JP2014230014A - Image processing device, image processing program and electronic camera - Google Patents

Abstract

PROBLEM TO BE SOLVED: To solve the problem that, in conventional interpolation processing, a difference may occur in a resolution for each pixel.SOLUTION: An image processing device comprises: an image input section; an interpolation target position setting section for setting an interpolation target position on an input image corresponding to a pixel of an output image obtained by deforming the input image; a reference pixel value extraction section for defining N (N is an integer equal to or more than 3) pieces of pixels, among a plurality of pixels forming the input image, in the vicinity of the interpolation target position as reference pixels and extracting pixel values of the reference pixels; an interpolation weighting factor setting section for setting an interpolation weighting factor for each reference pixel in accordance with positions of the N pieces of reference pixels; and an interpolation section for calculating a multiplication value by multiplying the pixel value of a reference pixel by the interpolation weighting factor of the reference pixel, and defining a total sum of the multiplication values for the N pieces of reference pixels as an interpolation value of the interpolation target position. The interpolation weighting factor setting section sets the interpolation weighting factor in such a manner that an evaluation value representing a degree of diffusion for the positions of the N pieces of reference pixels is fixed independently of the interpolation target position.

Description

  The present invention relates to an image processing technique for performing interpolation processing.

  In general, when performing deformation processing such as enlargement, reduction, rotation, translation, and distortion correction on a digital image taken with an electronic camera or the like, a pixel value that is newly generated from the pixel value before deformation is changed. Interpolation processing is performed. For example, in the process of enlarging 1.2 times, the pixel value of the coordinates (x / 1.2, y / 1.2) before enlargement is used as the pixel value of the coordinates (x, y) of the image after enlargement. Assign it. However, since the pixels before enlargement are arranged in integer coordinates, if the coordinates (x / 1.2, y / 1.2) do not become integers, the pixel values of the neighboring integer coordinates are used, for example. A pixel value of (x / 1.2, y / 1.2) is interpolated (see, for example, Patent Document 1).

Japanese Unexamined Patent Publication No. 64-051765

  However, conventionally, when a new pixel is interpolated between existing pixels, there is a problem that a difference in resolution (such as the degree of blur) occurs depending on the existing pixel and the interpolation target position. For example, when the coordinates of the interpolation target position are close to the integer coordinates of a specific existing pixel, the interpolation value generated by referring mainly to the pixel value of the integer coordinate is calculated, so that the resolution of the interpolation target position is hardly lowered. On the other hand, when the coordinates of the interpolation target position are in the middle of the pixel positions of a plurality of integer coordinates, the interpolation values of the interpolation target positions are calculated by uniformly referring to the pixel values of the integer coordinates. There arises a problem that the pixel value is smoothed and the resolving power at the interpolation target position is reduced.

  If the contour enhancement processing is performed after interpolation, the reduced resolution can be increased, but excessive enhancement processing is performed for pixels with a small decrease in the resolution, and the image quality deteriorates due to the occurrence of jaggies, etc. It becomes.

  In view of the above problems, an object of the present invention is to provide an image processing apparatus and an image processing capable of performing interpolation so as not to cause a difference in resolving power between pixels generated by interpolation when a new pixel is generated by interpolation. To provide a program and an electronic camera.

  An image processing apparatus according to the present invention includes an image input unit that inputs an image composed of a plurality of pixels arranged two-dimensionally, and interpolation on the input image corresponding to the pixels of the output image obtained by transforming the input image An interpolation target position setting unit that sets a target position, and N pixels (N is an integer of 3 or more) in the vicinity of the interpolation target position among a plurality of pixels constituting the input image, are used as reference pixels. A reference pixel value extracting unit for extracting a value, an interpolation weighting factor setting unit for setting an interpolation weighting factor for each reference pixel according to the position of the N reference pixels, and a reference to the pixel value of the reference pixel An interpolation unit that calculates a multiplication value obtained by multiplying the interpolation weighting coefficient of a pixel and uses the sum of the multiplication values for the N reference pixels as an interpolation value of the interpolation target position, and the interpolation weighting coefficient setting unit includes: , Expanding the position of the N reference pixels. Evaluation value representing the degree is, and sets the interpolation weighting factors to be constant regardless of the position to be interpolated.

  An image processing program according to the present invention is characterized in that the processing of each unit of the image processing apparatus is executed by a computer.

  An electronic camera according to the present invention is an electronic camera equipped with the image processing device, wherein an imaging unit that outputs an image obtained by imaging a subject to the image input unit, and the image processing device interpolated and generated. And a storage unit for recording an image.

  The image processing apparatus, the image processing program, and the electronic camera according to the present invention can interpolate so that a difference in resolving power does not occur between the interpolated pixels when a new pixel is interpolated between the pixels.

It is a block diagram which shows the structural example of the electronic camera 101 which concerns on this embodiment. It is a figure which shows the example of an expansion process. It is a figure which shows the example of interpolation of the pixel value before and behind expansion. It is a figure which shows the example of the interpolation object position. It is a figure which shows an example of the interpolation object position of an expansion process and a reduction process. It is a figure which shows the calculation procedure of the interpolation value of an interpolation object position. It is a figure which shows an example of the range of a reference pixel and an interpolation object position. It is a figure which shows the range and specific example of interpolation object position (DELTA) x. 6 is a diagram illustrating an example of a coefficient table 141. FIG. It is a figure which shows the range and specific example of interpolation object position (DELTA) y. FIG. 3 is a diagram illustrating a configuration example of an image processing unit 107 of the electronic camera 101 and a processing flow.

  Embodiments of an image processing apparatus, an image processing program, and an electronic camera according to the present invention will be described below in detail with reference to the drawings. In this embodiment, an example of an electronic camera equipped with the image processing apparatus according to the present invention will be described. However, a personal computer program or a single image processing apparatus that inputs a photographed image and performs image processing. It does not matter.

[Configuration Example of Electronic Camera 101]
FIG. 1 is a block diagram illustrating a configuration of the electronic camera 101. In FIG. 1, an electronic camera 101 includes an optical system 102, a mechanical shutter 103, an image sensor 104, an A / D conversion unit 105, an image buffer 106, an image processing unit 107, a control unit 108, and a memory 109. And a display unit 110, an operation unit 111, and a memory card IF (interface) 112. Here, the image processing unit 107 is a block corresponding to an image processing apparatus in which the image processing program according to the present invention is installed.

  In FIG. 1, the subject light incident on the optical system 102 is incident on the light receiving surface of the image sensor 104 via the mechanical shutter 103. Here, the optical system 102 includes a plurality of lenses such as a zoom lens and a focus lens, a lens driving unit, a diaphragm, and the like, and the zoom lens, the focus lens, or the diaphragm is photographed in accordance with a command from the control unit 108. It is controlled according to conditions.

  The image sensor 104 is configured by a CMOS sensor or the like. In the present embodiment, the signal output from the image sensor 104 may be a monochrome image or a color image. In addition, when the image sensor 104 is RAW data such as a Bayer array having color information of any one of RGB colors in each pixel, RGB data or YCbCr data having information of RGB three colors in each pixel by color interpolation processing. It is assumed that image data after being converted into luminance / color difference data such as is handled.

  The A / D conversion unit 105 converts the image signal output from the image sensor 104 into a digital value for each pixel, and temporarily stores image data for one captured image in the image buffer 106. For example, when the resolution of the image sensor 104 is 4000 pixels × 3000 pixels, image data for 12 million pixels is taken into the image buffer 106. Note that when the image data taken into the image buffer 106 is RAW data, the image processing unit 107 performs color interpolation processing for generating RGB three-color information for each pixel as described above. When the image sensor 104 is a monochrome image sensor or a three-plate image sensor corresponding to RGB three colors, color interpolation processing is not necessary.

  The image buffer 106 is configured by a volatile high-speed memory, and not only temporarily stores the captured image output from the A / D conversion unit 105 but also as a buffer memory when the image processing unit 107 performs image processing. used. Alternatively, it is also used as a display buffer when a captured image or a captured image stored in the memory card 112 a connected to the memory card IF 112 is displayed on the display unit 110.

  The image processing unit 107 performs general image processing such as noise removal processing, saturation enhancement processing, and contour enhancement processing on the image data captured in the image buffer 106. Further, according to the setting of the electronic camera 101, the captured image is compressed by an image compression method compliant with the JPEG standard or the like, and JPEG data is output. In particular, in the electronic camera 101 according to the present embodiment, the image processing unit 107 performs deformation processing such as enlargement, reduction, rotation, translation, and distortion correction. At this time, the image processing unit 107 performs a process of interpolating a pixel value newly generated after the deformation from the pixel value before the deformation. The interpolation process will be described in detail later.

  The control unit 108 includes a CPU that operates according to a program stored therein in advance, and controls the overall operation of the electronic camera 101. For example, the control unit 108 sets the shooting mode of the electronic camera 101 in accordance with the operation of the shooting mode selection dial and the release button constituting the operation unit 111, and controls the lens and diaphragm of the optical system 102 when the release button is pressed. The mechanical shutter 103 is opened and closed, and the subject image is captured by the image sensor 104. Then, the control unit 108 converts the analog image signal from the image sensor 104 into a digital value by the A / D conversion unit 105, and captures image data for one screen into the image buffer 106. Further, the control unit 108 instructs the image processing unit 107 to perform image processing such as noise removal processing and contour enhancement processing on the image data captured in the image buffer 106, and performs image processing (for example, a JPEG image). ) And a predetermined file name and header information are added to and stored in the memory card 112a via the memory card I / F 112. In particular, in the present embodiment, the control unit 108 performs deformation processing such as enlargement, reduction, rotation, translation, and distortion correction on the image data captured by the image sensor 104 with respect to the image processing unit 107. Command.

  The memory 109 is configured by a non-volatile semiconductor memory such as a flash memory, and stores parameters such as a shooting mode, exposure information, and focus information of the electronic camera 101. The control unit 108 controls the operation of the electronic camera 101 with reference to these parameters. In particular, in the electronic camera 101 according to the present embodiment, parameters and coefficient tables necessary for performing deformation processing such as enlargement, reduction, rotation, translation, and distortion correction are stored in the memory 109 at the time of manufacture. The unit 108 refers to these parameters and coefficient table and instructs the image processing unit 107 to perform deformation processing and interpolation processing. Note that these parameters and coefficient table can be appropriately changed by a user through a setting menu performed via the operation unit 111.

  The display unit 110 is configured by a liquid crystal monitor or the like, and the control unit 108 displays a captured image, a setting menu screen necessary for operating the electronic camera 101, and the like.

  The operation unit 111 includes a power button, a release button, a shooting mode selection dial, a cursor button, and the like. The user operates these operation buttons to use the electronic camera 101. The operation information of these operation buttons is output to the control unit 108, and the control unit 108 controls the operation of the entire electronic camera 101 according to the operation information input from the operation unit 111.

  The memory card IF 112 is an interface for connecting the memory card 112 a to the electronic camera 101, and the control unit 108 reads and writes image data with the memory card 112 a via the memory card IF 112.

The above is the configuration and basic operation of the electronic camera 101 according to the present embodiment. Note that in the case of configuring a single image processing apparatus according to the present invention instead of the electronic camera 101, the image capturing configured by the optical system 102, the mechanical shutter 103, the image sensor 104, and the A / D converter 105 in FIG. The control unit 108 (corresponding to the control unit of the image processing apparatus) reads the captured image data from the memory card 112a to the image buffer 106, and the image processing unit 107 performs deformation processing and interpolation processing. .
[Example of deformation processing]
In the electronic camera 101 according to the present embodiment, deformation processing such as enlargement, reduction, rotation, translation, and distortion correction is performed on an image captured by the image sensor 104 and a captured image stored in the memory card 112a. It can be carried out. In the deformation process, for example, a captured image or a captured image is displayed on the display unit 110, and the cursor button of the operation unit 111 is operated to select a deformation process such as enlargement, reduction, rotation, translation, distortion aberration correction, or the like. Run. For example, the magnification processing is executed for enlargement or reduction processing, and the transformation processing is executed by specifying the rotation center and angle for rotation processing. The present embodiment relates to a technique for interpolating the pixel values of the pixels constituting the output image obtained by transforming the input image, and therefore the operation of the transform process itself is omitted. When the touch panel is arranged on the display unit 110 as a man-machine interface of the operation unit 111, the control unit 108 enlarges / reduces according to operations of deformation processing such as enlargement / reduction / rotation performed on the touch panel. A command including information such as magnification and rotation angle is given to the image processing unit 107.

  Next, a case of enlargement processing will be described as an example of deformation processing. FIG. 2 shows an example in which an image 122 having the same pixel size and the same number of pixels is created by enlarging a bear image 121 composed of 4000 pixels × 3000 pixels k times (k is a real number excluding 0). Show. 2 indicates the positional relationship between the image and the image when the bear's eyes and nose of the image 121 and the image 122 are extracted, and 7 pixels in the horizontal direction with respect to the position of the left eye of the bear. It is the figure which showed the area | region of 5 pixels of vertical directions. Note that the number of pixels and the pixel size (pixel interval) of the image 121 and the image 122 are the same.

  Here, the image 121 before enlargement is indicated by pixels of xy coordinates, and the image 122 after enlargement by k times is indicated by pixels of ij coordinates. For ease of explanation, the pixel size is roughly drawn with respect to the size of the bear's eyes and nose, but in reality, the pixel size is assumed to be small enough to express the shape of the eyes and nose.

  The dotted frame 123 is a diagram arranged so that the left eye of the bear before and after enlargement can be easily compared with the same pixel position, and the left eye 121a of the bear in the image 121 before enlargement is the pixel (2, 2) of the xy coordinates. The bear's left eye 122a in the enlarged image 122 is also located at the pixel (2, 2) of the ij coordinate. On the other hand, the right eye of the bear is located at the pixel (4, 2) in the xy coordinates of the image 121 before enlargement (right eye 121b), but the right eye 122b is at the pixel (5, 2) in the ij coordinates of the image 122 after enlargement. In this case, the magnification is k = 1.5.

  Here, in a general enlargement process, the pixel value of the pixel (5, 2) after enlargement uses the pixel value of the pixel (4, 2) before enlargement as it is. If the pixel does not match the integer coordinate pixel, an interpolation process is used to obtain the enlarged pixel value. FIG. 3 depicts an x-coordinate pixel P (x) before enlargement only in the horizontal direction (only the x-axis or i-axis) and an i-coordinate pixel p (i) after enlargement of k = 4/3 times. It is a figure. As described with reference to FIG. 2, it is assumed that the pixel size is the same (pixel interval). In FIG. 3, the pixel P (0) of the image before enlargement is enlarged 4/3 times to the right in accordance with the pixel p (0) of the image after enlargement.

  In FIG. 3, the pixel value of the pixel p (0) before enlargement can be used as it is, but the pixel value of the pixel p (1) after enlargement is x Since the coordinate is 0.75 and is between the pixel P (0) and the pixel P (1) before enlargement, for example, the pixel value of the pixel P (0) and the pixel value of the pixel P (1) are set to 0.25. : It is necessary to obtain by adding at a ratio of 0.75. Similarly, since the pixel value of the next enlarged pixel p (2) is in the position of the real coordinate instead of the integer coordinate, for example, the pixel value of the pixel P (1) and the pixel P (1) The values need to be averaged. The pixel value of the pixel p (3) needs to be obtained by adding the pixel value of the pixel P (2) and the pixel value of the pixel P (3) at a ratio of 0.25: 0.75. The pixel value of the next pixel p (4) can be replaced with the pixel value of the pixel P (3) because there is a corresponding pixel P (3) of integer coordinates on the x-axis before enlargement.

  However, when the interpolation processing as described above is performed, the pixel p (0) and the pixel p (4) after enlargement are replaced with only the pixel values of the pixels P (0) and P (3) before the enlargement. Therefore, the resolving power (the degree of blur) is the same as that before the enlargement, but the pixel p (1), the pixel p (2), and the pixel p (3) after the enlargement have a plurality of pixels P (x ) Is included, the resolving power (the degree of blur) is deteriorated more than before enlargement. This means the same thing as generating an image (blurred image) with reduced sharpness by the averaging process. As a result, the pixel p (1), the pixel p (2), and the pixel p (3) after enlargement have lower resolution than the pixel p (0) and the pixel p (4), and the pixel before enlargement to be referred to The resolving power varies depending on the position of. In this way, the resolution of the enlarged image is not constant, and there is a problem that the image quality deteriorates due to the appearance of jaggy and the like. In the example of FIG. 3, an example of 4/3 times enlargement processing is given for easy understanding. However, even when deformation processing such as reduction processing, rotation processing, or distortion correction is performed, integer coordinates Since it is necessary to interpolate not only the pixel value of the pixel but also the pixel value at the position of the real number coordinate, an output image in which pixel values of various resolutions are mixed is generated.

  Therefore, in the electronic camera 101 according to the present embodiment, interpolation processing is performed so that the resolving power is constant, so that jaggies and the like do not appear in the output image.

  Next, a case where enlargement processing is performed with pixels arranged in a two-dimensional manner will be described. FIG. 4A is a diagram illustrating a pixel arrangement example before enlargement, and corresponds to, for example, the image 121 described with reference to FIG. In the example of FIG. 4A, a total of 9 pixels are arranged, with 3 pixels at integer positions in the x-axis direction and 3 pixels at integer positions in the y-axis direction. Here, the coordinates of each pixel are represented by pixel P (x, y), and in the example of FIG. 4A, nine pixels from pixel P (0, 0) to pixel P (2, 2) are drawn. It is.

  FIG. 4B is a diagram illustrating a pixel arrangement example after 1.25 times enlargement of FIG. 4A, and corresponds to the image 121 described with reference to FIG. In the example of FIG. 4B, a total of 9 pixels are arranged, with 3 pixels at integer positions in the i-axis direction and 3 pixels at integer positions in the j-axis direction. Here, it is assumed that the pixel size (pixel interval) before and after enlargement does not change. Further, the coordinates of each pixel after the enlargement are expressed by pixel p (i, j), and in the example of FIG. 4B, nine pixels from pixel p (0,0) to pixel p (2,2) are used. Is drawn.

  When the image 121 of the xy coordinate in FIG. 4A is enlarged 1.25 times to create the image 122 of the ij coordinate in FIG. 4B, the pixel P (0,0) and the pixel p (0,0) are created. ) And the pixel value of the pixel p (1, 1) of the ij coordinate after enlargement must be the same as the pixel value of the position (0.8, 0.8) on the xy coordinate before enlargement. Note that the position (0.8, 0.8) on the xy coordinate before the expansion corresponding to the pixel p (1, 1) of the ij coordinate after the expansion is referred to as an interpolation target position.

Here, when the magnification is k, the pixel p (i, j) of the ij coordinate after enlargement, and the coordinate (x0, y0) of the interpolation target position on the xy coordinate corresponding to the pixel p (i, j), There is a relationship.
x0 = i / k (Formula 1)
y0 = j / k (Formula 2)
In the above example, since the pixel p (1, 1) of the ij coordinate after expansion is k = 1.25, the coordinates of the interpolation target position (x0, y0) on the xy coordinate before expansion are as follows: I want.
x0 = 1 / 1.25 = 0.8
y0 = 1 / 1.25 = 0.8
When this interpolation target position (0.8, 0.8) is drawn on the xy coordinates in FIG. 4A, it corresponds to the coordinates (0.8, 0.8) of the black circles, and the interpolation value of this coordinate is If obtained, it becomes the pixel value of the pixel p (1, 1) of the ij coordinate after enlargement in FIG.

  Next, a method for obtaining the interpolation value of the interpolation target position will be described. The interpolation value at the interpolation target position is a process performed by the image processing unit 107. FIG. 5A corresponds to the enlargement processing described in FIGS. 4A and 4B, and it is necessary to obtain the interpolation value at the interpolation symmetry position (0.8, 0.8) indicated by the black circles. There is.

  First, in the present embodiment, a pixel closest to the interpolation target position (0.8, 0.8) is selected from the pixels of integer coordinates before enlargement. In the example of FIG. 5A, the pixel (1, 1) is selected as the central pixel. Then, the interpolation value at the interpolation target position (0.8, 0.8) is obtained from the pixel values of nine pixels (referred to as reference pixels) including the surrounding eight pixels centered on the pixel (1, 1). Here, the pixel of the integer coordinate closest to the interpolation target position (x0, y0) of the real number coordinate can be selected as the pixel of the integer coordinate obtained by rounding off the real number coordinate of the interpolation target position, for example. In the example of FIG. 5A, the integer coordinate pixel nearest to the interpolation target position (0.8, 0.8) before the enlargement is the integer coordinate 1 pixel P (1, 1). 4 is expanded from the pixel values of nine reference pixels from pixel P (0,0) to pixel (2,2) including pixel P (1,1) centered on pixel P (1,1). An interpolation value at the interpolation target position (0.8, 0.8) corresponding to the pixel value of the subsequent pixel p (1, 1) is obtained. The deviation (coordinate difference) from the pixel P (1, 1) closest to the interpolation target position (0.8, 0.8) is Δx in the x-axis direction and Δy in the y-axis direction. Here, Δx and Δy are referred to as fractional coordinates (Δx, Δy).

In the example of FIG. 5A, the fractional coordinates (Δx, Δy) are the interpolation target position (0.8, 0.8) and the pixel P (1) closest to the pixel P (0.8, 0.8). , 1) and the respective xy coordinates can be calculated as follows.
From Δx = 0.8-1, Δx = −0.2
From Δy = 0.8-1, Δy = −0.2
It becomes.

In FIG. 5A, the case of performing the enlargement process of k = 1.25 times has been described. However, in the case of the reduction process of k = 0.8 times, the pixel p (1, 1) after the reduction in FIG. The position to be interpolated in the xy coordinates before reduction corresponding to 1) is (1.25, 1.25), and is the largest position to be interpolated (1.25, 1.25) as shown in FIG. A pixel having a close integer coordinate is a pixel P (1, 1). The fractional coordinates (Δx, Δy) in this case can be calculated as follows.
From Δx = 1.25-1, Δx = 0.25
From Δy = 1.25-1, Δy = 0.25
It becomes.

Here, when the pixel of the integer coordinate closest to the interpolation target position P (x0, y0) is P (x, y), the fractional coordinate (Δx, Δy) can be expressed by (Expression 3).
(Δx, Δy) = (x0−x, y0−y) (Expression 3)
In the case of deformation processing such as rotation and distortion correction as well as enlargement and reduction, reference pixels can be similarly extracted from the interpolation target position (x0, y0) before the deformation processing.
[Calculation method of interpolation value of interpolation target position (x0, y0)]
Next, a method for obtaining the interpolation value of the interpolation target position (x0, y0) will be described. FIG. 6A is a diagram illustrating a generalized example of FIG. 5B. In FIG. 6A, the peripheral pixel P (x-1, y) including the pixel P (x, y) with the pixel P (x, y) of the integer coordinates closest to the interpolation target position (x0, y0) as the center. The interpolation value at the interpolation target position (x0, y0) is calculated from the pixel values of nine reference pixels from y-1) to the pixel P (x + 1, y + 1). Here, the deviation of the interpolation target position (x0, y0) from the integer coordinate pixel P (x, y) is a fractional coordinate (Δx, Δy). Further, the range of the fractional coordinates (Δx, Δy) corresponds to a region 131 indicated by hatching with the pixel P (x, y) as the center as shown in FIG. Shall be satisfied.
−0.5 ≦ Δx <0.5 (Formula 4)
−0.5 ≦ Δy <0.5 (Formula 5)
Hereinafter, the interpolation processing will be described in order. First, as shown in FIG. 6B, the interpolation value at the position where the fractional coordinates Δx of the interpolation target position (x0, y0) are projected on each row is obtained. In the example of FIG. 6B, the interpolation value at the interpolation target position (x0, y-1) in the y-1 row is set to the pixel P (x-1, y-1) and the pixel P ( It is calculated from three pixel values of x, y-1) and pixel P (x + 1, y-1). Similarly, the interpolation value of the interpolation target position (x0, y) in the y row is set to the pixel P (x−1, y), the pixel P (x, y), and the pixel P (x + 1, y) of integer coordinates in the same row. The pixel P (x-1, y + 1), the pixel P (x, y + 1), the pixel P (x-1, y + 1) of the integer coordinates in the same row are calculated from the three pixel values of y, and the interpolation value at the interpolation target position (x0, y + 1) in the y + 1 row is calculated. It is calculated from three pixel values of P (x + 1, y + 1). Here, the three pixels to be referred to are referred to as reference pixels.

  Next, how to obtain the interpolation value of the interpolation target position in each row will be described. In FIG. 8A, the interpolation value of the interpolation target position (x0, y) in the y row is set to three reference pixels (P (x-1, y), P (x, y), P (x + 1) in the same row. , Y)). Here, the range of the interpolation target position (x0, y) (the range of Δx) is a range of −0.5 ≦ Δx <0.5 with the pixel P (x, y) as the center. When the range of Δx is divided into 16, for example, any one of 17 positions including Δx = 0 is an interpolation target position (x0, y) to be obtained. Here, as an arithmetic method for dividing Δx into 17 interpolation target positions, for example, a value obtained by multiplying Δx by 16 can be obtained by rounding off. As a result, Δx is set to 17 levels of -8, -7, -6, -5, -4, -3, -2, -1, 0, 1, 2, 3, 4, 5, 6, 7, 8. It can be divided into division positions. For example, when Δx = −0.5, the division position = −8, and when Δx = 0.4999..., The division position = 8. In the example of FIG. 8B, since Δx of the interpolation target position (x0, y) is +0.3, 16 × 0.3 = 4.8, and the division position is 5 when rounded off.

  FIG. 9 multiplies the pixel values of three reference pixels (P (x-1, y), P (x, y), P (x + 1, y)) in the same row as the interpolation target position (x0, y). It is a coefficient table 141 showing numerical examples of three coefficients (A (Δ), B (Δ), C (Δ)). Here, since the coefficient table 141 is common to the x-axis direction and the y-axis direction, Δx and Δy are collectively expressed as Δ.

Then, using the coefficient table 141, the interpolation value of the interpolation target position (x0, y) can be obtained as shown in (Expression 6).
Interpolation value of interpolation target position (x0, y) = A (Δx) · P (x−1, y) + B (Δx) · P (x, y) + C (Δx) · P (x + 1, y) (formula 6)
For example, in the case of FIG. 8B, since Δx = 0.3, the division position of the coefficient table 141 in FIG. 9 is 16 × 0.3≈ + 5, and the coefficient A (Δ) = 0.018, B ( Δ) = 0.552 and C (Δ) = 0.330 are obtained. When this is applied to (Equation 6), the interpolation value at the interpolation target position (x0, y) is the reference pixel (P (x-1, y), P (x, y), P ( x + 1, y)) can be obtained by the following equation.
Interpolation value of interpolation target position (x0, y) = 0.018 · P (x−1, y) + 0.652 · P (x, y) + 0.330 · P (x + 1, y)
In this way, the interpolation value of the interpolation target position (x0, y) can be obtained. Similarly, the interpolation value of the interpolation target position (x0, y-1) in the y-1 row in FIG. 6B is also the reference pixel (P (x-1, y-1), P (x, y-1) and P (x + 1, y-1)). The interpolation value at the interpolation target position (x0, y + 1) in the y + 1th row is also the pixel value of the reference pixel (P (x-1, y + 1), P (x, y + 1), P (x + 1, y + 1)) in the same row. Can be obtained from

  In this way, the three interpolation target positions ((x0, y−1), (x0, y), (x0, y + 1)) at the x coordinate = x0 (x0 column) of each row in FIG. An interpolated value is determined.

  Next, as shown in FIG. 6C, from the interpolation values at the three interpolation target positions ((x0, y−1), (x0, y), (x0, y + 1)) in the x0 column of each row, the final values are obtained. The interpolation value at the correct interpolation target position (x0, y0) is obtained. In FIG. 6 (c), the interpolation value at the interpolation target position (x0, y0) is represented by three reference interpolation target positions ((x0, y-1), (x0) according to Δy on the y-axis of the x0 column. , Y), (x0, y + 1)), the coefficients (A (Δy), B (Δy), C (Δy)) to be multiplied are obtained from the coefficient table 141. Here, the range of the interpolation target position (x0, y0) (the range of Δy) is the pixel P (x0, y) as shown in FIG. 10A, as in the case of Δx in FIG. Is set to a range of −0.5 ≦ Δy <0.5. Then, the range of Δy is divided into 17 stages including, for example, Δy = 0 divided into 16. As in the case of Δx, the division position can be obtained by rounding a value obtained by multiplying Δy by 16, for example. In the example of FIG. 10B, since Δy of the interpolation target position (x0, y0) is +0.3, 16 × 0.3 = 4.8, and when rounded off, the division position = 5.

Then, using the coefficient table 141, the interpolation value of the interpolation target position (x0, y0) can be obtained as in (Equation 7).
Interpolation value of interpolation target position (x0, y0) = A (Δy) · P (x0, y−1) + B (Δy) · P (x0, y) + C (Δy) · P (x0, y + 1) (formula 7)
For example, in the case of FIG. 10B, since Δy = 0.3, the division position of the coefficient table 141 in FIG. 9 is 16 × 0.3≈ + 5, and the coefficient A (Δ) = 0.018, B ( Δ) = 0.552 and C (Δ) = 0.330 are obtained. When this is applied to (Expression 7), the interpolation value of the interpolation target position (x0, y0) is the three interpolation target positions ((x0, y−1), (x0) of each row obtained in FIG. 6B. , Y) and (x0, y + 1)) are P (x0, y-1), P (x0, y), and P (x0, y + 1), respectively, and can be obtained by the following equations.
Interpolated value of interpolation target position (x0, y0) = 0.018 · P (x0, y−1) + 0.652 · P (x0, y) + 0.330 · P (x0, y + 1)
In this way, the final interpolation value of the interpolation target position (x0, y0) shown in FIG. 6C can be obtained. Then, the above-described series of processing is executed in the same manner for all the pixels p (i, j) of integer coordinates that constitute the output image after the deformation processing.

Here, the above-described series of processing is processing executed by the image processing unit 107 in response to a command from the control unit 108 in FIG.
[Configuration Example of Image Processing Unit 107]
Next, the configuration and operation of the image processing unit 107 will be described. FIG. 11 is a diagram illustrating a configuration example of the image processing unit 107, and illustrates a flow of processing between blocks. In FIG. 11, an image processing unit 107 includes an image input unit 201, an interpolation target position setting unit 202, a reference pixel value extraction unit 203, an interpolation weighting coefficient setting unit 204, an interpolation unit 205, and an image output unit 206. And performing interpolation processing accompanying deformation processing such as enlargement, reduction, rotation, translation, and distortion correction. Here, as described in FIGS. 4A and 4B, the coordinates of the input image before the transformation process are represented by (x, y), and the pixels of the integer coordinates are P (x, y). . Further, the pixel of the integer coordinate of the output image generated by performing the interpolation process after the deformation process is assumed to be p (i, j).

  The image input unit 201 performs processing for capturing an image captured by the image sensor 104 or a captured image stored in the memory card 112 a into the image buffer 106. Then, the image input unit 201 performs a deformation process such as a scaling process on the image captured in the image buffer 106.

  The interpolation target position setting unit 202 sets an interpolation target position (x, y) on the input image corresponding to the pixel p (i, j) of the output image. This processing is performed by, for example, the interpolation target position (0.8,0) before scaling corresponding to the pixel p (i, j) = p (1,1) in the output image after scaling in FIG. .8) is set. The interpolation target position (x0, y0) before zooming can be obtained by (Expression 1) and (Expression 2) described above. The position to be interpolated in an arbitrary deformation process such as a translation process, a rotation process, or a distortion correction process is not limited to the scaling process, and may be calculated by a known method.

  Then, the interpolation target position setting unit 202 rounds the real number coordinates x0 and y0 of the interpolation target position (x0, y0) to the nearest decimal point to the integer coordinates on the input image (x0, y0) (x0, y0). x, y) is calculated, and the fractional coordinates (Δx, Δy) of the interpolation target position (x0, y0) are calculated according to (Equation 3) described above.

  The reference pixel value extraction unit 203 uses the x-coordinate (from (x−1) to (x + 1)) and the y-coordinate (from (y−1) to (y + 1)) of the input image to the interpolation target position (x0, Nine pixels (P (x-1, y-1), P (x, y-1), P (x + 1, y-) centered on integer coordinates (x, y) on the input image closest to y0) 1), P (x-1, y), P (x, y), P (x + 1, y), P (x-1, y + 1), P (x, y + 1), P (x + 1, y + 1)) Extract pixel values.

  The interpolation weighting coefficient setting unit 204 sets the interpolation weighting coefficients (A (Δx), B (Δx), C (Δx)) corresponding to Δx based on the coefficient table 141 of FIG. 9 described above. Then, Δx is multiplied by 16 and the value rounded off is searched from the leftmost column of the coefficient table 141, and the coefficients (A (Δx), B (Δx), C (Δx)) of the row are set. For example, when the value obtained by rounding off (16 × Δx) is −8, A (Δx) = 0.5, B (Δx) = 0.5, C (Δx) = 0, and the value is −7 Are A (Δx) = 0.439, B (Δx) = 0.559, and C (Δx) = 0.002. Similarly, interpolation weighting coefficients (A (Δy), B (Δy), C (Δy)) corresponding to Δy are also set.

  The interpolation unit 205 calculates an interpolation value at the interpolation target position (x0, y0) by the following calculation. Here, the interpolation target position (x0, y0) = (x + Δx, y + Δy).

First, the interpolation value P (x + Δx, y−1) of the interpolation target position (x0, y−1) in the (y−1) th row is calculated as follows using (Equation 6) described above.
P (x + Δx, y−1) = A (Δx) × P (x−1, y−1) + B (Δx) × P (x, y−1) + C (Δx) × P (x + 1, y−1)
Similarly, the interpolation value P (x + Δx, y) at the interpolation target position (x0, y) in the (y) th row is calculated as follows.
P (x + Δx, y) = A (Δx) × P (x−1, y) + B (Δx) × P (x, y) + C (Δx) × P (x + 1, y)
Similarly, the interpolation value P (x + Δx, y + 1) of the interpolation target position (x0, y + 1) in the (y + 1) th row is calculated as follows.
P (x + Δx, y + 1) = A (Δx) × P (x−1, y + 1) + B (Δx) × P (x, y + 1) + C (Δx) × P (x + 1, y + 1)
The interpolation value P (x + Δx) of the final interpolation target position (x0, y0) is used by using the interpolation values of the respective interpolation target positions in the (y-1) th line, the (y) th line, and the (y + 1) th line. , Y + Δy) is calculated as follows using (Equation 7) described above.
P (x + Δx, y + Δy) = A (Δy) × P (x + Δx, y−1) + B (Δy) × P (x + Δx, y) + C (Δy) × P (x + Δx, y + 1)
The image output unit 206 outputs the interpolation value P (x + Δx, y + Δy) of the interpolation target position (x0, y0) obtained by the interpolation unit 205 as the pixel value of the pixel p (i, j) of the output image. Note that the image processing unit 107 performs the above-described series of processing in the same manner for all pixels p (i, j) of integer coordinates constituting the output image after the deformation processing. The output image after the deformation process is temporarily stored in the image buffer 106, and after other image processing and further deformation processing are performed as necessary, image compression processing such as JPEG is performed. The file name is added and stored in the memory card 112a.

  In this way, the electronic camera 101 according to the present embodiment can perform enlargement processing, reduction processing, or deformation processing such as translation processing, rotation processing, and distortion correction processing as described with reference to FIG.

In the present embodiment, it is not specified whether the pixel value of each pixel is a monochrome image with only a luminance value or a color image including color difference and RGB data. However, in the case of a color image, for example, for each color of RGB An output image is generated by the same processing for each value (or each value of Y, Cr, and Cb).
[Effect of this embodiment]
As described with reference to FIG. 3, the electronic camera 101 according to the present embodiment has an effect that the resolving power of each pixel in the output image can be kept constant regardless of the interpolation target position. Hereinafter, the reason will be described in detail.

First, the conditions of the interpolation weighting coefficients (A (Δ), B (Δ), C (Δ)) illustrated in FIG. 9 are shown.
(1) Condition A: “The total of interpolation weighting coefficients is 1 (constant)”
When the condition A is expressed by a mathematical expression, A (Δ) + B (Δ) + C (Δ) = 1. By satisfying this condition A, the average brightness after interpolation becomes the same as before interpolation. That is, the average brightness does not change. In this embodiment, A (Δ) + B (Δ) + C (Δ) = 1, but A (Δ) + B (Δ) + C (Δ) may be constant. In this case, the average brightness after interpolation can be changed simultaneously. For example, if A (Δ) + B (Δ) + C (Δ) = 0.8, the brightness of the output image can be reduced. In the case of a color image, color correction or the like may be performed simultaneously by changing the value of condition A for each color.
(2) Condition B: “The weighted average of reference pixel positions is an interpolation target position.”
If the condition B is expressed by a mathematical formula for the x-axis case,
{(X-1) × A (Δx) + x × B (Δx) + (x + 1) × C (Δx)} / (A (Δx) + B (Δx) + C (Δx)) = (x + Δx)
In summary, using condition A, −A (Δx) + C (Δx) = Δx. Similarly for the y-axis case,
{(Y−1) × A (Δy) + y × B (Δy) + (y + 1) × C (Δy)} / (A (Δy) + B (Δy) + C (Δy)) = (y + Δy)
It becomes.

This is because, by satisfying the condition B, the interpolation value obtained by the interpolation weighting coefficients (A (Δ), B (Δ), C (Δ)) corresponds to the pixel value of the interpolation target position (x + Δx, y + Δy). It shows that it becomes a value. For example, in the case of FIG. 8B described above, since A (Δx) = 0.018 and C (Δx) = 0.330,
−A (Δx) + C (Δx) = − 0.018 + 0.330 = 0.312, and rounding off to the second decimal place results in +0.3. This is the same as Δx = + 0.3 in FIG. That is, the coefficient table 141 in FIG. 9 means that the coefficient value for accurately obtaining the interpolation value at the interpolation target position from the pixel value at the reference pixel position is set.
(3) Condition C: “The evaluation value indicating the extent of the reference pixel position is constant without depending on the interpolation target position.”
Here, the “evaluation value indicating the extent of the reference pixel position” is a value for evaluating how much the pixel values of the reference pixels at a plurality of positions are reflected in the interpolation value at the interpolation target position. The diffusivity of the pixel value of the reference pixel with respect to the position interpolation value is shown. For example, the weighted variance of the coordinates of the reference pixel is used as the “evaluation value representing the extent of the reference pixel position”. This is a sum obtained by multiplying (interpolation target position coordinates-reference pixel coordinates) ² by an interpolation weighting coefficient, as shown in the following equation.

尚、先に説明した図８（ｂ）の例はＮ＝３（Ｎは３以上の整数、ｎは１からＮまでの整数）の場合を示しており、（式８）において、補間加重係数（１）は図９のＡ（Δ）、同様に、補間加重係数（２）はＢ（Δ）、補間加重係数（３）はＣ（Δ）に対応する。ここで、第４の条件を追加して、Ｎ＝４の場合についても同様に考えることができる。

The example of FIG. 8B described above shows a case where N = 3 (N is an integer of 3 or more, and n is an integer from 1 to N). In (Equation 8), the interpolation weighting coefficient (1) corresponds to A (Δ) in FIG. 9, and similarly, the interpolation weighting coefficient (2) corresponds to B (Δ), and the interpolation weighting coefficient (3) corresponds to C (Δ). Here, the fourth condition can be added and the case of N = 4 can be considered similarly.

  In the example of FIG. 8B, since the reference pixel position (1) = − 1, the reference pixel position (2) = 0, the reference pixel position (3) = + 1, and the interpolation target position = + 0.3, From the coefficient table 141, when Δ = + 0.3, the interpolation weighting coefficient (1) = 0.018, the interpolation weighting coefficient (2) = 0.552, and the interpolation weighting coefficient (3) = 0.330. When these values are substituted into (Equation 8), the evaluation value is 0.2508 (about 0.25).

  Similarly, when the interpolation target position is −0.5, the evaluation values are: interpolation target position = −0.5, reference pixel position (1) = − 1, reference pixel position (2) = 0, reference pixel position ( 3) = + 1, and from the coefficient table 141, the interpolation weighting coefficient (1) = 0.500 when Δ = −0.5, the interpolation weighting coefficient (2) = 0.500, and the interpolation weighting coefficient (3) = 0. .000. If these values are substituted into (Equation 8), the evaluation value is 0.25.

  As another example, when the interpolation target position is −0.2, the evaluation values are: interpolation target position = −0.2, reference pixel position (1) = − 1, reference pixel position (2) = 0, reference pixel At position (3) = + 1, from the coefficient table 141, interpolation weight coefficient (1) = 0.049 when Δ = −0.2, interpolation weight coefficient (2) = 0.715, interpolation weight coefficient (3) = 0.236. When these values are substituted into (Equation 8), the evaluation value is 0.2502 (about 0.25).

  Other than the above, the evaluation value of the interpolation weighting coefficient in the coefficient table 141 in FIG. 9 corresponding to −0.5 ≦ Δ <0.5 is also set to be 0.25.

In this way, with respect to the interpolation value P (x + Δ) at the interpolation target position, using the interpolation weighting coefficients A (Δ), B (Δ), and C (Δ) corresponding to Δ,
P (x + Δ) = A (Δ) × P (x−1) + B (Δ) × P (x) + C (Δ) × P (x + 1)
The evaluation value E of the process to be interpolated by the following formula can be obtained by the following formula.
E = ((x + Δ) − (x−1)) ² × A (Δ) + ((x + Δ) −x) ² × B (Δ) + ((x + Δ) − (x + 1)) ² × C (Δ)
= (Δ + 1) ² × A (Δ) + Δ ² × B (Δ) + (Δ−1) ² × C (Δ)
By this evaluation value E, it is possible to appropriately express how much the coordinates of the reference pixel are spread with respect to the coordinates of the interpolation target position. Therefore, by setting the interpolation weighting coefficient so as to keep the evaluation value E constant, it is possible to keep the decrease in the resolution of the interpolation value constant.

As described above, in the present embodiment, the following three conditional expressions must be satisfied by the interpolation weighting coefficients (A (Δ), B (Δ), C (Δ)).
Condition A: A (Δ) + B (Δ) + C (Δ) = 1
Condition B: -A (Δ) + C (Δ) = Δ
Condition C: (Δ + 1) ² × A (Δ) + Δ ² × B (Δ) + (Δ−1) ² × C (Δ) = E
If the fractional coordinates Δ of the interpolation target position and the evaluation value E are given, three simultaneous equations for the interpolation weighting coefficients A (Δ), B (Δ), and C (Δ) are obtained. For example, the interpolation weighting coefficients A (Δ), B (Δ), and C (Δ) are uniquely determined. The coefficient table 141 in FIG. 9 shows the solutions of the above simultaneous equations (interpolation weighting coefficients A (Δ), B (Δ), C (Δ)) when the evaluation value E is set to 0.25. Further, this solution is obtained by interpolating weighting coefficients A (Δ) and B (Δ) when Δ is changed from −0.5 (−8/16) to +0.5 (+8/16) by 1/16 steps. , C (Δ).

In this way, by obtaining the interpolation value of the interpolation target position using the interpolation weight coefficient of the coefficient table 141, the resolution of the interpolation value does not depend on the interpolation target position as in the conventional example described in FIG. The resolving power of the interpolation value can be kept constant. As a result, since the resolving power is kept constant for all the pixels of the output image, it is possible to reduce the occurrence of jaggies even when a known contour enhancement process is strongly applied to the output image.
(Other examples)
In the above description, the weighted variance in which the difference between the coordinates of the reference pixel and the interpolation target position is squared and multiplied by the interpolation weighting coefficient is used as the “evaluation value representing the extent of the reference pixel position”. A primary norm that multiplies the absolute value of the difference between the reference pixel coordinates and the interpolation target position by an interpolation weighting coefficient as shown in FIG.

先の実施形態の例を（式９）に当てはめると、Ｎ＝３（Ｎは３以上の整数、ｎは１からＮまでの整数）として、評価値Ｅは次式で求めることができる。
Ｅ＝｜Δ＋１｜×Ａ（Δ）＋｜Δ｜×Ｂ（Δ）＋｜Δ−１｜×Ｃ（Δ）
尚、上記以外の方法で、参照画素位置の広がり具合を表す任意の評価値を求めるようにしてもよい。

When the example of the previous embodiment is applied to (Equation 9), the evaluation value E can be obtained by the following equation, assuming that N = 3 (N is an integer of 3 or more and n is an integer from 1 to N).
E = | Δ + 1 | × A (Δ) + | Δ | × B (Δ) + | Δ−1 | × C (Δ)
Note that an arbitrary evaluation value representing the extent of the reference pixel position may be obtained by a method other than the above.

  By using the evaluation value E thus obtained as the condition C and solving the simultaneous equations combined with the conditions A and B described above, interpolation weighting coefficients (A (Δ), B (Δ), C (Δ )). The process for generating the pixel value of the output image obtained by transforming the input image using the interpolation weighting coefficients (A (Δ), B (Δ), C (Δ)) is the same as in the previous embodiment, and will be described. Is omitted.

  As described above, the electronic camera 101 according to the present embodiment calculates an interpolation weighting coefficient that satisfies the conditions A, B, and C, and generates a pixel value of an output image using the interpolation weighting coefficient. As a result, the difference in resolving power for each pixel constituting the output image can be eliminated, and deterioration of image quality due to jaggies or the like can be prevented even when the edge enhancement processing is performed on the output image.

  In this embodiment, the example of the electronic camera equipped with the image processing apparatus according to the present invention has been described. However, an image processing program for executing the processing of the image processing unit 107 described in FIG. The same effect can be obtained even when transformation processing is performed on an image downloaded from a network or an image stored in the memory card 112a or the like after being installed on a telephone or a smartphone.

  Further, the image processing apparatus, the image processing program, and the electronic camera according to the present invention have been described by way of examples in the respective embodiments. can do. Therefore, the above-described embodiment is merely an example in all respects and should not be interpreted in a limited manner. The present invention is defined by the claims, and the present invention is not limited to the text of the specification. Further, all modifications and changes belonging to the equivalent scope of the claims are within the scope of the present invention.

DESCRIPTION OF SYMBOLS 101 ... Electronic camera; 102 ... Optical system; 103 ... Mechanical shutter; 104 ... Image pick-up element; 105 ... A / D conversion part; 106 ... Image buffer; Processing unit; 108 ... Control unit; 109 ... Memory; 110 ... Display unit; 111 ... Operation unit; 112 ... Memory card IF; 112a ... Memory card; Table ... 201 ... Image input part; 202 ... Interpolation target position setting part; 203 ... Reference pixel value extraction part; 204 ... Interpolation weight coefficient setting part; 205 ... Interpolation part;・ Image output unit

Claims (10)

An image input unit for inputting an image composed of a plurality of pixels arranged two-dimensionally;
An interpolation target position setting unit that sets an interpolation target position on the input image corresponding to a pixel of the output image obtained by transforming the input image;
A reference pixel value extraction unit that extracts N values (N is an integer of 3 or more) in the vicinity of the interpolation target position among a plurality of pixels constituting the input image as reference pixels, and extracts a pixel value of the reference pixels;
An interpolation weight coefficient setting unit that sets an interpolation weight coefficient for each reference pixel according to the position of the N reference pixels;
An interpolation unit that calculates a multiplication value obtained by multiplying the pixel value of the reference pixel by the interpolation weighting coefficient of the reference pixel, and uses a sum of the multiplication values for the N reference pixels as an interpolation value of the interpolation target position; Prepared,
The interpolation weighting coefficient setting unit
The image processing apparatus, wherein the interpolation weighting coefficient is set so that an evaluation value representing a diffusion degree at the position of the N reference pixels is constant without depending on an interpolation target position. The image processing apparatus according to claim 1.
The interpolation weighting coefficient setting unit
As condition A, the sum of the N interpolation weighting coefficients is constant,
Condition B is that the weighted average of the N reference pixel positions is the interpolation target position;
As the condition C, the evaluation value is constant without depending on the interpolation target position,
The image processing apparatus is characterized in that the interpolation weighting coefficient satisfying at least three conditions is set. The image processing apparatus according to claim 1 or 2,
The evaluation value is calculated by the following formula (n is an integer from 1 to N):

An image processing apparatus. The image processing apparatus according to claim 1 or 2,
The evaluation value is calculated by the following formula (n is an integer from 1 to N):

An image processing apparatus. The image processing apparatus according to claim 3.
When x is an integer coordinate, Δ is a fractional coordinate in a range from −0.5 to +0.5, an interpolation target position is (x + Δ), and the number N of reference pixels is 3,
The reference pixel value extraction unit extracts pixel values P (x−1), P (x), and P (x + 1) of coordinates (x−1), x, and (x + 1) of the three reference pixels,
The interpolation weighting coefficient setting unit
As the condition A, A (Δ) + B (Δ) + C (Δ) = 1,
As the condition B, −A (Δ) + C (Δ) = Δ,
As the condition C, (Δ + 1) ² × A (Δ) + Δ ² × B (Δ) + (Δ−1) ² × C (Δ) = constant,
An image processing apparatus characterized in that a solution of simultaneous equations comprising the three conditions is the interpolation weighting coefficients A (Δ), B (Δ), and C (Δ) for each of the three reference pixels. The image processing apparatus according to claim 4.
When x is an integer coordinate, Δ is a fractional coordinate in a range from −0.5 to +0.5, an interpolation target position is (x + Δ), and the number N of reference pixels is 3,
The reference pixel value extraction unit extracts pixel values P (x−1), P (x), and P (x + 1) of coordinates (x−1), x, and (x + 1) of the three reference pixels,
The interpolation weighting coefficient setting unit
As the condition A, A (Δ) + B (Δ) + C (Δ) = 1,
As the condition B, −A (Δ) + C (Δ) = Δ,
As the condition C, | Δ + 1 | × A (Δ) + | Δ | × B (Δ) + | Δ−1 | × C (Δ) = constant,
An image processing apparatus characterized in that a solution of simultaneous equations comprising the three conditions is the interpolation weighting coefficients A (Δ), B (Δ), and C (Δ) for each of the three reference pixels. The image processing apparatus according to claim 5 or 6,
A coefficient table in which the interpolation weighting coefficients A (Δ), B (Δ), and C (Δ) corresponding to the fractional coordinates Δ in the range of −0.5 to +0.5 are provided in advance;
The interpolation unit refers to the coefficient table to obtain the interpolation weighting coefficient corresponding to the fractional coordinate Δ, calculates a multiplication value obtained by multiplying the reference pixel value by the interpolation weighting coefficient of the reference pixel, and the N An image processing apparatus, wherein a sum total of the multiplication values of the reference pixels is used as an interpolation value at the interpolation target position. In the image processing device according to any one of claims 5 to 7,
An image processing apparatus, wherein the resolution of the fractional coordinate Δ in the range from −0.5 to +0.5 is set from 1/8 to 1/32.   9. An image processing program for executing processing of each unit of the image processing apparatus according to claim 1 by a computer. An electronic camera equipped with the image processing apparatus according to claim 1,
An imaging unit that outputs an image obtained by imaging a subject to the image input unit;
An electronic camera comprising: a storage unit that records an image interpolated by the image processing apparatus.

Images