KR101468358B1 - Anti aliasing method and apparatus - Google Patents

Abstract

Disclosed are an antialiasing method and an antialiasing apparatus. The antialiasing method comprises the following steps of: decomposing an inputted image into a plurality of subbands by performing a discrete wavelet transform for the inputted image; detecting an aliasing area with regard to the decomposed subbands by the discrete Fourier transform; and removing aliasing in each subband by performing adaptive wavelet shrinkage for the detected aliasing area.

Description

[0001] Anti-aliasing method and apparatus [0002]

The present invention relates to anti-aliasing, and more particularly, to an anti-aliasing method and apparatus for detecting aliasing through Fourier transform and eliminating aliasing using adaptive coefficient reduction of wavelets.

The development of imaging sensor technology has also increased the need for high resolution, high quality images. However, a mobile image system such as a smart phone, a camcorder, a digital camera, and the like has a problem that it is difficult to provide a high-quality image due to limitation of the light amount performance of the lens, non-linear sensor characteristic, . In addition, most mobile imaging systems have suffered from resampling and aliasing artifacts due to resampling.

Thus, over the last few decades a method has been proposed to effectively eliminate aliasing. The conventional method of removing aliasing has a disadvantage in that it is simple and easy to implement by using a low-pass filter in the spatial or frequency domain of an image, but does not completely remove aliasing artifacts. As a result, a method of removing aliasing by wavelet transform or eliminating aliasing by calculating local variance values in the frequency domain has been proposed.

However, these prior arts also have a limitation in providing high-quality images due to generation of artifacts such as traces due to damage to high-frequency components or removal of aliasing by performing filtering on the entire image.

The present invention is directed to an anti-aliasing method and apparatus that can detect aliasing through Fourier transform and remove aliasing using adaptive scaling of wavelets.

The present invention also provides an anti-aliasing method and apparatus that can effectively remove aliasing artifacts without damaging high-frequency components representing fine details of an image.

According to an aspect of the invention, there is provided an anti-aliasing method capable of detecting aliasing through Fourier transform and eliminating aliasing using adaptive coefficient reduction of wavelets.

According to an embodiment of the present invention, there is provided a method of transforming an input image into a plurality of subbands by performing discrete wavelet transform on an input image; (b) detecting an aliasing region through discrete Fourier transform for the plurality of decomposed subbands; And (c) eliminating aliasing in each subband through adaptive wavelet coefficient reduction for the detected aliasing region.

In step (b), the LL subband of the decomposed subband is subjected to patch-based filtering, and the remaining high frequency subbands (LH, HL, HH subband) Step can be preceded.

Wherein the patch-based filtering comprises: dividing the low frequency subband into n x n (natural number) patches; Calculating crosscorrelation coefficients for each of the divided patches with the remaining patches; Searching for at least one similar patch whose crosscorrelation coefficient calculated for each patch is equal to or greater than a threshold value; And reducing the aliasing for each patch by averaging the similar patches retrieved for the respective patches and the respective patches.

The step (b) includes: calculating an edge strength map using a local variable of the low frequency subband; And removing the aliasing for the low frequency subband using the boundary intensity map and the original low frequency subband decomposed according to the discrete wavelet transform and the low frequency subband where the aliasing is reduced.

In the step (b), the aliasing region may be detected by comparing the discrete Fourier transform coefficients of the original low frequency subband and the aliased low frequency subband.

The cross-correlation coefficient may be calculated using an average value of the patches, the divided patches, and an average value of the low-frequency sub-band and the low-frequency sub-band.

The cross-correlation coefficient for each patch is calculated using the following equation,

는 저주파 부대역에서의 n x n 패치를 나타내고,

는

의 평균값을 나타내며,

는 저주파 부대역(

)의 평균값을 나타내고, s,t는 n x n 패치를 나타내는 패치 인덱스를 나타낸다.here,

Represents the nxn patch in the low frequency subband,

The

, ≪ / RTI >

Low frequency subband (

), And s and t represent the patch index indicating the nxn patch.

The patches and the similar patches searched are averaged using the following equations,

는 저주파 부대역(

)의 영역을 나타내고,

는 각 패치에 대한 유사 패치들의 집합을 나타내고,

는 두 패치간의 가중치값으로, here,

Low frequency subband (

), ≪ / RTI >

Represents a set of similar patches for each patch,

Is a weight value between two patches,

는 0.1의 값을 포함하는 필터링의 정도(degree)를 나타내는 파라미터이다.
here,

Is a parameter indicating the degree of filtering including a value of 0.1.

According to another aspect of the present invention, there is provided an apparatus capable of detecting aliasing through a Fourier transform and eliminating aliasing using adaptive coefficient reduction of a wavelet.

According to an embodiment of the present invention, there is provided an image processing apparatus including a transform unit for performing discrete wavelet transform on an input image and decomposing the input image into a plurality of subbands; A detecting unit detecting an aliasing region through discrete Fourier transform on the plurality of decomposed subbands; And an aliasing remover that removes aliasing from each subband through adaptive wavelet coefficient reduction for the detected aliasing area.

Wherein the detector comprises: an aliasing detector for performing patch-based filtering on a low frequency subband of the decomposed subband and detecting aliasing using local variables of the remaining high frequency subbands (LH, HL, and HH subbands); A boundary line strength map generator for calculating an edge strength map using local variables of the low frequency subband; And removing aliasing for the low frequency subband using the boundary strength map and the original low frequency subband decomposed according to the discrete wavelet transform and the aliasing reduced low frequency subband, And an aliasing detector for detecting the aliasing region by comparing discrete Fourier transform coefficients for the low frequency subbands.

Wherein the patch-based filtering is performed by dividing the low frequency subband into nxn (natural number) patches, calculating a cross-correlation coefficient between the divided patches and the remaining patches, and then calculating a cross- At least one similar patch having a threshold value or more may be searched for and a similar patch searched for each patch and each patch may be averaged to reduce aliasing for each patch.

The aliasing removal unit may predict an edge direction using a predetermined direction filter and remove aliasing through adaptive wavelet coefficient reduction based on the predicted edge direction.

The directional filter may include a vertical direction, a horizontal direction, a 45 degree diagonal direction, and a 135 degree diagonal direction filter.

Wherein the aliasing removal unit adaptively reduces the wavelet coefficients for each high frequency subband using the following equation to eliminate aliasing,

,

,

는 수직, 수평, 대각선 방향의 웨이블릿 계수를 나타내고,

(x,y)는 2차원 웨이블릿 함수를 나타낸다.here,

,

,

Represents the wavelet coefficients in the vertical, horizontal, and diagonal directions,

(x, y) represents a two-dimensional wavelet function.

By providing an anti-aliasing method and apparatus according to an embodiment of the present invention, aliasing can be detected through Fourier transform and aliasing can be eliminated using adaptive wavelet coefficient reduction.

Further, the present invention can effectively remove aliasing artifacts without damaging high-frequency components representing fine details of an image.

1 is a flowchart illustrating an anti-aliasing method according to an embodiment of the present invention;
2 is a diagram illustrating a result of discrete wavelet transform for an input image according to an embodiment of the present invention;
3 is a diagram illustrating a process of detecting an aliasing area according to an embodiment of the present invention;
4 is a diagram illustrating patch-based filtering according to an embodiment of the present invention;
5 illustrates results of applying a directional filter to a low frequency subband according to an embodiment of the present invention.
Figure 6 compares anti-aliasing results according to one embodiment of the present invention.
7 is a graph comparing REMS according to an anti-aliasing method according to an embodiment of the present invention.
8 is a block diagram schematically showing an internal configuration of an anti-aliasing apparatus according to an embodiment of the present invention.
.

BRIEF DESCRIPTION OF THE DRAWINGS The present invention is capable of various modifications and various embodiments, and specific embodiments are illustrated in the drawings and described in detail in the detailed description. It is to be understood, however, that the invention is not to be limited to the specific embodiments, but includes all modifications, equivalents, and alternatives falling within the spirit and scope of the invention. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, the present invention will be described in detail with reference to the accompanying drawings.

The terms first, second, etc. may be used to describe various components, but the components should not be limited by the terms. The terms are used only for the purpose of distinguishing one component from another.

The terminology used in this application is used only to describe a specific embodiment and is not intended to limit the invention. The singular expressions include plural expressions unless the context clearly dictates otherwise. In the present application, the terms "comprises" or "having" and the like are used to specify that there is a feature, a number, a step, an operation, an element, a component or a combination thereof described in the specification, But do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, or combinations thereof.

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

FIG. 1 is a flowchart illustrating an anti-aliasing method according to an embodiment of the present invention. FIG. 2 is a diagram illustrating a result of discrete wavelet transform for an input image according to an embodiment of the present invention. 4 is a view for explaining patch-based filtering according to an embodiment of the present invention, and FIG. 5 is a view for explaining the process of detecting an aliasing area according to an embodiment of the present invention. FIG. 6 is a diagram illustrating a result of applying a directional filter to a low-frequency subband according to an exemplary embodiment of the present invention. FIG. 6 is a diagram illustrating an anti-aliasing result according to an embodiment of the present invention. 1 is a graph comparing REMS according to anti-aliasing according to an embodiment.

In step 110, the anti-aliasing device 100 receives an input image.

In step 115, the anti-aliasing apparatus 100 performs discrete wavelet transform (DWT) on the input image. DWT can analyze the input image into multiple frequency subbands. The DWT includes an analysis filter including a high-pass filter and a low-pass filter for analyzing an input image in a multi-frequency subband, and the input image is divided into a component representing a fine portion of the image, Is expressed as a component representing a part of the image.

This can be expressed as the following equation.

는 방향에 대한 인덱스로, H(horizontal)는 수평 방향, V(vertical)는 수직 방향, D(diagonal)는 대각선 방향을 각각 나타내며,

는 인공물시작 스케일을 나타내고,

는 스케일

에서 f(x,y)의 근사 계수를 나타내고,

는 세가지 방향에 대한 웨이블릿 계수를 나타내며,

는 2차원 스케일링 함수를 나타내고,

는 2차원 웨이블릿 함수를 나타내며,

는 입력 영상을 나타낸다.here,

H (horizontal) denotes a horizontal direction, V (vertical) denotes a vertical direction, and D (diagonal) denotes a diagonal direction,

Represents the artifact starting scale,

Scale

Represents an approximate coefficient of f (x, y)

Represents the wavelet coefficients for the three directions,

Represents a two-dimensional scaling function,

Represents a two-dimensional wavelet function,

Represents an input image.

The anti-aliasing device 100 can perform DWT on the input image and decompose it into LL sub-band, LH sub-band, HL sub-band, and HH sub-band.

2 illustrates the decomposed subbands (LL, LH, HL, and HH-220 in FIG. 2) by performing DWT on the input image (210 in FIG. 2).

After the DWT is performed on the input image to decompose the input image into a plurality of subbands, the anti-aliasing apparatus 100 detects aliasing through spatial frequency analysis of each subband in step 120.

A method for the anti-aliasing apparatus 100 to analyze aliasing through spatial frequency analysis will be described in more detail with reference to FIG.

First, in accordance with an embodiment of the present invention, the anti-aliasing device 100 may apply a different method of detecting aliasing in the sub-band for the sub-band and the three high-frequency components for the low-frequency component.

For example, the anti-aliasing device 100 may detect aliasing using adaptive patch-based filtering for a low frequency subband (LL subband) that represents a rough portion of the image.

Aliasing can also be detected by detecting a repetitive pattern using local variables for high frequency subbands (i.e., LH, HL, HH subbands) representing fine details of the image.

This will be described in more detail with reference to FIG.

First, in step 310, the anti-aliasing device 100 performs patch-based filtering of the low frequency sub-band among the plurality of sub-bands, and aliasing is detected using the local variable of the remaining high frequency sub-bands.

First, in order to facilitate understanding and explanation, a method of detecting aliasing by performing patch-based filtering on a low frequency subband will be described in detail with reference to FIG.

In step 410, the anti-aliasing device 100 divides the normalized coefficients of the low frequency subband into n x n patches. Here, n may be a natural number.

In step 415, the anti-aliasing device 100 calculates the cross-correlation coefficient with other patches for each patch. For example, the anti-aliasing device 100 may calculate a cross-correlation coefficient for each of the n x n patches with the other patches, respectively.

The anti-aliasing device 100 can calculate the cross-correlation coefficient with other patches for each patch using the following equation (2).

는 저주파 부대역에서의 n x n 패치를 나타내고,

는

의 평균값을 나타내며,

는 입력 영상에 대한 근사 계수(

)의 평균값을 나타내며, s,t는 n x n 패치를 나타내는 패치 인덱스이다.here,

Represents the nxn patch in the low frequency subband,

The

, ≪ / RTI >

Is an approximation coefficient for the input image (

), And s and t are patch indexes representing nxn patches.

In step 420, the anti-aliasing device 100 searches for at least one similar patch for each patch using the calculated cross-correlation coefficients.

For example, the anti-aliasing device 100 may search for similar patches of other patches whose cross-correlation coefficient calculated for each patch is equal to or greater than a threshold (for example, 0.7). It is a matter of course that there may be a plurality of similar patches retrieved for each patch.

Thus, if similar patches are retrieved for each patch, the anti-aliasing device 100 averages each patch and the retrieved similar patches in step 425 to reduce aliasing artifacts.

This can be expressed by the following equation (3).

는 저주파 부대역(

)의 영역을 나타내고,

는 각 패치의 유사 패치 집합을 나타내며,

는 두 패치들의 가중치값을 나타낸다. here,

Low frequency subband (

), ≪ / RTI >

Represents a similar patch set of each patch,

Represents the weight value of the two patches.

The weight value can be calculated using the following equation (4).

는 0.1의 값을 포함하는 필터링의 정도(degree)를 나타내는 파라미터이다.here,

Is a parameter indicating the degree of filtering including a value of 0.1.

Referring back to FIG. 3, in step 315, the anti-aliasing device 100 calculates an edge strength map using the local variables of each subband.

In the patch-based filtering performed in FIG. 4, a similar patch similar to each patch is averaged, and the high frequency detail of the high frequency component of each patch is damaged. Accordingly, the anti-aliasing device 100 can calculate the boundary strength map using local variables to preserve the fine details of the high frequency components lost in the low frequency subband. Here, the local variable represents the local variable of each subband. The anti-aliasing device 100 may calculate the boundary line intensity map using, for example, the number 5.

는 경계선강도맵의 분포를 제어하기 위한 튜닝 파라미터를 나타내고,

는 각 부대역의 로컬변수를 나타낸다.here,

Represents a tuning parameter for controlling the distribution of the boundary line intensity map,

Represents the local variable of each subband.

At step 320, the anti-aliasing device 100 removes aliasing from the subband using the border strength map.

For example, the anti-aliasing device 100 may use the number 6 to derive a subband for which aliasing has been removed.

는 필터링된 저주파 부대역(각 패치와 각 패치에 대한 유사 패치들을 에버리징하여 에일리어싱이 줄어든 저주파 부대역)을 나타내고,

는 경계선강도맵을 나타내며,

는 오리지널 저주파 부대역을 나타내고,

는 에일리어싱이 제거된 고주파의 세밀한 부분이 보존된 저주파 대역을 나타낸다.here,

Represents a filtered low frequency subband (a low frequency subband where each patch and a similar patch for each patch are averaged to reduce aliasing)

Represents a boundary strength map,

Represents the original low frequency subband,

Represents a low-frequency band in which a fine portion of high frequency without aliasing is preserved.

Then, in step 325, the anti-aliasing device 100 detects the aliasing region using the difference in frequency domain between the subbands for which aliasing has been removed and the subbands containing aliasing (i.e., discrete wavelet transformed subbands).

That is, the anti-aliasing device 100 performs discrete Fourier transform on the discrete wavelet-transformed sub-band (i.e., the original sub-band) and the aliased sub-band to derive a difference in the frequency domain, and then performs inverse discrete Fourier transform Thereby detecting an aliasing area.

 This can be expressed by the following equation (7).

는 이산 퓨리에 변환을 나타내고,

는 역이산 퓨리에 변환을 나타내며,

는 에일리어싱 영역을 나타낸다.here,

Represents a discrete Fourier transform,

Represents an inverse discrete Fourier transform,

Represents an aliasing area.

A method of detecting aliasing through spatial frequency analysis in discrete wavelet-transformed subbands has been described with reference to FIGS. 3 and 4. FIG.

Referring again to FIG. 1, in step 125, anti-aliasing device 100 removes the detected aliasing region using adaptive wavelet coefficient reduction.

The anti-aliasing device 100 according to an embodiment of the present invention can predict each edge direction using a directional filter to effectively remove aliasing artifacts generated in edge areas in different directions.

The directional filter is, for example, equal to the number 8.

는 수평 방향을 예측하기 위한 방향 필터이고,

는 45도 대각선 방향을 예측하기 위한 방향 필터이며,

는 135도 대각선 방향을 예측하기 위한 방향 필터를 나타내고,

는 수직 방향을 예측하기 위한 방향 필터를 나타낸다.here,

Is a directional filter for predicting the horizontal direction,

Is a directional filter for predicting the diagonal direction at 45 degrees,

Represents a directional filter for predicting the diagonal direction at 135 degrees,

Represents a direction filter for predicting the vertical direction.

FIG. 5 illustrates a result of applying a directional filter to a low frequency subband according to an embodiment of the present invention. 5A is a result of applying a vertical direction filter, FIG. 5B is a result of applying a 45-degree direction filter, FIG. 5C is a result of applying a 135-degree direction filter, Results. In this way, the anti-aliasing device 100 can adaptively remove aliasing in the sub-band by applying a directional filter. That is, the anti-aliasing device 100 uses the direction anticipated through the directional filter, By using adaptive wavelet coefficient reduction according to each station, the aliasing region is removed in each high frequency subband.

The anti-aliasing device 100 can remove the aliasing area of each high frequency subband using the following equation (9).

,

,

는 LH, HL, HH 부 대역의 에일리어싱이 제거된 영상이며,

,

,

는 LH, HL, HH 부 대역을 나타내고,

와

는 화소의 위치를 나타내며,

는 계수축소 시 참조할 인접 화소를 정하기 위한 가중치값을 나타낸다.here

,

,

Is an image in which the aliasing of the LH, HL, and HH subbands is removed,

,

,

Represents the LH, HL, and HH sub-bands,

Wow

Represents the position of the pixel,

Represents a weight value for determining the adjacent pixel to be referred to when decreasing the coefficient.

FIG. 6 is a result of comparing anti-aliasing results according to an embodiment of the present invention and FIG. 7 is a graph comparing RMSE values according to an anti-aliasing according to an embodiment of the present invention.

6A shows an input image, FIG. 6B shows an image added with aliasing by simulation, FIG. 6C shows an anti-aliasing result using conventional low-pass filtering (e) shows the result of anti-aliasing of Coulange, (f) shows the result of anti-aliasing of Ynag, and (g) shows the result of anti- (H) shows the result of anti-aliasing according to an embodiment of the present invention.

As shown in FIG. 6, it can be seen that aliasing artifacts are not removed even if a loss of fine detail of high-frequency components does not occur in the conventional five anti-aliasing methods. That is, in the conventional anti-aliasing, it is seen that damage is also caused to the fine portion of the high frequency component in order to remove the aliasing artifact.

On the other hand, the anti-aliasing according to an embodiment of the present invention minimizes the loss of fine portions of the high-frequency components and thus detects aliasing artifacts.

Also, as shown in FIG. 7, it can be seen that the anti-aliasing according to an embodiment of the present invention effectively detects and removes the aliasing artifacts compared to other conventional anti-aliasing methods.

8 is a block diagram schematically illustrating an internal configuration of an anti-aliasing apparatus according to an embodiment of the present invention.

Referring to FIG. 8, an anti-aliasing apparatus 100 according to an embodiment of the present invention includes a converting unit 810, a detecting unit 820, and an aliasing removing unit 830.

The converting unit 810 is means for discrete wavelet-transforming the input image and decomposing the input image into a plurality of subbands.

The detection unit 820 is means for detecting an aliasing region through spatial frequency analysis for a plurality of decomposed subbands.

8, the detection unit 820 includes an aliasing detection unit 821, a boundary line intensity map generation unit 823, and an aliasing area detection unit 825. The aliasing detection unit 820 includes a boundary line intensity map generation unit 823 and an aliasing area detection unit 825. [

The aliasing detection unit 821 is means for detecting aliasing in each subband.

The aliasing detector 821 performs patch-based filtering for low frequency subbands to detect aliasing, and aliasing can be reduced in low frequency subbands through patch-based filtering. In addition, the aliasing detector 821 can detect a repeated pattern using the local variables of each subband for the remaining three high-frequency subbands.

Since the method of reducing aliasing in the low frequency subband by performing the patch-based filtering is the same as that described above, a duplicate description will be omitted.

The boundary line intensity map generation unit 823 calculates a boundary line intensity map using local variables of each subband. This is the same as that described with reference to FIG. 3, so duplicate descriptions will be omitted.

The aliasing region detecting unit 825 detects the aliasing region by means of discrete Fourier transform after preserving the fine portion of the high frequency component using the boundary strength map in the original low frequency subband in which the aliasing is reduced and the aliasing is not reduced, to be.

For example, the aliasing region detection unit 825 may remove aliasing for the low frequency subband using a boundary strength map and a low frequency subband with reduced aliasing and an original low frequency subband decomposed according to the discrete wavelet transform. Next, the aliasing region detecting unit 825 can detect the aliasing region by comparing the discrete Fourier transform coefficients of the original low frequency subband and the low frequency subband that are aliased and removed.

The aliasing removal unit 830 estimates an edge direction using a predetermined direction filter for the aliasing region and reduces aliasing by reducing the wavelet coefficients adaptively according to each sub-band based on the predicted edge direction.

Meanwhile, the anti-aliasing method according to an embodiment of the present invention may be implemented in the form of a program command that can be executed through a variety of means for processing information electronically and recorded in a storage medium. The storage medium may include program instructions, data files, data structures, and the like, alone or in combination.

Program instructions to be recorded on the storage medium may be those specially designed and constructed for the present invention or may be available to those skilled in the art of software. Examples of storage media include magnetic media such as hard disks, floppy disks and magnetic tape, optical media such as CD-ROMs and DVDs, magneto-optical media such as floptical disks, magneto-optical media and hardware devices specifically configured to store and execute program instructions such as ROM, RAM, flash memory, and the like. Examples of program instructions include machine language code such as those produced by a compiler, as well as devices for processing information electronically using an interpreter or the like, for example, a high-level language code that can be executed by a computer.

The hardware devices described above may be configured to operate as one or more software modules to perform the operations of the present invention, and vice versa.

It will be apparent to those skilled in the art that various modifications and variations can be made in the present invention without departing from the spirit or scope of the invention as defined in the appended claims. It will be understood that the invention may be varied and varied without departing from the scope of the invention.

100: Anti-aliasing device
810:
820:
830: aliasing removal

Claims (15)

(a) performing discrete wavelet transform on an input image and decomposing it into a plurality of subbands;
(b) performing patch-based filtering on the low frequency sub-band among the plurality of sub-bands and detecting aliasing using the local variable in the remaining high frequency sub-bands (LH, HL, and HH subbands); And
(c) removing aliasing in each subband through adaptive wavelet coefficient reduction for the detected aliasing region. delete The method according to claim 1,
The patch-
Dividing the low frequency subband into nxn (natural number) patches;
Calculating crosscorrelation coefficients for each of the divided patches with the remaining patches;
Searching for at least one similar patch whose crosscorrelation coefficient calculated for each patch is equal to or greater than a threshold value; And
And reducing aliasing for each patch by averaging the similar patches retrieved for each patch and each patch. The method according to claim 1,
The step (b)
Calculating an edge strength map using local variables of the low frequency sub-bands among the plurality of sub-bands; And
Further comprising the step of removing aliasing for the low frequency subband using the boundary strength map and the original low frequency subband decomposed according to the low frequency subband and the discrete wavelet transform with aliasing reduced, Way. 5. The method of claim 4, wherein step (b)
Wherein the aliasing region is detected by comparing discrete Fourier transform coefficients of the original low frequency subband with the aliasing removed low frequency subband. The method of claim 3,
Wherein the cross-correlation coefficient is calculated using an average value of each of the patches, the divided patches, and an average value of the low-frequency sub-band and the low-frequency sub-band. The method according to claim 6,
Wherein the cross-correlation coefficient for each patch is calculated using the following equation.

here,

Represents the nxn patch in the low frequency subband,

The

, ≪ / RTI >

Low frequency subband (

), And s and t represent the patch index indicating the nxn patch. The method of claim 3,
Wherein each of the patches and the retrieved similar patches are averaged using the following equation.

here,

Low frequency subband (

), ≪ / RTI >

Represents a set of similar patches for each patch,

Is a weight value between two patches,

here,

Is a parameter indicating the degree of filtering including a value of 0.1. 9. A computer-readable recording medium having recorded thereon a program code for performing the method according to any one of claims 1 to 8. A transform unit for performing discrete wavelet transform on an input image and decomposing the input image into a plurality of subbands;
A detecting unit detecting an aliasing region through discrete Fourier transform on the plurality of decomposed subbands; And
And an aliasing elimination unit for predicting an edge direction using a predetermined directional filter and eliminating aliasing in each subband through adaptive wavelet coefficient reduction for the detected aliasing region based on the predicted edge direction, Device. 11. The method of claim 10,
Wherein:
Wherein the LL subband of the decomposed subband performs patch-based filtering and the rest of the high-frequency subbands (LH, HL, HH subband) detect aliasing using local variables;
A boundary line strength map generator for calculating an edge strength map using local variables of the low frequency subband; And
Removing aliasing for the low frequency subband using the boundary intensity map and the original low frequency subband decomposed according to the discrete wavelet transform and the aliasing reduced low frequency subband and the alienated low frequency subband And an aliasing detection unit for detecting the aliasing region by comparing discrete Fourier transform coefficients for subbands. 12. The method of claim 11,
Wherein the patch-based filtering is performed by dividing the low frequency subband into nxn (natural number) patches, calculating a cross-correlation coefficient between the divided patches and the remaining patches, and then calculating a cross- Searching for at least one similar patch having a threshold value or more and averaging the similar patches retrieved for each patch and each patch to reduce aliasing for each patch. delete 11. The method of claim 10,
Wherein the directional filter comprises a vertical direction, a horizontal direction, a 45 degree diagonal direction, and a 135 degree diagonal direction filter. 11. The method of claim 10,
Wherein the aliasing removal unit removes aliasing by adaptively reducing a wavelet coefficient for each high frequency subband using the following equation.

here,

,

,

Represents each high frequency subband,

Wow

Represents the position of the pixel,

Represents the weight value for determining the adjacent pixel to be referred to when decreasing the coefficient

Images