image. The task is to do.
[0009]
[Means for solving problems]
In order to solve this problem, a first aspect of the present invention includes a light source that irradiates a document, an optical unit that collects light from the document irradiated by the light source, and light received from the optical unit. An image sensor that generates an image signal for each color component; and a reading unit that forms and reads a reading line for each color component on a document surface with sensor lines in which a plurality of linearly arranged photoelectric conversion elements of the image sensor are used as pixels. A moving means for relatively moving the original and the reading line in a sub-scanning direction orthogonal to the main scanning direction as a reading line of the reading means, and converting the level of the image signal from the image sensor as a digital gradation value An A / D converter, an image memory for storing image data for each color component acquired by the plurality of sensor lines, and a color component stored in the image memory. An image reading apparatus including a color misregistration correction unit that performs correction based on image data of the first line, wherein the color misregistration correction unit includes image data of the first line read by the reading unit, and a document by the moving unit. Difference calculation means for calculating a difference of gradation values between two pixels corresponding to the main scanning direction between the image data of the second line read by relative movement and the difference calculation of the difference calculation means. It is determined that the difference value is large based on the result of this comparison means, the repetition means that repeats for each reading line in the sub-scanning direction, the comparison means that compares the difference calculation value of the difference calculation means with a predetermined value. Contrast judgment meansWhen,Calculating means for calculating a displacement amount of a pixel position between the image data of two colors based on a result of executing the difference calculation of the calculating means for at least two color image data of color components;WithThe calculating means determines the position of the first pixel determined by the contrast means to be large for the first color of the two colors of the color component, and the contrast means for the second color. Based on the position of the second pixel determined to have a large difference value, the amount of pixel position deviation between the two color image data is calculated.And a reading means for reading out data for each pixel from an image memory holding image data for each color component acquired by the plurality of sensor lines, and a pixel position calculated by the calculating means for the data read out by the reading means An image reading apparatus comprising: writing means for changing the position in accordance with the amount of deviation and writing to the image memory(Claim 1).
[0010]
  As a second aspect, the image reading apparatus includes a light source that irradiates a document, an optical unit that collects light from the document irradiated by the light source, and an image for each color component that receives light from the optical unit. An image sensor that generates a signal, a reading unit that forms a reading line for each color component on a document surface using a sensor line in which a plurality of linearly arranged photoelectric conversion elements of the image sensor are used as pixels, and a reading unit of the reading unit A moving means for relatively moving the original and the reading line in the sub-scanning direction orthogonal to the main scanning direction as a reading line, and an A / D converter for converting the level of the image signal from the image sensor as a digital gradation value An image memory for holding image data for each color component acquired by the plurality of sensor lines, and an image for each color component held in the image memory. An image reading apparatus including a color misregistration correction unit configured to correct based on data, wherein the color misregistration correction unit is configured to detect image data of a first line read by the reading unit and a document relative to the moving unit. Subtraction of the difference calculation means for calculating the difference of the gradation value between the two pixels corresponding to the main scanning direction and the difference calculation of the difference calculation means between the image data of the second line moved and read. Repetitive means for repeating each direction reading line, comparison means for comparing the difference calculated value of the difference calculation means with a predetermined value, and contrast determination for determining that the difference value is large based on the result of the comparison means meansWhen,Calculating means for calculating a displacement amount of a pixel position between the image data of two colors based on a result of executing the difference calculation of the calculating means for at least two color image data of color components;WithThe calculating means determines the position of the first pixel determined by the contrast means to be large for the first color of the two colors of the color component, and the contrast means for the second color. And calculating the shift amount of the pixel position between the image data of the two colors based on the position of the second pixel determined to have a large difference valueAnd a reading control means for reading out the pixel data from the image memory while controlling the position of reading out the pixel data based on the result of the calculating means.The image reading apparatus is characterized in that (Claim 2).
[0011]
  As a third aspect, the image reading apparatus transfers the image data read from the image memory controlled by the reading control means to an interface means for communicating with a host device.Claim 2An image reading apparatus (claim 3).
[0012]
  As a fourth aspect, the color misregistration correction unit performs correction of deviation by the reading correction unit after the distance correction between sensor lines by the inter-line correction unit.Claims 1 and 2The image reading apparatus described in 1. (Claim 4).
[0013]
  As a fifth aspect, the color misregistration correction unit performs correction of deviation by the reading correction unit after the distance correction between sensor lines by the inter-line correction unit.
Claim4(Claim 5).
[0014]
  As a sixth aspect, an image that obtains a gradation value for each pixel by moving relative to the document in a sub-scanning direction orthogonal to the main scanning direction in which the document is read by a color sensor having a sensor line in which pixels are arranged for each color component. A correction device that is directly or indirectly connected to a reading device by an interface means and corrects a pixel shift of received image data. The correction device performs main scanning on at least a part of the received image data for each color component. Between the first main scanning image data stored in the image memory and the second main scanning image data different in the sub scanning direction. A difference calculation means for calculating a difference in gradation value between two pixels corresponding to the above, and a repetition process for repeating the difference calculation of the difference calculation means for each reading line in the sub-scanning direction. When the difference calculation of the difference calculation means based on the result of relative two-color image data of color components, and calculating means for calculating a shift amount of the pixel positions between image data of two colors,Reading means for reading out data for each pixel from the image memory; and writing means for changing the position of the data read out by the reading means in accordance with the amount of deviation of the pixel position calculated by the calculating means and writing it into the image memory ,(6).
[0015]
  As a seventh aspect, an image that obtains a gradation value for each pixel by moving relative to the document in a sub-scanning direction orthogonal to the main scanning direction in which the document is read by a color sensor having a sensor line in which pixels are arranged for each color component. A correction device that is directly or indirectly connected to a reading device by an interface means and corrects a pixel shift of received image data. The correction device performs main scanning on at least a part of the received image data for each color component. Between the first main scanning image data stored in the image memory and the second main scanning image data different in the sub scanning direction. A difference calculation means for calculating a difference in gradation value between two pixels corresponding to the above, and a repetition process for repeating the difference calculation of the difference calculation means for each reading line in the sub-scanning direction. And a calculation means for calculating a shift amount of the pixel position between the two color image data based on the result of the difference calculation performed by the difference calculation means on the two color image data of the color component, and the pixel position Means for calculating the amount of deviationAnd reading based on the result of the calculation means while controlling the position where the pixel data from the image memory is read out, and controlling the reading control means to deliver to the interface means Correction device characterized by(Claim 7).
[0016]
  First8As an aspect of the present invention, the correction device reads out while controlling the position of reading out the pixel data from the image memory based on the result of the calculating means, and controls the interface by controlling the reading control means. The correction device according to claim 7, wherein the correction device is passed to the means.8).
[0017]
  As a ninth aspect, a reading process of forming and reading a reading line for each color component on the document surface with a sensor line in which a plurality of linearly arranged photoelectric conversion elements of the image sensor are used as pixels,
  A moving step of relatively moving the original and the reading line in the sub-scanning direction orthogonal to the main scanning direction as the reading line;
  An A / D conversion step of converting the level of the image signal from the image sensor as a digital gradation value;
  A holding step of holding image data for each color component acquired by the plurality of sensor lines in an image memory that stores the image data for each pixel according to the position of main scanning and sub-scanning;
The first read by the reading stepMain scanDifferent from the image data in the sub-scanning direction.Main scanA difference calculation step for calculating a difference in gradation value between two pixels corresponding to the main scanning direction between the image data of
Repeating means for repeating the difference calculation of the difference calculation means for each reading line in the sub-scanning direction;
A comparison step of comparing the difference calculation value of the difference calculation step with a predetermined value;
A contrast determination step for determining that the difference value is large based on the result of the comparison step;
A calculation step of calculating a shift amount of the pixel position between the image data of the two colors based on a result of executing the difference calculation of the calculation step on the image data of at least two colors of the color component;
In the calculation step, the position of the first pixel determined by the contrast means to be large for the first color out of the two colors of the color component and the contrast means for the second color. A calculation step of calculating a shift amount of the pixel position between the image data of the two colors based on the position of the second pixel determined to have a large difference value;
  A reading process of reading out data for each pixel from the image memory;
  A writing step of changing the position of the data read by the reading means according to the amount of deviation of the pixel position calculated by the calculating means and writing it into the image memory;(Claim 9).
[0022]
DETAILED DESCRIPTION OF THE INVENTION
Embodiments of an image reading apparatus with color misregistration correction to which the present invention can be applied will be described below with reference to the drawings.
[0023]
(Outline of color misregistration)
FIG. 1 is a diagram illustrating a conventional sheet-through color scanner that reads a color document while feeding a CCD having three lines. The scanner 1 is equipped with a three-line color CCD, reads a color while conveying a document, and transfers it to the computer 2. The scanner 1 incorporates a conventional 3-line CCD (not shown). The three primary color reading lines are each composed of 8 lines.
[0024]
Therefore, the scanner 1 reads three lines at different distant positions at the same time, temporarily places the read data on the memory, superimposes two colors after reading eight lines, and then reads the remaining when eight lines are read. The method of overlapping colors is used. The difference in position between the lines is always constant, and images can be superimposed without requiring any particular determination. This process is performed in the scanner 1. The computer 2 is used by displaying the received image on the display 3 or outputting it to the printer 4 for printing.
[0025]
Although not shown, it is possible to adopt a configuration in which the direction of the sensor line of the image sensor and the reading line for reading the document do not coincide with each other. If the optical path is changed using a mirror or the like, the installation direction of the image sensor can be arbitrarily selected. .
[0026]
FIG. 2 is also a conventional example in which an ADF 11 that separates and conveys an original is mounted on a flatbed scanner 10 that reads an original placed on a glass platen. For the sake of explanation, the three reading lines are extremely separated, but are separated by only 8 lines as in the case of FIG.
[0027]
A plurality of document bundles are placed on the sheet feed tray 14 upward. When the conveyance is started, first, the uppermost document is picked up by the pickup roller 16. The picked-up document is fed to the pre-reading roller 18 by the feeding roller 17. The document further fed between the back platen 25 and the reading window 13 by the pre-reading roller 18 is sent to the paper discharge roller. Further, the paper is discharged onto the paper discharge tray 15 by the paper discharge roller 19. The surface of the back platen 25 is white. Accordingly, in the image data read by the sensor 22 of the carriage 12, when there is no original, an extra read area such as the periphery of the original appears white.
[0028]
On the other hand, in the flat bed scanner 10, there is an optical carriage 12 that can move on a rail 30. The optical carriage 12 includes a three-line color CCD 22 (hereinafter referred to as an image sensor 22. This image sensor may be a MOS type or another type instead of a CCD), a lens 21, a mirror 20, and a light source 24. It has been. When reading a fixed document placed on the glass platen 26 with the document surface down, the carriage 12 reads the document surface while moving in the SW direction. When the entire original is read, the optical carriage 12 returns to the BK direction, and is positioned and stopped based on a home position sensor (not shown).
[0029]
On the other hand, when reading a document sent by the ADF 11 described above, the optical carriage 12 moves to the lower surface of the reading window 13 and stands by according to the reading window 13. When a signal is received by a position sensor near the pre-reading roller (not shown), the image processing and reading processing circuit is activated to read. In the example of FIG. 2, driving means such as a pulse motor omitted in the ADF 11 is provided, and conveyance control is performed in synchronization with the operation of the flat bed scanner 10.
[0030]
(Explanation of misalignment during document movement)
3 (a), (b), and (c) are enlarged so that the state that the document conveyed by the ADF 11 passes through the reading window 13 and is read by the image sensor 22 can be conceptually easily understood. ). FIG. 3A shows a state in which the rollers 18a and 18b of the roller pair 18 are held before the reading position and the rollers 19a and 19b of the roller pair 19 are held and conveyed after the reading position. Even in this state, the original document surface fluctuates and vibrations are applied. However, since the conveyance is more stable than the following (b) and (c), no large distortion appears on the image.
[0031]
On the other hand, FIG. 3B shows the moment when the trailing edge of the document is separated from the roller pair 18 before reading. The three sensors of the RGB three primary colors included in the image sensor 22 are not equidistant from the reading window 13 as in the state of FIG. 3A, but are inclined and changed in distance in FIG. Further, FIG. 3 (c) shows a state where a bounce occurs from the state of (b). At this time, the distance is different from (b). Here, the distance is the distance between the document positions read by the three RGB lines.
[0032]
In these FIGS. 2, 3A, 3B, and 3C used for the explanation, the reading lines of the three-color RGB are clearly separated for easy understanding. In practice, for example, when a scanner capable of reading an A4 size document having a reading resolution of 600 dpi is used, there are about 6000 pixels because the width of A4 is about 10 inches. The pixel-to-pixel pitch is 1/600 inch, that is, 42.5 μm.
[0033]
Therefore, in order to read at the same magnification in the main scanning direction of the reading line where the image sensor 22 reads the document and the sub-scanning direction in which the reading line is swept (scanned) in the length direction of the document, this sub-scanning is performed. The pitch in the direction must also be inscribed at 600 dpi (42.5 μm spacing). Therefore, it is not a large interval as shown in this figure, and even if the interval is 8 lines, it is a little less than 340 μm, which is an interval that cannot be seen with eyes. However, on the contrary, in order to read with a fine feed pitch, when a document whose position is shaken due to slight vibrations is read, there is sometimes a deviation of several lines.
[0034]
(Explanation of misalignment caused by carriage movement)
Next, FIGS. 4A and 4B are diagrams showing that a deviation may occur even when the flatbed scanner 10 reads the document while fixing the document. As in FIG. 3, the reading lines are extremely separated and the carriage 12 is extremely inclined, but this is emphasized for easy understanding. Actually, since the optical carriage 12 travels on the rail 30, the inclination is not so much. Similar to the description already given, in FIG. 4B, the lengths of AB and BC are different from the lengths of A'B 'and B'C'.
[0035]
When the carriage 12 reads at a horizontal fixed position, the three lines R, G, and B read the three positions A, B, and C. However, if the carriage 12 is tilted as shown by a solid line at a different time, A ', B', and C 'are read. The distance A-B is different from the distance A'-B '. Therefore, if it is tilted from the horizontal with 3 lines, the horizontal R and B colors are designed to overlap so that the R and B colors are shifted without overlapping after 3 lines. It is.
[0036]
As described above, a situation in which the overlapping of the three colors is lost due to the document rampage, vibration, or the reading carriage rampage occurs, resulting in color misregistration on the image. By the way, each line of the three primary colors of the image sensor 22 is arranged at a predetermined interval (in this case, every 8 lines), so that when this interval is corrected, a color image is realized. This is generally referred to as inter-line correction, but what occurs separately from the amount of inter-line correction is the color shift described here.
[0037]
Next, although illustration is omitted, the image when the three lines do not overlap when the uniformity of the line spacing is lost will be described in a little more detail. For example, when a pattern called a so-called pair line in which white and black are drawn as straight lines at equal intervals is acquired, even if the image on the document surface is a black and white image, What is output from the image sensor is color-separated image data, and the three primary colors are observed with varying intervals. Therefore, even if these three primary color image data are superimposed, the original black and white image is not reproduced, and the original boundary between white and black is expressed with color blurring. If the shift amount is large, it is observed as a color shift.
[0038]
However, it should be noted here that this phenomenon stands out at the boundary between the black and white areas. In the vicinity of a color region such as an intermediate color, even if there is a color shift, it may be difficult to understand whether it is such a color or a color shift. Therefore, the color shift can be most conspicuous at the boundary position of the achromatic region. That is to say, conversely, it is possible to perform division such that the color misregistration correction is omitted when it is determined that the region is not an achromatic region.
[0039]
Further, considering the density value itself, human visual characteristics tend to be saturated in the darker printed image (higher density, lower pixel value). In particular, it is easy to confirm in a bright area from the viewpoint of deviation of the three primary colors, but it is almost black in a dark area. On the basis of this, when the density value is low to some extent, it is possible to divide the color misregistration correction.
[0040]
Therefore, as described above, the data of each color is temporarily stored, and the contrast is calculated for the same pixel between the preceding and succeeding lines in each color. Further, by performing the above-mentioned division, it is possible to improve if the system employs a CPU that is repetitive and burdens processing, or if the system places importance on the processing speed of the scanner.
[0041]
(Outline explanation for simultaneous scanning on both sides)
Next, the scanner 50 provided with the color misregistration correction means and the ADF 90 with a reading function will be described with reference to FIGS. For convenience, the reading direction of the image sensor is referred to as main scanning. However, as described above, depending on the design of the optical system, the scanning direction of the photoelectric conversion element of the image sensor and the direction of the line for reading the document are different. Therefore, the direction of the document reading line is defined as main scanning, and the sub-scanning direction, which is the relative movement direction of the document and the reading line, is orthogonal to the main scanning direction.
[0042]
(Description of processing on the ADF side)
A document placed on the sheet feeding tray 67 with the upper surface as a reading surface is taken in by the pickup roller 52 and passes by the sheet feeding roller 53 from the pre-reading roller 54 below the contact image sensor 60 (hereinafter, CIS 60). After passing the CIS 60, the document passes under the back platen 65. Further, the document is sent out onto a paper discharge tray 66 by a paper discharge roller 55. At this time, the optical carriage 57 moving on the rails of the scanner 50 is in a readable state by moving the reading position below the back platen 65.
[0043]
The optical carriage 56 includes an image sensor 57 that is a 3-line CCD, a lens 58, a mirror 59, and a light source 69. The CIS 60 includes a light source (not shown), a SELFOC lens group 62, and a substrate 61 provided with three equal-magnification sensor arrays each including a color filter for three primary colors.
[0044]
The image data for the three primary colors extracted from the CIS 60 is sent to the ADF image processing board 70. On the ADF image processing substrate 70, color misregistration correction means, which will be described later, is placed. Similarly, the image data for the three primary colors extracted from the image sensor 57 is sent to the scanner image processing substrate 80. The scanner image processing substrate 80 is also provided with color misregistration correction means described later. Reference numeral 82 denotes a SCSI connector for interfacing with a computer.
[0045]
Next, the circuit configuration will be described with reference to FIG.
[0046]
First, image data for each color is input from the CIS 60 to the ADF processing substrate 70. Data digitized for each color by the A / D converter 110 is input to the shading correction means 111 and subjected to shading correction for each color. Data from the shading correction unit 111 is input to the inter-line correction unit 112, and superposition is performed every 8 lines using a work memory (not shown). The data for which the reading position has been corrected is stored in the page memory 113.
[0047]
(Description of color misregistration correction on the ADF side)
Since the detailed description using the flowchart will be described as a detailed description after the following description, the outline will be described based on FIG. 6 which is a block diagram. The image data of the three primary colors stored in the page memory 113 is managed by the CPU 119 and the address control 115 checks the value for each pixel from the first line of reading to the final line of reading, The contrast calculation unit 114 (contrast detection means) calculates the density gradient (pixel value difference value) of the pixels at the same position.
[0048]
Of course, even if adjacent pixels are used instead of the same position, substantially the same effect is obtained. As the distance increases, malfunctions occur in the main scanning direction. Therefore, the same position is most preferable. However, since the surrounding conditions to be determined shown in FIG. It is saved as deviation correction. After the calculation result of the contrast calculation unit 114 is determined by comparing with a predetermined value, it is classified into 0 and 1 by a binarizing unit (not shown), and then temporarily placed in the work memory in the region saturation determination unit 118. Is done.
[0049]
In parallel with this processing, the saturation is calculated from the determination of the density level together with the target pixel on which the contrast calculation (difference calculation) is performed and the surrounding values. Specifically, if all three colors have a density value (pixel value) greater than or equal to a predetermined value, it is regarded as an achromatic region, but it is not necessary to make a determination with the same value. Rather, color features may be used. possible. Then, the determination result in the region saturation determination unit 118 and the result of the contrast calculation unit 114 temporarily placed and stored in the work memory in the region saturation determination unit 118 are used to determine whether a color misregistration occurrence point is detected. The CPU 119 makes a determination and writes a flag in the work memory inside the memory shift control unit 117.
[0050]
If it is determined that the point is a color misregistration point, 1 is set, and if not, 0 is set. This operation is repeated for all pages of three colors, that is, one page for each of the three colors. Of course, the result of the contrast calculation unit 114 placed and stored in the work memory in the area saturation determination unit 118 may be written in the internal memory of the memory shift control unit 117, or a dedicated memory may be installed. Which memory is used for saving and calculation is the same and is simply a design change.
[0051]
The memory shift control unit 117 calculates the shift amount with reference to the flag recorded as a result of the contrast calculation unit 114 in the internal memory. The density value (pixel value) of each pixel used in this calculation is executed while being read from or written to the temporarily stored work memory or page memory 113. This control is performed by the CPU 119 and the address control unit. 115 and jointly manage.
[0052]
For example, FIG. 7 shows an example of the three primary color density data (pixel values) in a shifted state.
[0053]
For convenience, the part from K1 to K5 will be called the part S1, the part from L1 to L5 will be called the part S2, and the part from M1 to M5 will be called the part S3. Looking at the portion S1, the contrast is high for R at the line K1. That is, the density level (pixel value) changes greatly between adjacent lines. “Large” is a numerical value that should be considered for design because it affects the detection sensitivity, but is about 8 to 16. Therefore, the first level determination as a result of the contrast calculation unit 114 at this time is 1. Similarly, the contrast of G increases at the K3 line, and B increases at K5.
[0054]
In the situation shown in this figure, since the inherent deviation of the image sensor has already been corrected by the interline correction means 112, if there is no inherent color deviation, for example, at the position where R of K1 rises, G and B must also rise. Don't be. If it is projected on the display as it is, the color will appear to be shifted. Therefore, R and G are shifted by two lines. Similarly, G and B are shifted by two lines.
[0055]
Therefore, the G pixel data located at K3 is shifted to K1, and the G pixel data located at K4 is shifted to K2. In this state, when R and G are displayed in an overlapping manner, the color shift is not visible. Considering B in the same way, K5 is shifted by 4 lines up to K1. As a result, even if R, G, and B are overlapped, the color diagram at the position of K1 or K2 does not generate any image.
[0056]
Of course, the same effect can be obtained not only by shifting but also by copying the value of G located at K3 to the positions of K1 and K2, for example. Even if the operating position is not set to K1, even if K3 is used, color misregistration is not visible. Therefore, color misregistration correction can be performed if absolute image distortion is not considered. This is achieved simply by matching the rising positions of the R, G and B colors in S1.
[0057]
    If you don't take the next dark case,FIG.100 or less is a non-adopted area. That is, it is possible to use an algorithm in which the deviation determination is not performed if the value of each of the R, G, and B pixels is equal to or less than the previously determined value. Specifically, the saturation determination by the region saturation determination unit 118 is performed first before the contrast calculation by the contrast calculation unit 114 described above. Then, the processing time can be shortened by performing the contrast calculation only on the part that requires the calculation and determination by the contrast calculation means 114 according to the flag. In this case, since the S1 portion in FIG. 8 is not adopted, the color shift remains. This is the idea that it is divisible for inconspicuous parts.
[0058]
On the other hand, S2 and S3 are areas where a shift or color matching operation is performed. If L2 is shifted from L1 to S5 in S2, and M1 to M5 in S3, G is shifted by 2 lines and B is shifted by 4 lines, or G is shifted by 2 lines and B is shifted by 4 as in the description of S1 above. It will be whether line color filling is performed.
[0059]
In the case of shifting, it is how far to shift, but in fact, as in M8 to M12 of S3, a high-contrast edge may have a negative method of high contrast. A region where the absolute value of contrast is large to a certain extent is a region where there is no gentle change in density, and color misregistration is conspicuous. Therefore, the shift correction from M1 to M5 that is shifted in the rising direction of S3 and the shift correction in the falling direction from M8 to M12 may be considered separately. In this case, rather than shifting the whole, the shifted portion is shifted, or the density value (pixel value) is manipulated only at the shifted portion.
[0060]
In addition, since the eight lines M1 to M8 have a high contrast region width of R and can be regarded as being shifted to the same width as G and B, in this case, the pixel data for the eight lines of G is moved toward R. This means that 2 lines are shifted, and similarly, 8 lines of B are shifted 4 lines toward R.
[0061]
In the above description, the process in the sub-scanning (line direction) is described in detail for the pixel at a specific position in the main scanning direction. These processes are performed in the main scanning from the first pixel to the final of the main scanning. You have to do up to the pixel. However, the deviation in the sub-scanning direction is determined at a position that is considered to be relatively representative, for example, the 100th pixel, and the same behavior is observed in every pixel in the main scanning direction. It is also possible to shift it uniformly. This simplifies the process so that it is repeated only in the sub-scanning direction, and contributes to cases such as when using a CPU with low processing capability, a circuit with a reduced clock frequency, or when a high-speed reading device is required. it can.
[0062]
Returning to FIG. 6, the data that has been corrected for color misregistration by such an algorithm is transferred to the scanner processing substrate 80 via the I / F control unit 120 of the ADF processing substrate 70. The data is sent from the connector 82 to the computer connected as the host device via the SPC 139 which is a controller for the SCSI interface through the bus buffer or the like omitted from the scanner image processing board 80 on the receiving side. At this time, since the order of sending out to the computer or the order of colors depends on the software running on the computer, the computer has a buffer memory 141 for temporarily buffering so that it can be exchanged. These series of processes are properly synchronized by the clock control unit 116.
[0063]
(Explanation of color misregistration on the scanner side)
The processing in the scanner image processing substrate 80 is basically the same as that of the ADF processing substrate 70, and thus detailed description thereof is omitted. However, the video signal from the image sensor 57, which is a 3-line color CCD, is digitally converted by the A / D converter 130. The shading correction unit 131 performs shading correction. This output is made to coincide with three lines by the interline correction means 132 and temporarily stored in the page memory 133. Since this data is used by the region saturation determination unit 138 and the contrast calculation unit 134, color misregistration determination is performed in cooperation with the CPU 139 and the address control 135, and color matching is performed by the memory shift control unit 137. The series of processes are synchronized by the clock control unit 136 and the process proceeds.
[0064]
The finally color-matched image data is stored in the buffer memory 142 together with the image data received from the ADF processing board 70. Similar to the image data received from the ADF processing board 70, the CPU 139 controls the transmission in accordance with the application on the computer side, and the SPC 140 controls it to send it out from the connector 82.
[0065]
In the following description, for the sake of easy understanding, an example in which the color misalignment correction processing is performed after the interline correction of the three lines is performed by the interline correction unit 132 is shown. In the example, it is a fixed constant such as 8 lines. Therefore, in the pixel shifting loop shown in FIG. 15, for example, a process for moving 8 lines (distance between two colors) is added before S288. The interline correction 112 can be omitted. That is, it is possible to incorporate inter-line correction for all lines in the process for performing color misregistration correction regardless of the position of the color misregistration.
[0066]
(Detailed description of color misregistration correction method)
Hereinafter, the correction processing method will be described with reference to FIGS. 8 to 22 which are flowcharts of the present invention.
[0067]
FIG. 8 is a flowchart of an outline of a reading procedure when the reading device (scanner or ADF with reading) reads the document surface while conveying the document by sheet through, and reads by the apparatus of FIGS. 5 and 6. Since a close contact sensor (CIS) 60 for reading a document is provided in the ADF 51, both sides of a color document can be read as a color image together with reading by the scanner 50 from below. It is a configuration.
[0068]
For example, when a document reading start command is received from a connected computer, processing starting from S199 is executed. When reading with the reading device on the ADF 51 side and reading with the device on the scanner side, the present invention is basically the same in the present invention. Therefore, in FIG. The description will be made assuming that the scanner 50 reads one side of the document in accordance with the feeding of the ADF 51 to be performed. The same is true when the opposite side of the original is read by the CIS 60 built in the ADF 51, as already described in the description of the block diagram in FIG.
[0069]
The original cueing at S200 is a reading position at which the CPU 139 detects a signal from one sensor in the vicinity of the pre-reading roller 54 (not shown) when the ADF 51 starts conveyance and recognizes it after driving the motor with a predetermined pulse. This determination is S201. When it is recognized that the leading edge of the document has reached the reading position, the CCD scan start of S202 is executed. Here, the CCD scan starts, but the image sensor drive clock is not started here. Actually, the clock itself is always given for the stability of the device, and the output from the image sensor 57 is selected. This is implemented by controlling a gate circuit for determining whether to enable or a terminal for enabling the A / D converter 110.
[0070]
If reading starts from S202, the A / D converter output is temporarily stored in the page memory for three colors according to the processing of S203, S204, and S205, and then the position correction by the three sensor intervals in S206 is performed. Correction processing is performed while reading from the memory, and storage is performed again in the page memory. The end of S205 means whether one page of the original has been completed.
[0071]
In FIG. 8, the correction is performed between lines after the storage in the page memory is completed. However, it is possible to write in the page memory while performing processing such as correction between lines in accordance with the procedure described above by accumulating in the work memory. Is possible. At this time, there are 8 lines for R-G and 8 lines for G-B, so at least 16 lines are used as work memory.
[0072]
After the image data for the three primary colors for which the inter-line correction has been completed is stored in the page memory according to the flow of FIG. 8, the correction start processing S210 shown in FIGS. 9, 10, 11, and 12 is executed. The flow described in FIG. 9 to FIG. 12 is to calculate the rising and falling positions for each color as pre-processing for determining the shift amount. Then, after the rising and falling positions are described in the memory for determination, a process of capturing the change width in S262 (counting the number of color-shifted lines) is performed.
[0073]
In the following description, as already described, the image data is described as having been stored in the page memory 133 for all three primary colors in the page memory 133, and the sensor line direction of the image sensor 57 ( 1 is used in the main scanning direction which is the direction of the reading line) and i is used in the sub-scanning direction which is the document feed direction of the ADF 51. Therefore, substituting 1 in S211 and S212 indicates that it starts from the main scanning direction 1 and the sub-scanning direction 1, and as shown by the entire large loops X and Y, sequential processing is performed in the sub-scanning direction. This is further repeated in the main scanning direction.
[0074]
However, as in this procedure, it is not necessary to perform processing after placing all the data on the page memory 133 as a whole, but at least by performing memoization to the extent that the contextual pixel reference described below can be executed. It should be mentioned in advance that this is possible. In addition, it is basically an amount of memory with an appropriate number of lines that can be referred for calculation. Furthermore, it is a simple design change range that other memory means such as a dedicated work memory may be prepared and adopted regardless of the page memory 133.
[0075]
In S213, S214, and S215, the calculation for the l-th pixel of interest in the main scanning direction is shown in the i-th line in the sub-scanning direction. Therefore, the calculation for determination on the first pixel is started first in the first line. . Accordingly, S213 will be described in detail. The value of the first pixel in the second line is subtracted from the value of the first pixel in the first line in R (red), and the absolute value is calculated and stored in the temporary variable MR. It will be that. In S214 and S215, the same calculation is performed for G (green) and B (blue). In terms of the macro, the MR, MG, and MB decision branch points in S216, S228, and S240 are the start points of the blocks for judging the contrast in the R, G, and B colors.
[0076]
In S216, for the red color, the difference in pixel value between adjacent lines is taken to determine the magnitude of contrast. If the contrast is low, it is determined that the current position is not the target position, and 0 is substituted for the flag value, and the process proceeds to the green determination S228. Whether it is higher or lower is a binary determination as to whether or not it exceeds a predetermined value, but in this embodiment, it is about 6 to 18. If it is determined that the contrast is larger than this value, how much the pixel value changes in the vicinity of the pixel to be determined for red by S218, S219, S220, S221, S222, S223, S224, and S225. Calculation is performed. The maximum value and the minimum value of the pixel values are calculated for each of the e pixels before and after the i-th pixel. And it records so that it can judge later by S226. In S227, the flag that is the target position to be determined is entered as 1. Here, it is added that the position with high contrast is specified, and the shift amount is not yet determined and specified.
[0077]
After S228, the same calculation is performed for green. If it is not the target position, 0 is recorded as a non-target in S229, and if it is the target position, S230, S231, S232, S233, S234, S235, S236, and S237 are used to change the nearby density (pixel value) change width. The calculation is performed and recording is performed in S238, and the flag that is the target pixel is recorded in S239.
[0078]
After S240, the same calculation is performed for blue. If the target position is not the target position, 0 is recorded as a non-target in S241, and if it is the target position, S242, S243, S244, S245, S246, S247, S248, and S249 are used to change the nearby density (pixel value) change width. The calculation is performed and recording is performed in S250, and the flag that is the target pixel is recorded in S251.
[0079]
Since the contrast determination and neighborhood value calculation for the three primary colors are completed as described above, it is determined in S252, S254, and S256 whether or not the surrounding area to which the pixel belongs for each color is a gentle change area. , G, and B3 colors, when there is no significant change in the periphery, flags SRL (l, i), SG () indicating the target positions for color misregistration determination (high contrast positions) in S253, S255, and S257, respectively. l, i) and SB (l, i) are substituted with 0 and are not targeted. This is determined using a loop from S258, S259, S260, and S261 to X and Y, and determination flags are determined for all pages of each of the three primary colors.
[0080]
When the rising / falling determination block is completed, a routine S262 for determining a change width (an amount of continuous high contrast state) is executed again based on these flags. Then, the process ends at S263. This routine S262 is described in detail in FIGS.
[0081]
An example of a method for examining the color shift amount is a procedure starting from S262. S264 and S265 are for repeating the internal processing for one page corresponding to S281, S282, S283, and S284 in the repetitive procedures V and W. Internally, referring to the flags SR (l, i), SG (l, i), and SB (l, i) that are contrast flags for red, green, and blue, S267, S272, and S277 are referred to. By this determination, it is determined for each color how many lines each high contrast position occurs.
[0082]
Since each color has the same algorithm, for example, when red is described, m counts how many lines of pixel positions where SR (l, i) is 1, that is, where the contrast is high in the sub-scanning direction. Then, the count is completed, and the value of m is substituted into SFR (l, i) in S270. When the loop of V and W is completed, all the determinations for one page are completed.
[0083]
Next, pixel shift (color shift correction) started from S285 in FIG. 15 will be described. Similar to the above-described determination loop, the processing for one page is repeated in a loop surrounded by S286, S287 and S291, S292, S293, S294. Details of the processing will be described below.
[0084]
As already described, the fact that the flags SR (l, i), SG (l, i), and SB (l, i) have a value of 1 at this stage means that the contrast is high and the density changes sufficiently in the vicinity. This is a region where the so-called color shift is conspicuous. This determination order is G, R, and B, but it is not necessary to be R, G, and B.
[0085]
The determination target position is detected in S288, S289, and S290. For example, if the high-contrast position starts from green, S288 branches first in this. Then, the process proceeds to S296. Here, the states of red and blue other than green are examined.
[0086]
CR and CB in S296 are variables for counting how many color shifts exist between colors. Red is checked in S297 and blue is checked in S301. If it is 1 from the beginning, there is no deviation. If red and green are shifted, the process branches from S297 to S298. S299 is a branch for escaping an error if the deviation is too large. In S300, the counter is incremented and repeated.
[0087]
After checking between red and green, check between blue and green. This is S303, S304, and S305. Since the shift amounts between green and red and green and blue are obtained as CR and CB values, the shift to image data is performed based on these values. This is after S306 in FIG.
[0088]
S306 and S311 mean that processing is not performed when CR and CB are 0, that is, when there is no shift between colors. In S307, an initial value is given. In S308, the red image data is shifted, and the determination target flag corresponding to the position is also moved. S309 is a change of the repeat counter, but data shift is performed according to the branch of S310 by the number of times of adding the color misregistration amount CR and the high contrast continuous amount SFG (l, i) and 2. First, since CR is the amount of color misregistration, if it does not move at least this much, the misplaced portion will not be filled.
[0089]
Since the value of SFG (l, i) is where the contrast determination value SR (l, i) has a value of 1, SR (l, i) is 1 by the flag movement described together with the data movement of S308. It is important to reset the flag where it was, and if this is not done, there is a possibility of duplicate determination at other determinations, which is important. The reason for adding 2 is that the boundary of the determination target position is a delicate position, and it is preferable that the boundary outside is also implemented in order to stabilize the image. For example, if the shift is about half a pixel and the shift affects two pixels, it is preferable to treat both the two pixels.
[0090]
As described above, the red data is moved to the green position and color misregistration correction is performed. In the same manner, the initial value of the counter is introduced in S312, and the blue image data and the determination target flag associated therewith are shifted in S313. The counter operation S314 and branch S315 are repeatedly moved as many times as necessary.
[0091]
When the shift is completed, the process moves to S316. However, in S317, S318, S319, S320, S321, S322, S323, and S324, since the color misregistration correction process is completed, the determination target flags SR (l, i) and SB (l , I) is set to 0 over the high contrast continuous amount SFG (l, i). And it returns to the flow which performs the flag test | inspection of the determination target position of the main flowchart 15. FIG.
[0092]
In the foregoing, the explanation has been given of color correction by moving the position of red and blue when green has a high contrast first. Then, S325, S326, S327, S328, S329, S330, S331, S332, S333, S334, and S335 in FIG. 19, which are flows when red has a high contrast first, S336, S337, S338 in FIG. S339, S340, S341, S342, S343, S344, S345, and S346, S347, S348, S349, S350, S351, S352, S353, S354 of FIG. 21, which is the flow when blue has a high contrast first. , S355, S356, S357, S358, S359, S360, S361, S362, S363, S364, S365, S366 in FIG. 22 are basically the same in the case of the green, and the variable names are changed. Even so, the idea and flow are the same Since, the detailed description thereof is omitted, green, close to the wavelength of particularly sensitive According to human retina characteristics (human visibility characteristics). For this reason, it is meaningful to move on the basis of green because there is a tendency that green is concerned even with the same shift amount, and there is a concern that the image contour line in the vicinity region is deteriorated by moving.
[0093]
(Second form)
5 shows an example in which this correction processing is mounted inside the scanner, but as shown in FIG. 23, a device 5 specialized for this correction processing is formed, and the latter stage of the scanner is formed. It may be attached. Of course, this color misregistration correction function does not have to be present anywhere, and may be configured between the printer 4 and the computer 2 or in the printer 4.
[0094]
Store this correction process as a program in a recording medium such as a CDROM, distribute it, make it possible to supply a single program on a network such as the Internet, download it as needed, and provide program upgrades Is common sense nowadays.
[0095]
(Other forms)
In the above description, a method of performing color correction on a pixel value at a certain position and performing position correction in the line direction (sub-scanning direction) has been described. Accordingly, determination is made from the first pixel to the end pixel in the main scanning direction, and shift correction based on the color shift in the sub-scanning direction is performed for each pixel position. However, only a representative position is determined in the main scanning direction, and it is regarded as the same in the line direction (main scanning direction), and position correction (shift) in the sub-scanning direction can be performed uniformly. It is possible as a method.
[0096]
As described above in detail, it is also possible to write a predetermined density value at a pixel position that is shifted other than the shift (movement of the pixel value) for the alignment after detecting the color shift amount. . Furthermore, although two colors shifted from one color are moved, any color may be moved, or all three colors may be moved to a predetermined position.
[0097]
Further, in the step of determining the color misregistration position, the contrast determination is used by subtracting two different pixel values in the sub-scanning direction, but it is also possible to filter a differential matrix having the same function as the subtraction. As long as the CPU and circuit configuration have sufficient processing capability, various function operations can be used, so that the configuration can be made without being subtracted.
[Brief description of the drawings]
FIG. 1 is a diagram of a conventional sheet-through color scanner.
FIG. 2 is a view in which a sheet through ADF is attached to a conventional flat bed scanner.
3 is a diagram conceptually showing the behavior of a document by the ADF and the scanner in FIG.
4 is a diagram conceptually showing the reading behavior of a carriage with the ADF and the scanner in FIG. 2. FIG.
FIG. 5 is a diagram of an ADF scanner that reads both sides in color with an ADF and a scanner equipped with color misregistration correction according to the present invention.
6 is a processing unit block diagram of the ADF scanner of FIG. 5. FIG.
FIG. 7 is an example in which a low density region is not adopted.
FIG. 8 is a flowchart showing a processing flow of the present invention.
FIG. 9 is a flowchart showing a processing flow of the present invention.
FIG. 10 is a flowchart showing a processing flow of the present invention.
FIG. 11 is a flowchart showing a processing flow of the present invention.
FIG. 12 is a flowchart showing a processing flow of the present invention.
FIG. 13 is a flowchart showing a processing flow of the present invention.
FIG. 14 is a flowchart showing a processing flow of the present invention.
FIG. 15 is a flowchart showing a processing flow of the present invention.
FIG. 16 is a flowchart showing a processing flow of the present invention.
FIG. 17 is a flowchart showing a processing flow of the present invention.
FIG. 18 is a flowchart showing a processing flow of the present invention.
FIG. 19 is a flowchart showing a processing flow of the present invention.
FIG. 20 is a flowchart showing a processing flow of the present invention.
FIG. 21 is a flowchart showing a processing flow of the present invention.
FIG. 22 is a flowchart showing the processing flow of the present invention.
FIG. 23 illustrates an example of a configuration that can be corrected outside an image reading apparatus as a color misregistration correction apparatus.
[Explanation of symbols]
1 Conventional sheet-through scanner
2 Computer
3 Display
4 Printer
5 External color shift correction device
10 Conventional flatbed scanner
11 Auto Document Feeder
12 Carriage for reading
13 Reading window
22 Image sensor with 3 color lines
50 Flatbed scanner with color misregistration correction of the present invention
57 CCD sensor with 3 color lines
60 Contact image sensor with 3 color lines
70 Image processing board for ADF
71 Color misregistration correction means for ADF
80 Image processing board for scanner
81 Color misregistration correction means for a scanner
82 Interface connector to computer
90 ADF with reading means
112 ADF side 3-line reading position correction (line-to-line correction)
114 Contrast judgment on the ADF side
118 ADF side saturation determination
119 CPU controlling the ADF side
120 Interface controller for sending image data from ADF side to scanner side
132 Three-line reading position correction on the scanner side (interline correction)
134 Contrast judgment on the scanner side
138 Area saturation determination on the scanner side
139 CPU controlling the scanner side
140 SCSI Controller for Sending Scanned Image Corrected for ADF and Scanner Color Shift to Computer
 
 

Claims (8)

A light source for illuminating the document;
Optical means for condensing light from the document irradiated with this light source;
An image sensor that receives light from the optical means and generates an image signal for each color component;
Reading means for reading by forming a reading line for each color component on a document surface with a sensor line in which a plurality of linearly arranged photoelectric conversion elements of this image sensor are arranged as pixels,
Moving means for relatively moving the original and the reading line in the sub-scanning direction orthogonal to the main scanning direction which is a reading line of the reading means;
An A / D converter that converts a level of an image signal from the image sensor as a digital gradation value;
An image memory for holding image data for each color component acquired by the plurality of sensor lines;
An image reading apparatus comprising color misregistration correction means for correcting based on image data for each color component held in the image memory,
The color misregistration correction means includes
Between the image data of the first line read by the reading unit and the image data of the second line read by moving relative to the original by the moving unit, a step is performed between two pixels corresponding to the main scanning direction. A difference calculating means for calculating a difference between key values;
Repeating means for repeating the difference calculation of the difference calculation means for each reading line in the sub-scanning direction;
Comparing means for comparing the difference calculated value of the difference calculating means with a predetermined value;
Contrast determination means for determining that the difference value is large based on the result of the comparison means;
Calculation means for calculating a shift amount of the pixel position between the image data of the two colors based on the result of performing the difference calculation of the calculation means for at least the two color image data of the color components;
The calculation means uses the contrast means for the second color and the position of the first pixel determined to have a large difference value by the contrast means for the first color of the two colors of the color component. Based on the position of the second pixel determined to have a large difference value, and calculating the amount of pixel position deviation between the two color image data,
Reading means for reading out data for each pixel from an image memory holding image data for each color component acquired by the plurality of sensor lines;
A writing means for changing the position of the data read by the reading means according to the amount of deviation of the pixel position calculated by the calculating means and writing it into the image memory;
An image reading apparatus comprising: A light source for illuminating the document;
Optical means for condensing light from the document irradiated with this light source;
An image sensor that receives light from the optical means and generates an image signal for each color component;
Reading means for reading by forming a reading line for each color component on a document surface with a sensor line in which a plurality of linearly arranged photoelectric conversion elements of this image sensor are arranged as pixels,
Moving means for relatively moving the original and the reading line in the sub-scanning direction orthogonal to the main scanning direction which is a reading line of the reading means;
An A / D converter that converts a level of an image signal from the image sensor as a digital gradation value;
An image memory for holding image data for each color component acquired by the plurality of sensor lines;
An image reading apparatus comprising color misregistration correction means for correcting based on image data for each color component held in the image memory,
The color misregistration correction means includes
Between the image data of the first line read by the reading unit and the image data of the second line read by moving relative to the original by the moving unit, a step is performed between two pixels corresponding to the main scanning direction. A difference calculating means for calculating a difference between key values;
Repeating means for repeating the difference calculation of the difference calculation means for each reading line in the sub-scanning direction;
Comparing means for comparing the difference calculated value of the difference calculating means with a predetermined value;
Contrast determination means for determining that the difference value is large based on the result of the comparison means;
Calculation means for calculating a shift amount of the pixel position between the image data of the two colors based on the result of performing the difference calculation of the calculation means for at least the two color image data of the color components;
The calculation means uses the contrast means for the second color and the position of the first pixel determined to have a large difference value by the contrast means for the first color of the two colors of the color component. Based on the position of the second pixel determined to have a large difference value, and calculating the amount of pixel position deviation between the two color image data,
An image reading apparatus comprising reading control means for reading out the pixel data from the image memory while controlling the position for reading out the pixel data based on the result of the calculating means. The image data read from the image memory under the control of the reading control means is transferred to an interface means for communicating with a host device.
The image reading apparatus according to claim 2. The color misregistration correction means includes
Inter-line correction means for correcting the distance between the plurality of sensor lines structurally included in the three-line sensor;
3. The image reading apparatus according to claim 1, further comprising reading correction means for correcting a shift during reading caused by relative movement during reading. The color misregistration correction means includes
After the distance correction between the sensor lines by the interline correction unit, the deviation correction by the reading correction unit is performed,
The image reading apparatus according to claim 4. Directly or indirectly to an image reading device that obtains a gradation value for each pixel by moving relative to the document in the sub-scanning direction orthogonal to the main scanning direction for reading the document with a color sensor having a sensor line in which pixels are arranged for each color component A correction device for correcting pixel shift of received image data while being connected by an interface means,
This corrector is
An image memory for storing at least a part of the received image data for each color component according to the position of main scanning and sub-scanning;
Between the first main scanning image data on the image memory and the second main scanning image data different in the sub-scanning direction, a difference calculation of gradation values is performed between two pixels corresponding to the main scanning direction. Difference calculating means to
Repeating means for repeating the difference calculation of the difference calculation means for each reading line in the sub-scanning direction;
Calculation means for calculating the amount of displacement of the pixel position between the two color image data based on the result of the difference calculation performed by the difference calculation means on the two color image data of the color components;
Reading means for reading out data for each pixel from the image memory;
A writing means for changing the position of the data read by the reading means according to the amount of deviation of the pixel position calculated by the calculating means and writing it into the image memory;
A correction apparatus comprising: Directly or indirectly to an image reading device that obtains a gradation value for each pixel by moving relative to the document in the sub-scanning direction orthogonal to the main scanning direction for reading the document with a color sensor having a sensor line in which pixels are arranged for each color component A correction device for correcting pixel shift of received image data while being connected by an interface means,
This corrector is
An image memory for storing at least a part of the received image data for each color component according to the position of main scanning and sub-scanning;
Between the first main scanning image data on the image memory and the second main scanning image data different in the sub-scanning direction, a difference calculation of gradation values is performed between two pixels corresponding to the main scanning direction. Difference calculating means to
Repeating means for repeating the difference calculation of the difference calculation means for each reading line in the sub-scanning direction;
Calculation means for calculating the amount of displacement of the pixel position between the two color image data based on the result of the difference calculation performed by the difference calculation means on the two color image data of the color components ;
On the basis of the calculation of results means, and a read-out control means reads out while controlling the position of reading the data of pixels from the image memory,
A correction apparatus , wherein the image data read from the image memory under the control of the reading control means is transferred to the interface means. A reading process for forming and reading a reading line for each color component on the document surface with a plurality of linearly arranged sensor image photoelectric conversion elements of the image sensor;
A moving step of relatively moving the original and the reading line in the sub-scanning direction orthogonal to the main scanning direction as the reading line;
An A / D conversion step of converting the level of the image signal from the image sensor as a digital gradation value;
A holding step of holding image data for each color component acquired by the plurality of sensor lines in an image memory that stores the image data for each pixel according to the position of main scanning and sub-scanning;
Between the first main scanning image data read in the reading step and the second main scanning image data different in the sub-scanning direction, a gradation value difference calculation is performed between two pixels corresponding to the main scanning direction. A difference calculation step to perform,
Repeating means for repeating the difference calculation of the difference calculation means for each reading line in the sub-scanning direction;
A comparison step of comparing the difference calculation value of the difference calculation step with a predetermined value;
A contrast determination step for determining that the difference value is large based on the result of the comparison step;
A calculation step of calculating a shift amount of the pixel position between the image data of the two colors based on a result of executing the difference calculation of the calculation step on the image data of at least two colors of the color component;
In the calculation step, the position of the first pixel determined by the contrast means to be large for the first color out of the two colors of the color component and the contrast means for the second color. A calculation step of calculating a shift amount of the pixel position between the image data of the two colors based on the position of the second pixel determined to have a large difference value;
A reading process of reading out data for each pixel from the image memory;
A writing step of changing the position of the data read by the reading means according to the amount of deviation of the pixel position calculated by the calculating means and writing it into the image memory;
A correction method characterized by comprising:

Images