WO2012114574A1 - 画像拡大装置及び方法 - Google Patents
画像拡大装置及び方法 Download PDFInfo
- Publication number
- WO2012114574A1 WO2012114574A1 PCT/JP2011/073222 JP2011073222W WO2012114574A1 WO 2012114574 A1 WO2012114574 A1 WO 2012114574A1 JP 2011073222 W JP2011073222 W JP 2011073222W WO 2012114574 A1 WO2012114574 A1 WO 2012114574A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- pixel
- change amount
- resolution image
- value
- interpolation
- Prior art date
Links
- 238000000034 method Methods 0.000 title claims description 53
- 230000008859 change Effects 0.000 claims description 242
- 230000002596 correlated effect Effects 0.000 claims 2
- 230000006870 function Effects 0.000 description 14
- 230000008569 process Effects 0.000 description 14
- 238000010586 diagram Methods 0.000 description 12
- 230000007423 decrease Effects 0.000 description 6
- 238000001514 detection method Methods 0.000 description 5
- 230000004069 differentiation Effects 0.000 description 4
- 230000000694 effects Effects 0.000 description 3
- 101100004933 Arabidopsis thaliana CYP79F1 gene Proteins 0.000 description 1
- 230000000052 comparative effect Effects 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 238000003708 edge detection Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformations in the plane of the image
- G06T3/40—Scaling of whole images or parts thereof, e.g. expanding or contracting
- G06T3/4007—Scaling of whole images or parts thereof, e.g. expanding or contracting based on interpolation, e.g. bilinear interpolation
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N1/00—Scanning, transmission or reproduction of documents or the like, e.g. facsimile transmission; Details thereof
- H04N1/387—Composing, repositioning or otherwise geometrically modifying originals
- H04N1/393—Enlarging or reducing
- H04N1/3935—Enlarging or reducing with modification of image resolution, i.e. determining the values of picture elements at new relative positions
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N1/00—Scanning, transmission or reproduction of documents or the like, e.g. facsimile transmission; Details thereof
- H04N1/40—Picture signal circuits
- H04N1/409—Edge or detail enhancement; Noise or error suppression
- H04N1/4092—Edge or detail enhancement
Definitions
- the present invention relates to an image enlargement apparatus and method.
- an image processing apparatus that performs image enlargement that switches an interpolation method according to an edge shape is known.
- the edge portion is prevented from being deteriorated due to the interpolation calculation by switching the interpolation method when a predetermined edge shape is detected.
- detection of a predetermined edge shape is performed by comparing an edge pattern with a predetermined pattern.
- the edge shape detection result may not be optimal depending on the edge shape.
- An image enlarging apparatus includes: An image enlarging device that generates a high resolution image from a low resolution image, When the high-resolution image is superimposed on the low-resolution image, a position where pixels in the high-resolution image overlap is set as a target position, and a pixel value when it is assumed that a pixel exists at the position for each target position
- An interpolation calculation unit obtained by interpolation calculation using pixel values of a plurality of pixels in the low-resolution image and an interpolation coefficient for the plurality of pixels; Among the plurality of pixels in the low-resolution image, an interpolation coefficient calculation unit that obtains the interpolation coefficient having a larger value as to a pixel having a strong correlation with the target pixel, and outputs the interpolation coefficient to the interpolation calculation unit,
- the interpolation coefficient calculation unit A first change amount calculation unit for obtaining a first change amount with respect to a first direction around the pixel of each pixel of the low resolution image
- An image enlargement method includes: An image enlargement method for generating a high resolution image from a low resolution image, When the high-resolution image is superimposed on the low-resolution image, a position where pixels in the high-resolution image overlap is set as a target position, and a pixel value when it is assumed that a pixel exists at the position for each target position
- An interpolation calculation step that is determined by an interpolation calculation using pixel values of a plurality of pixels in the low-resolution image and an interpolation coefficient for the plurality of pixels; Among the plurality of pixels in the low-resolution image, there is an interpolation coefficient calculation step for obtaining the interpolation coefficient having a larger value for a pixel having a strong correlation with the target pixel, and outputting the interpolation coefficient to the interpolation calculation step.
- the interpolation coefficient calculating step includes A first change amount calculating step for obtaining a first change amount with respect to a first direction around the pixel of each pixel of the low resolution image; A second change amount calculating step for obtaining a second change amount in a second direction around the pixel for each pixel of the low resolution image; Direction indication data representing a direction with strong correlation using the first interpolation change amount obtained by interpolating the first change amount and the second interpolation change amount obtained by interpolating the second change amount.
- the angle of the edge or the shape of the edge is not classified into any of the predetermined patterns, so that appropriate interpolation calculation can always be performed regardless of the shape of the edge.
- FIG. 1 is a block diagram showing an image enlargement apparatus according to Embodiment 1 of the present invention. It is a figure which shows roughly the arrangement
- A) And (B) is a figure showing the example of the function f (r) which decreases monotonously.
- A), (B), and (C) are diagrams showing examples of images in which at least one of the absolute values of the variation estimation values D3A51 and D3A52 of FIG. 4 is small.
- FIG. 10 is a flowchart illustrating a processing procedure of an image enlargement method according to the second embodiment. It is a flowchart which shows the process sequence of interpolation coefficient calculation step ST3A of FIG. It is a flowchart which shows the process sequence of direction calculation step ST3A5 of FIG. 12 is a flowchart showing a processing procedure of a change amount comparison step ST3A53 of FIG.
- FIG. 1 is a diagram showing the configuration of an image enlargement apparatus according to Embodiment 1 of the present invention.
- the image enlargement apparatus according to Embodiment 1 includes an interpolation coefficient calculation unit 3A and an interpolation calculation unit 3B, and uses one low-resolution image as an input image and one high-resolution image as an output image.
- a low resolution image D01 is shown as an input image
- a high resolution image D30 is shown as an output image.
- FIG. 2 is a diagram showing the low-resolution image D01, in which a part of the pixels constituting the low-resolution image D01 and the low-resolution image D01 is schematically shown.
- the smallest square represents each pixel.
- horizontal coordinates and vertical coordinates are defined along the horizontal direction and the vertical direction of the low-resolution image D01, and the pixel value of the pixel located at the coordinates represented by x and y is represented by D01 ( x, y).
- the interval between the lattice points of these coordinate axes is assumed to be equal to the interval between the pixels constituting the input image D0.
- the number of pixels in the horizontal direction of the low resolution image D01 is W
- the number of pixels in the vertical direction is H.
- the interpolation calculation unit 3B obtains the pixel value of each pixel constituting the high resolution image D30 by interpolation calculation. That is, when the low resolution image D01 and the high resolution image D30 are overlapped, that is, when the high resolution image D30 is overlapped on the plane representing the low resolution image D01, the position where the pixels of the high resolution image D30 overlap is set as the target position. Assuming that there is a pixel of the high-resolution image D30 at the target position, the pixel value of a plurality of pixels existing in the vicinity of the target position among the pixels of the low-resolution image D01 and the interpolation for the plurality of pixels Calculation is performed by interpolation using coefficients.
- the pixel value when the pixel of the high resolution image D30 is at a position where the horizontal coordinate is s and the vertical coordinate is t on the coordinate plane representing the low resolution image D01 is represented by D30 (s, t).
- D30 s, t
- the coordinate value is given on the coordinate plane representing the low resolution image D01.
- the target position where the pixels of the high resolution image D30 overlap is based on the ratio between the number of pixels of the low resolution image D01 and the number of pixels of the high resolution image D30. Desired. That is, when the number of pixels of the high resolution image D30 is U in the horizontal direction and V in the number of vertical pixels, the horizontal coordinate s of the target position is expressed by the following equation using a variable i that takes an integer value between 1 and U. (1) The vertical coordinate t of the target position can be expressed by the following equation (2) using a variable j that takes an integer value between 1 and V. Can be represented by
- FIG. 3 is a diagram showing the position of interest on the coordinate space shown in FIG. In FIG. 3, the smallest square represents each pixel of interest or a pixel of the high-resolution image D30 that overlaps each pixel of interest.
- the pixel value calculation unit 30 obtains a pixel value when it is assumed that there is a pixel at each position of interest, and sets it as the pixel value of the high resolution image D30. For example, D30 (x, y), D30 (x + 0.25, y), D30 (x, y + 0.75), D30 (x + 0.5, y + 1.5) The pixel value is obtained.
- s is a discrete value in units of W / U times the pixel pitch in the low resolution image, for example, 0.25 times
- t is H / V times the pixel pitch in the low resolution image, for example, 0.25. It is a discrete value in units of double.
- the interpolation coefficient calculation unit 3A obtains an interpolation coefficient D3A having a larger value for a pixel having a strong correlation with the target pixel among pixels existing in the low resolution image D01. Also, pixels located in a direction with strong correlation are handled as pixels with strong correlation.
- FIG. 4 is a diagram illustrating a configuration example of the interpolation coefficient calculation unit 3A.
- the interpolation coefficient calculation unit 3A includes a first change amount calculation unit 3A1, a second change amount calculation unit 3A2, a third change amount calculation unit 3A3, a fourth change amount calculation unit 3A4, a direction calculation unit 3A5, and a coefficient calculation. Part 3A6 is provided.
- the first change amount calculation unit 3A1 calculates the first derivative of the pixel value in the horizontal direction for each pixel of the low-resolution image D01, and outputs the calculation result as the first change amount D3A1.
- D3A1 (x, y) represents the first derivative of the pixel value in the horizontal direction obtained for a pixel whose horizontal coordinate is x and whose vertical coordinate is y
- D3A1 (x, y) is Calculated by equation (3).
- the second change amount calculation unit 3A2 calculates the first derivative of the pixel value in the vertical direction for each pixel of the low resolution image D01, and outputs the calculation result as the second change amount D3A2.
- D3A2 (x, y) represents the first derivative of the pixel value in the vertical direction obtained for a pixel represented by x in the horizontal coordinate and y in the vertical coordinate
- D3A2 (x, y) is expressed as follows: Calculated by equation (4).
- the third change amount calculation unit 3A3 calculates the second derivative of the pixel value in the horizontal direction for each pixel of the low resolution image D01, and outputs the calculation result as the third change amount D3A3.
- D3A3 (x, y) is expressed as D3A3 (x, y), where D3A3 (x, y) represents the second-order derivative of the pixel value in the horizontal direction obtained for the pixel represented by x in the horizontal coordinate and y in the vertical coordinate. (5).
- the fourth change amount calculation unit 3A4 calculates the second derivative of the pixel value in the vertical direction for each pixel of the low resolution image D01, and outputs the calculation result as the fourth change amount D3A4.
- D3A4 (x, y) is expressed as D3A4 (x, y), where D3A4 (x, y) represents the second derivative of the pixel value in the vertical direction obtained for the pixel represented by x in the horizontal coordinate and y in the vertical coordinate. (6).
- the direction calculation unit 3A5 uses the first change amount D3A1, the second change amount D3A2, the third change amount D3A3, and the fourth change amount data D3A5 indicating which direction the correlation is strong around the target position. Is obtained from the amount of change D3A4.
- the direction having a strong correlation with the target position means a direction in which pixels having pixel values close to the pixel values calculated when it is assumed that there is a pixel existing at the target position.
- the direction calculation unit 3A5 includes a first change amount estimation unit 3A51, a second change amount estimation unit 3A52, and a change amount comparison unit 3A53.
- the first change amount estimation unit 3A51 obtains the value of the first change amount at the target position from the first change amount D3A1 obtained for the pixels of the first low-resolution image D01 existing around the target position.
- the estimation result is output as a first change amount estimation value D3A51.
- linear interpolation is used to estimate the change amount estimation value D3A51.
- the estimated change amount D3A51 (s, t) of the target position is expressed by the following equation (7), Calculated by
- sdec and tdec represent the value of the decimal part of the coordinate value s and the value of the decimal part of the coordinate value t, respectively.
- the second change amount estimation unit 3A52 determines the value of the second change amount at the target position from the second change amount D3A2 obtained for the pixels of the first low-resolution image D01 existing around the target position.
- the estimation result is output as a second change amount estimation value D3A52.
- linear interpolation is used to estimate the change amount estimated value D3A52.
- the estimated change amount D3A52 (s, t) of the target position is expressed by the following equation (8), Calculated by
- the change amount comparison unit 3A53 for each position of interest, based on the values of the first change amount estimated value D3A51, the second change amount estimated value D3A52, the third change amount D3A3, and the fourth change amount D3A4.
- Direction indication data D3A5 is determined and output.
- the direction indication data D3A5 is a two-dimensional value composed of a first direction indication value D3A5x and a second direction indication value D3A5y.
- the change amount comparison unit 3A53 includes a first comparison determination unit 3A51, a second comparison determination unit 3A52, and a direction indication value calculation unit 3A533.
- the first comparison determination unit 3A531 has an absolute value of the first change amount estimated value D3A51 larger than the first threshold value D3A51t, and an absolute value of the second change amount estimated value D3A52 is larger than the second threshold value D3A52t.
- the first value for example, “1”
- the second value for example, “0”
- the second comparison determination unit 3A532 uses the absolute value and the fourth change amount of the third change amount D3A3 obtained for the pixel.
- the magnitude relationship of the absolute value of D3A4 is examined.
- the absolute value of the third change amount D3A3 is greater than the fourth change amount D3A4.
- the number of pixels having the third change amount D3A3 is the first count value D3A532N1, and the absolute value of the fourth change amount D3A4 is the third change value.
- the number of pixels larger than the amount D3A3 is output as the second count value D3A532N2. Note that data including the first count value D3A532AN1 and the second count value D3A532N2 may be referred to as a second comparison result D3A532.
- the direction instruction value calculation unit 3A533 uses the first change amount estimated value D3A51, the second change amount estimated value D3A52, the first comparison result D3A531, and the second comparison result D3A532 as follows.
- Data D3A5 is generated and output.
- the direction indication data D3A5 is a two-dimensional value composed of the first direction indication value D3A5x and the second direction indication value D3A5y.
- the direction instruction calculation unit 3A533 first checks what the value of the first comparison result D3A531 is in step ST1.
- the value is the first value (“1”) (that is, when both the first change amount estimated value D3A51 and the second change amount estimated value D3A52 are large)
- the process proceeds to step ST2.
- the first direction instruction value D3A5x and the second direction instruction value D3A5y are expressed by the following equation (9).
- step ST3 the difference between the first count value D3A532N1 and the second count value D3A532N2 is compared. If the absolute value of the difference is smaller than the third threshold value D3A532t, the process proceeds to step ST4.
- step ST4 the first direction instruction value D3A5x and the second direction instruction value D3A5y are expressed by the following equation (10). Calculate according to
- step ST5 the first count value D3A532N1 compares the second count value D3A532N2. If the first count value D3A532N1 is larger than the second count value D3A532N2, the process proceeds to step ST6.
- step ST6 the first direction instruction value D3A5x and the second direction instruction value D3A5y are expressed by the following equation (11). Calculate according to
- step ST7 the first direction instruction value D3A5x and the second direction instruction value D3A5y are expressed by the following equation (12). Calculate according to
- the coefficient calculation unit 3A6 calculates and outputs an interpolation coefficient D3A for the pixels constituting the low resolution image D01 based on the direction instruction data D3A5.
- the interpolation coefficient D3A is calculated so as to have a larger value for a pixel located in a direction closer to the direction in which the correlation is determined to be strong with respect to the target position.
- the interpolation coefficient D3A calculated for the interpolation coefficient D3A is represented by D3A (p, q, s, t), and a specific example of a method for calculating the interpolation coefficient D3A (p, q, s, t) is shown.
- p and q are discrete values having a unit of 0.25 pixel (0.25 times the pixel pitch in the low resolution image) in the example of FIG. 3, while s and t are as described above. These are discrete values each having a unit of W / U times, for example, 0.25 times, H / V times, for example, 0.25 times the pixel pitch in the low resolution image.
- the interpolation coefficient D3A (p, q, s, t) is expressed by the following equation (13).
- the interpolation coefficient D3A (p, q, s, t) is Is calculated.
- D3A5x (s, t) and D3A5y (s, t) are respectively calculated as the first position calculated with respect to the position of interest represented by the horizontal coordinate s and the vertical coordinate t.
- the direction indication data D3A5x and the second direction indication data D3A5y are represented.
- FIG. 6A and 6B show examples of the function f (r) that monotonously decreases with respect to the variable r.
- the function f (r) monotonously decreasing with respect to the variable r is a function satisfying f (r1) ⁇ f (r2) if r1> r2, as shown in FIG. 6A, or FIG. If r1> r2, a function satisfying f (r1) ⁇ f (r2) can be used.
- interpolation coefficient D3A can be calculated by the above-described method so that the value for a pixel located in a direction closer to the direction in which the correlation is determined to be strong with respect to the target position will be larger.
- the operation and configuration of the interpolation coefficient calculation unit 3A are as described above.
- the interpolation calculation unit 3B obtains the pixel value D30 (s, t) by interpolation calculation. This interpolation calculation is performed based on the pixel value D01 (p, q) of each pixel of the low resolution image D01 and the interpolation coefficient D3A (p, q, s, t) for each pixel. It is expressed.
- D01 (p, q) is the pixel that is present at the position represented by the horizontal coordinate p and the vertical coordinate q among the pixels constituting the low resolution image D01. Represents a given pixel value.
- the change amount estimation values D3A51 and D3A52 will be described.
- the x axis corresponds to the horizontal coordinate of the high resolution image D30
- the y axis corresponds to the vertical coordinate of the high resolution image D30
- the z axis corresponds to the luminance of the high resolution image D30.
- the two-dimensional vector having the variation estimated value D3A51 as the first component and the variation estimated value D3A52 as the second component is on the curved surface representing the high-resolution image D30.
- a curved surface can be defined as a two-variable function in a coordinate space consisting of an x-axis, a y-axis, and a z-axis.
- the value taken by f (x, y) is a pixel value of a pixel existing at a position where the horizontal coordinate is x and the vertical coordinate is y.
- a two-dimensional vector (D3A1, D3A2) having the first change amount D3A1 represented by the expression (3) as the first component and the second change amount D3A2 represented by the expression (4) as the second component.
- this two-dimensional vector becomes the gradient of the curved surface representing the low image degree image D01.
- the contour lines on the curved surface are obtained by connecting pixels having the same pixel value in the image represented by the curved surface. Therefore, it can be considered that pixels having the same pixel value as the pixels existing at the position where the gradient is calculated are arranged in the direction orthogonal to the direction indicated by the gradient at each point on the curved surface representing the image. If the gradient at the position of interest on the curved surface representing the high-resolution image D30 can be calculated from the above properties, the pixel value close to the pixel value calculated for the pixel existing at the position of interest is calculated based on the value. It is possible to determine the direction in which the pixels are arranged, that is, the direction in which the correlation with the target pixel is strong.
- the image enlargement apparatus approximately obtains the gradient of the high resolution image D30 from the gradient of the low resolution image D01.
- the reason why the gradient of the high resolution image D30 can be approximately obtained from the gradient of the low resolution image D01 is as follows.
- the two-variable function corresponding to a certain image can be regarded as a continuous function whose value changes continuously.
- its first derivative value also changes continuously, so the first derivative value of any point on the continuous function is interpolated with the first derivative value given to multiple points located near that point. It will be almost the same value as the one.
- the gradient of the curved surface is expressed by the first derivative of the function representing the curved surface, the gradient of each point on the curved surface is also given to multiple points located near that point. It can be seen that approximation is possible by interpolating the gradient obtained.
- the two images are the same if the difference in the number of pixels is ignored, and the curved surfaces representing the two in the coordinate space are almost the same. Therefore, the gradient of the high resolution image D30 can be obtained approximately by interpolating the gradient of the low resolution image D01 as in equations (7) and (8).
- the estimated change amount D3A51 and the estimated change amount D3A52 obtained by interpolating the first change amount D3A1 and the second change amount D3A2 representing the gradient of the low-resolution image D01 are respectively referred to as the first component and the second component.
- the two-dimensional vector to be used is a vector that approximately represents the gradient of the high-resolution image D30.
- the first direction instruction value D3A5x and the comparison result of the third change amount D3A3 and the fourth change amount D3A4 The second direction instruction value D3A5y is obtained. This has the following actions and effects.
- FIGS. 7A to 7C show typical images in which at least one of the absolute values of the change amount estimation values D3A51 and D3A52 is a small value.
- FIG. 7A shows a case where a vertical stripe pattern exists in the image
- FIG. 7B shows a case where a horizontal stripe pattern exists in the image
- FIG. 7C shows a pixel value in the image. This is a case where no change is observed.
- the absolute value of the second change amount D3A2 does not become a very large value, and thus the change obtained by interpolating the second change amount D3A2
- the absolute value of the quantity estimated value D3A52 is not too large.
- the absolute value of the first change amount D3A1 does not become a very large value, so the change obtained by interpolating the first change amount D3A1.
- the absolute value of the quantity estimated value D3A51 does not become too large.
- the absolute value of both the first change amount D3A1 and the second change amount D3A2 is not so large.
- the absolute values of D3A51 and D3A52 are not so large.
- the absolute value of the first-order horizontal differential of the pixel value of the low-resolution image D01 is expected to be a large value to some extent.
- the value of the first derivative can take a positive value or a negative value.
- the first derivative calculated for the target position in the high resolution image D30 is given by interpolation of the first derivative calculated for each pixel of the low resolution image D01.
- D3A1 (s-sdec, t-tdec) and D3A1 (s-sdec, t-tdec + 1) are positive values
- D3A1 (s-sdec + 1, t-tdec) and D3A1 (s-sdec + 1, t-tdec + 1) If is a negative value, the positive value and the negative value are added together, and the change amount estimation value D3A51 obtained as a result of the interpolation calculation can be a value close to zero, that is, a small value.
- the first direction indication value D3A5x should take a value close to 1 and the second direction indication value D3A5y should take a value close to 0.
- the absolute value of the change amount estimation value D3A51 needs to be much larger than the absolute value of the change amount estimation value D3A52, but as described above, the absolute value of the change amount estimation value D3A51 takes a small value. As such, there is no guarantee that such a relationship will be satisfied. In other words, when there is a vertical stripe pattern, the first direction instruction value D3A5x and the second direction instruction value D3A5y may not be appropriately obtained from the change amount estimation values D3A51 and D3A52.
- the first direction indication value D3A5x should take a value close to “0”, and the second direction indication value D3A5y should take a value close to “1”. From the amount estimated values D3A51 and D3A52, the first direction indication value D3A5x and the second direction indication value D3A5y may not be appropriately obtained.
- the second derivative of the pixel value calculated for each pixel of the low resolution image D01. The absolute value in the vertical direction is small, but the absolute value of the second-order differential in the horizontal direction is somewhat large. Therefore, when the periphery of the target position is a vertical stripe pattern, the absolute value of the horizontal second-order derivative calculated for each pixel of the low-resolution image D01 existing around the target position and the vertical secondary When comparing the absolute values of the derivatives, the number of pixels having a larger absolute value of the second-order differential in the horizontal direction is larger. Therefore, the first count value D3A532N1 is larger than the second count value D3A532N2.
- the first count value D3A532N2 is larger than the second count value D3A532N1 from the same discussion. Value.
- both the horizontal direction and the vertical direction of the image calculated for the pixels around the target position are displayed.
- the absolute value of the second derivative does not become too large, and it is determined by chance which one takes the larger value. Therefore, when comparing the absolute value of the second-order differential in the horizontal direction and the absolute value of the second-order differential in the vertical direction calculated for the pixels around the target position, the pixel whose absolute value of the second-order differential in the vertical direction is larger There is no significant difference between the number of pixels and the number of pixels for which the absolute value of the second-order differential in the horizontal direction is larger. Therefore, the first count value D3A532N1 and the second count value D3A532N2 are substantially the same value.
- the direction indication value D3A5y can be obtained appropriately. That is, when D3A532N1> D3A532N2, the first direction indication value D3A5x is set to “1”, the second direction indication value D3A5y is set to “0”, and conversely when D3A532N1 ⁇ D3A532N2, Since the direction has a strong correlation, the first direction indication value D3A5x is set to “0” and the second direction indication value D3A5y is set to “1”.
- the interpolation coefficient D3A changes based on the first direction instruction value D3A5x and the second direction instruction value D3A5y, the first direction instruction value D3A5x and the second direction instruction value D3A5y are appropriately set. Obtaining leads to finding the interpolation coefficient D3A appropriately.
- the reason why both the first direction indication value D3A5x and the second direction indication value D3A5y are zero is as follows. This is because the interpolation coefficient can be obtained so that the pixels around the target position are equally weighted when there is no direction.
- the first direction indication value D3A5x and the second direction indication value D3A5y or the direction indication data D3A5 are data representing a direction having a strong correlation.
- the first change amount D3A1 is the first-order horizontal differential of the change in the pixel value of the low-resolution image D01
- the second change amount D3A2 is the vertical direction of the change in the pixel value of the low-resolution image D01.
- the absolute value of the second-order differential of the pixel value change of the low-resolution image D01 as the third change amount D3A3
- the vertical direction of the change of the pixel value of the low-resolution image D01 as the fourth change amount D3A4.
- the absolute value of the second derivative of is used, the values that can be used as the first change amount D3A1, the second change amount D3A2, the third change amount D3A3, and the fourth change amount D3A4 are shown in the above example. Not exclusively.
- the first change amount D3A1 and the third change amount D3A3, and the second change amount D3A2 and the fourth change amount D3A4 may have different properties or may be obtained by different methods.
- the first change amount D3A1 and the third change amount D3A3, and the second change amount D3A2 and the fourth change amount D3A4 having different properties, the first change amount D3A1 and the second change amount are used. Even when it is difficult to determine a strong correlation direction based on the amount D3A2, it is possible to determine a strong correlation direction based on the third change amount D3A3 and the fourth change amount D3A4. That is, a direction with a strong correlation can be obtained more accurately.
- the interpolation coefficient D3A will be described. A pixel located in a direction with a strong correlation is treated as a pixel with a strong correlation, and a larger interpolation coefficient D3A is given.
- the interpolation coefficient D3A is given by a function that decreases monotonously with respect to r, the value decreases as the distance from the contour line passing through the target position increases.
- the longer the distance from the contour line the weaker the correlation with the target position.
- the interpolation coefficient D3A is required to have a smaller value as the pixel is located in a direction where the correlation is weaker. In other words, it is required that the pixels located in the direction of strong correlation have a larger value. That is, by calculating the interpolation coefficient D3A from the first direction instruction value D3A5x and the second direction instruction value D3A5y, the correlation with the target position can be taken into account when calculating the interpolation coefficient D3A.
- the first direction instruction value D3A5x and the second direction instruction value D3A5y are determined by the equation (10), the above argument does not hold, but in this case, the value of r is set to the values of p and q. It is always zero regardless. Therefore, the same value is used as the interpolation coefficient for all pixels.
- the first direction indication value D3A5x and the second direction indication value D3A5y are determined by the equation (10) when there is no specific direction having a strong correlation. Therefore, the interpolation coefficient need not be a large value in a specific direction. In other words, when there is no direction with strong correlation, the interpolation coefficient can be obtained so that all directions can be weighted equally.
- the target pixel when the pixel value of the pixel existing at the target position (hereinafter referred to as the target pixel) is obtained by interpolating the pixel values of the pixels positioned around the target pixel, Since a direction having a strong correlation with respect to the target pixel is determined and an interpolation coefficient is obtained according to the result, an interpolation coefficient for a pixel having a strong correlation with the target pixel has a higher weight. Further, since the interpolation calculation using the interpolation coefficient determined as described above is performed, the pixel value of the target pixel can be obtained more accurately.
- the target pixel is included in an area with low luminance.
- the pixel value of the target pixel is higher than that of the surrounding low-luminance pixels, and a sense of discomfort occurs.
- the pixel value of the pixel of interest is obtained by interpolating the pixel values around the pixel of interest, if the interpolation coefficient is calculated in consideration of the strength of the correlation with the pixel of interest, the weight for the pixel with high luminance is small. Thus, weighting for pixels with low luminance is increased. Therefore, the pixel value of the pixel of interest becomes the same value as the surrounding low-brightness pixel, and the above-mentioned uncomfortable feeling does not occur.
- the direction of strong correlation obtained for the target pixel is represented by a continuous value. Therefore, since the edge angle or edge shape is not classified into any of the predetermined patterns, appropriate interpolation calculation can always be performed regardless of the edge shape.
- the method of obtaining the correlation for the target pixel is not limited to the above example.
- a pixel at a position distant from the target pixel is likely to have a pixel value different from that of the target pixel. Therefore, in addition to r shown in Expression (13), the interpolation coefficient may be changed according to the distance from the target position. For example, the interpolation coefficient may be reduced as the distance from the position of interest increases. By considering the distance from the target pixel, the correlation between the target pixel and surrounding pixels is more strictly considered, and the pixel value of the target pixel can be obtained more appropriately.
- Equation (13) depends on (ps) and (qt), and f (r) in equation (14) decreases with increasing r, so the distance is The larger the value, the smaller the interpolation coefficient. Therefore, if the interpolation coefficient is reduced in accordance with the distance from the target pixel in addition to r, the distance from the target pixel is considered twice.
- the first change amount D3A1 may be a value corresponding to the first derivative in the horizontal direction of the change in the pixel value of the low resolution image D01. Therefore, the method for obtaining it is not limited to Equation (3) as long as it is a method that can approximately calculate the first derivative in the horizontal direction of the change in the pixel value of the low-resolution image D01.
- the second change amount D3A2 only needs to be a value corresponding to the first-order differentiation in the vertical direction of the change in the pixel value of the low-resolution image D01. Therefore, the method for obtaining the pixel value is not limited to the equation (4) as long as it is a method that can approximately calculate the first derivative in the vertical direction of the change in the pixel value of the low resolution image D01.
- the first change amount D3A1 and the second change amount D3A2 are a combination of a first-order derivative in the horizontal direction and a first-order derivative in the vertical direction of the change in the pixel value of the low resolution image D01. May be any combination of first-order differentiations of pixel value changes in different directions. Therefore, in general terms, the first change amount D3A1 is the first derivative of the change in the pixel value of the low resolution image D01 in the first direction, and the second change amount D3A2 is the change in the pixel value of the low resolution image D01. What is necessary is just the primary differentiation regarding a 2nd direction.
- the third change amount D3A3 may be a value corresponding to the second order differential in the horizontal direction of the change in the pixel value of the low resolution image D01. Therefore, the method for obtaining it is not limited to Equation (5) as long as it is a method that can approximately calculate the second derivative in the horizontal direction of the change in the pixel value of the low resolution image D01.
- the fourth change amount D3A4 may be a value corresponding to the second derivative in the vertical direction of the change in the pixel value of the low resolution image D01. Therefore, the method for obtaining it is not limited to Expression (6) as long as it is a method that can approximately calculate the second derivative in the vertical direction of the change in the pixel value of the low resolution image D01.
- the third change amount D3A3 and the fourth change amount D3A4 are a combination of the second-order derivative in the horizontal direction and the second-order derivative in the vertical direction of the change in the pixel value of the low resolution image D01. Both may be a combination of the second derivative of the change in the pixel value with respect to different directions.
- the direction of change of the third change amount D3A3 and the direction of change of the fourth change amount are respectively the direction of change of the first change amount (first direction) and the direction of change of the second change amount. (The second direction) may not be the same.
- the third change amount D3A3 is the second derivative of the change in the pixel value of the low resolution image D01 in the third direction
- the fourth change amount D3A4 is the change in the pixel value of the low resolution image D01. It is sufficient that the second-order derivative is related to the fourth direction.
- the image enlargement apparatus can be used as a part of an image display apparatus.
- it can be used when the number of pixels of the input image for the image display device is smaller than the number of pixels of the display unit of the image display device.
- FIG. FIG. 8 shows an arithmetic unit for executing the image enlargement method according to the second embodiment of the present invention.
- the image enlargement method of the present invention generates a high resolution image D30 from the low resolution image D01.
- An arithmetic unit for executing the image enlargement method according to the present invention includes an input interface IF1, an output interface IF2, a processor CPU1, a program memory MEM1, and a data memory MEM2, and a data bus BUS1 connecting them.
- the processor CPU1 operates according to a program stored in the program memory MEM1.
- Various data generated in the course of operation is stored in the data memory MEM2.
- the low resolution image D01 and the low resolution image D02 are input to the arithmetic unit via the input interface IF1.
- the high resolution image D30 generated by the image enlargement method according to the present invention is output to the outside of the arithmetic unit via the output interface IF2.
- FIG. 9 is a diagram showing the processing procedure of the image enlargement method according to the present invention.
- the image enlargement method according to the present invention includes an interpolation coefficient calculation step ST3A and an interpolation calculation step ST3B.
- the pixel value of each pixel constituting the high resolution image D30 is obtained by interpolation calculation.
- FIG. 10 is a diagram illustrating a processing procedure of the interpolation coefficient calculation step ST3A.
- the interpolation coefficient calculation step ST3A includes a first change amount calculation step ST3A1, a second change amount calculation step ST3A2, a third change amount calculation step ST3A3, a fourth change amount calculation step ST3A4, a direction calculation step ST3A5, and a coefficient.
- Calculation step ST3A6 is included.
- the first change amount calculation step ST3A1 calculates the first derivative of the pixel value in the horizontal direction for each pixel of the low resolution image D01 by the same process as the first change amount calculation unit 3A1 described in the first embodiment, A first change amount D3A1 is obtained.
- the second change amount calculation step ST3A2 calculates the first derivative of the pixel value in the vertical direction for each pixel of the low resolution image D01 by the same process as the second change amount calculation unit 3A2 described in the first embodiment. A second change amount D3A2 is obtained.
- the third change amount calculation step ST3A3 calculates a second derivative of the pixel value in the horizontal direction for each pixel of the low resolution image D01 by the same processing as the third change amount calculation unit 3A3 described in the first embodiment.
- the third change amount D3A3 is obtained.
- the fourth change amount calculation step ST3A4 calculates the second derivative of the pixel value in the vertical direction for each pixel of the low resolution image D01 by the same process as the fourth change amount calculation unit 3A4 described in the first embodiment.
- the fourth change amount D3A4 is obtained.
- the direction calculation step ST3A5 will be described with reference to FIG.
- FIG. 11 is a diagram illustrating a processing procedure of the direction calculation step ST3A5.
- the direction calculation step ST3A5 includes a first change amount estimation step ST3A51, a second change amount estimation step ST3A52, and a change amount comparison step ST3A53.
- the first change amount estimation value D3A51 is obtained by the same processing as that of the first change amount estimation unit 3A51 described in the first embodiment.
- the first change amount estimation value D3A51 is obtained by the same process as the second change amount estimation unit 3A52 described in the first embodiment.
- the change amount comparison step ST3A53 will be described with reference to FIG.
- FIG. 12 is a diagram illustrating a processing procedure of the change amount comparison step ST3A53.
- the change amount comparison step ST3A53 includes a first comparison determination step ST3A531, a second comparison determination step ST3A532, and a direction instruction value step ST3A533.
- the first comparison result D3A531 is obtained by the same process as the first comparison determination unit 3A531 of the first embodiment.
- the second comparison result D3A532 composed of the first count value D3A532AN1 and the second count value D3A532N2 calculated by the same processing as that of the second comparison determination unit 3A532 of the first embodiment is used. Ask.
- Direction indication value calculation step ST3A533 obtains direction indication data D3A5 by the same processing as the direction indication value calculation unit 3A533 of the first embodiment.
- the operation of the change amount comparison step ST3A53 is as described above, and the operation is the same processing as the change amount comparison unit 3A53 described in the first embodiment.
- the operation of the direction calculation step ST3A5 is as described above, and the operation is the same as that of the direction calculation unit 3A5 described in the first embodiment.
- the coefficient calculation step ST3A6 calculates the interpolation coefficient D3A by the same process as the coefficient calculation unit 3A6 described in the first embodiment.
- the operation of the interpolation coefficient calculation step ST3A is as described above, and the operation is the same as that of the interpolation coefficient calculation unit 3A described in the first embodiment.
- the interpolation calculation step ST3B will be described.
- the pixel value of each pixel of the high resolution image D30 is obtained by the same processing as the interpolation calculation unit 3B described in the first embodiment.
- the operation of the image enlargement method according to the second embodiment of the present invention is as described above. Since the image enlargement method of the present invention can perform the same processing as the image enlargement apparatus of the first embodiment of the present invention, it has the same effect as the image enlargement apparatus of the first embodiment of the present invention.
- the modification of the image enlargement apparatus according to the first embodiment of the present invention can also be applied to the image enlargement method according to the second embodiment of the present invention.
- the image enlargement method according to the present embodiment or the arithmetic device that performs the method can be used as a part of the image display device.
- it can be used when the number of pixels of the input image for the image display device is smaller than the number of pixels of the display unit of the image display device.
- 3A interpolation coefficient calculation unit 3B interpolation calculation unit, D01 low resolution image, D3A interpolation coefficient, D30 high resolution image.
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Image Processing (AREA)
- Editing Of Facsimile Originals (AREA)
Abstract
Description
低解像度画像から高解像度画像を生成する画像拡大装置であって、
前記高解像度画像を前記低解像度画像上に重ねた場合に前記高解像度画像内の画素が重なる位置を注目位置とし、前記注目位置ごとにその位置に画素が存在すると仮定した場合の画素値を前記低解像度画像内の複数の画素の持つ画素値と、該複数の画素についての補間係数とを用いた補間演算によって求める補間演算部と、
前記低解像度画像内の複数の画素のうち、前記注目画素との相関が強い画素に対するものほど大きな値となる前記補間係数を求め、前記補間演算部に出力する補間係数算出部を有し、
前記補間係数算出部は、
前記低解像度画像の各画素に対してその画素を中心とした第1の方向に関する第1の変化量を求める第1の変化量算出部と、
前記低解像度画像の各画素に対してその画素を中心とした第2の方向に関する第2の変化量を求める第2の変化量算出部と、
前記第1の変化量を補間して得た第1の補間変化量と、前記第2の変化量を補間して得た第2の補間変化量を用いて相関の強い方向を表す方向指示データを求める方向算出部と、
前記方向指示データから前記補間係数を求める係数算出部を含む
ことを特徴とする。
低解像度画像から高解像度画像を生成する画像拡大方法であって、
前記高解像度画像を前記低解像度画像上に重ねた場合に前記高解像度画像内の画素が重なる位置を注目位置とし、前記注目位置ごとにその位置に画素が存在すると仮定した場合の画素値を前記低解像度画像内の複数の画素の持つ画素値と、該複数の画素についての補間係数とを用いた補間演算によって求める補間演算ステップと、
前記低解像度画像内の複数の画素のうち、前記注目画素との相関が強い画素に対するものほど大きな値となる前記補間係数を求め、前記補間演算ステップに出力する補間係数算出ステップを有し、
前記補間係数算出ステップは、
前記低解像度画像の各画素に対してその画素を中心とした第1の方向に関する第1の変化量を求める第1の変化量算出ステップと、
前記低解像度画像の各画素に対してその画素を中心とした第2の方向に関する第2の変化量を求める第2の変化量算出ステップと、
前記第1の変化量を補間して得た第1の補間変化量と、前記第2の変化量を補間して得た第2の補間変化量を用いて相関の強い方向を表す方向指示データを求める方向算出ステップと、
前記方向指示データから前記補間係数を求める係数算出ステップを含む
ことを特徴とする。
図1は本発明の実施の形態1による画像拡大装置の構成を表す図である。実施の形態1による画像拡大装置は、補間係数算出部3A及び補間演算部3Bを備え、1枚の低解像度画像を入力画像とし、1枚の高解像度画像を出力画像とする。図1では入力画像として低解像度画像D01が記され、出力画像として高解像度画像D30が記されている。
画素値算出部30は各注目位置に画素が存在すると仮定した場合の画素値を求め、高解像度画像D30の画素値とする。例えば、
D30(x,y)、
D30(x+0.25,y)、
D30(x,y+0.75)、
D30(x+0.5,y+1.5)
といった画素値を求めることになる。
即ち、sは、低解像度画像における画素ピッチのW/U倍、例えば0.25倍を単位とする離散値であり、tは、低解像度画像における画素ピッチのH/V倍、例えば0.25倍を単位とする離散値である。
まず、補間係数算出部3Aの動作、構成について説明する。以下に詳しく述べるように、補間係数算出部3Aは、低解像度画像D01内に存在する画素のうち、注目画素との相関が強い画素に対するものほど大きな値となる補間係数D3Aを求めるものであり、また、相関の強い方向に位置する画素を相関の強い画素として扱うものである。
変化量推定値D3A52の推定には例えば線形補間を用いる。この場合、水平座標がs、垂直座標がtで表される位置が注目位置であったとした場合、該注目位置の変化量推定値D3A52(s,t)は、下記の式(8)、
ここで、p、qは、図3の例では、0.25画素(低解像度画像における画素ピッチの0.25倍)を単位とする離散値であり、一方、s、tは上記のように、それぞれ低解像度画像における画素ピッチのW/U倍、例えば0.25倍、H/V倍、例えば0.25倍を単位とする離散値である。
補間係数算出部3Aの動作、構成は以上の通りである。
補間演算部3Bは、画素値D30(s,t)を補間演算によって求める。この補間演算は、低解像度画像D01の各画素の画素値D01(p,q)と、各画素についての補間係数D3A(p,q,s,t)とに基づいて行われるものであり、例えば
まず、変化量推定値D3A51及びD3A52について説明する。以下に説明するように、高解像度画像D30を、x軸が高解像度画像D30の水平座標に対応し、y軸が高解像度画像D30の垂直座標に対応し、z軸が高解像度画像D30の輝度値に対応する座標空間内の曲面で表した場合、変化量推定値D3A51を第1成分とし、変化量推定値D3A52を第2成分とする2次元ベクトルは、高解像度画像D30を表す曲面上の各点での勾配に相当するベクトルとなる。
以上の性質から、高解像度画像D30を表す曲面上の注目位置での勾配を計算することが出来れば、その値によって、注目位置に存在する画素に対して計算される画素値に近い画素値をもった画素が並ぶ方向、即ち、注目画素に対して相関が強い方向を判断することが出来る。
図7(A)のように画像内に垂直方向の縞模様がある場合、第2の変化量D3A2の絶対値はあまり大きな値にならないため、第2の変化量D3A2を補間して得られる変化量推定値D3A52の絶対値もあまり大きな値にならない。
図7(B)のように画像内に水平方向の縞模様がある場合、第1の変化量D3A1の絶対値はあまり大きな値にならないため、第1の変化量D3A1を補間して得られる変化量推定値D3A51の絶対値もあまり大きな値にはならない。
図7(C)のように画像内で画素値の変化が見られない場合、第1の変化量D3A1及び第2の変化量D3A2ともにその絶対値はあまり大きな値にならないため、変化量推定値D3A51及びD3A52ともにその絶対値はあまり大きな値にならない。
一方、高解像度画像D30内の注目位置に対して計算される一次微分は低解像度画像D01の各画素に対して計算される一次微分の補間によって与えられる。
相関が強い方向に位置する画素ほど相関の強い画素であるとして扱われ、より大きな値の補間係数D3Aが与えられる。
一方、第1の方向指示値D3A5x及び第2の方向指示値D3A5yが式(10)によって決定されるのは、相関の強い特定の方向が存在しない場合である。よって補間係数としても、特定の方向に対して大きな値にする必要はない。逆に言えば相関の強い方向がない場合、全ての方向に対して等しく重み付けできるよう補間係数を求めることが出来る。
なお、式(13)に示したrは、(p-s)、(q-t)に依存し、式(14)のf(r)は、rの増加に対して減少するので、距離が大きくなるほど補間係数が小さくなる。従って、rに加えて注目画素からの距離にも応じて補間係数が小さくなるようにすると、注目画素からの距離を二重に考慮することになる。
D30(x,y)=D01(x,y)
D30(x-1,y)=D01(x-1,y)
D30(x,y+1)=D01(x,y+1)
等としてもよい。
図8は本発明の実施の形態2による画像拡大方法を実行するための演算装置を表す。本発明の画像拡大方法は低解度画像D01から高解像度画像D30を生成する。
プロセッサCPU1はプログラムメモリMEM1に記憶されたプログラムに従って動作する。動作の過程で発生する様々なデータはデータメモリMEM2に記憶される。低解度画像D01、低解像度画像D02は入力インターフェースIF1を介して演算装置へ入力される。本発明による画像拡大方法によって生成される高解像度画像D30は出力インターフェースIF2を介して演算装置の外部へ出力される。
高解像度画像D30を構成する各画素の画素値を補間演算によって求める。
図10は、補間係数算出ステップST3Aの処理手順を表す図である。補間係数算出ステップST3Aは、第1の変化量算出ステップST3A1、第2の変化量算出ステップST3A2、第3の変化量算出ステップST3A3、第4の変化量算出ステップST3A4、方向算出ステップST3A5、及び係数算出ステップST3A6を含む。
図11は方向算出ステップST3A5の処理手順を表す図である。方向算出ステップST3A5は、第1の変化量推定ステップST3A51、第2の変化量推定ステップST3A52、及び変化量比較ステップST3A53を含む。
図12は変化量比較ステップST3A53の処理手順を表す図である。変化量比較ステップST3A53は、第1の比較判定ステップST3A531、第2の比較判定ステップST3A532、及び方向指示値ステップST3A533を含む。
また、方向算出ステップST3A5の動作は以上の通りであり、その動作は、実施の形態1で説明した方向算出部3A5と同様である。
補間演算ステップST3Bは実施の形態1で説明した補間演算部3Bと同様の処理で高解像度画像D30の各画素の画素値を求める。
Claims (10)
- 低解像度画像から高解像度画像を生成する画像拡大装置であって、
前記高解像度画像を前記低解像度画像上に重ねた場合に前記高解像度画像内の画素が重なる位置を注目位置とし、前記注目位置ごとにその位置に画素が存在すると仮定した場合の画素値を前記低解像度画像内の複数の画素の持つ画素値と、該複数の画素についての補間係数とを用いた補間演算によって求める補間演算部と、
前記低解像度画像内の複数の画素のうち、前記注目画素との相関が強い画素に対するものほど大きな値となる前記補間係数を求め、前記補間演算部に出力する補間係数算出部を有し、
前記補間係数算出部は、
前記低解像度画像の各画素に対してその画素を中心とした第1の方向に関する第1の変化量を求める第1の変化量算出部と、
前記低解像度画像の各画素に対してその画素を中心とした第2の方向に関する第2の変化量を求める第2の変化量算出部と、
前記第1の変化量を補間して得た第1の補間変化量と、前記第2の変化量を補間して得た第2の補間変化量を用いて相関の強い方向を表す方向指示データを求める方向算出部と、
前記方向指示データから前記補間係数を求める係数算出部を含む
ことを特徴とする画像拡大装置。 - 前記補間係数算出部は、
相関の強い方向に位置する画素が前記相関の強い画素であるとして処理するものであり、
前記相関の強い方向を、前記低解像度画像から求める
ことを特徴とする請求項1に記載の画像拡大装置。 - 前記第1の変化量は前記低解像度画像の各画素を中心として前記第1の方向に関する画素値の変化の一次微分を求めたものであり、
前記第2の変化量は前記低解像度画像の各画素を中心として前記第2の方向に関する画素値の変化の一次微分を求めたものである
ことを特徴とする請求項1に記載の画像拡大装置。 - 前記補間係数算出部は、
前記低解像度画像の各画素に対してその画素を中心とした第3の方向に関する第3の変化量を求める第3の変化量算出部と、
前記低解像度画像の各画素に対してその画素を中心とした第4の方向に関する第4の変化量を求める第4の変化量算出部をさらに含み、
前記方向算出部は、前記第1から第4の変化量をもとに、前記方向指示データを求める
ことを特徴とする請求項3に記載の画像拡大装置。 - 前記第3の変化量は前記低解像度画像の各画素を中心として前記第3の方向に関する画素値の変化の二次微分の絶対値を求めたものであり、
前記第4の変化量は前記低解像度画像の各画素を中心として前記第4の方向に関する画素値の変化の二次微分の絶対値を求めたものである
ことを特徴とする請求項4に記載の画像拡大装置。 - 前記方向算出部は、
前記第1の補間変化量の絶対値が第1の閾値以下もしくは前記第2の補間変化量の絶対値が第2の閾値以下の場合、
前記第3の変化量及び前記第4の変化量に応じて前記相関の強い方向を判断する
ことを特徴とする請求項5に記載の画像拡大装置。 - 前記方向算出部は、
前記注目画素の近傍に存在する、前記低解像度画像の画素ごとに前記第3の変化量の絶対値と前記第4の変化量の絶対値を比較した結果、
前記第3の変化量の絶対値より前記第4の変化量の絶対値の方が大きい画素の数が多い場合、前記第3の方向に相関が強いと判断し、
前記第4の変化量の絶対値より前記第3の変化量の絶対値の方が大きい画素の数が多い場合、前記第4の方向に相関が強いと判断する
ことを特徴とする請求項6に記載の画像拡大装置。 - 前記方向算出部は、
前記注目画素の近傍に存在する、前記低解像度画像の画素ごとに前記第3の変化量の絶対値と前記第4の変化量の絶対値を比較した結果、
前記第3の変化量の絶対値より前記第4の変化量の絶対値の方が大きい画素の数と、前記第4の変化量の絶対値より前記第3の変化量の絶対値の方が大きい画素の数の差が第3の閾値より小さい場合、
相関が強い方向は存在しないと判断する
ことを特徴とする請求項6に記載の画像拡大装置。 - 前記係数算出部は、
前記低解像度画像内に存在する画素と前記注目画素の距離及び前記方向指示データに基づいて、前記補間係数を求める
ことを特徴とする請求項1に記載の画像拡大装置。 - 低解像度画像から高解像度画像を生成する画像拡大方法であって、
前記高解像度画像を前記低解像度画像上に重ねた場合に前記高解像度画像内の画素が重なる位置を注目位置とし、前記注目位置ごとにその位置に画素が存在すると仮定した場合の画素値を前記低解像度画像内の複数の画素の持つ画素値と、該複数の画素についての補間係数とを用いた補間演算によって求める補間演算ステップと、
前記低解像度画像内の複数の画素のうち、前記注目画素との相関が強い画素に対するものほど大きな値となる前記補間係数を求め、前記補間演算ステップに出力する補間係数算出ステップを有し、
前記補間係数算出ステップは、
前記低解像度画像の各画素に対してその画素を中心とした第1の方向に関する第1の変化量を求める第1の変化量算出ステップと、
前記低解像度画像の各画素に対してその画素を中心とした第2の方向に関する第2の変化量を求める第2の変化量算出ステップと、
前記第1の変化量を補間して得た第1の補間変化量と、前記第2の変化量を補間して得た第2の補間変化量を用いて相関の強い方向を表す方向指示データを求める方向算出ステップと、
前記方向指示データから前記補間係数を求める係数算出ステップを含む
ことを特徴とする画像拡大方法。
Priority Applications (3)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US13/983,523 US9020303B2 (en) | 2011-02-21 | 2011-10-07 | Image magnification device and method |
JP2013500828A JP5701374B2 (ja) | 2011-02-21 | 2011-10-07 | 画像拡大装置及び方法 |
CN201180068016.0A CN103380615B (zh) | 2011-02-21 | 2011-10-07 | 图像放大装置及方法 |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2011-034294 | 2011-02-21 | ||
JP2011034294 | 2011-02-21 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2012114574A1 true WO2012114574A1 (ja) | 2012-08-30 |
Family
ID=46720378
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2011/073222 WO2012114574A1 (ja) | 2011-02-21 | 2011-10-07 | 画像拡大装置及び方法 |
Country Status (4)
Country | Link |
---|---|
US (1) | US9020303B2 (ja) |
JP (1) | JP5701374B2 (ja) |
CN (1) | CN103380615B (ja) |
WO (1) | WO2012114574A1 (ja) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US9390471B1 (en) | 2015-02-05 | 2016-07-12 | Synaptics Display Devices Gk | Device and method for image scaling |
CN107256377A (zh) * | 2012-09-12 | 2017-10-17 | 威智伦富智堡公司 | 用于检测视频中的对象的方法、设备和系统 |
US10192286B2 (en) | 2014-09-11 | 2019-01-29 | Synaptics Japan Gk | Device and method for image enlargement and display panel driver using the same |
Families Citing this family (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP6276580B2 (ja) * | 2013-12-20 | 2018-02-07 | 株式会社メガチップス | 画素補間処理装置、撮像装置、プログラムおよび集積回路 |
CN104021523B (zh) * | 2014-04-30 | 2017-10-10 | 浙江师范大学 | 一种基于边缘分类的图像超分辨率放大的方法 |
CN106454288B (zh) | 2016-11-29 | 2018-01-19 | 广东欧珀移动通信有限公司 | 控制方法、控制装置、成像装置及电子装置 |
CN106454054B (zh) | 2016-11-29 | 2019-03-19 | Oppo广东移动通信有限公司 | 控制方法、控制装置及电子装置 |
CN106507068B (zh) | 2016-11-29 | 2018-05-04 | 广东欧珀移动通信有限公司 | 图像处理方法及装置、控制方法及装置、成像及电子装置 |
CN106341670B (zh) | 2016-11-29 | 2017-09-22 | 广东欧珀移动通信有限公司 | 控制方法、控制装置及电子装置 |
CN106604001B (zh) * | 2016-11-29 | 2018-06-29 | 广东欧珀移动通信有限公司 | 图像处理方法、图像处理装置、成像装置及电子装置 |
CN106504218B (zh) | 2016-11-29 | 2019-03-12 | Oppo广东移动通信有限公司 | 控制方法、控制装置及电子装置 |
CN109523469B (zh) * | 2018-11-16 | 2022-12-27 | 深圳朗田亩半导体科技有限公司 | 图像缩放方法和装置 |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH08274983A (ja) * | 1995-03-30 | 1996-10-18 | Sharp Corp | 画素補間方法 |
JP2005094120A (ja) * | 2003-09-12 | 2005-04-07 | Canon Inc | 画像処理装置 |
JP2005094119A (ja) * | 2003-09-12 | 2005-04-07 | Canon Inc | 画像処理装置 |
Family Cites Families (19)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH0750752A (ja) | 1993-08-06 | 1995-02-21 | Fuji Xerox Co Ltd | 画像密度変換方法及び装置 |
JPH07107273A (ja) | 1993-10-07 | 1995-04-21 | Canon Inc | 画像処理装置及び方法 |
US6882364B1 (en) | 1997-12-02 | 2005-04-19 | Fuji Photo Film Co., Ltd | Solid-state imaging apparatus and signal processing method for transforming image signals output from a honeycomb arrangement to high quality video signals |
JP4607265B2 (ja) | 1998-01-08 | 2011-01-05 | 富士フイルム株式会社 | 固体撮像装置および信号処理方法 |
JP2001008037A (ja) | 1999-06-23 | 2001-01-12 | Nec Corp | 画素補間方法および回路 |
JP2002024815A (ja) | 2000-06-13 | 2002-01-25 | Internatl Business Mach Corp <Ibm> | 拡大画像データに変換するための画像変換方法、画像処理装置、および画像表示装置 |
US6987893B2 (en) * | 2001-01-05 | 2006-01-17 | Lg Electronics Inc. | Image interpolation method and apparatus thereof |
US7286721B2 (en) * | 2003-09-11 | 2007-10-23 | Leadtek Research Inc. | Fast edge-oriented image interpolation algorithm |
US7477803B2 (en) | 2003-09-12 | 2009-01-13 | Canon Kabushiki Kaisha | Image processing apparatus |
TWI220363B (en) * | 2003-10-06 | 2004-08-11 | Sunplus Technology Co Ltd | Directional interpolation method and device for increasing resolution of an image |
US20050094899A1 (en) * | 2003-10-29 | 2005-05-05 | Changick Kim | Adaptive image upscaling method and apparatus |
JP4762115B2 (ja) | 2006-11-10 | 2011-08-31 | シャープ株式会社 | 映像処理装置 |
US8538203B2 (en) * | 2007-07-24 | 2013-09-17 | Sharp Laboratories Of America, Inc. | Image upscaling technique |
JP4129284B1 (ja) | 2007-08-30 | 2008-08-06 | 三菱電機株式会社 | 拡大画像生成装置 |
KR100952667B1 (ko) * | 2008-03-20 | 2010-04-13 | 중앙대학교 산학협력단 | 저역 통과 필터링을 기반으로 한 영상 보간 장치 및 방법 |
US8380011B2 (en) * | 2008-09-30 | 2013-02-19 | Microsoft Corporation | Fast directional image interpolator with difference projection |
JP2010140460A (ja) * | 2008-11-13 | 2010-06-24 | Sony Corp | 画像処理装置および方法、並びにプログラム |
US8417049B2 (en) * | 2009-10-14 | 2013-04-09 | Hewlett-Packard Development Company, L.P. | Fast image resolution enhancement with de-pixeling |
US8330827B2 (en) * | 2009-11-19 | 2012-12-11 | Eastman Kodak Company | Increasing image resolution using combined differential image |
-
2011
- 2011-10-07 WO PCT/JP2011/073222 patent/WO2012114574A1/ja active Application Filing
- 2011-10-07 JP JP2013500828A patent/JP5701374B2/ja active Active
- 2011-10-07 US US13/983,523 patent/US9020303B2/en active Active
- 2011-10-07 CN CN201180068016.0A patent/CN103380615B/zh not_active Expired - Fee Related
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH08274983A (ja) * | 1995-03-30 | 1996-10-18 | Sharp Corp | 画素補間方法 |
JP2005094120A (ja) * | 2003-09-12 | 2005-04-07 | Canon Inc | 画像処理装置 |
JP2005094119A (ja) * | 2003-09-12 | 2005-04-07 | Canon Inc | 画像処理装置 |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107256377A (zh) * | 2012-09-12 | 2017-10-17 | 威智伦富智堡公司 | 用于检测视频中的对象的方法、设备和系统 |
US10192286B2 (en) | 2014-09-11 | 2019-01-29 | Synaptics Japan Gk | Device and method for image enlargement and display panel driver using the same |
US10510138B2 (en) | 2014-09-11 | 2019-12-17 | Synaptics Japan Gk | Device and method for image enlargement and display panel driver using the same |
US9390471B1 (en) | 2015-02-05 | 2016-07-12 | Synaptics Display Devices Gk | Device and method for image scaling |
Also Published As
Publication number | Publication date |
---|---|
US9020303B2 (en) | 2015-04-28 |
US20130315506A1 (en) | 2013-11-28 |
CN103380615B (zh) | 2015-09-09 |
JP5701374B2 (ja) | 2015-04-15 |
CN103380615A (zh) | 2013-10-30 |
JPWO2012114574A1 (ja) | 2014-07-07 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
JP5701374B2 (ja) | 画像拡大装置及び方法 | |
JP5487106B2 (ja) | 画像処理装置および方法、ならびに、画像表示装置 | |
JP3829862B1 (ja) | 3次元モデル変形システム及びプログラム | |
JP5653184B2 (ja) | 画像処理装置及び方法 | |
CN104700360B (zh) | 基于边缘自适应的图像缩放方法及系统 | |
WO2014126093A1 (ja) | 画像処理装置 | |
WO2013114675A1 (ja) | 画像処理装置、歪補正マップ作成装置、および半導体計測装置 | |
Lavatelli et al. | A motion blur compensation algorithm for 2D DIC measurements of deformable bodies | |
WO2013031418A1 (ja) | 線分および円弧検出装置 | |
JP4635652B2 (ja) | 画像処理装置、画像処理方法、画像処理方法のプログラム及び画像処理方法のプログラムを記録した記録媒体 | |
JP5388779B2 (ja) | 画像処理装置、画像処理方法、および画像処理プログラム | |
KR101122279B1 (ko) | 베지어 곡선에 따른 오프셋 곡선을 생성하는 방법 | |
JP5388780B2 (ja) | 画像処理装置、画像処理方法、および画像処理プログラム | |
JP2013175143A (ja) | ニューラルネットワーク設計方法、プログラム及びデジタルアナログフィッティング方法 | |
JP3906201B2 (ja) | 補間画像生成装置および隠蔽領域推定方法 | |
JP5333791B2 (ja) | 画像処理方法、画像処理装置及び画像処理プログラム | |
JP6049000B2 (ja) | 線分および円弧検出装置 | |
JP5728164B2 (ja) | 高炉装入物の目標分布変換装置、この装置に用いられる高炉装入物の目標分布変換方法、及び高炉装入物の目標分布変換プログラム | |
JP2004282593A (ja) | 輪郭補正装置 | |
JP5464135B2 (ja) | ケーブル曲率分布生成システム及びそのコンピュータプログラム、並びにケーブルの曲率分布生成方法 | |
WO2009116586A1 (ja) | 画像処理方法、画像処理装置及び画像処理プログラム | |
Knas et al. | A Modification of the Parallel Spline Interpolation Algorithms | |
Schmid et al. | High-order curved mesh generation by using a fine linear target mesh | |
JP2009104228A (ja) | 画像位置合わせ方法及び画像位置合わせプログラム | |
US20150086136A1 (en) | Image scaling processor and image scaling processing method |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 11859289 Country of ref document: EP Kind code of ref document: A1 |
|
ENP | Entry into the national phase |
Ref document number: 2013500828 Country of ref document: JP Kind code of ref document: A |
|
WWE | Wipo information: entry into national phase |
Ref document number: 13983523 Country of ref document: US |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
122 | Ep: pct application non-entry in european phase |
Ref document number: 11859289 Country of ref document: EP Kind code of ref document: A1 |