KR20190080719A - Image magnifying apparatus - Google Patents

Abstract

According to one embodiment of the present invention, provided is an image magnifying apparatus, capable of selecting and applying an interpolation method according to localized features of an input image. According to one embodiment of the present invention, the image magnifying apparatus comprises: an input means configured to receive image data; a first interpolation means configured to generate an interpolation pixel by applying a first interpolation method which highlights a high-band spectrum between pixels; a second interpolation means configured to generate an interpolation pixel by applying a second interpolation method which does not highlight a high-band spectrum between pixels; a selection means configured to identify a pattern of pixels by extracting peripheral pixels of an interpolation object position in the image data that the input means has received and select whether to apply the first interpolation method or the second interpolation method to the interpolation object position; and an output means configured to output one of the interpolation pixel generated by the first interpolation means and the interpolation pixel generated by the second interpolation means, as an output interpolation pixel, based on the selection of the selection means.

Description

[0001] IMAGE MAGNIFYING APPARATUS [

The present invention relates to an image magnifying apparatus.

There has been a demand for a technique of expanding a low-resolution image to a high-resolution image in accordance with the spread of a high-resolution imaging device. In order to enlarge a low-resolution image to a high-resolution image, an image enlarging device for interpolating and outputting pixels of an image constituting the image is provided in the image device. The image enlarging apparatus enlarges and outputs an input image by applying a process of interpolating pixels.

When a natural image is input as an image to an image enlarging device, an interpolation method suitable for a natural image is applied. An interpolation method suitable for a natural image is an interpolation method for emphasizing a change of a waveform of an input signal. As a result, the image can be sharpened and the bouquet phenomenon that may occur at the time of image enlargement can be reduced. On the other hand, a graphic image having a sharp edge may be input to the image enlarging device. In this case, if an interpolation method for emphasizing the change of the waveform of the input signal is applied, ringing occurs at the edge portion, and the output image becomes unnatural. The graphic image having a sharp edge may include, for example, an image including a caption such as a character, an image of a PC, and the like. For such a graphic image, it is preferable to apply an interpolation method which does not emphasize the change of the waveform of the input signal.

The image enlarging apparatus can change the interpolation method based on the histogram of the input image. Thus, deterioration of image quality due to interpolation can be suppressed in the graphic image while suppressing the bokeh phenomenon that may occur in enlargement in the natural image. However, in this case, if a natural image and a graphic image are mixed in one image, it may be difficult to distinguish them sufficiently. Therefore, an optimal interpolation method may not be applied to the area of each image. In addition, the boundary between the natural image and the graphic image can be unnatural.

Also, when an image includes a certain pattern such as a constant houndstooth lattice, a character, and an isolated point, if the interpolation method suitable for the natural image is applied to the area, the pattern may not be properly enlarged. Therefore, an image enlarging device capable of selecting an interpolation method suitable for an image needs to be proposed.

An object of the present invention is to provide an image enlarging apparatus capable of selecting and applying an interpolation method according to local characteristics of an input image.

An image enlarging apparatus according to an embodiment of the present invention is an image enlarging apparatus that receives image data including a plurality of pixels arranged in a lattice form and applies an interpolation method to the image data to output enlarged image data Input means for receiving the image data; First interpolation means for generating an interpolation pixel to which a first interpolation method for emphasizing high band is applied between pixels of the image data inputted by the input means; Second interpolation means for generating an interpolation pixel to which a second interpolation method which does not emphasize high band is applied between the pixels of the image data inputted by said input means; Wherein the first interpolation method is applied to the interpolation object position and the second interpolation method is applied to the interpolation object position by extracting peripheral pixels of the interpolation object position from the image data input by the input means, Selecting means for selecting whether to apply the two interpolation method; And output means for outputting, as an output interpolation pixel, either the interpolation pixel inserted by the first interpolation method or the interpolation pixel inserted by the second interpolation method on the basis of the selection of the selection means; .

According to the embodiment of the present invention, it is possible to provide an image enlarging device capable of outputting an enlarged image of high image quality while suppressing unnaturalness by selecting an appropriate interpolation method according to the local characteristic of the input image.

The various and advantageous advantages and effects of the present invention are not limited to the above description, and can be more easily understood in the course of describing a specific embodiment of the present invention.

1 is a block diagram showing a configuration of an image enlarging apparatus according to a first embodiment of the present invention.
2 is a diagram showing the arrangement relationship of the pixels included in the image data and the interpolation object positions.
3 is a flowchart for explaining a method of selecting an interpolation method.
4 is a diagram showing a range of a reference area in a case where the interpolation object position is adjacent to the pixel in the lateral direction.
5 is a diagram showing the range of the reference area when the interpolation object position is adjacent to the pixel in the longitudinal direction.
Fig. 6 is a diagram showing the range of the reference area when the interpolation object position is adjacent to the pixel in the oblique direction. Fig.
7 is a block diagram showing a configuration of an image enlarging apparatus according to a second embodiment of the present invention.
FIG. 8 is a view showing a normalized direction as a direction with respect to a specific pixel.
9 is an explanatory diagram showing a first interpolation method according to the second embodiment of the present invention.
10 is an explanatory diagram showing a second interpolation method according to the second embodiment of the present invention.
11 is a flowchart for explaining interpolation method selection in the second embodiment of the present invention.
12 is a flowchart for explaining the interpolation method selection in the third embodiment of the present invention.
13 is a diagram showing a range of a reference area in a case where the interpolation object position is adjacent to the pixel in the lateral direction in the third embodiment of the present invention.
14 is a diagram showing a range of a reference area in a case where the interpolation object position is adjacent to the pixel in the longitudinal direction, according to the third embodiment of the present invention.
Fig. 15 is an explanatory view showing contents of an operation performed on a reference area in the third embodiment of the present invention. Fig.
Fig. 16 is a flowchart for explaining interpolation method selection in the fourth embodiment of the present invention. Fig.
FIG. 17 is an explanatory view showing contents of an operation performed on a reference area in the fourth embodiment of the present invention. FIG.
18 is an explanatory view showing the contents of an operation performed on a reference area in the case where one unit constituting the hilltop pattern is two neighboring pixels.
Fig. 19 is a flowchart for explaining interpolation method selection in the fifth embodiment of the present invention. Fig.
20 is a diagram showing a range of a reference area in the case where the interpolation object position is adjacent to the pixel in the lateral direction in the fifth embodiment of the present invention.
21 is a diagram showing a range of a reference area in the case where the interpolation object position is adjacent to the pixel in the longitudinal direction, according to the fifth embodiment of the present invention.
22 is a diagram showing a range of a reference area when the interpolation object position is adjacent to the pixel in the oblique direction, according to the fifth embodiment of the present invention.
23 is a flowchart for explaining interpolation method selection in the sixth embodiment of the present invention.
Fig. 24 is an explanatory view showing contents of an operation performed on a reference area in the sixth embodiment of the present invention. Fig.

Hereinafter, preferred embodiments of the present invention will be described with reference to the accompanying drawings.

1 is a block diagram showing a configuration of an image enlarging apparatus according to a first embodiment of the present invention. 2 is a diagram showing the arrangement relationship of the pixels included in the image data and the interpolation object positions.

First, the configuration of the image enlarging apparatus 1 according to the first embodiment of the present invention will be described with reference to Fig. In the image enlarging apparatus 1, image data including a plurality of pixels arranged in a lattice form can be input. The image enlarging apparatus 1 can insert interpolation pixels between respective pixels of input image data and output enlarged image data.

The image enlarging apparatus 1 may include input means 10 for receiving an input signal including image data. Further, the image enlarging apparatus 1 may include a selecting means 11 for selecting an interpolation method based on the input image data. The image enlarging apparatus 1 further includes a first interpolating means 12 for generating interpolation pixels by a first interpolation method with respect to the pixels of the image data and a second interpolating means 12 for generating interpolation pixels by a second interpolating method, (13). The first interpolation method may be an interpolation method for emphasizing a high band of an image and the second interpolation method may be an interpolation method which does not emphasize a high band of an image. The output means 14 can output the pixel generated by the interpolation method selected by the selection means 13 as an interpolation pixel.

The image enlarging apparatus 1 can insert interpolation pixels for all the pixels included in the image data and generate enlarged image data. The arrangement relationship of the interpolation pixels will be described with reference to Fig. In the present embodiment, it is assumed that input image data is enlarged twice in both the horizontal direction and the vertical direction.

The interpolation pixel for the pixel A can be generated at the interpolation object position a1 between the pixel A and the pixel B and at the interpolation object position a2 between the pixel A and the pixel C. [ In addition, the interpolation pixel for the pixel A can also be generated at the interpolation target position a3 between the pixel A and the pixel D. The interpolation object position a1 may be adjacent to the left and right arranged pixels A and B in the lateral direction. The interpolation object position a2 may be vertically adjacent to the pixel A and the pixel C arranged vertically. The interpolation object position a3 may be adjacent to the surrounding pixels A, B, C, and D in the oblique direction.

The first interpolation method may be an interpolation method for emphasizing the high band of the image as described above. This interpolation method adds overshoot or undershoot by emphasizing the change of the waveform in the image data. As a result, the image becomes clear and the bouquet phenomenon that may occur upon enlargement can be reduced. On the other hand, the first interpolation method can cause ringing at an edge portion in an image. The first interpolation method is suitable for a natural image. Specifically, a Bi-Cubic method or a Lanczos method can be used. However, the first interpolation method is not limited to these, and may be other interpolation methods.

The second interpolation method may be an interpolation method that does not emphasize the high band of the image as described above. In this interpolation method, the effect of reducing the bokeh in a natural image may not be obtained. On the other hand, the second interpolation method can be suitable for a graphic image because it does not cause ringing at a sharp edge portion. As the second interpolation method, a nearest neighbor method can be employed. However, the second interpolation method is not limited to this, and may be another interpolation method.

3 is a flowchart for explaining a method of selecting an interpolation method. Fig. 4 is a view showing a range of a reference area when the interpolation object position is adjacent to the pixel in the lateral direction, Fig. 5 is a view showing a range of the reference area when the interpolation object position is adjacent to the pixel in the longitudinal direction to be. On the other hand, Fig. 6 is a diagram showing the range of the reference area when the interpolation object position is adjacent to the pixel in the oblique direction.

The selection of the interpolation method in the present embodiment will be described with reference to FIG. The selection means 13 can first extract the reference region from the periphery of the interpolation object position (S1-1). The reference area will be described with reference to Figs. 4 to 6 together.

In the case of generating the interpolation pixel with respect to the interpolation object position a1, the reference area may be the four areas shown in Fig. The reference area R1-1 in Fig. 4 (a) is a region including the left pixel A of the interpolation subject position a1 and the lower pixel of the pixel A, and two pixels in the horizontal direction and two pixels in the vertical direction, . The reference area R1-2 in Fig. 4 (b) may be an area in which the reference area R1-1 is shifted upward by one pixel. The reference area R1-3 in Fig. 4 (c) includes two pixels in the horizontal direction, two pixels in the vertical direction, and four pixels in total, including the pixel B on the right side of the interpolation object position a1 and the pixel on the lower side of the pixel B . ≪ / RTI > The reference area R1-4 in Fig. 4 (d) may be an area in which the reference area R1-3 is shifted upward by one pixel.

When an interpolation pixel is generated for the interpolation object position a2, the reference area may be the four areas shown in Fig. The reference area R2-1 in Fig. 5 (a) is a region including the lower pixel C of the interpolation object position a2 and the left pixel of the pixel C, and is composed of a total of four pixels, i.e., two pixels in the horizontal direction and two pixels in the vertical direction Lt; / RTI > The reference area R2-2 in Fig. 5 (b) may be an area in which the reference area R2-1 is shifted to the right by one pixel. The reference area R2-3 in Fig. 5 (c) includes an upper pixel A of the interpolation object position a2 and a left pixel of the pixel A, and includes two pixels in the horizontal direction, two pixels in the vertical direction, Lt; / RTI > The reference area R2-4 in Fig. 5 (d) may be an area in which the reference area R2-3 is shifted to the right by one pixel.

When an interpolation pixel is generated for the interpolation object position a3, the reference area may include eight areas shown in Fig. The reference area R3-1 shown in Fig. 6A includes a left upper pixel A and a lower left pixel C of the interpolation target position a3, and includes two pixels in the horizontal direction and two pixels in the vertical direction, that is, Lt; / RTI > The reference area R3-2 in Fig. 6 (b) may be an area in which the reference area R3-1 is shifted upward by one pixel. The reference area R3-3 in Fig. 6 (c) may be an area in which the reference area R3-1 is moved downward by one pixel. The reference area R3-4 in Fig. 6 (d) is a region including the upper right pixel B and the lower right pixel D in the interpolation object position a3, and is a region including two pixels in the horizontal direction and two pixels in the vertical direction. . The reference area R3-5 in Fig. 6 (e) may be an area in which the reference area R3-4 is shifted upward by one pixel. The reference area R3-6 in Fig. 6 (f) may be an area in which the reference area R3-4 is moved downward by one pixel. The reference area R3-7 in Fig. 6 (g) may be an area including the upper left pixel A and the upper right pixel B of the interpolation object position a3 and a total of four pixels in the width of 2 pixels in the width and 2 pixels in the height of 2 pixels. The reference area R3-8 shown in FIG. 6 (h) may be a region consisting of four pixels in total, two pixels in the horizontal direction and two pixels in the vertical direction, including the lower left pixel C and the lower right pixel D of the interpolation object position a3.

When the reference region is extracted, the selection means 13 can calculate the difference value between adjacent pixels in each reference region (S1-2). As shown in Fig. 4 (a), there may be four pairs of pixels adjacent to each other in the horizontal or vertical direction in each reference area. The absolute value of the difference is calculated for each of them.

Next, it is possible to judge whether or not all of the values calculated in step S1-2 are equal to or less than a predetermined set value (S1-3). When the absolute values of the differences are all equal to or less than the set value, the pixels constituting the reference area are almost the same and are judged to be a flat area. The flat area is estimated as the area of the graphic image. If at least one of the absolute values of the differences is larger than the set value, the reference area may be determined as a non-flat area. The non-planar region is estimated as a region of the natural image.

The determination in step S1-3 may be made as long as it is possible to detect a small difference among the four pixel values included in the reference area. Therefore, the method is not limited to the above-described method, and other methods may be used, such as determining whether the sum of the absolute values of the differences calculated by the four pixel pairs is equal to or less than a predetermined set value.

In the step S1-3, the selecting means 13 can judge whether or not the absolute values of the differences are all equal to or lower than the set values for all the reference regions extracted in the step S1-1. As a result, when it is determined that any one reference area is a flat area, the selecting means 13 can select the second interpolation method (S1-4). If it is determined that all the reference areas are not flat, the selecting means 13 can select the first interpolation method (S1-5).

As described above, in the present embodiment, the interpolation method can be selected whether or not the reference region around the interpolation object position is judged to be a flat region. Accordingly, a first interpolation method for emphasizing the high band is applied to the natural image area, and a second interpolation method which does not emphasize the high band is applied to the graphic image area. According to the selection of the interpolation method as described above, it is possible to reduce the bokeh phenomenon that may appear in the enlargement of the natural image area, and to prevent the edge part from becoming unnatural in the area of the graphic image.

7 is a block diagram showing a configuration of an image enlarging apparatus according to a second embodiment of the present invention. FIG. 8 is a view showing a normalized direction as a direction with respect to a specific pixel.

A second embodiment of the present invention will be described with reference to Figs. 7 and 8. Fig. As shown in Fig. 7, the image enlarging apparatus 20 according to the second embodiment has input means 21 for receiving an input signal. The image data included in the input signal received by the input means 21 can be transmitted to the direction discriminating means 22. [ The direction determination means 22 can determine the directionality and direction of the interpolation object position. The image enlarging apparatus 20 may include a first interpolating means 24 and a second interpolating means 25 and a selecting means 23 for selecting any one of them. Furthermore, the image enlarging apparatus 20 may include an arithmetic interpolation means 26 employed when the directional interpolation is not performed.

The directional interpolation may be a method of detecting a correlation direction from image data and selecting a pixel for interpolation from a direction parallel to the correlation direction. As a result, it is possible to suppress the refinement of the edge of the image that follows the enlargement of the image. As a method of detecting the directionality at the interpolation object position, a Sobel filter can be used. However, the directionality detection method is not limited to this, and other methods such as a prewitt filter may be used.

The first interpolation method applied by the first interpolation means 24 and the second interpolation method applied by the second interpolation means 25 may all be methods of directional interpolation. When directional interpolation is performed, information of a correlation direction may be required to select a pixel to be used for interpolation. Since the pixels of the image data are arranged at regular intervals in the vertical and horizontal directions, the angle information necessary for the selection of the pixels can be standardized on the basis of the arrangement of the pixels. As shown in Fig. 8, the direction toward one pixel in the horizontal direction and the pixel P1 separated by two pixels in the vertical direction with reference to the pixel A can be defined as " 1 ". The direction toward the pixel P2 separated by two pixels in the lateral direction and three pixels in the longitudinal direction based on the pixel A may be " 2/3 ". Similarly, the direction from the pixel A to the pixel P3 may be defined as "2", and the direction from the pixel A to the pixel P4 may be defined as "5/2". Further, the direction from the pixel A to the pixel P5 may be defined as "3", the direction from the pixel A to the pixel P6 as "4", and the direction from the pixel A to the pixel P7 as "5". The angular range from the horizontal direction with respect to the pixel A to the direction " 1 " can be expressed by the reciprocal of these numerical values. In addition, the direction described so far is the direction toward the first upper limit (quadrant) and the third upper limit. On the other hand, the directions toward the second upper limit and the fourth upper limit can be indicated as negative directions.

Fig. 9 is an explanatory diagram showing a first interpolation method in the second embodiment of the present invention, and Fig. 10 is an explanatory diagram showing a second interpolation method in the second embodiment of the present invention.

First, a first interpolation method will be described with reference to FIG. The first interpolation method is an interpolation method for emphasizing high-band as in the first embodiment. In the present embodiment, it is assumed that the direction of the interpolation target position a1 is the direction " 3 ". In order to generate the interpolation pixel at the interpolation object position a1 by the first interpolation method, four pixels may be required.

First, a line L1 in the direction "3" passing through the interpolation object position a1 and a line L2 which is a normal line of the line L1 can be generated. Next, as a line parallel to the line L1, a line passing through the center position of another pixel can be generated, and a plurality of such lines can exist. A line L3 passing through the centers of the pixels E and E 'and a line L4 passing through the centers of the pixels C and C' can be generated. The intersection of line L3 and line L2 can be defined as interpolation point q1. The pixel value of the interpolation point q1 can be calculated from the pixel values of the pixel E and the pixel E '. Further, the intersection point of the line L4 and the line L2 is defined as the interpolation point q2, and the pixel value of the interpolation point q2 can be calculated from the pixel values of the pixel C and the pixel C '. It is possible to generate a line L5 passing through the center of the pixel F and the pixel F 'and a line L6 passing through the center of the pixel G and the pixel G', which is next to the line L1 after the line L3 or the line L4. The intersection point of the line L5 and the line L2 is defined as the interpolation point q3 and the pixel value of the interpolation point q3 can be calculated from the pixel values of the pixel F and the pixel F '. The intersection of the line L6 and the line L2 is defined as an interpolation point q4. The pixel value of the interpolation point q4 can be calculated from the pixel values of the pixel G and the pixel G '. For the calculation of each interpolation point, a Bi-Linear method can be used.

Then, from the four points of the interpolation points q1, q2, q3, and q4, the pixel of the interpolation object position a1 can be generated by the bicubic method. As described above, the pixel values of the four interpolation points can be calculated using the actually existing pixels in the correlation direction detected at the interpolation object position. This makes it possible to generate the interpolation pixel using the bicubic method. By using the bicubic method for the generation of the interpolation pixel, it is possible to generate a slant having a high sharpness. On the other hand, in the case of other correlation directions and interpolation target positions, similarly, four interpolation points are calculated, and an interpolation pixel can be generated by the bicubic method.

Next, the second interpolation method will be described with reference to FIG. The second interpolation method may be an interpolation method that does not emphasize high-band as in the first embodiment. In this embodiment also, it is assumed that the direction of correlation of the interpolation target position a1 is the direction " 3 ". In order to generate the interpolation pixel at the interpolation object position a1 by the second interpolation method, two pixels may be required.

First, a line L1 in the direction "3" passing through the interpolation object position a1 and a line L2 which is a normal line of the line L1 can be generated. Next, among the lines passing through the center position of the other pixel, a line closest to the line L1 can be generated as a line parallel to the line L1. 8, it may be a line L3 passing through the centers of the pixels E and E 'and a line L4 passing through the centers of the pixels C and C'. When the intersection of the line L3 and the line L2 is defined as the interpolation point q1, the pixel value of the interpolation point q1 can be calculated from the pixel values of the pixel E and the pixel E '. When the intersection of the line L4 and the line L2 is defined as the interpolation point q2, the pixel value of the interpolation point q2 can be calculated from the pixel values of the pixel C and the pixel C '. For calculation of each interpolation point, a bi-linear method can be used.

Next, from the two points of interpolation points q1 and q2, the pixel of the interpolation object position a1 can be generated by the bilinear method. As described above, the pixel values of two interpolation points can be calculated using the actually existing pixels in the correlation direction detected at the interpolation object position. This makes it possible to generate the interpolation pixel using the bilinear method. By using the bi-linear method to generate the interpolation pixel, smooth slant lines can be generated.

The interpolation method used in the arithmetic interpolation means 26 may be a normal interpolation method other than the directional interpolation. The arithmetic interpolation means 26 may adopt any interpolation method such as a bilinear method or a bicubic method. In addition, the arithmetic interpolation means 26 may be provided with two interpolation means and selection means, and the interpolation method may be selected depending on whether the reference region is a flat region or not, as in the first embodiment.

11 is a flowchart for explaining interpolation method selection in the second embodiment of the present invention.

The selection of the interpolation method in the present embodiment will be described with reference to Fig. First, the direction determination means 22 can detect the directionality and direction of the interpolation object position (S2-1). When the directionality is detected in step S2-1, the selecting means 23 can extract the reference area from the periphery of the interpolation object position (S2-2). The reference area extracted in step S2-2 may be the same as that in the first embodiment.

When the reference area is extracted, the selection means 23 can calculate the difference value between adjacent pixels of each reference area and determine whether the reference area is a flat area (S2-3). Whether or not the flat region is discriminated may be the same as in the case of the first embodiment. Further, the judgment in step S2-3 can be executed for all extracted reference areas.

If it is determined in step S2-3 that any one reference area is a flat area, the selecting means 23 can select the second interpolation method (S2-4). If it is determined that all the reference areas are not flat, the selecting means 23 can select the first interpolation method (S2-5).

If directionality is not detected in step S2-1, the arithmetic interpolation means 26 can be selected (S2-6).

As described above, in the present embodiment, the directional interpolation method can be selected whether or not the reference region around the interpolation object position is judged to be a flat region. Thus, in the region of the natural image, a diagonal line which suppresses the ringing can be generated in the area of the graphic image while generating the sharp diagonal line.

12 is a flowchart for explaining the interpolation method selection in the third embodiment of the present invention. 13 is a diagram showing a range of a reference area in a case where the interpolation object position is adjacent to the pixel in the lateral direction in the third embodiment of the present invention. Fig. 8 is a diagram showing a range of a reference area when the object position is adjacent to the pixel in the longitudinal direction. Fig. Fig. 15 is an explanatory view showing contents of an operation performed on a reference area in the third embodiment of the present invention. Fig.

A third embodiment of the present invention will be described with reference to Figs. 12 to 15. Fig. The configuration of the image enlarging apparatus 1 of the present embodiment may be the same as that of the first embodiment. In the present embodiment, the selection means 13 can determine whether there is a gradient pattern around the interpolation target position.

In the present embodiment, as the first interpolation method, a bicubic method or a Rancho method can be used. However, the first interpolation method is not limited to these, and other interpolation methods for emphasizing the high band may be used.

In the present embodiment, the second interpolation method is a near-list neighbor method or a bilinear method. However, the second interpolation method is not limited to these, and other interpolation methods that do not emphasize the high band may be used.

First, the selection of the interpolation method in the present embodiment will be described with reference to FIG. The selection means 13 can first extract the reference region from the periphery of the interpolation object position (S3-1). The reference area will be described with reference to Figs. 13 to 14. Fig.

First, an interpolation object position a1 adjacent to both pixels in the lateral direction will be described. In the case of generating the interpolation pixel at the interpolation object position a1, as shown in Fig. 13, a transverse field reference area including six pixels in the lateral direction and two pixels in the longitudinal direction can be extracted. The reference area of Fig. 13A is defined as the position of the left upper end of the reference area at the left pixel of the correction target position a1. As shown in Figs. 13 (b) to 13 (f), a reference area in which the reference area of Fig. 13 (a) is shifted by one pixel in the left direction can also be extracted. In addition, a reference area in which each of the six reference areas shown in Fig. 13 is shifted upward by one pixel can also be extracted.

Next, a description will be given of the interpolation object position a2 adjacent to pixels on both the longitudinal sides. In the case of generating the interpolation pixel at the interpolation object position a2, as shown in Fig. 14, a reference area having two pixels in the lateral direction and six pixels in the longitudinal direction can be extracted. In the reference area of Fig. 14A, the upper pixel of the correction target position a2 is located at the upper left of the reference area. On the other hand, as shown in Figs. 14 (b) to 14 (f), a reference area in which the reference area of Fig. 14 (a) is shifted by one pixel in the upward direction is also extracted. In addition, a reference area in which the six reference areas shown in Fig. 14 are shifted toward the left side by one pixel is also extracted.

Next, the interpolation object position a3 adjacent to the four surrounding pixels in the oblique direction will be described. In the case of generating the interpolation pixel at the interpolation object position a3, the transverse field reference area in Fig. 13 and the reference area in the longitudinal direction in Fig. 14 can be extracted in both directions. As in the case of the interpolation object position a1, twelve reference areas of the transverse field are extracted. Also, as in the case of the interpolation object position a2, twelve reference areas are extracted.

Next, it is determined whether or not the extracted reference area conforms to the three conditions. The first condition is whether the pixel values of two columns along the long side direction of the reference area all monotonously increase or monotonously decrease toward one direction as shown in Fig. 15 (a). Judgment of the first condition may be executed in step S3-2. The second condition is whether the six pixel values along the short side direction of the reference area increase or decrease toward one direction as shown in Fig. 15 (b). The determination of the second condition may be executed in step S3-3. As shown in Fig. 15 (c), the third condition is to calculate difference values between adjacent pixels in the reference area, and judge whether or not the sum is equal to or less than a predetermined set value. The judgment on the third condition is executed in step S3-4. In steps S3-2, S3-3, and S3-4, it is possible to determine whether the first condition, the second condition, and the third condition are satisfied for all the extracted reference areas.

When at least one of the first to third conditions is satisfied with respect to any one reference area, the periphery of the interpolation target position is determined to be a pattern of gradation. In this case, the second interpolation method is selected (S3-5). On the other hand, when the first to third conditions are not satisfied with respect to all the reference areas, it is determined that the periphery of the interpolation object position is not a pattern of gradation. In this case, the first interpolation method is selected (S3-6).

Applying the first interpolation method emphasizing the high band to the pattern of gradation can emphasize the minute luminance difference existing in the gradation. As a result, a stripe called banding occurs. In the present embodiment, it is determined whether or not there is a gradient pattern around the interpolation target position, and in the case of gradation, the second interpolation method which does not emphasize the high band is selected. Thus, occurrence of banding can be suppressed.

Further, by extracting a plurality of reference areas around the interpolation object position, it is possible to reliably discriminate the end area of the gradation.

Fig. 16 is a flowchart for explaining interpolation method selection in the fourth embodiment of the present invention. Fig. On the other hand, FIG. 17 is an explanatory view showing contents of an operation performed on a reference area in the fourth embodiment of the present invention, and FIG. 18 shows a case where one unit constituting the shorebird pattern is two neighboring pixels , And the contents of the operation performed on the reference area.

A fourth embodiment of the present invention will now be described. The configuration of the image enlarging device 1 of this embodiment is the same as that of the first embodiment. In the present embodiment, the selection means 13 determines whether or not there is a pattern of a float pattern around the interpolation object position. In the present embodiment, as the first interpolation method, a bicubic method or a Rancho method can be used. However, the first interpolation method is not limited to these, and other interpolation methods for emphasizing the high band may be used. In the present embodiment, the second interpolation method is a near list neighbor method. However, the second interpolation method is not limited to the above method, and other interpolation methods not emphasizing the high band may be used.

The selection of the interpolation method in this embodiment will be described with reference to Fig. The selection means 13 first extracts the reference region from the periphery of the interpolation object position (S4-1). The reference area to be extracted is the same as that of the third embodiment, and a description thereof will be omitted.

Next, it is determined whether or not the extracted reference area satisfies two conditions. The first condition is whether or not the absolute values of the differences between the pixels adjacent in the oblique direction within the reference area all become equal to or less than a predetermined set value. As shown in Fig. 17 (a), the pairs of pixels adjacent in the oblique direction in the reference area are 10 pairs in total. When the reference area is a hillock shape, the difference between adjacent pixels in the oblique direction can be small. It can be detected as the first condition.

The second condition relates to whether or not the differential values of the two sets of pixels located at the center in the long side direction of the reference area and along the short side direction are respectively greater than or equal to a predetermined set value. As shown in Fig. 17 (b), when the reference area is a hillock shape, the difference between adjacent pixels in the short side direction can be large. It can be detected as the second condition.

Fig. 17 shows the reference region of the transverse field. The first condition and the second condition are also discriminated in the reference region of the longitudinal direction.

As a result of judging the first condition at step S4-2, if the reference area satisfies the first condition, the second condition can be judged at S4-3. If it is determined in step S4-3 that the second condition is satisfied, the second interpolation method can be selected. If the reference area does not satisfy the first condition or the second condition in step S4-2 or step S4-3, the first interpolation method may be selected. Since a plurality of reference regions are extracted, a determination is made for each reference region. The selection means 13 can select the second interpolation method when any one reference region satisfies the first condition and the second condition.

Fig. 17 shows a case in which one unit constituting the hilltop shape is one pixel. Alternatively, as shown in Fig. 18, one unit constituting the hillock shape may be two neighboring pixels. In the present embodiment, it is preferable to additionally add a condition assuming this case. One unit constituting the hillock shape in the reference area may be two pixels neighboring in the long-side direction like the pixel R and the pixel S in Fig. 18 (a). In this case, the first condition is that the absolute value of the difference between the pixels in one unit is equal to or less than the set value, and the absolute value of the difference between the pixels in the unit in the oblique direction is equal to or less than the set value. For example, in Fig. 18 (a), the pixel S and the pixel T are adjacent to each other in the oblique direction. In the reference area, there are six pairs of adjacent units in the oblique direction. As in the case of Fig. 17, the second condition is that the absolute values of the differences between two sets of pixels located at the center in the long side direction of the reference area and along the short side direction are all equal to or more than the set value. If all of these conditions are satisfied, it is judged that there is a hillock pattern.

On the other hand, in the case where one unit constituting the hilltop shape has two pixels in the horizontal direction and two or more pixels in the vertical direction, it can be discriminated as the flat area described in the first embodiment.

If the first interpolation method for mixing neighboring pixels of the interpolation object position with respect to the regularly arranged hillock pattern is applied, the pattern of the hillock pattern can be collapsed. In this embodiment, it is judged whether or not the interpolation object position is included in the shorebird shape, and in the case of the shorebird shape, the second interpolation method which does not mix the pixels can be selected. Thus, the image can be enlarged while maintaining the hillock shape.

In addition, by extracting a plurality of reference areas around the interpolation object position, a reliable judgment can be made also for the end area in the trough shape.

Fig. 19 is a flowchart for explaining interpolation method selection in the fifth embodiment of the present invention. Fig. FIG. 20 is a diagram showing a range of a reference area in the case where the interpolation object position is adjacent to the pixel in the lateral direction in the fifth embodiment of the present invention. FIG. Fig. 8 is a diagram showing a range of a reference area when the object position is adjacent to the pixel in the longitudinal direction. Fig. On the other hand, Fig. 22 is a diagram showing the range of a reference area in the case where the interpolation object position is adjacent to the pixel in the oblique direction, according to the fifth embodiment of the present invention.

A fifth embodiment of the present invention will be described. The configuration of the image enlarging device 1 of this embodiment is the same as that of the first embodiment. In the present embodiment, the selection means 13 can determine whether the pixel adjacent to the interpolation object position is an isolated point or not.

In the present embodiment, as the first interpolation method, a bicubic method or a Rancho method can be used. However, the first interpolation method is not limited to these, and may be another interpolation method emphasizing the high band.

In the present embodiment, the second interpolation method may be a near list neighbor method. However, the second interpolation method is not limited to the above method, and may be another interpolation method that does not emphasize the high band.

The selection of the interpolation method in this embodiment will be described below with reference to Fig. The selecting means 13 can first extract the reference area from the periphery of the interpolation object position (S5-1). The reference area will be described with reference to FIGS. 20 to 22 together.

First, an interpolation object position a1 adjacent to pixels on both lateral sides will be described. In this case, an area including three pixels in the horizontal direction and three pixels in the vertical direction is defined as the reference area, with the pixels adjacent in the transverse direction to the interpolation subject position a1 as the center pixel. Since there are two pixels which are adjacent to each other in the transverse direction with respect to the interpolation target position a1, two reference areas are extracted as shown in Fig.

Next, a description will be given of the interpolation object position a2 adjacent to pixels on both the longitudinal sides. In this case, an area including three pixels in the horizontal direction and three pixels in the vertical direction is defined as the reference area, with the pixels adjacent to the interpolation object position a2 in the longitudinal direction as the center pixel. Since there are two pixels adjacent in the longitudinal direction to the interpolation object position a2, two reference areas are extracted as shown in Fig.

Next, the interpolation object position a3 adjacent to the surrounding four pixels in the oblique direction will be described. In this case, an area including three pixels in the horizontal direction and three pixels in the vertical direction is set as the reference area with the pixel adjacent to the interpolation object position a3 in the oblique direction as the center pixel. As shown in FIG. 22, four reference areas are extracted because there are four pixels adjacent in the oblique direction to the interpolation target position a3.

When the reference area is extracted, the selection unit 13 calculates the maximum value and the minimum value of the reference area and determines whether the center pixel value of the reference area is the maximum value or the minimum value (S5-2). If the center pixel value of the reference area is the maximum value or the minimum value and the difference from the surrounding pixel value is sufficiently large, it can be determined that the center pixel is an isolated point. Accordingly, when the reference area conforms to the condition of S5-2, it can be judged whether or not the following conditions are met (S5-3). α is a constant set value.

If the center pixel value of the reference area is the maximum value:

[Center pixel value]> (([maximum value] - [minimum value]) / 2+ [minimum value]) [

If the center pixel value of the reference area is the minimum value:

[Center pixel value] < (([maximum value] - [minimum value]) / 2+ [minimum value]) /

If the reference area satisfies both the condition of step S5-2 and the condition of step S5-3, the selecting means 13 can select the second interpolation method (S5-4). If the reference area does not satisfy either the condition of step S5-2 or the condition of step S5-3, the selecting means can select the first interpolation method (S5-5). The selection means (13) can select the second interpolation method when the second interpolation method is selected in at least one of the plurality of extracted reference areas.

When the pixel to be interpolated is an isolated point, it is necessary to be an isolated point also in the enlarged image. However, since the first interpolation method for emphasizing the high band mixes the isolated point and the peripheral pixels, a problem may arise. In the present embodiment, it is determined whether or not the interpolation object position is adjacent to the isolated point, and in the case of the isolated point, the second interpolation method is selected, in which pixels are not mixed. Thus, the isolated point can be enlarged as it is.

23 is a flowchart for explaining interpolation method selection in the sixth embodiment of the present invention. Fig. 24 is an explanatory view showing contents of an operation performed on a reference area in the sixth embodiment of the present invention. Fig.

Next, the sixth embodiment will be described. The configuration of the image enlarging device 1 of this embodiment is the same as that of the first embodiment. In the present embodiment, the selection means 13 determines whether or not there is a line of characters around the interpolation object position.

In the present embodiment, as the first interpolation method, a bicubic method or a Rancho method can be used. However, the first interpolation method is not limited to these, and may be another interpolation method emphasizing the high band. On the other hand, in the present embodiment, the second interpolation method is the near list neighbor method. However, the second interpolation method is not limited to the above method, and may be another interpolation method not emphasizing the high frequency band.

The selection of the interpolation method in the present embodiment will be described with reference to Fig. The selection means 13 first extracts the reference region from the periphery of the interpolation object position (S6-1). As shown in Fig. 24A, the reference area is an area including 11 pixels in the horizontal direction and 5 pixels in the vertical direction, with the pixel A to be interpolated as the center pixel. In addition, eight identical reference areas can be extracted with eight pixels surrounding each other in the lateral, longitudinal, and oblique directions as the center pixel.

Next, the selecting means 13 can calculate the difference value DR between the maximum value and the minimum value of the reference area (S6-2). Further, the selecting means 13 can calculate the histogram of the reference area (S6-3). Again, the selecting means 13 can select a central region Q having three pixels in the horizontal direction and three pixels in the vertical direction, with the center pixel as the center, from the reference region. Then, as shown in Fig. 24 (b), the selection means 13 can calculate the difference value between the pixel value of the central pixel and its surrounding pixels in the central region Q, respectively. Here, the selection means 13 calculates the sum H1 of the absolute values of the differences of the pixels in the lateral direction including the central pixel of the central region Q. Further, the selecting means 13 calculates the sum H2 of the absolute values of the differences of the pixels in the vertical direction including the central pixel of the central region Q. Again, the selecting means 13 calculates the sum H3 of absolute differences between the center pixel of the center region Q and the pixel in the oblique direction including the upper left pixel of the center pixel. The selecting means 13 calculates the sum H4 of the absolute differences between the center pixel of the central region Q and the pixel in the oblique direction including the right lower pixel of the center pixel (S6-4).

From the histogram of the reference region calculated in S6-3, the selecting means 13 can judge whether the reference region is two or not (S6-5). The peripheral area of the character line indicates that the pixel value is a binarized value. Thus, in step S6-5, it is determined whether or not the peripheral area of the pixel A to be interpolated is the peripheral area of the character line.

If it is determined in S6-5 that the reference area is a binarized value, the selecting means 13 determines whether two or more of the four values H1 to H4 are equal to or larger than DR x? (S6-6) . ? Is a constant set value, and? = 1.5 is assumed in the present embodiment. However, the present invention is not limited to this, and? Can be selected as any number satisfying 1 <? &Lt; 2. This makes it possible to determine whether or not the center pixel of the reference area is on a line of a character.

If the reference area satisfies the condition of step S6-6, the selecting means 13 can select the second interpolation method (S6-7). If the reference area does not satisfy the condition of step S6-5 or step S6-6, the selecting means 13 can select the first interpolation method (S6-8). Nine reference regions are extracted as described above. When the second interpolation method is selected in any one of the reference areas, the second interpolation method is selected. When the first interpolation method is selected in all the reference areas, the first interpolation method is selected.

In an image, when a character is displayed with a clear line, it is also necessary to display the character with a clear line even in an enlarged image. If the line of characters is enlarged by the first interpolation method for emphasizing the high band, the lines blend with the pixels around the characters. In the present embodiment, it is determined whether or not the interpolation object position is a line of characters, and in the case of a line of characters, a second interpolation method which does not mix pixels is selected. Thus, the image can be enlarged while keeping the lines of characters in a clear state.

If there is only one reference area, there is a possibility that the end of the line of characters is erroneously determined. Thus, for the eight neighboring pixels of the pixel to be interpolated, a reference area having eight pixels as the center pixel is extracted. Thus, it is possible to reliably determine the end of the character line.

The present invention is not limited to the above-described embodiment and the accompanying drawings, but is intended to be limited by the appended claims. It will be apparent to those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention as defined by the appended claims. something to do.

1: Image magnifying device
10: input means
11: Selection means
12: first interpolation means
13: second interpolation means
14: output means
15: Direction discriminating means
16: Arithmetic interpolation means

Claims (17)

An image enlarging apparatus which receives image data including a plurality of pixels arranged in a lattice, and applies an interpolation method to the image data to output enlarged image data,
Input means for receiving the image data;
First interpolation means for generating an interpolation pixel to which a first interpolation method for emphasizing high band is applied between pixels of the image data inputted by the input means;
Second interpolation means for generating an interpolation pixel to which a second interpolation method which does not emphasize high band is applied between the pixels of the image data inputted by said input means;
Wherein the first interpolation method is applied to the interpolation object position and the second interpolation method is applied to the interpolation object position by extracting peripheral pixels of the interpolation object position from the image data input by the input means, Selecting means for selecting whether to apply the two interpolation method; And
Output means for outputting, as an output interpolation pixel, either the interpolation pixel inserted by the first interpolation method or the interpolation pixel inserted by the second interpolation method on the basis of the selection of the selection means; And an image enlarging device.
The method according to claim 1,
The selection means calculates the difference value between adjacent pixels with respect to the reference region including a plurality of pixels adjacent to the interpolation object position and determines that the reference region is not a flat region by the difference value , The first interpolation method is selected and the second interpolation method is selected when the reference area is determined as a flat area by the difference value.
3. The method of claim 2,
Wherein the selection means includes a first reference region including a left pixel of the interpolation object position and a lower pixel of the left pixel when the interpolation object position is adjacent to the left and right two pixels in a lateral direction, A second reference area in which the reference area is shifted upward by one pixel, a third reference area including a pixel on the right side of the interpolation object position, a pixel on the lower side of the right pixel, Refers to a fourth reference area moved upward by a pixel,
Wherein the selection means includes a first reference region including a lower pixel of the interpolation object position and a left pixel of the lower pixel when the interpolation target position is adjacent to the upper and lower two pixels in the longitudinal direction, A second reference area in which the reference area is shifted rightward by one pixel, a third reference area including an upper pixel of the interpolation object position, a left pixel of the upper pixel, and a third reference area, Refers to a fourth reference area shifted rightward by a pixel,
Wherein the selection means includes a first reference region including an upper left pixel of the interpolation object position and a lower left pixel of the interpolation object position when the interpolation target position is adjacent to the four surrounding pixels in an oblique direction, A second reference area in which the first reference area is shifted upward by one pixel, a third reference area in which the first reference area is shifted downward by one pixel, A fourth reference area including the right lower pixel of the interpolation object position, a fifth reference area in which the fourth reference area is shifted upward by one pixel, and a fourth reference area in which the fourth reference area is shifted by one pixel A seventh reference area including a pixel on the upper left side of the interpolation object position and a pixel on the upper right side of the interpolation object position; Pixel and an eighth reference area including a lower right pixel of the interpolation target position,
Wherein the selecting means selects the second interpolation method when any one of the plurality of reference areas to be referred to is determined as a flat area by the differential value.
4. The method according to any one of claims 1 to 3,
Wherein the first interpolation method is a bicubic method or a Rancho method, and the second interpolation method is a near list method.
4. The method according to any one of claims 1 to 3,
A direction determination means for determining a direction and a direction in which the interpolation object position correlates with surrounding pixels; Further comprising:
The first interpolation means generates the interpolation pixel by using a bicubic method when the direction is detected and the directional interpolation is performed by the direction determination means,
Wherein the second interpolating means comprises an image enlarging device for generating the interpolation pixel by using a bilinear method when the direction is detected and directional interpolation is performed by the direction judging means,
The method according to claim 1,
Wherein the selection means selects one of the six pixels in the horizontal direction from the periphery of the interpolation object position, the reference area in the horizontal direction that includes two pixels in the vertical direction, two pixels in the horizontal direction, Extracts a reference region of &lt; RTI ID = 0.0 &
The selecting means
A first condition in which pixel values of two columns along the long side direction of the reference area all monotonously increase or monotonically decrease toward one direction,
A second condition in which all six sets of pixel values along the short side direction of the reference area increase or decrease toward one direction,
A third condition that the total sum of difference values between adjacent pixels in the reference area is equal to or less than a predetermined set value
, The second interpolation method is selected.
The method according to claim 6,
Wherein the selecting means extracts a plurality of reference areas in such a manner that the reference areas are not overlapped with each other, and when any one of the plurality of reference areas conforms to any one of the first condition, the second condition, and the third condition, And selects the second interpolation method.
8. The method according to claim 6 or 7,
Wherein the second interpolation method is a near list neighbor method or a bi-linear method image enlarging device.
The method according to claim 1,
Wherein the selecting means selects one of the six pixels in the horizontal direction from the periphery of the interpolation target position, the horizontal reference area including two pixels in the vertical direction, or two pixels in the horizontal direction and six pixels in the vertical direction Extracts the reference area,
The selecting means
And a second condition in which all of the difference values are equal to or less than a predetermined set value,
A difference value between two sets of pixels located at the center in the long side direction of the reference region and along the short side direction is calculated,
And selects the second interpolation method when both of the first interpolation method and the second interpolation method coincide with each other.
10. The method of claim 9,
Wherein the selecting unit extracts, as a first lateral reference area, a reference area of the transverse field in which the left pixel of the correction target position is located at the upper left side when the interpolation target position is adjacent to the left and right two pixels in the lateral direction, And extracting the second to sixth transverse reference regions shifted by one pixel from the first transverse direction reference region to the left side and extracting the first to sixth transverse direction reference regions from one pixel Extracts the seventh to twelfth lateral reference regions shifted by a predetermined distance,
Wherein the selection unit extracts, as the first longitudinal direction reference area, the reference area in which the upper side pixel of the correction target position is located at the upper left side when the interpolation target position is adjacent to the upper and lower two pixels in the longitudinal direction, And extracting second through sixth longitudinal reference regions shifted by one pixel from the first longitudinal reference region upward, and extracting the first through sixth longitudinal reference regions by one pixel toward the left Extracts the shifted seventh to twelfth directional reference areas,
Wherein the selecting means sets the upper left pixel of the correction target position as the position of the upper left side and the first to sixth lateral direction reference marks Region, the seventh to twelfth lateral direction reference regions, the first to sixth longitudinal direction reference regions, and the seventh to twelfth direction reference regions,
Wherein the selecting means selects the second interpolation method when any one of the plurality of reference areas matches any one of the first condition and the second condition.
11. The method according to claim 9 or 10,
The second interpolation method is an image enlargement apparatus which is a near-list Naver corporation.
The method according to claim 1,
Wherein the selection means extracts a reference area having three pixels in the horizontal direction and three pixels in the vertical direction with the pixel adjacent to the interpolation object position as the center pixel,
Wherein the selecting means calculates a maximum value and a minimum value of the nine pixels included in the reference area and determines whether the center pixel value of the reference area is the maximum value and the [center pixel value] > ([Maximum value] - [minimum value]) / 2 + [minimum value]) / 2+ [minimum value]) x is satisfied or the center pixel value of the reference area is the minimum value [center pixel value] [Minimum value]) / [alpha] is established, the second interpolation method is selected.
13. The method of claim 12,
Wherein the selection means comprises a first reference region having a left pixel as a center pixel of the interpolation object position and a second reference region having a center pixel as a center of the right pixel of the interpolation object position when the interpolation object position is adjacent to the left and right two pixels in the lateral direction, Extracts a second reference area as a pixel,
Wherein the selection means includes a first reference region in which an upper pixel of the interpolation object position is a center pixel and a second reference region in which a lower pixel of the interpolation object position is centered Extracts a second reference area as a pixel,
Wherein the selection means includes a first reference region having a left upper pixel of the interpolation target position as a central pixel and a second reference region having a left upper pixel of the interpolation target position as an upper right A third reference area having a center pixel at the lower left pixel of the interpolation object position and a fourth reference area having a center pixel at the lower right pixel of the interpolation object position Extraction,
Wherein the selecting means selects the second interpolation method when any one of the plurality of reference regions matches the condition.
The method according to claim 12 or 13,
Wherein the second interpolation method is a near list neighbor method.
The method according to claim 1,
Wherein the selection means extracts a reference region having eleven pixels in the horizontal direction and five pixels in the vertical direction from the periphery of the interpolation target position,
Wherein the selection means calculates a difference value DR between a maximum value and a minimum value of 55 pixels included in the reference region and calculates a histogram and includes three pixels each consisting of a central pixel of the reference region and its surrounding pixels, A sum H1 of absolute values of differences of pixels in the lateral direction including the center pixel and a sum H2 of absolute values of differences of the pixels in the vertical direction including the center pixel, A sum H4 of absolute values of difference between the pixels in the oblique directions including the pixels on the upper left side of the center pixel and a sum H4 of absolute values of differences between the pixels in the oblique direction including the center pixel and the lower right pixel of the center pixel and,
The selecting means
A first condition in which a pixel value is a binarized value from the histogram,
When the second condition that two or more out of the sum H1, the sum H2, the sum H3, and the sum H4 is equal to or larger than? Times the differential value DR is satisfied for the constant value?, The second interpolation method An image enlarging device to be selected.
16. The method of claim 15,
Wherein the selection means sets the difference value DR, the histogram, the sum H1, the sum H2, the sum H3, the sum H3, the sum H3, And selects the second interpolation method when any one of the reference area and the eight additional reference areas satisfies the first condition and the second condition.
17. The method according to claim 15 or 16,
The second interpolation method is an image enlargement apparatus which is a near-list Naver corporation.

Images