JP2006155518A - Geometric correction method and device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a geometric correction method simultaneously realizing both a real-time property and reliability of pixel data. <P>SOLUTION: This geometric correction method geometrically correcting the pixel data acquired by an imaging means, having corresponding error information has: an error complementing process 100 for deciding whether the pixel data corresponds to an error pixel or not on the basis of the error information corresponding to the inputted pixel data, replacing the pixel data decided to correspond to the error pixel with the other pixel data present on the periphery of the pixel data when setting the pixel data as an output image, and decided not to correspond to the error pixel, and outputting them; and a rearranging process 31 for obtaining the output image by weighting of the inputted pixel data. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a geometric correction method and apparatus for geometrically correcting pixel data acquired by an imaging means.

Conventionally, a geometric correction method for geometrically correcting pixel data acquired by an imaging means such as a camera has been known. For example, it is as described in Patent Document 1 and Patent Document 2.
Here, the “imaging means” is means for acquiring pixel data, and specifically includes a CCD camera and a CMOS camera.
In addition, “pixel data” is information acquired by individual solid-state image sensors (pixels) included in the imaging unit, and is usually a luminance value in an imaging unit that captures a black and white image, or an imaging that captures a color image. The luminance value of each RGB in a means is pointed out.
In addition, “geometric correction” generally refers to image processing for correcting geometric distortion included in an image that is a collection of pixel data captured by an imaging unit such as a camera.
The geometric distortion included in the image includes (1) internal distortion caused by the imaging means, and (2) external distortion caused by the platform that is the part holding the imaging means and the imaging target. A typical example of the internal distortion is distortion of a lens provided in a camera or the like that is an imaging unit.

The geometric correction generally includes a rearrangement process for obtaining an output image by weighting input pixel data.
In the rearrangement processing, more specifically, the address of the pixel data is corrected to correct distortion of the input image that is a set of input pixel data, and the address is corrected based on the pixel data whose address is corrected. The position at which becomes an integer value, that is, the luminance value of the pixel data of the grid point is determined by weighting.
Here, a specific example of the method for correcting the distortion of the input image includes non-systematic correction represented by high-order polynomial transformation.
As a specific example of the weighting method, (1) the luminance value of the pixel data of the grid point is set to the position where the Euclid distance is the shortest among the pixel data subjected to the four-point geometric transformation in the vicinity of the grid point. Nearest neighbor method for replacement with the luminance value of pixel data, (2) 4-neighbor bilinear for determining the luminance value of pixel data of grid points by proportional calculation from pixel data subjected to geometric transformation of 4 points located in the vicinity of the grid points And a cubic convolution method in which the luminance value of the pixel data at the lattice points is determined by convolution from the pixel data that has undergone 16-point geometric transformation in the vicinity of the lattice points.
Examples of the 4-neighbor bilinear method include the method described in Patent Document 3.

Geometric correction is real-time, that is, pixel data is subjected to geometric correction and the pixel data subjected to geometric correction is output in order to be able to make various determinations instantaneously based on the image captured by the imaging means. In some cases, it is required to reduce the time required for the process.
As an example of the geometric correction that requires real-time characteristics, an obstacle on a driving lane or a road is recognized based on pixel data acquired by an imaging unit provided in an automobile or the like, and an automobile is detected based on the recognition result. There are cases where automatic driving and driving assistance are performed.

Specifically, a CCD camera or a CMOS camera, which is a specific example of the image pickup means, performs various processes such as noise component removal, gain control, knee correction, and pre-gamma correction on an analog signal acquired by a solid-state image pickup device (pixel). An analog signal processing unit to perform, an analog / digital conversion circuit that converts the analog signal corrected by the analog signal processing unit into a digital signal, gamma correction, white / black level balance processing to the converted digital signal, And a digital signal processing unit that performs various processes such as shading correction and pixel defect correction.
Among the processes performed by the digital signal processing unit, the pixel defect correction is a process for correcting pixel data of a pixel corresponding to the pixel defect, more strictly, a luminance value of the pixel data.
“Pixel defect” refers to a display defect in pixel units caused by a solid-state imaging device of a CCD camera or a CMOS camera. Examples of the pixel defect include a white defect and a black defect.
As a specific example of the pixel defect correction, there is a method of determining the luminance value of the pixel data of the pixel corresponding to the pixel defect based on the pixel data related to a plurality of pixels adjacent to the pixel corresponding to the pixel defect. For example, as described in Patent Document 4.

Such pixel defect correction is indispensable in terms of aesthetics when displaying an image as an aggregate of pixel data on a display screen or the like.
Japanese Patent Laid-Open No. 10-307925 JP-A-10-283495 JP 2004-140495 A JP-A-10-210365

However, it is possible to geometrically correct the pixel data after the pixel defect correction by the imaging unit as described above and perform various determinations based on the output image that is an aggregate of the pixel data after the geometric correction. There is a problem in terms.
That is, the pixel data after the pixel defect correction is performed by the conventional imaging means is output in a state in which it is impossible to specify the pixel data actually corrected for the pixel defect. For this reason, it is inappropriate in some cases to make various determinations based on the pixel data without specifying which pixel data of the pixel data after geometric correction has been corrected for pixel defects.
As an example of the inappropriate determination, an obstacle on a driving lane or a road is recognized based on an image that is a collection of pixel data acquired by an imaging unit provided in the automobile, and the automobile is based on the recognition result. There are cases where automatic driving and driving assistance are performed. This is because there is a possibility that an object that does not actually exist exists, or that an object that actually exists does not exist.

  Therefore, when it is inappropriate to make various determinations based on an image as an aggregate of pixel data without specifying which pixel data has been corrected for pixel defects among the pixel data after geometric correction. (A) Reliability of pixel data after geometric correction, that is, pixel data related to a pixel defect among pixel data after geometric correction, or (B) It is required that the pixel data after geometric correction is previously subjected to correction processing that is considered appropriate due to the nature of various judgments made after geometric correction.

  In view of the circumstances as described above, the present invention provides a geometric correction method that achieves both the reliability of pixel data and real-time performance.

  The problem to be solved by the present invention is as described above. Next, means for solving the problem will be described.

That is, in claim 1,
A geometric correction method for geometrically correcting pixel data obtained by an imaging means and having corresponding error information,
Based on the error information corresponding to the input pixel data, it is determined whether or not the pixel data corresponds to the error pixel. When the pixel data determined to correspond to the error pixel is used as an output image, the pixel data Error complement processing that is output in place of another pixel data that is determined to be incompatible with the error pixel,
Rearrangement processing to obtain an output image by weighting the input pixel data;
It comprises.

In claim 2,
Another pixel data determined not to correspond to the error pixel is:
With respect to pixel data determined to correspond to an error pixel, the imaging means is located in a direction substantially orthogonal to the scanning direction for acquiring pixel data,
In addition, it is located at four neighboring positions including pixel data determined to correspond to the error pixel.

In claim 3,
Another pixel data determined not to correspond to the error pixel is:
With respect to the pixel data determined to correspond to the error pixel, the imaging means is located in a direction substantially coinciding with the scanning direction for acquiring the pixel data,
In addition, it is located at four neighboring positions including pixel data determined to correspond to the error pixel.

In claim 4,
An apparatus for performing geometric correction for geometric correction of pixel data obtained by an imaging unit and having corresponding error information,
Based on the error information corresponding to the input pixel data, it is determined whether or not the pixel data corresponds to the error pixel. When the pixel data determined to correspond to the error pixel is used as an output image, the pixel data An error complementing processing unit that outputs a pixel data that is replaced with another pixel data that is determined not to correspond to the error pixel,
A rearrangement processing unit for obtaining an output image by weighting input pixel data;
It comprises.

In claim 5,
Another pixel data determined not to correspond to the error pixel is:
With respect to pixel data determined to correspond to an error pixel, the imaging means is located in a direction substantially orthogonal to the scanning direction for acquiring pixel data,
In addition, it is located at four neighboring positions including pixel data determined to correspond to the error pixel.

In claim 6,
Another pixel data determined not to correspond to the error pixel is:
With respect to the pixel data determined to correspond to the error pixel, the imaging means is located in a direction substantially coinciding with the scanning direction for acquiring the pixel data,
In addition, it is located at four neighboring positions including pixel data determined to correspond to the error pixel.

  As effects of the present invention, the following effects can be obtained.

  In claim 1, it is possible to achieve both the reliability of pixel data and the real-time property.

  In claim 2, when the luminance value of the pixel data after geometric correction is differentiated in the scanning direction of the image pickup means and the boundary line of the side surface of the object substantially orthogonal to the scanning direction is identified, It is possible to accurately identify the boundary line of the side surface of the pixel, and particularly the reliability of the pixel data after correction is improved.

  According to a third aspect of the present invention, the luminance value of the pixel data after geometric correction is differentiated in a direction orthogonal to the scanning direction of the image pickup means, and a process of identifying the boundary line of the side surface of the object that substantially matches the scanning direction is performed. In addition, it is possible to accurately identify the boundary line on the side surface of the object, and in particular, the reliability of the corrected pixel data is improved.

  In claim 4, it is possible to achieve both the reliability of pixel data and the real-time property.

  According to claim 5, when the luminance value of the pixel data after geometric correction is differentiated in the scanning direction of the image pickup means and the boundary line of the side surface of the object substantially orthogonal to the scanning direction is identified, It is possible to accurately identify the boundary line of the side surface of the pixel, and particularly the reliability of the pixel data after correction is improved.

  According to another aspect of the invention, the luminance value of the pixel data after geometric correction is differentiated in a direction orthogonal to the scanning direction of the image pickup means, and the process of identifying the boundary line of the side surface of the object that substantially matches the scanning direction is performed. In addition, it is possible to accurately identify the boundary line on the side surface of the object, and in particular, the reliability of the corrected pixel data is improved.

  Below, the Example of the geometric correction method of this invention is described using FIG. 1 and FIG.

Below, the pixel data acquisition process 10 is demonstrated using FIG. 1 and FIG.
The pixel data acquisition process 10 is a process for acquiring pixel data by an imaging means such as a CCD camera or a CMOS camera and outputting the pixel data.
The pixel data acquisition process 10 of this embodiment performs various processes such as noise component removal, gain control, knee correction, and pre-gamma correction on the analog signal acquired by the solid-state imaging device (pixel), and further converts the analog signal to digital. It may be converted into a signal, and various processing such as gamma correction, white / black level balance processing, shading correction, etc. may be performed on the digital signal, but pixel defect correction is performed on the pixel data. It is required not to apply.
Then, instead of not performing pixel defect correction on the pixel data, error information corresponding to the pixel data is output. In other words, the pixel data of this embodiment includes corresponding error information.
“Error information” is information relating to the reliability of individual pixel data. In the present embodiment, whether or not pixel defects have occurred, that is, whether or not pixel defects have occurred in pixels corresponding to individual pixel data. Is included.
Hereinafter, pixels related to pixel defects and other errors are referred to as “error pixels”.

As shown in FIG. 2A, each piece of pixel data acquired in the pixel data acquisition process 10 has a unique luminance value.
In FIG. 2, the address of each pixel data is expressed by the x-coordinate and the y-coordinate for convenience of explanation. However, in this embodiment, the order in which the pixel data is acquired by the image pickup means is the address of the pixel data. To express.

As shown in FIG. 2A, the description will be made by paying attention to four adjacent pixel data A1, A2, A3, and A4 among the pixel data acquired in the pixel data acquisition process 10.
In addition, the x coordinate of the pixel data A1 is x1, the y coordinate is y1, and the luminance value is z1, and is hereinafter referred to as pixel data A1 (x1, y1, z1) as necessary. Similarly, pixel data A2 (x2, y2, z2), pixel data A3 (x3, y3, z3), and pixel data A4 (x4, y4, z4) will be hereinafter described as necessary.
The x-coordinate and y-coordinate of the pixel data A1, A2, A3, and A4 that are the pixel data acquired in the pixel data acquisition process 10 are all represented by integers, and x1 = x3, x2 = x4 = x1 + 1, y1 = y2, The relationship y3 = y4 = y1 + 1 is established.
Here, it is assumed that the pixel data A2 is pixel data corresponding to an error pixel among the four pieces of pixel data A1, A2, A3, and A4.

Hereinafter, the error complementing process 100 will be described with reference to FIGS. 1 and 2.
The error complementing process 100 determines whether or not the pixel data corresponds to an error pixel based on error information corresponding to the input pixel data, that is, the pixel data after the pixel data acquisition process 10 is performed. In this process, the pixel data determined to correspond to the error pixel is replaced with another pixel data that is in the vicinity of the pixel data and determined not to correspond to the error pixel when the output image is output. is there.

As shown in FIG. 2B, in the error complementing process 100, the pixel data A1, A2, A3, and A4 are converted into pixel data B1, B2, B3, and B4, respectively.
At this time, the addresses of the pixel data B1, B2, B3, and B4 are the same as the addresses of the pixel data A1, A2, A3, and A4, respectively, but the luminance values correspond to the pixel data A1, A2, A3, and A4. It is determined whether or not to change based on the error information.

That is, in the error complementing process 100, among the pixel data A1, A2, A3, and A4, it is determined that the pixel data A1, A3, and A4 do not correspond to error pixels, and thus the luminance of the pixel data B1, B3, and B4 The values are the same as the luminance values of the pixel data A1, A3, and A4, respectively.
Further, in the error complementing process 100, since it is determined that the pixel data A2 corresponds to the error pixel, the imaging means uses the pixel data A1 and A2 for the luminance value (z2) of the pixel data B2 with respect to the pixel data A2. Pixel data located in a direction substantially orthogonal to the scanning direction (usually the horizontal direction) for acquiring A3 and A4 and located in four neighboring positions including the pixel data A2, that is, the luminance value of the pixel data A4 ( Replace with z4).

  As a result, as shown in FIG. 2B, in the error complementing process 100, the pixel data A1 (x1, y1, z1) is converted into the pixel data B1 (x1, y1, z1) and the pixel data A2 (x2, y2). , Z2) is the pixel data B2 (x2, y2, z4), the pixel data A3 (x3, y3, z3) is the pixel data B3 (x3, y3, z3), and the pixel data A4 (x4, y4, z4) is Each pixel data is converted into B4 (x4, y4, z4).

FIG. 3 is a schematic diagram showing an error compensation circuit 1000 which is an embodiment of a circuit that performs the error compensation processing 100.
The error complementing circuit 1000 includes four FIFO memories 1001, 1002, 1003, and 1004, and four multiplexers 1011, 1012, 1013, and 1014.
The error interpolation circuit 1000 stores the pixel data A1, A2, A3, and A4 input from the imaging unit 1200 in the FIFO memories 1001, 1002, 1003, and 1004, respectively, and then outputs them.
At this time, the pixel data A1, A2, A3, and A4 output from the FIFO memories 1001, 1002, 1003, and 1004 are input to the multiplexers 1011, 1012, 1013, and 1014 as follows.
That is, pixel data A 1 and A 3 are input to the multiplexer 1011, pixel data A 2 and A 4 are input to the multiplexer 1012, pixel data A 1 and A 3 are input to the multiplexer 1013, and pixel data A 2 and A 4 are input to the multiplexer 1014. Is entered.
Error information corresponding to the pixel data A1, A2, A3, and A4 output from the imaging unit 1200 is also input to the multiplexers 1011, 1012, 1013, and 1014, respectively.

The multiplexer 1011 determines that the pixel data A1 does not correspond to the error pixel based on the error information corresponding to the pixel data A1, and uses the luminance value of the pixel data A1 to output the pixel data B1.
The multiplexer 1012 determines that the pixel data A2 corresponds to the error pixel based on the error information corresponding to the pixel data A2, and uses the luminance value of the pixel data A4 to output the pixel data B2.
The multiplexer 1013 determines that the pixel data A3 does not correspond to the error pixel based on the error information corresponding to the pixel data A3, and outputs the pixel data B3 using the luminance value of the pixel data A3.
The multiplexer 1014 determines that the pixel data A4 does not correspond to the error pixel based on the error information corresponding to the pixel data A4, and outputs the pixel data B4 using the luminance value of the pixel data A4.

In FIG. 3, an error detection circuit 1100 having a counter is provided between the imaging unit 1200 and the error compensation circuit 1000.
The error detection circuit 1100 compares a pixel counter 1101 that counts pixel data output from the imaging unit 1200, a count of a counter circuit (not shown) built in the imaging unit 1200 and a count of the pixel counter 1101, and both And a pixel counter comparison circuit 1102 that outputs error information indicating that the pixel data corresponds to an error pixel.
With this configuration, even when a communication error occurs between the imaging unit 1200 and the error complement circuit 1000, the error complement circuit 1000 does not correspond to the error pixel in the luminance value of the pixel data related to the communication error. It is possible to replace the brightness value of the pixel data.

Below, the rearrangement process 31 is demonstrated using FIG. 1 and FIG.
The rearrangement process 31 is a process for obtaining an output image by weighting input pixel data.
More specifically, the rearrangement process 31 corrects the address of the pixel data to correct distortion of the input image that is a collection of input pixel data, and the address is determined based on the pixel data with the corrected address. The position at which becomes an integer value, that is, the luminance value of the pixel data of the grid point is determined by weighting.

In this embodiment, as a method of correcting the distortion of the input image, non-systematic correction based on a high-order polynomial specific to a lens such as a camera as an imaging unit is performed.
In this embodiment, the 4-neighbor bilinear method is adopted as the weighting method. The “4-neighbor bilinear method” is a method for determining the luminance value of pixel data at grid points from the pixel data subjected to geometric transformation of four points located in the vicinity of the grid points by proportional calculation.
Here, the position of the pixel data subjected to the four-point geometric transformation is defined as “four neighboring positions”. The 4-neighbor position indicates an address from pixel data having the closest Euclidean distance to the address of the grid point to pixel data having the fourth closest position in the input pixel data.

  As shown in FIG. 2C, in the rearrangement process 31, the pixel data B1, B2, B3, and B4 that have been subjected to the error compensation process 100 are converted into pixel data C1, C2, C3, and C4, respectively. The position where the address is an integer value, that is, the pixel data D (X, Y, Z) of the grid point is determined based on the pixel data C1, C2, C3, and C4 located at positions near the grid point.

  As a result, as shown in FIG. 2C, in the rearrangement process 31, the pixel data B1 (x1, y1, z1) is converted into the pixel data C1 (X1, Y1, z1) and the pixel data B2 (x2, y2). , Z4) is pixel data C2 (X2, Y2, z4), pixel data B3 (x3, y3, z3) is pixel data C3 (X3, Y3, z3), and pixel data B4 (x4, y4, z4) is The pixel data is converted into pixel data C4 (X4, Y4, z4), and pixel data D (X, Y, Z) is determined based on the pixel data C1, C2, C3, and C4.

The x coordinate (X) of the address of the pixel data D is an integer in the range from the minimum value to the maximum value of the x coordinate (X1, X2, X3, and X4) of the address of the pixel data C1, C2, C3, and C4. It is.
The y coordinate (Y) of the address of the pixel data D is an integer in the range from the minimum value to the maximum value of the y coordinate (Y1, Y2, Y3, Y4) of the address of the pixel data C1, C2, C3, C4. It is.
Further, the following relationship is established between the luminance value Z of the address of the pixel data D and the luminance values z1, z2, z3, and z4 of the pixel data C1, C2, C3, and C4.
(R1 + R2 + R3 + R4) / Z = R1 / z1 + R2 / z2 + R3 / z3 + R4 / z4
Here, R1, R2, R3, and R4 indicate Euclidean distances between the pixel data D and the pixel data C1, C2, C3, and C4, respectively.

Hereinafter, the other image processing 40 will be described with reference to FIG.
The other image processing 40 refers to processing for performing various image processing or determination using the pixel data after geometric correction.
As another example of the image processing 40, the pixel data acquired by the image pickup unit provided in the automobile is subjected to geometric correction, and another pixel existing in the traveling direction of the automobile based on the pixel data after the geometric correction is performed. Recognizing obstacles such as automobiles.

As described above, the embodiment of the geometric correction method of the present invention is
A geometric correction method for geometrically correcting pixel data obtained by an imaging means and having corresponding error information,
Based on the error information corresponding to the input pixel data, it is determined whether or not the pixel data corresponds to the error pixel. When the pixel data determined to correspond to the error pixel is used as an output image, the pixel data Error complementing processing 100 that outputs a pixel data that is replaced with another pixel data determined to not correspond to an error pixel,
A rearrangement process 31 for obtaining an output image by weighting the input pixel data;
It comprises.

Such a configuration has the following advantages.
That is, the pixel data that has been subjected to geometric correction by the geometric correction method includes pixel data corresponding to an error pixel (more strictly speaking, the luminance value of the pixel data) at the time when the geometric correction method is used as an output image. In order to replace the pixel data with another pixel data (more precisely, the luminance value of the pixel data) that is determined to be around the pixel data and does not correspond to the error pixel, Correction processing appropriate for the nature of the judgment is performed in advance. Therefore, the reliability of the pixel data after geometric correction is high.
In addition, it is easy to replace pixel data corresponding to an error pixel with another pixel data that is in the vicinity of the pixel data when it is used as an output image and is determined not to correspond to an error pixel. Is possible. Therefore, the real-time property of the geometric correction is ensured.
As a result, the geometric correction method can achieve both the reliability of pixel data and the real-time property.

In the embodiment of the geometric correction method of the present invention, another pixel data determined not to correspond to the error pixel is:
With respect to pixel data determined to correspond to an error pixel, the imaging means is located in a direction substantially orthogonal to the scanning direction for acquiring pixel data,
In addition, it is located at four neighboring positions including pixel data determined to correspond to the error pixel.

  With this configuration, the pixel data after geometric correction (more precisely, the luminance value of the pixel data) is differentiated in the scanning direction of the imaging unit, and the boundary of the side surface of the object that is substantially orthogonal to the scanning direction. When the process of identifying a line is performed, it is possible to accurately identify the boundary line on the side surface of the object, and in particular, the reliability of the pixel data after correction is improved.

As another example, another pixel data determined not to correspond to an error pixel is
With respect to the pixel data determined to correspond to the error pixel, the imaging means is located in a direction substantially coinciding with the scanning direction for acquiring the pixel data,
In addition, it is also possible to adopt a configuration located at four neighboring positions including pixel data determined to correspond to the error pixel.

  With this configuration, an object that is substantially coincident with the scanning direction by differentiating the pixel data after geometric correction (more precisely, the luminance value of the pixel data) in a direction orthogonal to the scanning direction of the imaging means. When performing the process of identifying the boundary line of the side surface of the object, it is possible to identify the boundary line of the side surface of the object with high accuracy, and in particular, the reliability of the corrected pixel data is improved.

The flowchart which shows the Example of the geometric correction method of this invention. The schematic diagram of the pixel data in the Example of the geometric correction method of this invention. The schematic diagram of the error complementation circuit which concerns on the Example of the geometric correction method of this invention.

Explanation of symbols

31 Rearrangement processing 100 Error compensation processing

Claims (6)

A geometric correction method for geometrically correcting pixel data obtained by an imaging means and having corresponding error information,
Based on the error information corresponding to the input pixel data, it is determined whether or not the pixel data corresponds to the error pixel. When the pixel data determined to correspond to the error pixel is used as an output image, the pixel data Error complement processing that is output in place of another pixel data that is determined to be incompatible with the error pixel,
Rearrangement processing to obtain an output image by weighting the input pixel data;
A geometric correction method comprising: Another pixel data determined not to correspond to the error pixel is:
With respect to pixel data determined to correspond to an error pixel, the imaging means is located in a direction substantially orthogonal to the scanning direction for acquiring pixel data,
The geometric correction method according to claim 1, wherein the geometric correction method is located at four neighboring positions including pixel data determined to correspond to an error pixel. Another pixel data determined not to correspond to the error pixel is:
With respect to the pixel data determined to correspond to the error pixel, the imaging means is located in a direction substantially coinciding with the scanning direction for acquiring the pixel data,
The geometric correction method according to claim 1, wherein the geometric correction method is located at four neighboring positions including pixel data determined to correspond to an error pixel. An apparatus for performing geometric correction for geometric correction of pixel data obtained by an imaging unit and having corresponding error information,
Based on the error information corresponding to the input pixel data, it is determined whether or not the pixel data corresponds to the error pixel. When the pixel data determined to correspond to the error pixel is used as an output image, the pixel data An error complementing processing unit that outputs a pixel data that is replaced with another pixel data that is determined not to correspond to the error pixel,
A rearrangement processing unit for obtaining an output image by weighting input pixel data;
The apparatus characterized by comprising. Another pixel data determined not to correspond to the error pixel is:
With respect to pixel data determined to correspond to an error pixel, the imaging means is located in a direction substantially orthogonal to the scanning direction for acquiring pixel data,
The apparatus according to claim 4, wherein the apparatus is located at four neighboring positions including pixel data determined to correspond to an error pixel. Another pixel data determined not to correspond to the error pixel is:
With respect to the pixel data determined to correspond to the error pixel, the imaging means is located in a direction that substantially matches the scanning direction for acquiring the pixel data,
The apparatus according to claim 4, wherein the apparatus is located at four neighboring positions including pixel data determined to correspond to an error pixel.

Images