KR101106613B1 - Apparatus and method for converting resolution of image using edge profile - Google Patents

Abstract

An apparatus and method for converting resolution of an image using an edge profile are disclosed. The resolution converter converts the low resolution image into a high resolution image. The image corrector modifies the edge information of the high resolution image, and includes an edge profile generator, a fill region determiner, and an edge information restorer. The edge profile generator generates an edge profile corresponding to the high resolution image based on the edge direction of each pixel constituting the high resolution image. The filling region determiner calculates the connection information for each pixel constituting the high resolution image, and determines the pixels as the filling region whose calculated connection information is equal to a preset value. The edge information reconstructor reconstructs edge information by modifying pixel values of pixels determined as filling regions based on the edge profile generated for the high resolution image. The edge sharpening unit sharpens the edges of the high resolution image in which the edge area is corrected. According to the present invention, in the process of converting a low resolution image into a high resolution, an image having sharp edges can be obtained simply and quickly despite the inaccuracy of the edge information.

Description

Apparatus and method for converting resolution of image using edge profile}

The present invention relates to an apparatus and method for converting resolution of an image, and more particularly, to an apparatus and method for converting an image having a function for restoring edge information of a high resolution image obtained from a low resolution image.

When converting a low resolution image into a high resolution image, a high resolution image is obtained through matching, reconstruction, blurring, and noise cancellation for a low resolution image sequence deteriorated by various causes. In this case, a low resolution original image is degraded due to optical distortion, frequency overlap, blur, noise, or the like, and the quality of the image is degraded due to interpolation due to incorrect pixel information when converting to a high resolution image. This degradation in image quality is particularly pronounced in the edge region, which leads to a deterioration in visual quality.

In order to solve such a problem, a technique for defining a spatial characteristic of an image in units of blocks constituting the image frame and restoring the image based on this has been proposed. Techniques for reconstructing images by performing directional interpolation based on simplified edge models show good performance in areas with flat and simple edge characteristics. However, this technique has a problem in that the image quality of a reconstructed image in a detailed or complicated region of image contents is not excellent. In addition, a technique of dividing an edge into a plurality of directional components and sharpening the edge based on the directional component has a problem in that the edge information is missing or damaged, which is not good.

SUMMARY OF THE INVENTION The present invention has been made in an effort to provide an apparatus and method for converting an image resolution capable of recovering lost edge information in a simple and rapid manner.

Another object of the present invention is to provide a computer-readable recording medium having recorded thereon a program for executing a method of converting a resolution of an image, which can be easily and quickly lost edge information.

In order to achieve the above technical problem, an apparatus for converting a resolution of an image according to the present invention includes: a resolution converting unit converting a low resolution image into a high resolution image; An image correction unit for modifying edge information of the high resolution image; And an edge sharpening unit for sharply processing an edge of the high resolution image in which the edge information is corrected, wherein the image correcting unit corresponds to the high resolution image based on an edge direction of each pixel constituting the high resolution image. An edge profile generator for generating an edge profile; A filling region determining unit configured to calculate connection information for each pixel constituting the high resolution image and to determine pixels as the filling region whose calculated connection information is equal to a preset value; And an edge information reconstructing unit configured to reconstruct the edge information by modifying pixel values of pixels determined as the filling region based on the edge profile generated for the high resolution image.

According to another aspect of the present invention, there is provided a resolution converting method of an image, the method comprising: converting a low resolution image into a high resolution image; Modifying edge information of the high resolution image; And sharply processing an edge of the high resolution image in which the edge information is corrected, wherein the edge information modification step comprises: an edge corresponding to the high resolution image based on an edge direction of each pixel constituting the high resolution image; Creating a profile; Calculating connection information for each pixel constituting the high resolution image, and determining pixels as the filling region whose calculated connection information is equal to a preset value; And restoring the edge information by modifying pixel values of pixels determined as the filling region based on the edge profile generated for the high resolution image.

According to the apparatus and method for converting an image resolution using an edge profile according to the present invention, an image having a sharp edge can be obtained simply and quickly despite the inaccuracy of edge information in the process of converting a low resolution image into a high resolution. In particular, when converting a low resolution image into a high resolution image, there is an advantage of reducing the number of repetitions in a similar peak signal to noise ratio (PSNR) performance.

1 is a view showing the configuration of a preferred embodiment of an apparatus for converting a resolution of an image according to the present invention;
2 is a flowchart illustrating a process of performing a preferred embodiment of a method for converting resolution of an image according to the present invention;
3 is a diagram showing a detailed configuration of the image correction unit 120;
4 is a diagram illustrating a process for the edge profile generator 310 to derive edge characteristics for each pixel.
5 is a diagram for describing a process of determining, by the edge information reconstructor 330, pixel values of pixels constituting an input image;
6 is a view showing an image of an enlarged rectangular region after the input image, an enlarged image of the rectangular region of the input image, and the modification of the input image according to the present invention;
7 is a diagram illustrating an edge profile of an image in which a rectangular region of an input image is enlarged and an edge profile of an image of an enlarged rectangular region after the input image is corrected according to the present invention;
8 is a view illustrating an input image having incorrect edge information and an image of which edge information is modified by the image reconstruction device according to the present invention;
9 is a flowchart illustrating a process of reconstructing edge information by the image correction unit 130, and
FIG. 10 is a diagram illustrating a result of converting a Lena image by a conventional resolution conversion method and a resolution conversion method according to the present invention according to a factor estimation using the L-Curve method.

Hereinafter, with reference to the accompanying drawings will be described in detail a preferred embodiment of the device and method for adaptive adaptive image reconstruction using an edge profile according to the present invention, and an image resolution conversion device and method using the same.

1 is a view showing the configuration of a preferred embodiment of the resolution conversion apparatus of the image according to the present invention, Figure 2 is a flow chart showing a process of performing a preferred embodiment of the resolution conversion method of the image according to the present invention.

1 and 2, a preferred embodiment of an image resolution converting apparatus according to the present invention includes a resolution converting unit 110, an image corrector 120, and an edge sharpening unit 130.

The resolution converting unit 110 converts a low resolution image into a high resolution image (S200). The detailed configuration and function of the resolution converting unit 110 is well known to those skilled in the art to which the present invention pertains, and thus detailed description thereof will be omitted.

The image corrector 120 corrects edge information of an image having a higher resolution by the resolution converter 110 (S210). 3 is a diagram showing the detailed configuration of the image correction unit 120. As shown in FIG. Referring to FIG. 3, the image corrector 120 includes an edge profile generator 310, a fill region determiner 320, and an edge information restorer 330.

The edge profile generator 310 determines an edge characteristic corresponding to each pixel constituting the input image and generates an edge profile corresponding to the input image based on the edge characteristic. In this embodiment, the edge characteristic is determined using the direction adaptive edge search. At this time, the edge profile generator 310 divides the pixels constituting the input image into eight directions of 0 °, 45 °, 90 °, 135 °, 180 °, 225 °, 270 °, and 315 °, respectively. An edge profile is generated by assigning a 3-bit unique code for the direction.

4 is a diagram illustrating a process for the edge profile generator 310 to derive edge characteristics for each pixel. Referring to FIG. 4, edge characteristics of arbitrary pixels X _{i and j} constituting the input image are obtained by comparison with each of the adjacent pixels. To this end, the edge profile generator 310 performs a first operation value (| X _{i , j-1 that} is an absolute value of a difference between pixel values of pixels positioned in a horizontal direction with respect to any pixel X _{i , j} constituting the input image. -X _{i , j + 1} |) is compared with a preset reference value M (S400). If the first operation value is larger than M, the edge direction is determined as the pixel value is higher (S405). Therefore _, if X _{i , j-1} -X _{i , j + 1} > 0, the edge direction is determined to be 180 °. On the contrary, if the first operation value is smaller than M, the edge profile generator 310 performs a second operation value (| X _{i −)} , which is an absolute value of the difference between pixel values of pixels located in the vertical direction with respect to pixels X _{i , j} . _{1, j} -X _{i + 1, j} |) is compared with a preset reference value M (S410). If the second operation value is larger than M, the edge direction is determined as the pixel value is higher (S415). Therefore, if X _{i-1, j} −X _{i + 1, j} > 0, the edge direction is determined to be 90 °. Next, when the second operation value is smaller than M, the edge profile generator 310 determines the third operation value (| X) which is the absolute value of the difference between pixel values of pixels located in the upper left and lower directions with respect to the pixel X _{i , j} . _{i -1, j-1} -X _{i + 1, j + 1} |) is compared with a preset reference value M (S420). If the third operation value is larger than M, the edge direction is determined as the pixel value is higher (S425). Therefore _, if X _{i -1, j-1} -X _{i + 1, j + 1} > 0, the edge direction is determined to be 135 °. Next, when the third operation value is smaller than M, the edge profile generator 310 performs a fourth operation value (| X) which is an absolute value of the difference between pixel values of pixels located in the top-right direction from the top left with respect to the pixel X _{i , j} . _{i + 1, j-1} -X _{i -1, j + 1} |) is compared with a preset reference value M (S430). If the fourth operation value is larger than M, the edge direction is determined as the pixel value is higher (S435). Therefore, if X _{i + 1, j-1} -X _{i -1, j + 1} > 0, the edge direction is determined to be 225 °. If the fourth operation value is smaller than M, the corresponding pixel is defined as a non-edge pixel (S440). Finally, the edge profile generator 310 performs the process as shown in FIG. 4 for all pixels constituting the input image (S445), and then applies the edge characteristic codes as described in Table 1 for each direction. In operation S450, the edge profile is generated. In this case, a separate code is not assigned to non-edge pixels.

Edge direction 0 ° 45 ° 90 ° 135 ° 180 ° 225 ° 270 ° 315 ° 3-bit code 000 001 010 011 100 101 110 111

The filling area determiner 320 calculates connection information for each pixel constituting the input image, and determines the filling area as a pixel having the calculated connection information equal to a preset value. In this case, the connection information is defined as the number of pixels in which there is a difference between a pixel value of a pixel X _{i , j} constituting the input image and a pixel value of each of the nine pixels adjacent to the pixel. For example, if any pixel X _{i , j} has the same pixel value as four pixels among the surrounding nine pixels, the connection information of the pixel becomes five. Therefore, when the setting value is 5, the corresponding pixel is determined as the filling area. In this case, a plurality of setting values may be set. For example, when 8 and 7 are setting values, the filling region determiner 320 determines a pixel having connection information of 8 or 7 as the filling region.

The edge information reconstructor 330 reconstructs edge information by modifying pixel values of pixels determined as filling regions based on the edge profile generated for the input image. As shown in FIG. 5, when the edge direction of the pixel X _{i , j} is 001 (that is, 45 °), the edge information reconstructor 330 is adjacent to the pixel X _{i -1, j +} Give it a pixel value of ₁ . In contrast, the edge information restoration unit 330 may determine the pixel value of the pixel X _i, the pixel based on the pixel values of the pixels having the highest frequency direction from the edge direction of the pixels adjacent to the _j X _{i, j.} In this case _, the pixel values of pixel X _{i , j} are pixels (X _{i -1, j-1} , X _{i -1, j} , X _{i , j-} with edge characteristics having the highest frequency of 010 (ie, 90 °)). A pixel value arbitrarily selected from among pixel values of ₁ , X _{i +1, j-1} , and corresponding pixels (X _{i -1, j-1} , X _{i -1, j} , X _{i , j-1} , X _{i + 1, j-1} ) may be determined as the average value of the pixel value.

FIG. 6 is a diagram illustrating an input image, an enlarged image of a rectangular region of the input image, and an image of an enlarged square region after the input image is corrected according to the present invention. Referring to FIG. 6, after the input image (left) is corrected by the image restoring apparatus according to the present invention, the image (right) in which the rectangular area 610 is enlarged (right) is an image in which the rectangular image is enlarged (center). It can be seen that the edge is clearer. In particular, the edges of the rectangular image 620 of the center image are blurred, whereas the edges of the rectangular image 630 of the right image have sharp edges. The edge profile (left image of FIG. 7) of the image in which the rectangular region of the input image shown in FIG. 7 is enlarged and the edge profile of the image in which the square region is enlarged after the input image is corrected according to the present invention (right side of FIG. 7). This can be seen clearly by comparing the images. The edge profile shown in FIG. 7 represents an edge characteristic of each pixel in a 3-bit code and gives each code a different color to represent an image. 8 illustrates an input image having incorrect edge information and an image in which edge information is modified by the image reconstruction device according to the present invention. Referring to FIG. 8, although edge information is lost in various places of the input image (left), the reconstructed image (right) may confirm that the edge information that is substantially lost is restored.

9 is a flowchart illustrating a process of restoring edge information by the image corrector 120.

Referring to FIG. 9, the edge profile generator 310 determines an edge direction of each pixel constituting the input image and generates an edge profile corresponding to the input image by applying a code corresponding to each edge direction. (S900). Next, the filling region determiner 320 calculates connection information for each pixel constituting the input image, and determines the filling region in which the calculated connection information is the same as a preset value (S910). Next, the edge information reconstructor 330 reconstructs the edge information by correcting pixel values of pixels determined as filling regions based on the edge profile generated for the input image (S930).

The edge sharpening unit 130 sharply processes the edge of the high resolution image in which the edge region is restored by the image corrector 120 (S220). Such edge sharpening may be performed by a steerable filter capable of adjusting the direction. The steerable filter is generally used to sharpen the edge by dividing 360 ° into eight directional components, and the filter characteristic can be expressed by the following equation.

이다.Where F is an edge sharpened image, I is an input image,

to be.

Steps S200 to S220 as described above are repeatedly performed until an image having a desired resolution is obtained. In the resolution conversion process of such an image, if the coefficient of the steerable filter is appropriately set through the normalization factor estimation using the L-Curve method, the resolution conversion apparatus of the image according to the present invention can obtain a conventional image at a similar PSNR performance. Compared to the resolution converter, the conversion process of the super resolution image can reduce the number of repetitions by four times or more. FIG. 10 shows the results of converting a Lena image by a conventional resolution conversion technique (upper) and a Lena image by a resolution conversion technique according to the present invention according to a factor estimation using the L-Curve method. The result (bottom) is shown. Referring to FIG. 10, a conventional resolution conversion technique and an L-Curve technique (edge mode = 5,) performed on a Lena image having a size of 256 × 256 using an L-Curve technique (edge mode = 5, λ = 0.01). The resolution conversion method according to the present invention performed on a Lena image of size 256 × 256 using λ = 0.047) has almost the same PSNR, but in terms of the number of repetitions, it can be seen that there is a large difference between 32 times and 7 times. .

The present invention can also be embodied as computer-readable codes on a computer-readable recording medium. The computer-readable recording medium includes all kinds of recording devices in which data that can be read by a computer system is stored. Examples of the computer-readable recording medium include a ROM, a RAM, a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, and the like, and may be implemented in the form of a carrier wave (for example, transmission via the Internet) . The computer readable recording medium can also be distributed over network coupled computer systems so that the computer readable code is stored and executed in a distributed fashion.

Although the preferred embodiments of the present invention have been shown and described above, the present invention is not limited to the specific preferred embodiments described above, and the present invention belongs to the present invention without departing from the gist of the present invention as claimed in the claims. Various modifications can be made by those skilled in the art, and such changes are within the scope of the claims.

Claims (13)

A resolution converting unit converting the low resolution image into the high resolution image;
An image correction unit for modifying edge information of the high resolution image; And
And an edge sharpening unit for sharply processing an edge of the high resolution image in which the edge information is corrected.
The image correction unit,
An edge profile generator configured to generate an edge profile corresponding to the high resolution image based on an edge direction of each pixel constituting the high resolution image;
A filling region determining unit configured to calculate connection information for each pixel constituting the high resolution image and to determine pixels as the filling region whose calculated connection information is equal to a preset value; And
And an edge information restoring unit for restoring the edge information by modifying pixel values of the pixels determined as the filling region based on the edge profile generated for the high resolution image. The method of claim 1,
The edge profile generator determines the edge direction of the pixels constituting the high resolution image in one of a predetermined number of different directions, and based on the edge direction determined for each of the pixels constituting the high resolution image. An apparatus for converting resolution of an image comprising generating an edge profile. The method of claim 1,
The edge profile generation unit determines an edge direction with respect to any pixel X _{i, j} constituting the high resolution image by the following conditional expression:
Condition 1: among the pixel X _i, is greater than the reference value, the absolute value of the first calculated value of the difference between pixel values of adjacent pixels in the horizontal direction to the _j is set in advance, the pixel X _i, pixels adjacent in the horizontal direction from the _j The edge direction is determined in a direction in which the pixel value is high. If the first operation value is smaller than the reference value, condition 2 is examined.
Condition 2: the pixel X _i, absolute value of the second calculated value of the difference between pixel values of adjacent pixels in the vertical direction to the _j is larger than the reference value, the pixel X _i, the pixel value among the vertical adjacent pixels in a direction from the _j The edge direction is determined in this high direction. If the second operation value is smaller than the reference value, condition 3 is examined.
Condition 3: If the third operation value which is the absolute value of the difference of the pixel values of the pixels adjacent to the pixel X _{i , j} in the upper left direction is greater than the reference value, the pixel adjacent to the upper right in the lower left direction from the pixel X _{i , j} Among them, the edge direction is determined in a direction in which the pixel value is higher. If the third operation value is smaller than the reference value, condition 4 is examined.
Condition 4: If the fourth operation value which is the absolute value of the difference of the pixel values of the pixels adjacent to the pixel X _{i , j} in the upper left direction is greater than the reference value, the pixel adjacent in the upper left to right direction from the pixel X _{i , j} Among them, the edge direction is determined in a direction in which the pixel value is high, and when the fourth operation value is smaller than the reference value, the pixel X _{i , j} is determined as the non-edge pixel. 4. The method according to claim 2 or 3,
And the edge profile generator generates an edge profile by assigning a 3-bit unique code to an edge direction determined for each pixel constituting the high resolution image. The method of claim 1,
The connection information may be defined as the number of pixels having a difference between a pixel value of a pixel X _{i , j} constituting the high resolution image and a pixel value of each of the nine pixels adjacent to the pixel. Resolution conversion device. 6. The method of claim 5,
And the set value is determined by at least one of natural numbers of 9 or less. The method of claim 1,
The edge information reconstructor detects an edge direction of a pixel determined as the filling region from an edge profile corresponding to the high resolution image, and corrects the pixel value of the pixel determined as the filling region to the pixel value of a pixel adjacent to the determined edge direction. An apparatus for converting a resolution of an image comprising: The method of claim 1,
The edge information reconstructor determines an edge direction of pixels adjacent to the pixel determined as the filling region from an edge profile corresponding to the high resolution image, and determines a direction having a high frequency among the edge directions of pixels adjacent to the pixel determined as the filling region. And a pixel value of the pixel determined as the filling region is corrected based on the pixel values of the pixels. Converting the low resolution image into the high resolution image;
Modifying edge information of the high resolution image; And
And sharply processing an edge of the high resolution image in which the edge information is corrected.
The edge information modification step,
Generating an edge profile corresponding to the high resolution image based on an edge direction of each pixel constituting the high resolution image;
Calculating connection information for each pixel constituting the high resolution image, and determining pixels as the filling region whose calculated connection information is equal to a preset value; And
And reconstructing the edge information by modifying pixel values of pixels determined as the filling region based on the edge profile generated for the high resolution image. The method of claim 9,
In the edge profile generation step, the edge direction of the pixels constituting the high resolution image is determined as one of a preset number of different directions, and based on the edge direction determined for each of the pixels constituting the high resolution image. And generating the edge profile. The method of claim 9,
The connection information may be defined as the number of pixels having a difference between a pixel value of a pixel X _{i , j} constituting the high resolution image and a pixel value of each of the nine pixels adjacent to the pixel. Resolution conversion method. The method of claim 9,
In the step of restoring the edge information, the edge direction of the pixel determined as the filling region is determined from an edge profile corresponding to the high resolution image, and the pixel value of the pixel adjacent to the determined edge direction is determined by the pixel value of the pixel determined as the filling region. The resolution conversion method of the image, characterized in that for correcting. A computer-readable recording medium having recorded thereon a program for causing a computer to execute the method for converting a resolution of an image according to any one of claims 9 to 12.

Images