JP2010113554A - Image processing apparatus and image processing method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an image processing apparatus and an image processing method, having a high processing speed and capable of reducing a circuit area and cost. <P>SOLUTION: When calculating an interpolation value P of a luminance component, an interpolation processing part 223 calculates intermediate values PA to PD based on sixteen image data P<SB>11</SB>to P<SB>44</SB>and a weight coefficient received from a coefficient part 224. An output data control part 225 adds the intermediate values PA to PD to calculate an interpolation value P. When calculating an interpolation value of a color difference component, the interpolation processing part 223 respectively calculates the intermediate values PA to PD as Cb and Cr components of a first interpolation point and Cb and Cr components of a second interpolation point based on four Cb data PA<SB>22</SB>to PA<SB>33</SB>and four Cr data PB<SB>22</SB>to PB<SB>33</SB>of the first interpolation point, four Cb data PC<SB>22</SB>to PC<SB>33</SB>and four Cr data PD<SB>22</SB>to PD<SB>33</SB>of the second interpolation point, and the weight coefficient received from the coefficient part 224. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to an image processing apparatus and an image processing method for performing processing for deforming an image.

  When the image processing apparatus displays the image by enlarging, reducing, rotating, or transforming it, the image processing apparatus performs geometric conversion on the image data representing the image and displays it. Geometric transformation is transformation for projecting image data to another coordinate system using a predetermined coordinate transformation formula. When the image data is vector type data, geometric transformation can be performed only by performing coordinate transformation processing. However, when the image data is raster type data, in addition to coordinate transformation processing, image rearrangement processing and It is necessary to perform image data interpolation processing.

  The coordinate conversion process is a process of converting the image data of the input image into image data in the coordinate system of the output image, or a process of converting the image data of the output image into image data in the coordinate system of the input image. In the image rearrangement process, the input image is set so that each pixel of the image indicated by the image data after the coordinate conversion process corresponds to a grid-like pixel position (hereinafter referred to as a “grid point”) determined by the coordinate system of the output image. This is a process for rearranging the pixels.

  In the image data interpolation process, if the position of each pixel of the output image indicated by the image data converted by the coordinate conversion process does not match the position of the grid point, the value of the pixel placed at the grid point is calculated by interpolation. It is processing to do. There are several methods of interpolation, and it is necessary to select a method according to the purpose in consideration of influences such as simplicity of processing, accuracy, and occurrence of distortion.

  Along with diversification of image processing apparatuses, many image processing apparatuses have a function of performing enlarged display or reduced display of an image. As an image processing apparatus, for example, a digital still camera can perform enlargement / reduction of a color image by digital signal processing by a digital zoom function. Image scaling greatly affects image quality and processing speed depending on the method used to calculate the pixel value (hereinafter referred to as “interpolation value”) of the point to be interpolated (hereinafter referred to as “interpolation point”). .

  Examples of the interpolation method include nearest neighbor interpolation (hereinafter also referred to as “nearest neighbor method”), bilinear interpolation (hereinafter also referred to as “bilinear method”), and cubic convolution interpolation ( (Hereinafter also referred to as “bicubic method”). The nearest neighbor method and the bilinear method are simpler and faster than the bicubic method, but the interpolation accuracy is low and the image quality is inferior. The bicubic method has higher interpolation accuracy and better image quality than the nearest neighbor method and the bilinear method, but the calculation processing is complicated and takes time. Therefore, it is necessary to select an interpolation method depending on whether the required specification emphasizes processing speed or image quality.

  FIG. 14 is a block diagram illustrating a configuration of an image processing apparatus 100 according to a conventional technique. The image processing apparatus 100 is an apparatus that enlarges or reduces an image using two interpolation methods, and includes an input image memory 101, an interpolation calculation unit 102, and an output image memory 103. The input image memory 101 is a memory that stores image data representing an input image, and stores image data for each type of color component, such as a luminance component and a color difference component. The interpolation calculation unit 102 includes a first interpolation processing unit 102a and a second interpolation processing unit 102b. The output image memory 103 is a memory that stores the interpolation value calculated by the interpolation calculation unit 102 for each type of color component.

  The first interpolation processing unit 102a and the second interpolation processing unit 102b calculate interpolation values by different interpolation methods. For example, the first interpolation processing unit 102a calculates the interpolated value of the luminance component that is the first component by the bicubic method, and the second interpolation processing unit 102b has the second component and the third component by the bilinear method. An interpolated value of the color difference component is calculated. The interpolation calculation unit 102 illustrated in FIG. 14 includes two interpolation processing units. When there are three or more types of color components, by providing an interpolation processing unit corresponding to each component, Interpolated values can be calculated in parallel.

  For example, an image processing apparatus described in Patent Literature 1 includes a first pixel density conversion unit that performs high-quality pixel density conversion such as a projection method, and a second pixel density conversion that performs high-speed pixel density conversion such as simple thinning. Including parts. The first pixel density conversion unit performs pixel density conversion on a Y signal obtained from YUV image data, that is, a luminance signal. The second pixel density conversion unit includes two pixel density conversion units, and performs pixel density conversion on the U signal and the V signal obtained from the YUV image data, that is, two color difference signals.

  The first pixel density conversion unit corresponds to the interpolation processing unit 102a shown in FIG. 14, that is, the interpolation processing unit that calculates the interpolation value by the bicubic method, and the second pixel density conversion unit is the interpolation shown in FIG. This corresponds to the processing unit 102b, that is, an interpolation processing unit that calculates an interpolation value by the bilinear method.

  The image processing apparatus described in Patent Document 2 performs luminance processing on a luminance signal by performing signal processing using an interpolation method of a calculation area of 4 × 4 pixels or more, for example, a three-dimensional convolution interpolation method, that is, a third-order convolution interpolation method. A signal processing unit, and a color difference signal processing unit that performs signal processing using another interpolation method that is an interpolation method of a calculation area within 2 × 2 pixels that is different from the three-dimensional convolution interpolation of the color difference signal The image signal is divided into a luminance signal and a color difference signal, and signal processing is performed using different interpolation methods. The signal processing is processed by executing an interpolation program corresponding to the interpolation method.

  Other interpolation methods are, for example, bilinear interpolation, nearest neighbor interpolation, simple decimation or simple interpolation. It is also possible to store an interpolation program corresponding to each interpolation method in a ROM (Read Only Memory) and select and execute the interpolation program corresponding to the target interpolation method. Since the interpolation method with a narrow calculation area is used for the color difference signal, it is possible to suppress the color difference caused by the distant position and to prevent color misregistration.

Japanese Patent Laid-Open No. 9-298660 Japanese Patent Laid-Open No. 2001-14454

  The image processing apparatus described in Patent Document 1 utilizes the fact that human eyes are more sensitive to luminance signals than color difference signals, and reduces the amount of computation to convert the images into those with less image quality degradation. Can do. The image processing apparatus disclosed in Patent Document 1 can perform high-speed processing, but requires a circuit that performs pixel density conversion for each of the luminance signal and the two color difference signals, which increases the circuit area.

  In the image processing apparatus described in Patent Document 2, since signal processing is performed by a predetermined interpolation method by executing an interpolation program, it takes more time than signal processing by hardware and is not suitable for real-time processing. . In order to apply to an apparatus that requires real-time processing, it is necessary to implement interpolation processing by interpolation using hardware. However, when the interpolation processing is realized by hardware, there is a problem that the circuit area increases as in the image processing apparatus of Patent Document 1.

  An object of the present invention is to provide an image processing apparatus and an image processing method capable of achieving high processing speed and reducing circuit area and cost.

The present invention comprises an input means for inputting image data representing an image constituted by pixels composed of mutually different first to third components;
An instruction means for instructing a magnification for changing the size of the image;
Interpolation value calculating means for calculating an interpolation value by a cubic convolution interpolation method or a bilinear interpolation method;
Conversion means for converting the image data input by the input means into image data representing an image having a size changed by the magnification indicated by the instruction means, using the interpolation value calculation means;
Output means for outputting image data converted by the conversion means,
When the conversion means causes the interpolation value calculation means to calculate the interpolation values of the first to third components at the interpolation points that are positions corresponding to the pixels of the image indicated by the image data after conversion in the coordinate system before conversion,
For the first component, for each interpolation point, among the pixels of the input image indicated by the image data input by the input means, four pixels surrounding the interpolation point and twelve pixels surrounding the four pixels are included. Based on the first component of the 16 pixels including, the interpolation value of the first component is calculated by the third-order convolution interpolation method,
For the second component and the third component, the interpolation value of the second component at the first interpolation point of the two interpolation points is simultaneously calculated for each of the two interpolation points, and the first of the pixels of the input image. Based on the second component of the four pixels surrounding the interpolation point, the interpolation value of the third component at the first interpolation point is based on the third component of the four pixels surrounding the first interpolation point. Interpolation values of the second component at the second interpolation point different from the first interpolation point of the two interpolation points are the four pixels surrounding the second interpolation point of the pixels of the input image. And the interpolated value of the third component at the second interpolation point based on the third component of the four pixels surrounding the second interpolation point by bilinear interpolation Let me calculate
The interpolation value of the first component calculated by the interpolation value calculation means is the first component, the interpolation value of the second component calculated by the interpolation value calculation means is the second component, and the third component is calculated by the interpolation value calculation means. The image processing apparatus is characterized in that image data representing an image constituted by pixels having the third interpolation value as the third component is converted image data.

According to another aspect of the present invention, there is provided an input step for inputting image data representing an image constituted by pixels including first to third components different from each other;
An instruction step for instructing a magnification for changing the size of the image;
An image having a size obtained by changing the image data input in the input step by the magnification specified in the instruction step by the interpolation value calculation unit that calculates the interpolation value by the cubic convolution interpolation method or the bilinear interpolation method. A conversion step for converting into image data representing,
When the interpolation value calculation unit calculates the interpolation values of the first to third components at the interpolation point that is the position corresponding to the pixel of the image indicated by the image data after conversion in the coordinate system before conversion,
For the first component, for each interpolation point, among the pixels of the input image indicated by the image data input in the input step, four pixels surrounding the interpolation point and twelve pixels surrounding the four pixels are included. Calculating an interpolated value of the first component by a third-order convolution method based on the first component of the 16 pixels included;
For the second component and the third component, the interpolation value of the second component at the first interpolation point of the two interpolation points is simultaneously calculated for each of the two interpolation points, and the first of the pixels of the input image. Based on the second component of the four pixels surrounding the interpolation point, the interpolation value of the third component at the first interpolation point is based on the third component of the four pixels surrounding the first interpolation point. Interpolation values of the second component at the second interpolation point different from the first interpolation point of the two interpolation points are the four pixels surrounding the second interpolation point of the pixels of the input image. And the interpolated value of the third component at the second interpolation point based on the third component of the four pixels surrounding the second interpolation point by bilinear interpolation Calculate
The interpolation value of the first component calculated by the interpolation value calculator is the first component, the interpolation value of the second component calculated by the interpolation value calculator is the second component, and the third component calculated by the interpolation value calculator A conversion step of converting image data representing an image constituted by pixels having the interpolation value of the third component into image data after conversion;
And an output step for outputting the image data converted in the conversion step.

  According to the present invention, image data representing an image composed of mutually different first to third components is input by the input means, and the magnification for changing the size of the image is instructed by the instruction means. . The interpolation value calculation means calculates the interpolation value by the cubic convolution interpolation method or the bilinear interpolation method. The conversion means converts the image data input by the input means using the interpolation value calculation means into image data representing an image having a size changed at the magnification indicated by the instruction means, and the output means converts the image data by the conversion means. The converted image data is output.

  Further, the interpolation value calculation means calculates the interpolation values of the first to third components at the interpolation points at the positions corresponding to the pixels of the image indicated by the converted image data in the coordinate system before the conversion. For the first component, for each interpolation point, among the pixels of the input image indicated by the image data input by the input means, four pixels surrounding the interpolation point and 12 pixels surrounding the four pixels are included. Based on the first component of the 16 pixels including the pixel, the interpolation value of the first component is calculated by the third-order convolution interpolation method.

  As for the second component and the third component, the interpolation value of the second component at the first interpolation point of the two interpolation points is the first of the pixels of the input image at the same time every two interpolation points. Based on the second component of the four pixels surrounding the interpolation point, the third component interpolation value at the first interpolation point is based on the third component of the four pixels surrounding the first interpolation point. Four pixels whose second component interpolation value at a second interpolation point different from the first interpolation point of the two interpolation points surrounds the second interpolation point of the pixels of the input image. And the interpolated value of the third component at the second interpolation point based on the third component of the four pixels surrounding the second interpolation point by bilinear interpolation Calculated.

  Then, the interpolation value of the first component calculated by the interpolation value calculation means is the first component, the interpolation value of the second component calculated by the interpolation value calculation means is the second component, and the first value calculated by the interpolation value calculation means. Image data representing an image composed of pixels having the third component interpolation value as the third component is the converted image data.

  Therefore, since the interpolation value calculation means for calculating the interpolation value by the cubic convolution interpolation method or the bilinear interpolation method has means configured by hardware, for example, a program for calculating the interpolation value is provided. It is possible to calculate the interpolation value by the third-order convolution interpolation method in a shorter time than the execution and sequential calculation, and the interpolation value calculation means can be used to calculate the interpolation value calculated by the bilinear interpolation method. Since it can be calculated, hardware for calculating an interpolation value by bilinear interpolation is not required, and the circuit area and cost can be reduced. That is, the image can be interpolated with a circuit having a high processing speed and a small circuit area and cost.

  According to the invention, in the input step, image data representing an image composed of mutually different first to third components is input, and in the instruction step, a magnification for changing the size of the image is instructed. .

  In the conversion step, the image data input in the input step is changed by the magnification specified in the instruction step by the interpolation value calculation unit that calculates the interpolation value by the cubic convolution interpolation method or the bilinear interpolation method. When converting to image data representing a size image, the interpolation values of the first to third components at the interpolation point at the position corresponding to the pixel of the image indicated by the image data after conversion in the coordinate system before conversion are interpolated values When calculating by the calculation unit, for the first component, for each interpolation point, among the pixels of the image indicated by the image data input in the input step, four pixels surrounding the interpolation point and the four pixels are included. Based on the first component of the 16 pixels including the 12 surrounding pixels, the interpolated value of the first component is calculated by the cubic convolution interpolation method.

  For the second component and the third component, the interpolation value of the second component at the first interpolation point of the two interpolation points is simultaneously calculated for each of the two interpolation points, and the first of the pixels of the input image. Based on the second component of the four pixels surrounding the interpolation point, the interpolation value of the third component at the first interpolation point is based on the third component of the four pixels surrounding the first interpolation point. Interpolation values of the second component at the second interpolation point different from the first interpolation point of the two interpolation points are the four pixels surrounding the second interpolation point of the pixels of the input image. And the interpolated value of the third component at the second interpolation point based on the third component of the four pixels surrounding the second interpolation point by bilinear interpolation calculate.

  Then, the interpolation value of the first component calculated by the interpolation value calculation unit is the first component, the interpolation value of the second component calculated by the interpolation value calculation unit is the second component, and the first value calculated by the interpolation value calculation unit is Image data representing an image composed of pixels having the third component interpolation value as the third component is defined as converted image data. Further, in the output step, the image data converted in the conversion step is output.

  Therefore, when the image processing method according to the present invention is applied, a circuit configured by hardware, for example, is used as an interpolation value calculation unit that calculates an interpolation value by a cubic convolution interpolation method or a bilinear interpolation method. Therefore, it is possible to calculate the interpolated value by the third-order convolution interpolation method in a shorter time than executing the program for calculating the interpolated value and sequentially calculating, and using the interpolated value calculating unit, Since the interpolation value calculated by the interpolation method can be calculated, hardware for calculating the interpolation value by the bilinear interpolation method is unnecessary, and the circuit area and cost can be reduced. That is, the processing speed is high, and the circuit area and cost can be reduced.

  FIG. 1 is a block diagram showing a schematic configuration of an image processing apparatus 1 according to an embodiment of the present invention. The image processing device 1 is a device that performs processing for transforming an image, such as enlargement, reduction, or rotation, and includes an input image memory 11, an interpolation calculation unit 12, and an output image memory 13. The image processing method according to the present invention is processed by the image processing apparatus 1.

  The input image memory 11 is a memory that stores image data representing an image input by an input device (not shown) that is an input unit such as an image sensor. The input image memory 11 is a component of each pixel constituting the image indicated by the image data, such as a luminance component and a color difference component when the image is represented by a YCbCr (Luma Chroma-blue Croma-red) method. Remember for each type. The image data A shown in FIG. 1 is data representing, for example, a luminance component, and the image data B is data representing, for example, one color difference component of two color difference components.

  The interpolation calculation unit 12 selects any one of a bicubic method that is a cubic convolution interpolation method, a bilinear method that is a bilinear interpolation method, and a nearest neighbor method, and the selected interpolation method Can be used to calculate the interpolation value. By using these interpolation methods, the interpolation calculation unit 12 converts the input image into an output image at a magnification indicated by an operation unit (not shown) that is an instruction unit.

  The interpolation calculation unit 12 includes an input data control unit 121, an interpolation processing unit 122, an output data control unit 123, and an input / output control unit 124. The input data control unit 121 reads out and reads out from the input image memory 11 pixel data representing a pixel instructed by the input / output control unit 124 and a component instructed from the input / output control unit 124 among the components of the pixel. The pixel data or the component data whose value instructed from the input / output control unit 124 is “0” and the interpolation coefficient instructed from the input / output control unit 124 are input to the interpolation processing unit 122. The interpolation coefficient is a coefficient necessary for calculating the interpolation value, and includes a bicubic interpolation coefficient, a bilinear interpolation coefficient, and a nearest neighbor interpolation coefficient. Hereinafter, the interpolation coefficient is also referred to as a weighting coefficient.

  The interpolation processing unit 122 is a logic circuit that calculates an intermediate value for calculating an interpolation value, calculates an intermediate value based on input data and a weighting factor input by the input data control unit 121, and calculates the calculated intermediate value. A value is input to the output data control unit 123. The output data control unit 123 calculates an interpolation value by performing an operation corresponding to the pixel component instructed from the input / output control unit 124 on the intermediate value input from the interpolation processing unit 122, and calculates the calculated interpolation value. Are written in the output image memory 13 as converted image data. When there is no need for calculation, the intermediate value input from the interpolation processing unit 122 is treated as an interpolation value, and the interpolation value is written in the output image memory 13 as converted image data.

  The input / output control unit 124 controls the input data control unit 121 and the output data control unit 123. The input data control unit 121 is instructed with an image and a component of pixel data read from the input image memory 11, a component whose value is “0”, and a weighting coefficient. The output data control unit 123 is instructed with a pixel component for which an interpolation value is calculated.

  The output image memory 13 is a memory that stores the interpolation value calculated by the interpolation calculation unit 12, that is, the converted image data, for each image component. For example, the image data A ′ is image data representing a luminance component, and the image data B ′ is image data representing one color difference component of two color difference components. Image data stored in the output image memory 13 is output to an output device (not shown) which is an output means such as a liquid crystal display or a printer.

  The input image memory 11, the input data control unit 121, the interpolation processing unit 122, the output data control unit 123, and the output image memory 13 are interpolation value calculation means or an interpolation value calculation unit. The input / output control unit 124 is a conversion unit.

  FIG. 2 is a diagram illustrating the relationship between the position of each pixel 51 and the position of the interpolation point 52 that form the image indicated by the image data stored in the input image memory 11. Each pixel 51 is arranged in a grid, and one pixel 51 is arranged at each grid point. The position of each lattice point on the XY coordinate axis is an integer value position for both the X coordinate and the Y coordinate.

  The position of the interpolation point 52 is a position that represents the pixel of the output image in the coordinate system of the input image when the pixel of the input image is converted into the output image, and is represented by coordinates (u, v). The interpolation value at the interpolation point 52 is P (u, v). When the interpolation value P (u, v) (hereinafter also simply referred to as “P”) is calculated using the bicubic method, 16 pixels in the region 53, that is, four pixels surrounding the interpolation point 52, and When calculating based on 16 pixels including 12 pixels surrounding these 4 pixels and calculating the interpolation value P using the bilinear method, four pixels in the region 54, that is, interpolation points The calculation is performed based on the four pixels surrounding 52.

FIG. 3 is an enlarged view of the region 53 shown in FIG. Image data _{P 11, P 12, P 13} , P 14, P 21, P 22, P 23, P 24, P 31, P 32, P 33, P 34, P 41, P 42, P 43, P 44 ( (Hereinafter, also referred to as “P _{11 to} P ₄₄ ”) indicates the calculation target components of each pixel in the region 53. When the bicubic method is used, the interpolation value P (u, v) can be calculated from the equations (1) to (3). Variables x _{1 to} x ₄ are distances in the X-axis direction from the interpolation point to 16 pixels in the region 53, and variables y _{1 to} y ₄ are 16 pixels in the region 53 from the interpolation point. Is the distance in the Y-axis direction. The weighting factors f (x ₁ ) to f (x ₄ ) and f (y ₁ ) to f (y ₄ ) are the variables t ₁ to x ₄ and the variables t ₁ to x ₄ and It is a coefficient obtained by substituting variables y _{1 to} y ₄ . [] In Equations (2) and (3) is a Gaussian symbol that rounds off the decimal part and takes the integer part.

x ₁ = 1 + (u- [u]), x ₂ = (u- [u]), x ₃ = 1- (u- [u]), x ₄ = 2- (u- [u]),
y ₁ = 1 + (v- [v]), y ₂ = (v- [v]), y ₃ = 1- (v- [v]), y ₄ = 2- (v- [v])… (2)
f (t) = sin (πt) / (πt) (3)

  Since the sine function included in Expression (3) cannot be calculated digitally, the interpolation calculation unit 12 calculates using the approximate expression expressed only by the product-sum operation shown in Expression (4). The constant a is a real number, and the interpolation characteristic can be controlled by changing the value of the constant a. In the image processing apparatus 1, the value of the constant a is “−1”, for example.

FIG. 4 is an enlarged view of the region 54 shown in FIG. In FIG. 2, the components of the four pixels in the region 54 are P ₂₂ , P ₂₃ , P ₃₂ , and P ₃₃ , but in FIG. 4, they are generalized as P _ij , P _{ij + 1} , P _{i + 1j} , and P _{i + 1j + 1} . ing. Variables i and j are positive integers. When the bilinear method is used, the interpolation value P can be calculated from equations (5) and (6) using a weighting factor corresponding to the distance between the four pixels surrounding the interpolation point and the interpolation point. Compared with the bicubic method, the bilinear method has an advantage that the amount of calculation is small although the interpolation accuracy is low. [] In Equation (6) is a Gaussian symbol.
P = {(i + 1) -u} {(j + 1) -v} P _ij + {(i + 1) -u} (vj) P _{ij + 1}
+ {(ui) {(j + 1) -v} P _{i + 1j} + (ui) (vj) P _{i + 1j + 1} (5)
i = [u], j = [v] (6)

FIG. 5 is a diagram for explaining an example in which an interpolation value is obtained using the nearest neighbor method. In the nearest neighbor method, an interpolation value is calculated by Expressions (7) and (8) based on pixels surrounding an interpolation point. The nearest neighbor method has the advantage that the position accuracy is maximum and the distance of ½ of the distance between the pixels is shifted, but the value of the original pixel component is applied as it is and does not change. Furthermore, the amount of calculation is small compared to the bilinear method. [] In Equation (8) is a Gaussian symbol.
P = P _mn (7)
m = [u + 0.5], n = [v + 0.5] (8)

In order to compare the calculation amount of the bicubic method and the calculation amount of the bilinear method, the respective calculation formulas are modified. For the bicubic method, Formula (1) is expanded and transformed into Formulas (9) to (13). PA to PD calculated by the equations (9) to (12) are intermediate values that are intermediate results for obtaining the final interpolation value P, and the 16 pixels shown in FIG. These are the calculation results for each of the four pixels equally divided into four pixels, four upper right pixels, and four lower right pixels. The interpolation value P is the sum of the intermediate values PA to PD, and is calculated by the equation (13).
PA = P ₁₁ f (x ₁ ) f (y ₁ ) + P ₂₁ f (x ₂ ) f (y ₁ ) + P ₁₂ f (x ₁ ) f (y ₂ ) + P ₂₂ f (x ₂ ) f ( y ₂ ) (9)
PB = P ₃₁ f (x ₃ ) f (y ₁ ) + P ₄₁ f (x ₄ ) f (y ₁ ) + P ₃₂ f (x ₃ ) f (y ₂ ) + P ₄₂ f (x ₄ ) f ( y ₂ )… (10)
PC = P ₁₃ f (x ₁ ) f (y ₃ ) + P ₂₃ f (x ₂ ) f (y ₃ ) + P ₁₄ f (x ₁ ) f (y ₄ ) + P ₂₄ f (x ₂ ) f ( y ₄ )… (11)
PD = P ₃₃ f (x ₃ ) f (y ₃ ) + P ₄₃ f (x ₄ ) f (y ₃ ) + P ₃₄ f (x ₃ ) f (y ₄ ) + P ₄₄ f (x ₄ ) f ( y ₄ )… (12)
P = PA + PB + PC + PD… (13)

For the bilinear method, equations (5) and (6) are transformed into equations (14) and (15).
f (x ₁ ) = (i + 1) -u, f (x ₂ ) = ui, f (y ₁ ) = (j + 1) -v, f (y ₂ ) = vj… (14)
P = f (x ₁ ) f (y ₁ ) P _ij + f (x ₁ ) f (y ₂ ) P _{ij + 1} + f (x ₂ ) f (y ₁ )) P _{i + 1j} + f (x ₂ ) f (y ₂ ) P _{i + 1j + 1}
… (15)

  As for the weighting coefficient, the bilinear method is calculated according to the equation (14) based on the four pixels surrounding the interpolation point. The bicubic method calculates the four pixels surrounding the interpolation point and the four pixels. Based on 16 pixels including 12 surrounding pixels, the calculation is performed using equations (2) and (4). That is, the amount of calculation for calculating the weighting coefficient by the bicubic method is larger than the amount of calculation for calculating the weighting coefficient by the bilinear method.

  These weighting factors can be calculated in advance if the magnification and interpolation accuracy are determined in advance. By storing the weighting factors calculated in advance as tabulated data and performing interpolation calculation using the stored weighting factors, the amount of calculation can be reduced and the processing time can be shortened.

  As for the interpolation calculation, when the formulas (9) to (12) that are the calculation formulas of the bicubic method are compared with the formula (15) that is the calculation formula of the bilinear method, the configuration of the term of the formula (15) is (9) It is the same as the structure of each term of (12). Therefore, even if the weighting factor is tabulated, the amount of computation in the bicubic method is four times the amount of computation in the bilinear method, and depending on the configuration of the logic circuit, the bicubic method requires more processing time than the bilinear method. Take it.

  FIG. 6 is a block diagram illustrating a configuration of the interpolation processing unit 122. In FIG. 6, a coefficient unit 125 is added in which data in which weight coefficients are calculated in advance is tabulated and stored. In FIG. 1, the weighting factor is calculated by the input / output control unit 124, and the calculated weighting factor is input from the input / output control unit 124 to the interpolation processing unit 122 via the input data control unit 121. In FIG. The weighting factor stored in the coefficient unit 125 is input to the interpolation processing unit 122 to shorten the processing time.

The interpolation processing unit 122 includes eight logic circuit units (hereinafter referred to as “CAL”) 31 to 38. When the interpolation value is calculated by the bicubic method, the input data control unit 121 includes 16 pixels of image data P _{11 to} P ₄₄ including four pixels surrounding the interpolation point and twelve pixels surrounding the four pixels. Is input to the interpolation processing unit 122. The coefficient unit 125 inputs weighting coefficients f (x ₁ ) to f (x ₄ ) and f (y ₁ ) to f (y ₄ ) to the interpolation processing unit 122. The interpolation processing unit 122 outputs the intermediate values PA to PD calculated by the CALs 31 to 38 and inputs them to the output data control unit 123. The output data control unit 123 adds the intermediate values PA to PD and outputs the interpolation value P.

  FIG. 7 is a logic circuit diagram of the interpolation processing unit 122. FIG. 8 is a logic circuit diagram of a portion of the interpolation processing unit 122 that calculates the intermediate value PA. The CAL 31 includes three multipliers 311, 312, 314 and one adder 313, and the CAL 33 includes three multipliers 331, 332, 334 and one adder 333. Multipliers 311, 312, 314, 331, 332, and 334 each multiply two inputs, and adders 313 and 333 each add two inputs.

The multiplier 311 receives the image data P ₁₁ and the weighting factor f (x ₁ ), and the multiplier 312 receives the image data P ₂₁ and the weighting factor f (x ₂ ). The output of the multiplier 311 and the output of the multiplier 312 are input to the adder 313, and the output of the adder 313 and the weighting factor f (y ₁ ) are input to the multiplier 314. The multiplier 331 receives the image data P ₁₂ and the weighting factor f (x ₁ ), and the multiplier 332 receives the image data P ₂₂ and the weighting factor f (x ₂ ). The adder 333 receives the output of the multiplier 331 and the output of the multiplier 332, and the multiplier 334 receives the output of the adder 333 and the weighting factor f (y ₂ ). The output of CAL31 and the output of CAL33 are input to an adder 391, and the value added by the adder 391 is output as an intermediate value PA.

  Combination of CAL32, CAL34 and adder 392 for outputting intermediate value PB, combination of CAL35, CAL37 and adder 393 for outputting intermediate value PC, and CAL36, CAL38 and addition for outputting intermediate value PD The combination of the units 394 has the same logic circuit configuration as the combination of the CAL31, CAL33, and the adder 391, and a description thereof will be omitted to avoid duplication.

Intermediate values PA to PD are calculated based on equations (16) to (19) obtained by modifying equations (9) to (12).
PA = {P ₁₁ f (x ₁ ) + P ₂₁ f (x ₂ )} f (y ₁ ) + {P ₁₂ f (x ₁ ) + P ₂₂ f (x ₂ )} f (y ₂ )… (16 )
PB = {P ₃₁ f (x ₃ ) + P ₄₁ f (x ₄ )} f (y ₁ ) + {P ₃₂ f (x ₃ ) + P ₄₂ f (x ₄ )} f (y ₂ )… (17 )
PC = {P ₁₃ f (x ₁ ) + P ₂₃ f (x ₂ )} f (y ₃ ) + {P ₁₄ f (x ₁ ) + P ₂₄ f (x ₂ )} f (y ₄ )… (18 )
PD = {P ₃₃ f (x ₃ ) + P ₄₃ f (x ₄ )} f (y ₃ ) + {P ₃₄ f (x ₃ ) + P ₄₄ f (x ₄ )} f (y ₄ )… (19 )

  CAL31 is a logic circuit that calculates the first term of Equation (16), CAL32 is a logic circuit that adds the first term of Equation (17), and CAL33 is the second term of Equation (16). CAL34 is a logic circuit that adds the second term of Equation (17), CAL35 is a logic circuit that adds the first term of Equation (18), and CAL36 is 19) is a logic circuit for adding the first term of the equation, CAL37 is a logic circuit for adding the second term of the equation (18), and CAL38 is a logic circuit for adding the second term of the equation (19). is there.

  Each of the CALs 31 to 38 includes three multipliers and one adder, and the interpolation processing unit 122 includes four adders. That is, it is composed of a total of 24 multipliers and 12 adders. On the other hand, if the interpolation processing unit 122 is logically configured based on the equations (9) to (12), 32 multipliers and 12 adders are required. Therefore, by configuring the logic based on the equations (16) to (19), the number of multipliers can be reduced by 8 as compared with the case where the logic configuration is based on the equations (9) to (12). Can be reduced.

  FIG. 9 is a diagram illustrating an arrangement of pixel data used when one interpolation value by the bilinear method is calculated using the interpolation calculation unit 12. When obtaining one interpolation value by the bilinear method, the interpolation calculation unit 12 uses pixel data in the region 54, that is, pixel data of four pixels surrounding the interpolation point, and sets the values of the other 12 pixel data to “0”. "

Specifically, the interpolation calculation unit 12 supplies pixel data P ₂₂ , P ₃₂ , P ₂₃ , P ₃₃ (hereinafter referred to as “P _{22 to} P ₃₃ ”) of four pixels surrounding the interpolation point to the interpolation processing unit 122. Then, “0” data is input to the interpolation processing unit 122 as the other 12 pieces of pixel data excluding the four pieces of pixel data P _{22 to} P ₃₃ among the 16 pieces of pixel data P _{11 to} P ₄₄ . Then, the weighting factors f (x ₁ ), f (x ₂ ), f (y ₁ ), and f (y ₂ ) obtained by the equation (14) are inputted among the terms of the equations (9) to (12). The weighting coefficient of the term whose pixel data value is not “0”, that is, the weighting factors f (x ₂ ), f (x ₃ ), f (y ₂ ), and f () of the term including the pixel data P _{22 to} P _33. y ₃ ) is input to the interpolation processing unit 122, and the interpolation value P is calculated by equations (9) to (13). Since the values of the weighting factors f (x ₁ ), f (x ₄ ), f (y ₁ ), and f (y ₄ ) in the equations (9) to (12) are 0, any value can be used. Good.

First, in Expressions (9) to (12), four pixel data P _{22 to} P ₃₃ and 12 pixel data P _{11 to} P ₄₄ other than the four pixel data P _{22 to} P ₃₃ are excluded. When “0” is substituted as pixel data, Expressions (9) to (12) are expressed as Expressions (20) to (23), respectively.
PA = 0 ・ f (x ₁ ) f (y ₁ ) +0 ・ f (x ₂ ) f (y ₁ ) +0 ・ f (x ₁ ) f (y ₂ ) + P ₂₂ f (x ₂ ) f ( y ₂ )… (20)
PB = 0 ・ f (x ₃ ) f (y ₁ ) +0 ・ f (x ₄ ) f (y ₁ ) + P ₃₂ f (x ₃ ) f (y ₂ ) +0 ・ f (x ₄ ) f ( y ₂ )… (21)
PC = 0 ・ f (x ₁ ) f (y ₃ ) + P ₂₃ f (x ₂ ) f (y ₃ ) +0 ・ f (x ₁ ) f (y ₄ ) +0 ・ f (x ₂ ) f ( y ₄ )… (22)
PD = P ₃₃ f (x ₃ ) f (y ₃ ) +0 ・ f (x ₄ ) f (y ₃ ) +0 ・ f (x ₃ ) f (y ₄ ) +0 ・ f (x ₄ ) f ( y ₄ )… (23)

Next, equations (20) to (23) are transformed into equations (24) to (27).
PA = P ₂₂ f (x ₂ ) f (y ₂ )… (24)
PB = P ₃₂ f (x ₃ ) f (y ₂ )… (25)
PC = P ₂₃ f (x ₂ ) f (y ₃ )… (26)
PD = P ₃₃ f (x ₃ ) f (y ₃ )… (27)

Further, the intermediate values PA to PD calculated by the equations (24) to (27) are substituted into the equation (13), and the interpolation value P is calculated by the equation (28).
P = PA + PB + PC + PD = P ₂₂ f (x ₂ ) f (y ₂ ) + P ₃₂ f (x ₃ ) f (y ₂ ) + P ₂₃ f (x ₂ ) f (y ₃ ) + P ₃₃ f (x ₃ ) f (y ₃ )
… (28)

  Expression (28) is an expression obtained by replacing the four pixel data and weighting coefficient of Expression (15) for calculating the interpolation value by the bilinear method. In other words, if the value of 12 pieces of image data is set to “0” and the weighting factor is a weighting factor of the bilinear method, the interpolation value calculated by the bicubic method is used by the interpolation calculation unit 12 to calculate the interpolation value by the bilinear method. Can be calculated.

  For example, when the pixel is represented by the YCbCr system, the interpolation value of the Y component that is the luminance component is calculated by the bicubic method, and the interpolation value of the Cb component and the Cr component that are the color difference components is calculated by the bilinear method, 12 first calculates an interpolated value of the Y component that is the interpolated value of the first component by the bicubic method, then calculates an interpolated value of the Cb component that is the interpolated value of the second component by the bilinear method, and finally The Cr component interpolation value, which is the third component interpolation value, is calculated by the bilinear method.

FIG. 10 is a diagram illustrating an arrangement of pixel data used when four interpolation values by the bilinear method are calculated simultaneously using the interpolation calculation unit 12. When the interpolation operation unit 12 obtains four interpolation values simultaneously for two pigment components for two interpolation points, that is, a total of four pigment components, by the bilinear method, the four pixel data P ₁₁ , P ₂₁ , P ₁₂ , and P ₂₂ are used to obtain one color difference component, and the four pixel data P ₃₁ , P ₄₁ , P ₃₂ , and P ₄₂ in the area 56 are used to obtain one color difference component, The four pieces of pixel data P ₁₃ , P ₂₃ , P ₁₄ , and P ₂₄ in the region 57 are used to obtain one color difference component, and the four pieces of pixel data P ₃₃ , P ₄₃ , P ₃₄ , and P in the region 58 are used. ₄₄ is used to obtain one color difference component.

Specifically, among the pixels at the two interpolation points, the Cb components of the four pieces of pixel data for obtaining the Cb component of the pixel at the first interpolation point are referred to as PA ₂₂ , PA ₂₃ , PA ₃₂ , PA ₃₃ (hereinafter “ PA _{22 to} PA ₃₃ ), and the Cr components of the four pieces of pixel data for obtaining the Cr component of the pixel at the first interpolation point are PB ₂₂ , PB ₂₃ , PB ₃₂ , PB ₃₃ (hereinafter referred to as “PB ₂₂ ˜ PB ₃₃ ”) and the Cb components of the four pieces of pixel data for obtaining the Cb component of the pixel at the second interpolation point are PC ₂₂ , PC ₂₃ , PC ₃₂ , PC ₃₃ (hereinafter“ PC _{22 to} PC ₃₃ ”). And the Cr components of the four pieces of pixel data for obtaining the Cr component of the pixel at the second interpolation point as PD ₂₂ , PD ₂₃ , PD ₃₂ , PD ₃₃ (hereinafter referred to as “PD _{22 to} PD ₃₃ ”). You If, interpolation operation unit 12, the pixel data _PA 22 _{~PA 33} of Cb components of the four pixels surrounding the first interpolation point, sixteen four pixel data of the pixel data _P 11 _{to P 44} P ₁₁ , P ₂₁ , P ₁₂ , P ₂₂ are input to the interpolation processing unit 122, and the four pixel Cr component pixel data PB _{22 to} PB ₃₃ surrounding the first interpolation point are converted into 16 pixel data P ₁₁ input as four pixel data _{P 31, P 41, P 32} , P 42 of the to P ₄₄ to the interpolation processing unit 122, pixel data PC of Cr components of the four pixels surrounding the second interpolation points _{22 to} PC ₃₃ are input to the interpolation processing unit 122 as four pieces of pixel data P ₁₃ , P ₂₃ , P ₁₄ and P ₂₄ out of the 16 pieces of pixel data P _{11 to} P ₄₄ , and the second interpolation point is set as the second interpolation point. Pixel component of Cr component of four surrounding pixels The data _PD 22 _{~PD 33,} and inputs to the interpolation processing unit 122 as four pixel data _{P 33, P 43, P 34} , P 44 of the 16 pieces of pixel data _P 11 _{to P 44.}

Then, the weighting factors f (x ₁ ), f (x ₂ ), f (y ₁ ), and f (y ₂ ) obtained by the equation (14) for the Cb component at the first interpolation point are expressed by the equation (16). The weighting factors f (x ₁ ), f (x ₂ ), f (y ₁ ), and f (y ₂ ) are input to the interpolation processing unit 122, and the Cr component at the first interpolation point is obtained by Expression (14). The weighting factors f (x ₁ ), f (x ₂ ), f (y ₁ ), and f (y ₂ ) are replaced with the weighting factors f (x ₃ ), f (x ₄ ), and f (y ₁ ) in Expression (17). ) And f (y ₂ ) are input to the interpolation processing unit 122, and the weighting coefficients f (x ₁ ), f (x ₂ ), f (y ₁ ) obtained by the expression (14) for the Cb component of the second interpolation point ) and f a _{(y 2),} the weighting factor _f (x 1 of the formula _{(18)), f (x} 2), and f _{(y 3)} and f _{(y 4)} Input to the interpolation processing unit 122, a second weighting coefficient determined by equation (14) for the Cr component of the interpolation point _f (x _1), f a (x 2), f _{(y 1)} and f _{(y 2)} The weighting coefficients f (x ₃ ), f (x ₄ ), f (y ₃ ), and f (y ₄ ) of Expression (19) are input to the interpolation processing unit 122. When these values are substituted into the equations (16) to (19) and transformed, the equations (16) to (19) are expressed as equations (29) to (32), respectively.
PA = PA ₂₂ f (x ₁ ) f (y ₁ ) + PA ₃₂ f (x ₂ ) f (y ₁ ) + PA ₂₃ f (x ₁ ) f (y ₂ ) + PA ₃₃ f (x ₂ ) f ( y ₂ )… (29)
PB = PB ₂₂ f (x ₃ ) f (y ₁ ) + PB ₃₂ f (x ₄ ) f (y ₁ ) + PB ₂₃ f (x ₃ ) f (y ₂ ) + PB ₃₃ f (x ₄ ) f ( y ₂ )… (30)
PC = PC ₂₂ f (x ₁ ) f (y ₃ ) + PC ₃₂ f (x ₂ ) f (y ₃ ) + PC ₂₃ f (x ₁ ) f (y ₄ ) + PC ₃₃ f (x ₂ ) f ( y ₄ )… (31)
PD = PD ₂₂ f (x ₃ ) f (y ₃ ) + PD ₃₂ f (x ₄ ) f (y ₃ ) + PD ₂₃ f (x ₃ ) f (y ₄ ) + PD ₃₃ f (x ₄ ) f ( y ₄ )… (32)

  Expressions (29) to (32) are expressions in which the four pixel data and the weighting coefficient in Expression (15) for calculating the interpolation value by the bilinear method are replaced. In other words, if the image data is the pixel data of the Cb component and the Cr component at the two interpolation points, and the weighting factor is the weighting factor of the bilinear method for calculating the Cb component and the Cr component at the two interpolation points. The interpolation calculation unit 12 that calculates the interpolation value by the bicubic method can simultaneously calculate the interpolation values by the bilinear method of the Cb component and the Cr component that are color difference components for two interpolation points.

  For example, when the pixel is represented by the YCbCr system, the Y component interpolation value that is the luminance component is calculated by the bicubic method, and the Cb component and Cr component interpolation values that are the color difference components are calculated by the bilinear method, the interpolation calculation unit 12 First, for each of the first interpolation point and the second interpolation point, the interpolation value of the Y component that is the first component is calculated by the bicubic method, and then the second component Cb and the second component are calculated by the bilinear method. The interpolation value of the Cr component, which is the three components, is calculated simultaneously for the first interpolation point and the second interpolation point. Therefore, when calculating the interpolation values for two interpolation points, in the case shown in FIG. 9, the bicubic method required two times and the bilinear method required four times in total, but the interpolation calculation unit In the case of 12, it can be calculated by the bicubic method twice and the bilinear method once in a total of three times, and the processing time can be reduced to ½.

  FIG. 11 is a block diagram showing a configuration of a digital still camera 2 according to another embodiment of the present invention. A digital still camera 2 that is an image processing apparatus is an example in which the image processing apparatus 1 shown in FIG. 1 is applied to a digital still camera, and includes an input device 20, an image memory 21, a size conversion processing unit 22, an output device 23, and a control. A processor 24 is included. The image processing method according to the present invention is processed by the digital still camera 2.

  The input device 20 that is an input means is constituted by, for example, an image sensor, and specifically represents image data of a captured image. Specifically, when the image is expressed in the YCbCr system, the Y component that is a luminance component and the Cb component that is a color difference component And converted to Cr component image data, and sent to the image memory 21 for storage. The image memory 21 is configured by a storage device such as a semiconductor memory or a hard disk device, for example, and stores image data received from the input device 20 and image data received from the size conversion processing unit 22.

  When the image is represented by the YCbCr system, the digital still camera 2 calculates the interpolation value of the Y component that is the luminance component by the bicubic method, and calculates the interpolation values of the Cb component and the Cr component that are the color difference components by the bilinear method. This utilizes the fact that the human eye is more sensitive to the luminance component than the color difference component, and is very effective in reducing image quality degradation and reducing the amount of computation. Furthermore, in order to calculate the interpolated values of the Cb component and the Cr component which are color difference components, four pixels surrounding the interpolation points necessary for the calculation by the bicubic method and 12 pixels surrounding the four pixels are calculated. Since the calculation is performed with the components of less than 16 pixels, that is, four pixels surrounding the interpolation point, it is possible to suppress the color difference caused by the distant positions and to prevent color misregistration.

  The size conversion processing unit 22 includes an image input unit 221, an input data control unit 222, an interpolation processing unit 223, a coefficient unit 224, an output data control unit 225, an image output unit 226, and a controller unit 227.

  The image input unit 221 includes an input buffer 221a configured by, for example, a semiconductor memory, reads image data instructed from the controller unit 227 from the image memory 21, and stores the read image data in the input buffer 221a. Specifically, out of the image data stored in the image memory 21, the pixel data for the pixel instructed from the controller unit 227 is selected and read out from the pixel data of the component instructed from the controller unit 227. Pixel data is stored in the input buffer 221a.

For example, when a luminance component is instructed for one interpolation point from the controller unit 227, the image input unit 221 determines 16 pixels, that is, interpolation points, instructed from the controller unit 227 from the luminance component, that is, Y component pixel data. The pixel data P _{11 to} P ₄₄ of 4 × 4 pixels including the surrounding four pixels and the 12 pixels surrounding the four pixels are selected and read out, and the read out 16 pixel data P _{11 to} P _44. Is stored in the input buffer 221a. Then, the 16 pieces of pixel data P _{11 to} P ₄₄ stored in the input buffer 221 a are sent to the input data control unit 222.

Alternatively, when the color difference component is instructed for one interpolation point from the controller unit 227, the image input unit 221 determines the Cb of the four pixels instructed from the controller unit 227 from the color difference component, that is, the pixel data of the Cb component and the Cr component. Component pixel data, that is, Cb component pixel data of 2 × 2 pixels surrounding the interpolation point, and four Cr component pixel data, that is, Cb component pixel data of 2 × 2 pixels surrounding the interpolation point, for example, Four Cb component pixel data PA _{22 to} PA ₃₃ and four Cr component PB ₂₂ to PB ₃₃ are selected and read out, and the read four Cb component pixel data PA _{22 to} PA ₃₃ and four pieces of Cb component pixel data PA _{22 to} PA ₃₃ are selected. The Cr component pixel data PB _{22 to} PB ₃₃ are stored in the input buffer 221a.

Furthermore, when the color difference component is instructed from the controller unit 227 for two interpolation points, the image input unit 221 uses the first interpolation point instructed from the controller unit 227 from the color difference component, that is, the pixel data of the Cb component and the Cr component. 4 Cb component pixel data and 4 Cr component pixel data, for example, 4 Cb component pixel data PA _{22 to} PA ₃₃ and 4 Cr component PB ₂₂ to PB ₃₃ , and Cb component pixel data and four Cr component pixel data of four pixels surrounding the second interpolation point instructed by the controller unit 227, for example, four Cb component pixel data PC _{22 to} PC ₃₃ and 4 Select and read out PD ₂₂ to PD ₃₃ of Cr components. Then, four Cb component pixel data PA _{22 to} PA ₃₃ and four Cr component pixel data PB _{22 to} PB _{33 for} the read first interpolation point, and four Cb component pixel data PB _{22 to} PB ₃₃ for the second interpolation point are obtained. The Cb component pixel data PC _{22 to} PC ₃₃ and the four Cr component pixel data PD _{22 to} PD ₃₃ are stored in the input buffer 221a.

  The input data control unit 222 converts the pixel data received from the image input unit 221 into input data to the interpolation processing unit 223 and sends it to the interpolation processing unit 223. The interpolation processing unit 223 has the same circuit configuration as the interpolation processing unit 122 illustrated in FIG. 1, and detailed description of the circuit configuration is omitted to avoid duplication. The interpolation processing unit 223 calculates intermediate values PA to PD based on the 16 pieces of image data received from the input data control unit 222 and the weighting coefficient received from the coefficient unit 224, and outputs the calculated intermediate values PA to PD as output data. The data is sent to the control unit 225.

  The coefficient unit 224 that is a storage unit is configured by, for example, a non-volatile memory, and calculates and stores in advance a weighting coefficient obtained by the equations (2), (4), and (14). Then, the weighting factor instructed from the controller unit 227 is read from the table which is interpolation coefficient information stored in the coefficient unit 224, and the read weighting factor is input to the interpolation processing unit 223.

Specifically, when the luminance component is instructed from the controller unit 227, the coefficient unit 224 weights coefficients f (x ₁ ) to f (x ₄ ) and f (y) obtained from the equations (2) and (4). ₁ ) to f (y ₄ ) are read from the table stored in the coefficient unit 224, and the read weight coefficients f (x ₁ ) to f (x ₄ ) and f (y ₁ ) to f (y ₄ ) are interpolated. 223. Alternatively, when the color difference component is instructed from the controller unit 227, the weighting coefficients f (x ₁ ), f (x ₂ ), f (y ₁ ), and f (y ₂ ) obtained from the equation (14) are used as the coefficient unit 224. The weighting factors f (x ₁ ), f (x ₂ ), f (y ₁ ), and f (y ₂ ) read out from the table stored in FIG. x ₁ ) to (x ₄ ) and f (y ₁ ) to f (y ₄ ) are sent to the interpolation processing unit 223.

  The output data control unit 225 calculates an interpolation value based on the intermediate value received from the interpolation processing unit 223, and sends the calculated interpolation value to the image output unit 226. For the luminance component, the intermediate values PA to PD received from the interpolation processing unit 223 are added to obtain the interpolation value P, and the obtained interpolation value P is sent to the image output unit 226. For the color difference component, the intermediate value PA received from the interpolation processing unit 223 is used as the interpolation value PA, the intermediate value PB received from the interpolation processing unit 223 is used as the interpolation value PB, and the intermediate value PC received from the interpolation processing unit 223 is used as the interpolation value PC. The intermediate value PD received from the interpolation processing unit 223 is sent to the image output unit 226 as the interpolation value PD.

  The image output unit 226 includes an output buffer 226a configured by, for example, a semiconductor memory, and the interpolation value PA and the interpolation value received from the output data control unit 225 using the interpolation value P received from the output data control unit 225 as Y component pixel data. The interpolated value PC and the interpolated value PD received from the output data control unit 225 are the pixel data of the Cb component for the second interpolation point and the interpolated value PC and the interpolated value PD received from the output data control unit 225 as the Cb component pixel data and the Cr component pixel data for the first interpolation point. The pixel data of the Cr component is stored in the output buffer 226a. Then, the Y component pixel data, or the Cb component pixel data and the Cr component pixel data stored in the output buffer 226a are sent to and stored in the image memory 21.

  The controller unit 227 includes a control register 227a configured by, for example, a semiconductor memory, and based on parameters stored in the control register 227a, an image input unit 221, an input data control unit 222, an interpolation processing unit 223, a coefficient unit 224, The output data control unit 225 and the image output unit 226 are controlled.

  The parameters stored in the control register 227a are set by the control processor 24 and the controller unit 227. Parameters set by the control processor 24 include processing component designation information indicating components to be processed, that is, components to be processed among luminance components and color difference components, input image size information indicating the size of the input image, and enlargement or reduction. It includes information such as enlargement / reduction ratio information indicating the magnification.

  The control processor 24 generates information such as input image size information and enlargement / reduction ratio information based on information input by the user from an operation unit (not shown) which is an instruction unit, and the generated information is used as a parameter in the control register 227a. Set to and memorize. The parameters stored in the control register 227a are referred to by the image input unit 221, the input data control unit 222, the interpolation processing unit 223, the coefficient unit 224, the output data control unit 225, and the image output unit 226.

  When instructed to start processing from the control processor 24, the controller unit 227 starts size conversion processing. The controller unit 227 calculates the output image size and output image coordinates based on information such as input image size information and enlargement / reduction ratio information stored in the control register 227a, and stores the calculation result in the control register 227a. The image input unit 221, the input data control unit 222, the interpolation processing unit 223, the coefficient unit 224, the output data control unit 225, and the image output unit 226 are instructed to start operations.

  The image input unit 221, the input data control unit 222, the interpolation processing unit 223, the coefficient unit 224, the output data control unit 225, and the image output unit 226 instructed to start the operation refer to parameters stored in the control register 227a. In response to the instruction from the controller unit 227, the process is performed.

  When calculating the interpolation value, the controller unit 227 calculates the interpolation value in units of two interpolation points. Specifically, a Y component interpolation value that is a luminance component is calculated for the first interpolation point of the two interpolation points by the bicubic method, and then the Y component interpolation is performed for the second interpolation point. The values are calculated by the bicubic method, and finally, the interpolation values of the Cb component and the Cr component, which are the pigment components, are calculated by the bilinear method shown in FIG. 10 for the first interpolation point and the second interpolation point. If the number of interpolation points is an odd number and one remains at the end, the interpolation value of the Y component that is the luminance component is calculated by the bicubic method for the interpolation point, and then the Cb component and Cr that are the pigment components The component interpolation value is set as the first interpolation point, and the coefficient of the second interpolation point is set to 0, which is calculated by the bilinear method shown in FIG.

  The output device 23 which is an output means is configured by a display device such as a liquid crystal display, reads image data instructed from the control processor 24 from the image memory 21, and displays the read image data on the display device. The control processor 24 includes, for example, a storage device that stores a program and data, and a central processing unit (hereinafter referred to as “CPU”) that executes a program stored in the storage device. The CPU controls the input device 20, the size conversion processing unit 22, and the output device 23 by executing a program stored in the storage device.

  The image memory 21, the image input unit 221, the input data control unit 222, the interpolation processing unit 223, the output data control unit 225, and the image output unit 226 are interpolation value calculation means or an interpolation value calculation unit. The controller unit 227 and the control processor 24 are conversion means.

  FIG. 12 is a flowchart showing the processing procedure of the size conversion process executed by the control processor 24 and the interpolation calculation unit 22. The size conversion process is a process for calculating four interpolation values for two interpolation points at the same time as shown in FIG. In the preprocessing step of the flowchart shown in FIG. 12, image data is input by the input device 20, the input image data is stored in the image memory 21, and a magnification for changing the image size is instructed from an operation unit (not shown). Then, the process proceeds to step A1.

  In step A1, parameters are set. Specifically, the control processor 24 generates information such as processing component designation information, input image size information, and enlargement / reduction ratio information, and sets and stores the generated information in the control register 227a as a parameter. After setting the parameters in the control register 227a, the control processor 24 instructs the controller unit 227 to start processing. When instructed to start processing from the control processor 24, the controller unit 227 starts processing.

  In step A2, coordinate calculation is performed. Specifically, the controller unit 227 calculates the output image size and the coordinates of the output image based on information such as the input image size information and the enlargement / reduction ratio information stored in the control register 227a, and outputs the calculation result to the control register. After the data is stored in 227a, the image input unit 221, the input data control unit 222, the interpolation processing unit 223, the coefficient unit 224, the output data control unit 225, and the image output unit 226 are instructed to start operations.

  In step A3, input image data is acquired. Specifically, the image input unit 221 instructed to start the operation refers to the parameter stored in the control register 227a, reads the image data instructed by the referenced parameter from the image memory 21, and reads the read image data. Is stored in the input buffer 221a.

For example, when the referred processing component designation information designates a luminance component, the image input unit 221 includes four pixels surrounding an interpolation point for calculating an interpolation value and twelve pixels surrounding the four pixels. The 16 pixel Y component pixel data P _{11 to} P ₄₄ of the 16 pixels are read from the image memory 21, and the read 16 Y component pixel data P _{11 to} P ₄₄ are stored in the input buffer 221a. Or if the referenced processing component specifying information specifies a color difference component, the first pixel data _PA 22 of Cb components of the four pixels surrounding the interpolation point _{～PA 33} and Cr components of the pixel data _PB 22 _{~PB 33} , and second read pixel data _PC 22 _{to PC 33} and Cr components of the pixel data _PD 22 _{-PD 33} of Cb components of the four pixels surrounding the interpolation point, the four Cb component read pixel data _{PA 22} ˜PA ₃₃ and four Cr component pixel data PB _{22 to} PB ₃₃ , four Cb component pixel data PC _{22 to} PC ₃₃ and four Cr component pixel data PD _{22 to} PD ₃₃ are input buffer 221a. To remember.

Further, when the processing component designation information designates a luminance component, the image input unit 221 sends the 16 Y component pixel data P _{11 to} P ₄₄ stored in the input buffer 221 a to the input data control unit 222. . If the processing component specifying information specifies a color difference component, the input buffer 221a pixel data of the four Cb components stored in _PA 22 _{~PA 33} and four Cr components of the pixel data _PB 22 _{~PB 33,} and The four Cb component pixel data PC _{22 to} PC ₃₃ and the four Cr component pixel data PD _{22 to} PD ₃₃ are read, and the read four Cb component pixel data PA _{22 to} PA ₃₃ and the four Cr component data are read out. Component pixel data PB _{22 to} PB ₃₃ , four Cb component pixel data PC _{22 to} PC _33, and four Cr component pixel data PD _{22 to} PD ₃₃ are sent to the input data control unit 222.

In step A4, an interpolation calculation process is performed. Specifically, the input data control unit 222 sends 16 pieces of pixel data received from the image input unit 221 to the interpolation processing unit 223 as input data to the interpolation processing unit 223. If conversion to input data to the interpolation processing unit 223 is necessary, it is converted to input data to the interpolation processing unit 223 and sent to the interpolation processing unit 223. The interpolation processing unit 223 includes 16 pieces of image data P _{11 to} P ₄₄ received from the input data control unit 222 and weight coefficients f (x ₁ ) to f (x ₄ ) and f (y ₁ ) to f received from the coefficient unit 224. Based on f (y ₄ ), intermediate values PA to PD are calculated, and the calculated intermediate values PA to PD are sent to the output data control unit 225.

  The output data control unit 225 calculates an interpolation value based on the intermediate value received from the interpolation processing unit 223, and sends the calculated interpolation value to the image output unit 226. In the case of the luminance component, the intermediate values PA to PD are added to calculate the interpolation value P, and the calculated interpolation value P is sent to the image output unit 226. In the case of the color difference component, the intermediate values PA to PD received from the interpolation processing unit 223 are sent to the image output unit 226 as the interpolation values PA to PD. The interpolation calculation process of step A4 is processed by calling the interpolation calculation process shown in FIG.

  In step A5, output image data is transmitted. Specifically, the image output unit 226 stores the interpolation value P received from the output data control unit 225 in the output buffer 226a as the Y component pixel data in the case of the luminance component, and outputs data control in the case of the color difference component. Among the interpolation values PA to PD received from the unit 225, the interpolation value PA is used as the pixel data of the Cb component of the first interpolation point, the interpolation value PB is used as the pixel data of the Cr component of the first interpolation point, and the interpolation value PC is used. As the pixel data of the Cb component at the second interpolation point, the interpolation value PD is stored in the output buffer 226a as the pixel data of the Cr component at the second interpolation point. Then, the Y component pixel data or the Cb component pixel data and the Cr component pixel data stored in the output buffer 226a are sent to the image memory 21 and stored as output image data.

  In step A6, it is determined whether all output image data has been created. When all the pixels of the output image converted into the output image size are converted into the output image data for one color component, the controller unit 227 determines that all the output image data has been created, and the process proceeds to Step A7. move on. If it is not converted to output image data for all pixels of the output image converted to the output image size for one color component, it is determined that all output image data has not been created, and the process returns to step A2. Repeat step A2 to step A5.

  In step A7, it is determined whether or not all color components have been processed. When the control processor 24 determines that all the color components, that is, the luminance component and the color difference component have been processed, the control processor 24 ends the size conversion process, and when it determines that all the color components, that is, the luminance component and the color difference component have not been processed, The processing component designation information stored in 227a is changed to the next color component, the process returns to step A1, and steps A1 to A6 are repeated for the next color component.

  When the size conversion processing image shown in FIG. 12 is completed, output image data stored in the memory 21 is output by the output device 23 in a post-processing step not shown in FIG.

  FIG. 13 is a flowchart illustrating the processing procedure of the interpolation calculation process executed by the interpolation calculation unit 22. When step A4 shown in FIG. 12 is executed, the interpolation calculation process is called, and the process proceeds to step B1.

  In step B1, it is determined whether or not the process is a luminance component. Specifically, the input data control unit 222 and the output data control unit 225 refer to the processing component designation information stored in the control register 227a, and if the processing component designation information designates the luminance component, the luminance component It determines with it being a process and progresses to step B2. With reference to the processing component designation information stored in the control register 227a, if the processing component designation information designates a color difference component, it is determined that the processing is not a luminance component, and the process proceeds to step B3.

In step B2, a luminance component interpolation calculation is performed. Specifically, the input data control unit 222 sends the 16 pieces of pixel data P _{11 to} P ₄₄ received from the image input unit 221 to the interpolation processing unit 223 as input data without conversion. The interpolation processing unit 223 includes 16 pieces of image data P _{11 to} P ₄₄ received from the input data control unit 222 and weight coefficients f (x ₁ ) to f (x ₄ ) and f (y ₁ ) to f received from the coefficient unit 224. Based on f (y ₄ ), intermediate values PA to PD are calculated, and the calculated intermediate values PA to PD are sent to the output data control unit 225. The output data control unit 225 adds the intermediate values PA to PD received from the interpolation processing unit 223 to calculate the interpolation value P, sends the calculated interpolation value P to the image output unit 226, and ends the interpolation calculation processing.

In step B3, input data is generated. Specifically, the input data control unit 222 performs the four Cb component image data PA _{22 to} PA ₃₃ and the four Cr component image data received from the image input unit 221 according to Expression (33) which is a replacement expression. The PB _{22 to} PB ₃₃ , the four Cb component pixel data PC _{22 to} PC ₃₃ and the four Cr component pixel data PD _{22 to} PD ₃₃ are replaced with 16 pixel data P _{11 to} P ₄₄ .
P ₁₁ = PA ₂₂ , P ₁₂ = PA ₂₃ , P ₁₃ = PC ₂₂ , P ₁₄ = PC ₂₃ , P ₂₁ = PA ₃₂ , P ₂₂ = PA ₃₃ , P ₂₃ = PC ₃₂ , P ₂₄ = PC ₃₃ ,
P ₃₁ = PB ₂₂ , P ₃₂ = PB ₂₃ , P ₃₃ = PD ₂₂ , P ₃₄ = PD ₂₃ , P ₄₁ = PB ₃₂ , P ₄₂ = PB ₃₃ , P ₄₃ = PD ₃₂ , P ₄₄ = PD ₃₃ … (33 )

The input data control unit 222 sends the 16 pieces of pixel data P _{11 to} P ₄₄ after replacement to the interpolation processing unit 223 as input data.

In step B4, interpolation calculation of color difference components is performed. Specifically, the interpolation processing unit 223 includes 16 pieces of image data P _{11 to} P ₄₄ received from the input data control unit 222 and weighting factors f (x ₁ ) to f (x ₄ ) and f received from the coefficient unit 224. Based on (y ₁ ) to f (y ₄ ), intermediate values PA to PD are calculated, and the calculated intermediate values PA to PD are sent to the output data control unit 225.

  In step B5, output data is generated and the interpolation calculation process is terminated. Specifically, when the output data control unit 225 receives the intermediate values PA to PD from the interpolation processing unit 223, the output data control unit 225 sends the received intermediate values PA to PD to the image output unit 226 as the interpolation values PA to PD, and performs interpolation calculation processing. Exit. Steps A1 to A7 shown in FIG. 12 are conversion steps.

In the above-described embodiment, the bicubic method is applied to the luminance component and the bilinear method is applied to the chrominance component. However, the combination of interpolation methods to be applied may be freely changed according to the purpose. Further, the image is applied to the image data represented by the YCbCr method, that is, the YUV method. However, the image may be applied to other methods such as RGB (Red Green Blue) method or YIQ (Luma
The present invention can also be applied to image data represented by an In-phase Quadrature) method.

  In the above-described embodiment, the interpolation calculation unit 12 is applied to the image processing apparatus 1 and the interpolation calculation unit 22 is applied to the digital still camera 2. However, the interpolation calculation unit 12 or the interpolation calculation unit 22 performs an interpolation calculation. It can be applied to other devices.

  As described above, image data representing an image composed of pixels having different Y components, Cb components, and Cr components is input by the input device or the input device 20 (not shown), and the size of the image is input by the operation unit (not shown). The magnification for changing is indicated. Interpolation value calculating means, specifically, input image memory 11, input data control unit 121, interpolation processing unit 122, output data control unit 123 and output image memory 13 or image memory 21, image input unit 221, input data control unit 222, the interpolation processing unit 223, the output data control unit 225, and the image output unit 226 calculate an interpolation value by the cubic convolution interpolation method or the bilinear interpolation method.

  The input / output control unit 124 or the controller unit 227 and the control processor 24 use the interpolation value calculation means to convert image data input by an input device (not shown) or the input device 20 at a magnification indicated by an operation unit (not shown). It is converted into image data representing the image of the changed size. Image data converted by the input / output control unit 124 or the controller unit 227 and the control processor 24 is output by an output device or an output device 23 (not shown).

  Further, by the input / output control unit 124 or the controller unit 227 and the control processor 24, the Y component, the Cb component, and the Cr at the interpolation point that is the position corresponding to the pixel of the image indicated by the image data after conversion in the coordinate system before conversion When the interpolated value of the component is calculated by the interpolated value calculating means, for the Y component, for each interpolation point, among the pixels of the input image indicated by the input device (not shown) or the image data input by the input device 20 Based on the Y components of 16 pixels including four pixels surrounding the interpolation point and 12 pixels surrounding the four pixels, the interpolation value of the Y component is calculated by the cubic convolution interpolation method. The

  As for the second component and the third component, the interpolation value of the Cb component at the first interpolation point of the two interpolation points is the first interpolation among the pixels of the input image at the same time every two interpolation points. Based on the Cb component of the four pixels surrounding the point, the interpolation value of the Cr component of the first interpolation point is based on the Cr component of the four pixels surrounding the first interpolation point. The interpolation value of the Cb component of the second interpolation point different from the first interpolation point is based on the Cb components of the four pixels surrounding the second interpolation point of the pixels of the input image, and The interpolated value of the Cr component at the second interpolation point is calculated by bilinear interpolation based on the Cr components of the four pixels surrounding the second interpolation point.

  Then, the interpolation value of the first component calculated by the interpolation value calculation means is the first component, the interpolation value of the second component calculated by the interpolation value calculation means is the second component, and the first value calculated by the interpolation value calculation means. Image data representing an image composed of pixels having the third component interpolation value as the third component is the converted image data.

  The Y component interpolation value calculated by the interpolation value calculation means is the Y component, the Cb component interpolation value calculated by the interpolation value calculation means is the Cb component, and the Cr component is calculated by the interpolation value calculation means. The image data representing an image constituted by pixels having the interpolation value of Cr as the Cr component is the converted image data.

  Therefore, as an interpolation value calculation means for calculating an interpolation value by cubic convolution interpolation or bilinear interpolation, for example, hardware, specifically, input image memory 11, input data control unit 121, interpolation processing Section 122, output data control section 123 and output image memory 13 or image memory 21, image input section 221, input data control section 222, interpolation processing section 223, output data control section 225, and image output section 226. The interpolation value can be calculated by the third-order convolution interpolation method in a shorter time than when the program for calculating the interpolation value is executed and sequentially calculated. Furthermore, since the interpolation value calculated by the bilinear interpolation method can be calculated using the hardware that calculates the interpolation value by the cubic convolution interpolation method, the interpolation is performed by the bilinear interpolation method. Hardware for calculating the value is not required, and the circuit area and cost can be reduced. That is, the image can be interpolated with a circuit having a high processing speed and a small circuit area and cost.

  Further, the coefficient part 125 or the coefficient part 224 needs to calculate the interpolation coefficient of the Y component necessary for calculating the interpolation value by the cubic convolution interpolation method and the interpolation value by the bilinear interpolation method. A table representing interpolation coefficients of the Cb component and Cr component is stored.

  Input image memory 11, input data control unit 121, interpolation processing unit 122, output data control unit 123, and output image memory 13 or image memory 21, image input unit 221, input data control unit 222, interpolation processing unit 223, output data control The Y component interpolation value is calculated by the unit 225 and the image output unit 226 based on the Y component of the 16 pixels and the Y component interpolation coefficient indicated by the table stored in the coefficient unit 125 or the coefficient unit 224. The

  Based on the Cb component of the four pixels surrounding the first interpolation point or the second interpolation point, and the interpolation coefficient of the Cb component and the Cr component indicated by the table stored in the coefficient unit 125 or the coefficient unit 224, Cb Interpolated values of the components are calculated, the Cr components of the four pixels surrounding the first interpolation point or the second interpolation point, and the Cb component and the Cr component indicated by the table stored in the coefficient part 125 or the coefficient part 224 Based on the interpolation coefficient, the interpolation value of the Cr component is calculated.

  That is, since the interpolation coefficient necessary for calculating the interpolation value is calculated and stored in advance for each interpolation method, the interpolation coefficient can be selected according to the interpolation method for calculating the interpolation value. Processing time required to calculate the interpolation coefficient can be reduced.

  Furthermore, the input image memory 11, the input data control unit 121, the interpolation processing unit 122, the output data control unit 123, and the output image memory 13 or the image memory 21, the image input unit 221, the input data control unit 222, the interpolation processing unit 223, and the output When the position of the interpolation point is represented in the XY coordinate system, the data control unit 225 and the image output unit 226 have the first component of one of the 16 pixels and the X-axis direction corresponding to the first component. Two first multipliers, for example, multipliers 311 and 312, and a first adder for adding the outputs of the two first multipliers, for example, adder 313, Eight logic circuit portions CAL31 to CAL3 each including a second multiplier, for example, a multiplier 314, for multiplying the output of the first adder and the first interpolation coefficient in the Y-axis direction corresponding to the first component. If, when the divided eight logic circuit CAL31~38 into four groups, including four of the second adder 391 to 394 for adding the outputs of the two second multipliers of each group. Therefore, the number of multipliers can be reduced as compared with the case where the circuit is configured by the equations (9) to (12).

  Furthermore, the input image memory 11, the input data control unit 121, the interpolation processing unit 122, the output data control unit 123, and the output image memory 13 or the image memory 21, the image input unit 221, the input data control unit 222, the interpolation processing unit 223, and the output When the data control unit 225 and the image output unit 226 calculate an interpolation value by cubic convolution interpolation, the Y component of the 16 pixels and the first interpolation coefficient information stored in the storage unit are shown. Are input to the eight logic circuit units CAL31 to CAL38, and the outputs of the four adders of the second adder are added to calculate the interpolation value of the Y component.

  Further, when four interpolation values are calculated by the bilinear interpolation method, the Cb components of the four pixels surrounding the first interpolation point and the second interpolation coefficient information stored in the storage means are shown. The interpolation coefficient is input to the two logic circuit units of the first group of the four groups, and the output of the first group is used as the interpolation value of the Cb component at the first interpolation point. The Cr components of the four pixels surrounding the first interpolation point and the second interpolation coefficient indicated by the interpolation coefficient information stored in the storage means are the two logics of the second group of the four groups. Input to the circuit unit, and the output of the second group is the interpolation value of the Cr component of the first interpolation point. The Cb components of the four pixels surrounding the second interpolation point and the second interpolation coefficient indicated by the interpolation coefficient information stored in the storage means are two logics of the third group of the four groups. The output of the third group is input to the circuit unit, and the interpolation value of the Cb component of the second interpolation point is used. Then, the Cr components of the four pixels surrounding the second interpolation point and the second interpolation coefficient indicated by the interpolation coefficient information stored in the storage means are 2 in the fourth group of the four groups. The fourth group output is used as an interpolation value of the Cr component at the second interpolation point.

  Therefore, one interpolation value can be calculated for the cubic convolution method, and four interpolation values can be calculated simultaneously for the bilinear interpolation method.

  Furthermore, in the pre-processing step (not shown) in the flowchart shown in FIG. 12, image data representing an image composed of pixels having different Y components, Cb components, and Cr components is input, and the size of the image is changed. Specify the magnification.

  In steps A1 to A7, an interpolation value calculation unit that calculates an interpolation value by a cubic convolution interpolation method or a bilinear interpolation method, specifically, the input image memory 11, the input data control unit 121, and the interpolation processing unit 122. The output data control unit 123 and the output image memory 13 or the image memory 21, the image input unit 221, the input data control unit 222, the interpolation processing unit 223, the output data control unit 225, and the image output unit 226 are input in steps not shown. When the converted image data is converted into image data representing an image having a size changed at a magnification designated in a preprocessing step (not shown), a position corresponding to the pixel of the image indicated by the converted image data in the coordinate system before conversion When the interpolation value of the first to third components at the interpolation point is calculated by the interpolation value calculator, the first component Each of the interpolation points includes four pixels surrounding the interpolation point and twelve pixels surrounding the four pixels among the pixels of the input image indicated by the image data input in the preprocessing step (not shown). Based on the Y component of each pixel, an interpolated value of the Y component is calculated by cubic convolution interpolation.

  Further, with respect to the second component and the third component, the interpolation value of the Cb component at the first interpolation point of the two interpolation points is simultaneously set to the first of the pixels of the input image for every two interpolation points. Based on the Cb component of the four pixels surrounding the interpolation point, the interpolation value of the Cr component of the first interpolation point is calculated based on the Cr component of the four pixels surrounding the first interpolation point. The interpolation value of the Cb component of the second interpolation point different from the first interpolation point among the points is based on the Cb components of the four pixels surrounding the second interpolation point of the pixels of the input image. And the interpolated value of the Cr component of the second interpolation point is calculated by bilinear interpolation based on the Cr components of the four pixels surrounding the second interpolation point.

  Then, the interpolation value of the Y component calculated by the interpolation value calculation unit is the Y component, the interpolation value of the Cb component calculated by the interpolation value calculation unit is the Cb component, and Cr is calculated by the interpolation value calculation unit. Image data representing an image composed of pixels whose component interpolation values are Cr components is assumed to be converted image data. Further, in a post-processing step (not shown), the image data converted in steps A1 to A7 is output.

  Therefore, when the image processing method according to the present invention is applied, as an interpolation value calculation unit that calculates an interpolation value by a cubic convolution interpolation method or a bilinear interpolation method, for example, hardware, specifically, an input Image memory 11, input data control unit 121, interpolation processing unit 122, output data control unit 123 and output image memory 13 or image memory 21, image input unit 221, input data control unit 222, interpolation processing unit 223, output data control unit Since the image output unit 226 and the image output unit 226 are used, it is possible to calculate the interpolation value by the third-order convolution interpolation method in a shorter time than when the program for calculating the interpolation value is executed and sequentially calculated. Since the interpolation value calculated by the bilinear interpolation method can be calculated using the hardware that calculates the interpolation value by the cubic convolution interpolation method, the interpolation is performed by the bilinear interpolation method. Hardware for calculating the value is not required, and the circuit area and cost can be reduced. That is, the processing speed is high, and the circuit area and cost can be reduced.

1 is a block diagram illustrating a schematic configuration of an image processing apparatus 1 according to an embodiment of the present invention. FIG. 4 is a diagram illustrating a relationship between the position of each pixel 51 and the position of an interpolation point 52 that form an image indicated by image data stored in an input image memory 11. FIG. 3 is an enlarged view of a region 53 shown in FIG. 2. FIG. 3 is an enlarged view of a region 54 shown in FIG. 2. It is a figure for demonstrating the example which calculates | requires an interpolation value using the nearest neighbor method. 3 is a block diagram illustrating a configuration of an interpolation processing unit 122. FIG. 3 is a logic circuit diagram of an interpolation processing unit 122. FIG. FIG. 6 is a logic circuit diagram of a portion for calculating an intermediate value PA in the interpolation processing unit 122. It is a figure which shows arrangement | positioning of the pixel data used when calculating one interpolation value by a bilinear method using the interpolation calculating part. It is a figure which shows arrangement | positioning of the pixel data used when calculating four interpolation values by a bilinear method simultaneously using the interpolation calculating part. It is a block diagram which shows the structure of the digital still camera 2 which is the other embodiment of this invention. 5 is a flowchart showing a processing procedure of size conversion processing executed by a control processor 24 and an interpolation calculation unit 22; It is a flowchart which shows the process sequence of the interpolation calculation process which the interpolation calculating part 22 performs. It is a block diagram which shows the structure of the image processing apparatus 100 by a prior art.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Image processing apparatus 2 Digital still camera 11,101 Input image memory 12,102 Interpolation calculating part 13,103 Output image memory 20 Input apparatus 21 Image memory 22 Size conversion process part 23 Output apparatus 24 Control processor 31-38 CAL
51 pixels 52 interpolation points 221 image input unit 221a input buffer 121, 222 input data control unit 122, 223 interpolation processing unit 123, 225 output data control unit 124 input / output control unit 125, 224 coefficient unit 226 image output unit 226a output buffer 227 Controller unit 227a Control register 311, 312, 314, 331, 332, 334 Multiplier 313, 333, 391-394 Adder

Claims (5)

Input means for inputting image data representing an image constituted by pixels composed of mutually different first to third components;
An instruction means for instructing a magnification for changing the size of the image;
Interpolation value calculating means for calculating an interpolation value by a cubic convolution interpolation method or a bilinear interpolation method;
Conversion means for converting the image data input by the input means into image data representing an image having a size changed by the magnification indicated by the instruction means, using the interpolation value calculation means;
Output means for outputting image data converted by the conversion means,
When the conversion means causes the interpolation value calculation means to calculate the interpolation values of the first to third components at the interpolation points that are positions corresponding to the pixels of the image indicated by the image data after conversion in the coordinate system before conversion,
For the first component, for each interpolation point, among the pixels of the input image indicated by the image data input by the input means, four pixels surrounding the interpolation point and twelve pixels surrounding the four pixels are included. Based on the first component of the 16 pixels including, the interpolation value of the first component is calculated by the third-order convolution interpolation method,
For the second component and the third component, the interpolation value of the second component at the first interpolation point of the two interpolation points is simultaneously calculated for each of the two interpolation points, and the first of the pixels of the input image. Based on the second component of the four pixels surrounding the interpolation point, the interpolation value of the third component at the first interpolation point is based on the third component of the four pixels surrounding the first interpolation point. Interpolation values of the second component at the second interpolation point different from the first interpolation point of the two interpolation points are the four pixels surrounding the second interpolation point of the pixels of the input image. And the interpolated value of the third component at the second interpolation point based on the third component of the four pixels surrounding the second interpolation point by bilinear interpolation Let me calculate
The interpolation value of the first component calculated by the interpolation value calculation means is the first component, the interpolation value of the second component calculated by the interpolation value calculation means is the second component, and the third component is calculated by the interpolation value calculation means. An image processing apparatus characterized in that image data representing an image composed of pixels having a third component as an interpolation value is used as converted image data. Interpolation coefficient information representing the first interpolation coefficient necessary for calculating the interpolation value by the cubic convolution interpolation method and the second interpolation coefficient necessary for calculating the interpolation value by the bilinear interpolation method Further includes storage means for storing
The interpolation value calculating means includes
Based on the first component of the 16 pixels and the first interpolation coefficient indicated by the interpolation coefficient information stored in the storage means, the interpolation value of the first component is calculated,
Interpolation of the second component based on the second component of the four pixels surrounding the first interpolation point or the second interpolation point and the second interpolation coefficient indicated by the interpolation coefficient information stored in the storage means Calculate the value,
Interpolation of the third component based on the third component of the four pixels surrounding the first interpolation point or the second interpolation point and the second interpolation coefficient indicated by the interpolation coefficient information stored in the storage means The image processing apparatus according to claim 1, wherein a value is calculated. The interpolation value calculating means includes
When the position of the interpolation point is expressed in the XY coordinate system, the first component of one of the 16 pixels is multiplied by the first interpolation coefficient in the X-axis direction corresponding to the first component 2 A first adder for adding the outputs of the two first multipliers, and a first interpolation in the Y-axis direction corresponding to the output of the first adder and the first component Eight logic circuit units comprising a second multiplier for multiplying the coefficients;
3. The four second adders for adding the outputs of the two second multipliers of each group when the eight logic circuit units are divided into four groups. Image processing device. The interpolation value calculating means includes
When the interpolation value is calculated by the cubic convolution interpolation method, the eight logic circuit units represent the first interpolation coefficient indicated by the first component of the 16 pixels and the interpolation coefficient information stored in the storage means. And calculating the interpolated value of the first component by adding the outputs of the second adders of the four groups,
When four interpolation values are calculated by bilinear interpolation, the second component of the four pixels surrounding the first interpolation point and the second interpolation coefficient indicated by the interpolation coefficient information stored in the storage means Are input to two logic circuit units of the first group of the four groups, and the output of the first group is set as the interpolation value of the second component at the first interpolation point, and the first interpolation point The third component of the four pixels surrounding the second interpolation coefficient and the second interpolation coefficient indicated by the interpolation coefficient information stored in the storage means are input to the two logic circuit sections of the second group of the four groups, The output of the second group is the third component interpolation value at the first interpolation point, the second component of the four pixels surrounding the second interpolation point, and the interpolation coefficient information stored in the storage means Two theory of the third group out of the four groups indicating the second interpolation factor Input to the circuit unit, the output of the third group as the second component interpolation value at the second interpolation point, the third component of the four pixels surrounding the second interpolation point, and stored in the storage means The second interpolation coefficient indicated by the interpolation coefficient information to be input is input to the two logic circuit sections of the fourth group of the four groups, and the output of the fourth group is the third at the second interpolation point. The image processing apparatus according to claim 3, wherein an interpolated value of the component is used. An input step of inputting image data representing an image constituted by pixels composed of mutually different first to third components;
An instruction step for instructing a magnification for changing the size of the image;
An image having a size obtained by changing the image data input in the input step by the magnification specified in the instruction step by the interpolation value calculation unit that calculates the interpolation value by the cubic convolution interpolation method or the bilinear interpolation method. A conversion step for converting into image data representing,
When the interpolation value calculation unit calculates the interpolation values of the first to third components at the interpolation point that is the position corresponding to the pixel of the image indicated by the image data after conversion in the coordinate system before conversion,
For the first component, for each interpolation point, among the pixels of the input image indicated by the image data input in the input step, four pixels surrounding the interpolation point and twelve pixels surrounding the four pixels are included. Calculating an interpolated value of the first component by a third-order convolution method based on the first component of the 16 pixels included;
For the second component and the third component, the interpolation value of the second component at the first interpolation point of the two interpolation points is simultaneously calculated for each of the two interpolation points, and the first of the pixels of the input image. Based on the second component of the four pixels surrounding the interpolation point, the interpolation value of the third component at the first interpolation point is based on the third component of the four pixels surrounding the first interpolation point. Interpolation values of the second component at the second interpolation point different from the first interpolation point of the two interpolation points are the four pixels surrounding the second interpolation point of the pixels of the input image. And the interpolated value of the third component at the second interpolation point based on the third component of the four pixels surrounding the second interpolation point by bilinear interpolation Calculate
The interpolation value of the first component calculated by the interpolation value calculator is the first component, the interpolation value of the second component calculated by the interpolation value calculator is the second component, and the third component calculated by the interpolation value calculator A conversion step of converting image data representing an image constituted by pixels having the interpolation value of the third component into image data after conversion;
An image processing method comprising: an output step of outputting the image data converted in the conversion step.

Images