KR101158847B1 - Deinterlacing apparatus and method using edge map - Google Patents

Abstract

The present invention relates to a deinterlacing apparatus and method using an edge map, the deinterlacing apparatus according to an embodiment of the present invention includes a masking application unit for calculating the size of the gradient vector for the reference pixel of the input interlaced image; An edge detector for detecting an edge of the reference pixel by comparing the magnitude of the calculated gradient vector with a preset value; An edge determination unit that determines whether the interpolation target pixel is edged according to the edge of the reference pixel; An interpolation method selector which selects an interpolation method to be applied to the interpolation target pixel according to an edge of the interpolation target pixel; And an interpolation unit which interpolates the interpolation target pixel by using the selected interpolation method.

Description

Deinterlacing apparatus and method using edge map

The present invention relates to an apparatus and method for deinterlacing an image signal, and more particularly, to an apparatus and method for deinterlacing using an edge map.

Since the beginning of TV development, the image source has been interlaced scanning method, and the interlaced scanning method is widely adopted in the TV signal method due to the compatibility problem with the vast image source.

Interlaced scanning is widely used in many TV systems, including the National Television Systems Committee (NTSC), Phase Alternation by Line (PAL), and Secam (SECAM) to efficiently use limited bandwidth. The interlaced scanning method has the advantage of reducing the bandwidth, but the scanning method may cause unwanted visual deterioration such as interline flicker, image jaggedness, and line crawling. Can be.

In addition, many digital video display products (e.g. HDTVs, computer monitors, LCDs, PDP panels, etc.) that are recently released due to the development of video display have a progressive scanning method that can display an entire image at a time. I adopt a lot.

Thus, deinterlacing has been proposed to prevent image quality deterioration of the interlaced scanning method and to convert the interlaced scanning video signal into a sequential scanning video signal.

The deinterlacing technique can be divided into a single-field spatial algorithm using spatial correlation within one frame and an inter-field temporal algorithm using temporal correlation between frames. Single-field spatial algorithms perform intra-field interpolation using spatial correlation without using temporal information, and inter-field temporal algorithms perform inter-field interpolation using temporal correlation.

Inter-field temporal algorithms include a motion adaptive algorithm and a motion compensation algorithm. Inter-field temporal algorithms are known to be more efficient than single-field spatial algorithms, but require significant computational complexity.

Single-field spatial algorithms include line average (LA) algorithms, edge-based line average (ELA) algorithms, and efficient ELA (EELA) algorithms. Single-field spatial algorithms are suitable for real-time applications for reasons of clarity, low cost, and easy execution in hardware. In other words, the single-field spatial algorithm is low in complexity and can be applied to devices requiring real-time processing.

In general, the conventional methods included in the single-field spatial algorithm interpolate along the edge direction. Although the conventional methods belonging to the single-field spatial algorithm are widely used due to the simplicity and ease of execution, the number of edge directions to be considered is limited, which may cause noise or reconstructed pixels to appear as noise.

The ELA algorithm detects the direction of correlation between lines in an image, determines the direction with the highest correlation, and interpolates by taking the pixel average of two lines based on the direction. However, it is sensitive to small changes in pixel values and has the disadvantage of using wrong edge information for diagonals greater than or equal to 45 degrees. An algorithm to solve this problem is EELA.

The EELA algorithm classifies three spatial forms of the image at the current location to find more accurate orientation than the ELA algorithm. In other words, spatial directionality is classified into left, right, or other cases, and the ELA is modified and applied in the appropriate category in each classified category. However, the EELA algorithm is also difficult to realize optimal performance in terms of the estimated edge direction accuracy.

New deinterlacing algorithms are emerging to solve the problem of the existing deinterlacing algorithm. However, although the new deinterlacing algorithms have improved performance, there is a problem that the processing time of the deinterlacing is increased due to the increased computational amount.

The present invention has been made to solve the above problems, and proposes a deinterlacing apparatus and method using an edge map to effectively reduce the processing time of the deinterlacing while maintaining the performance of the deinterlacing.

Problems to be solved by the present invention are not limited to the above-mentioned problems, and other problems not mentioned will be clearly understood by those skilled in the art from the following description.

Deinterlacing apparatus according to an embodiment of the present invention for achieving the above object is a masking applying unit for calculating the size of the gradient vector for the reference pixel of the input interlace image; An edge detector for detecting an edge of the reference pixel by comparing the magnitude of the calculated gradient vector with a preset value; An edge determination unit that determines whether the interpolation target pixel is edged according to the edge of the reference pixel; An interpolation method selector which selects an interpolation method to be applied to the interpolation target pixel according to an edge of the interpolation target pixel; And an interpolation unit which interpolates the interpolation target pixel by using the selected interpolation method.

In addition, the deinterlacing method according to an embodiment of the present invention for achieving the above object comprises the steps of calculating the size of the gradient vector for the reference pixel of the input interlaced image; Detecting an edge of the reference pixel by comparing the magnitude of the calculated gradient vector with a preset value; Determining whether an interpolation target pixel is edged according to whether the reference pixel is edged; Selecting an interpolation method to be applied to the interpolation target pixel according to an edge of the interpolation target pixel; And interpolating the interpolation target pixel using the selected interpolation method.

The present invention also includes a computer-readable recording medium having recorded thereon a program for executing the deinterlacing method on a computer.

Other specific details of the invention are included in the detailed description and drawings.

According to the present invention, by using an edge map, a deinterlacing algorithm having a good deinterlacing performance but having a large amount of computation is applied to an edge region, and a deinterlacing algorithm having a short processing time is applied to a non-edge region to improve the performance of deinterlacing. In addition, there is an advantage that can significantly reduce the processing time of the deinterlacing.

1 is a block diagram of a deinterlacing apparatus according to an embodiment of the present invention.
Figure 2 shows a 3x3 Sobel mask.
3 illustrates an edge map generated using an original image and a Sobel mask.
4 is a diagram illustrating the size of a window centered on an interpolation target pixel.
5 is a diagram illustrating a relationship between an interpolation target pixel, a reference pixel, and adjacent pixels.
6 is a flowchart of a deinterlacing method according to an embodiment of the present invention.
7 is a configuration diagram for comparing an original image with an image to which deinterlacing is applied.
8 is a table showing the results of obtaining the PSNR values and the CPU time of eight test images using a deinterlacing algorithm.

Advantages and features of the present invention and methods for achieving them will be apparent with reference to the embodiments described below in detail with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below, but will be implemented in various forms, and only the present embodiments are intended to complete the disclosure of the present invention, and the general knowledge in the art to which the present invention pertains. It is provided to fully convey the scope of the invention to those skilled in the art, and the present invention is defined only by the scope of the claims. Like reference numerals refer to like elements throughout.

Although the first, second, etc. are used to describe various elements, components and / or sections, these elements, components and / or sections are of course not limited by these terms. These terms are only used to distinguish one element, component or section from another element, component or section. Therefore, the first device, the first component, or the first section mentioned below may be a second device, a second component, or a second section within the technical spirit of the present invention.

The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. In the present specification, the singular form includes plural forms unless otherwise specified in the specification. As used herein, the terms "comprises" and / or "made of" means that a component, step, operation, and / or element may be embodied in one or more other components, steps, operations, and / And does not exclude the presence or addition thereof.

Unless otherwise defined, all terms (including technical and scientific terms) used in the present specification may be used in a sense that can be commonly understood by those skilled in the art. Also, commonly used predefined terms are not ideally or excessively interpreted unless explicitly defined otherwise.

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

1 is a block diagram of a deinterlacing apparatus according to an embodiment of the present invention.

The deinterlacing apparatus 100 includes an input unit 110, a masking application unit 120, an edge detector 130, an edge determiner 140, an interpolation method selector 150, and an interpolator 160.

The input unit 110 inputs an interlaced image, that is, an interlaced image.

The masking application unit 120 calculates the magnitude of the gradient vector with respect to the reference pixel of the interlace image. In general, the Sobel operation is performed.

2 shows a 3x3 size Sobel mask. (a) shows the position of the pixel indicated by z, (b) shows the Sobel operator for performing the Sobel operation in the horizontal direction, and (c) shows the Sobel operator for performing the Sobel operation in the vertical direction.

The Sobel operation is the most representative edge detection method among several edge detection algorithms, and is an algorithm for detecting an edge using a second-order change in brightness between pixels.

The magnitude of the gradient vector in the Sobel operation is defined in Equation (1) below.

[Equation (1)]

Here, g _x and g _y are applied to the horizontal Sobel operator and the vertical Sobel operator, respectively.

Therefore, applying the 3x3 Sobel mask of FIG. 2 leads to the following equation (2).

[Equation (2)]

Here, z ₅ is the pixel position at which the gradient vector is calculated.

The edge detector 130 detects an edge by comparing the magnitude of the gradient vector calculated by the Sobel calculator 110 with a preset value.

가 미리 설정된 문턱값 T보다 큰 경우 해당 화소 z₅를 에지로 판별하여 검출한다. 이러한 방법으로

인 경우 해당 화소를 에지로 판별하고, 나머지는 에지가 아닌 것으로 판별한다.The preset value becomes a threshold, and when the threshold is T, the magnitude of the gradient vector

Is greater than the preset threshold T, the corresponding pixel z ₅ is determined as an edge and detected. In this way

In case of, the corresponding pixel is determined as an edge, and the rest is determined as not an edge.

Pixels detected as edges by the edge detector 130 set the pixel value to 255, and pixels not detected as edges set the pixel value to 0 to generate an edge map.

3 (a) and 3 (b) show original images of Lena and a boat, respectively, and (c) and (d) generate edge maps from the original images of Lena and the boat. The figure is shown. Compared to the original image, the edge map has half the heights.

The edge determiner 140 refers to the pixels included in the edge map by using the edge map, and determines whether to interpolate the target pixel according to whether the reference pixel is edged.

For example, the interpolation target pixel refers to one of the pixels adjacent to the interpolation target pixel to determine whether the interpolation target pixel is an edge region.

When the pixel referred to by the interpolation target pixel is referred to as a reference pixel, the reference pixel is generally located in the upper scan line of the interpolation target pixel adjacent to the interpolation target pixel. The reference pixel may be positioned diagonally to the interpolation target pixel, but is generally located vertically upward for convenience of operation. In addition, the reference pixel may be positioned in a lower scan line of the interpolation target pixel adjacent to the interpolation target pixel. When the reference pixel is positioned vertically downward among the lower scan lines of the interpolation target pixel, the amount of calculation will decrease.

From this, the edge determiner 140 determines the interpolation target pixel as an edge when the reference pixel is an edge, and determines that the interpolation target pixel is not an edge when the reference pixel is not an edge.

The interpolation method selector 150 selects an interpolation method to be applied to the interpolation target pixel according to the edge of the interpolation target pixel determined by the edge determination unit 140.

That is, the interpolation method applied depends on whether the interpolation target pixel is an edge. The interpolation method selector 150 selects an interpolation method using geometric duality when the interpolation target pixel is determined as an edge and interpolates using line averaging when it is determined that the interpolation target pixel is not an edge. Choose a method. Here, the interpolation method using the geometric duality has better interlacing performance than the interpolation method using the line average, but has a long processing time.

The interpolation method using line averaging includes line interpolation (LA) interpolation, edge-based line averaging (ELA) interpolation, efficient edge-based line averaging (EELA) interpolation, and modified edge-based line averaging (MELA) interpolation. It may be one of the. The MELA interpolation method is an interpolation method that is an improved interpolation method of ELA interpolation method considering spatial correlation in three directions of 90 degrees (vertical), 63 (anti-diagonal), and 117 degrees (diagonal).

In the present invention, a covariance-based adaptive deinterlacing method, also known as a covariance-based adaptive deinterlacing (CAD) interpolation method, is superior to the interpolation method using the line average as the interpolation method using geometric duality. Of course, the interpolation method using the geometric duality is not limited to the CAD interpolation method proposed by the present invention, and the CAD interpolation method is only one embodiment of the interpolation method using the geometric duality.

Hereinafter, a CAD interpolation method will be described with reference to FIGS. 4 and 5.

4 is a diagram illustrating a size of a window centered on an interpolation target pixel, and FIG. 5 is a diagram illustrating a relationship between an interpolation target pixel, a reference pixel, and adjacent pixels.

As shown in FIG. 4, as the window size increases, the number of pixels increases, and each pixel may be represented as a data vector for convenience of calculation. The number of pixels included in the window is 2n vertical scan line and 2n + 1 horizontal scan line, respectively, where n is any natural number. WS in FIG. 2 is an abbreviation of Window Size and means any natural number n. Pixels located in the center of the window represent interpolation target pixels, vertical scan lines are scan lines displayed vertically, and horizontal scan lines are scan lines displayed horizontally.

In FIG. 5, a pixel located at (2i + 1, j + 1) coordinates represents an interpolation target pixel to be interpolated through a deinterlacing algorithm, and a pixel located at (2i, j) coordinates is a reference for calculating reference weights in various directions. The reference pixel to be shown. Subscripts i and j are for indicating the position of the pixel.

The horizontal scan line is divided into an upper scan line and a lower scan line based on the interpolation target pixel. The upper scan line is a scan line positioned above and adjacent to the interpolation target pixel, and the lower scan line is a scan line positioned below and adjacent to the interpolation target pixel, and the interpolation target line is formed by using the upper scan line and the lower scan line. Means a set of pixels to be interpolated. 2i-1 and 2i + 1 indicated by dotted lines in the coordinates are interpolation target lines. When 2i + 1 is used, 2i lines become upper scan lines and 2i + 2 lines become lower scan lines.

In Fig. 5, reference weight coefficient α in eight directions with respect to pixel N ₁ (reference pixel) is indicated, and an optimal coefficient α can be obtained from Equation (3) below.

[Equation (3)]

는, 인터레이스 이미지(즉, 저해상도 이미지)에서, 로컬 윈도우(local window) S안에 위치 m의 화소에 인접한 8개의 화소값을 의미한다. 물론, m의 화소에 인접한 화소는 6개일 수도 있다.From here,

Denotes eight pixel values adjacent to the pixel at position m in the local window S in the interlaced image (ie, a low resolution image). Of course, there may be six pixels adjacent to m pixels.

Two methods to obtain Equation (3) are by using least squares approximation or Wiener filter.

In the present invention, the optimal reference weight coefficient α is obtained by using the least square method.

The reference weight is calculated according to the following equations (4) and (5) using the least square method.

[Equation (4)]

[y] = [C] · [α]

[Equation (5)]

[α] = ([C] ^T · [C]) ^-1 ([C] ^T · [y])

Here, [α] denotes [α ₁ ,... , α _2m ], the reference weight vector, [y] is [y ₁ , y ₂ ,. , y _{2n x (2n + 1)} ] ^T , a data vector, [C] represents a data matrix of [{2n X (2n + 1)} X 2m]. N and m are any natural numbers, in particular m will have a value of 3 or 4. Therefore, six edge directions when m is 3 and eight edge directions when m is 4 are considered. Since [C] is not a square matrix, it is converted from equation (4) to equation (5) using a transpose matrix.

FIG. 5 illustrates a case where eight edge directions having a window size of 1 are considered, and WS = n = 1 and m = 4. At this time, N ₁ (Y _{2i , j} coordinate) is selected as the reference pixel in the window (when WS = n = 1), and eight pixels D ₁ adjacent to the reference pixel N ₁ (when m = 4). The reference weight coefficients α ₁ to α ₈ for each of the directions of ˜ D ₈ are calculated. In addition, since the window size WS is 1, the data matrix [C] is a matrix of 6 X 8, and the data vector [y] is [y1,... , y6] ^T. Therefore, the k-th row vector of [C] becomes eight pixels adjacent to y _k .

Of course, when the window size is 1 and m is 3, α ₁ to α ₆ are represented by the weight vector [α], the data matrix [C] is a matrix of 6 X 6, and the edge direction is 6 directions. Only consideration will be given. When considering the six edge directions, reference weights for D ₄ and D ₅ located in the horizontal scan line of N ₁ in FIG. 3 are excluded from consideration.

Based on the weight of the eight edge direction with respect to the reference pixels N ₁ is applied weight of the eight edge direction for the interpolation subject pixel corresponding to the α ₁ α ~ ₈ is a β ₁ ~ β _8. The general natural image has a characteristic that the value of adjacent pixels does not change significantly. Therefore, the direction of the edge also has the same tendency within a certain range. Thus, the correlation between the two variables α and β represents a positive correlation, and the covariance of the two variables α and β is also positive. Therefore, it can be said that the relationship between the existing pixels is the same as the relationship between the pixels to be interpolated. This relationship may be referred to as geometric duality. That is, it may be assumed that pixels having the same direction have weights in proportional relations. That is, the relationship of β ∝ α is established.

Therefore, it is also possible to assume that the weight of each direction of the reference pixel is the same as the weight of each direction of the interpolation target pixel, and the pixel value of the interpolation target pixel is calculated based on the reference weight of each direction of the reference pixel. .

Therefore, the pixel value of the interpolation target pixel can be calculated according to the following equations (6) and (7).

[Equation 6]

[α] = [β]

[Equation (7)]

Here, [α] is a reference weight vector for each direction of the reference pixel, [β] is an applied weight vector for each direction of the compensation target pixel, N is a pixel value of a plurality of pixels adjacent to the interpolation target pixel, and Y is Pixel value of the interpolation target pixel.

The pixel values of the interpolation target pixel are the reference weights α ₁ ,..., Obtained from the reference pixel N ₁ located on the upper left of the interpolation target pixel. , N _1, adjacent to the α ₈ to the interpolation subject pixel ... , N ₈ . Therefore, the pixel value of the pixel to be interpolated by substituting α to β is α ₁ · N _1. + Α ₂ · N ₂ + Α ₃ · N ₃ + Α ₄ · N ₄ + Α ₅ · N ₅ + Α ₆ · N ₆ + Α ₇ · N ₇ + Α ₈ · N ₈ . N ₄ and N ₅ are located on the scan line to be interpolated, but the average of pixel values of pixels located on the upper scan line and the lower scan line may be obtained.

In addition, when only six edge directions are considered, N ₄ and N ₅ may be excluded from the pixel value calculation of the interpolation target pixel, and thus the pixel of the interpolation target scan line may not be used.

The reference weight may be used as the application weight by multiplying the reference weight by the user's preset threshold value rather than the same value as the application weight.

The interpolator 160 interpolates the interpolation target pixel by using the interpolation method determined by the interpolation method determiner 150. The interpolated pixel is output as a deinterlaced image.

6 is a flowchart of a deinterlacing method according to an embodiment of the present invention.

An interlaced image is input (S10), an operation for calculating a gradient vector with respect to the reference pixel of the interlaced image is performed (S20), and the size of the gradient vector is set to a preset value, that is, The edge of the reference pixel is detected by comparing with the threshold value (S30), and whether or not the interpolation target pixel is an edge according to whether the reference pixel is an edge using the detected edges (S40, S50), and the interpolation target pixel is an edge In this case, an interpolation method using geometric duality is applied, and an interpolation method using a line average is applied if not an edge (S60), and a deinterlaced image is output by interpolating an interpolation target pixel using an interpolation method (S70). ).

Sobel operation of the gradient vector may be performed by applying a 3 × 3 Sobel mask.

In addition, the reference pixel included in the edge map referred to by the interpolation target pixel may be positioned at any one of a vertical upper portion or a vertical lower portion of the interpolation target pixel.

When selecting an interpolation method, if the interpolation target pixel is determined to be an edge, an interpolation method using geometric duality is selected. If the interpolation target pixel is determined to be not an edge, an interpolation method using a line average is selected.

The interpolation method using the line average may be one of an LA interpolation method, an ELA interpolation method, an EELA interpolation method, and a MELA interpolation method.

When the interpolation target pixel is determined as an edge, an interpolation method using geometric duality is a method of interpolating the interpolation target pixel according to Equations (5) and (7) described above.

[Equation (5)]

[α] = ([C] ^T · [C]) ^-1 ([C] ^T · [y])

[Equation (7)]

Here, [α] is a reference weight vector for each direction of the reference pixel which is the reference for interpolation of the interpolation target pixel, [y] is a data vector, and [C] is [{2n X (2n + 1)} X 6] or [{2n X (2n + 1)} X 8], wherein n is any natural number, N is a pixel value of a plurality of pixels adjacent to the pixel to be interpolated, and is Y or Y. Is the pixel value of the interpolation target pixel.

Finally, we compare the performance and processing time of the various deinterlacing algorithms.

7 is a diagram illustrating a comparison between an original image and an image to which a deinterlacing algorithm is applied.

Peak signal to noise ratio (PSNR) and CPU time (processing time), which are widely used as objective elements of comparative image quality, were calculated by applying various intra field deinterlacing algorithms. PSNR is obtained from Mean Squared Error (MSE), and is represented by Equation (8) below, and CPU time represents processing time of deinterlacing.

[Equation (8)]

Here, X _org and Xr _ec represent an original image and a reconstructed image having a size of width X height.

8 shows LA, ELA, EELA, MELA, Direction-Oriented Interpolation (DOI), Low-Complexity Interpolation Deinterlacing (LCID), Bayesian Network based line Interpolation (BNI), Edge Nap Deinterlacing (EMD), Modified Covariance-based The PSNR values and the CPU time of eight test images were obtained by using Adaptive Deinterlacing and Covariance-based Adaptive Deinterlacing using Edge Map algorithms. The MCAD algorithm is an improvement on the CAD algorithm proposed in the present invention. In addition, it is a CADEM algorithm to apply a CAD algorithm only to edges.

Referring to FIG. 8, the average PSNR value obtained using the MCAD algorithm is 33.274, which is higher than the LA, ELA, EELA, and MELA algorithms, but the CPU time is 6.260 long. In the CADEM algorithm, the PSNR value is slightly lower than the PSNR value obtained using the CAD algorithm, but the processing time is shorter than that of the CAD algorithm with a CPU time of 0.630. The processing time is not significantly different from other algorithms such as the LA algorithm. Therefore, it is advantageous to apply a CAD algorithm when high performance is required, and it is advantageous to apply a CADEM algorithm when fast processing time is required with high performance.

The invention can also be embodied as computer readable code on a computer readable recording medium. The computer-readable recording medium includes all kinds of recording devices in which data read by a computer system is stored. For example, there are ROM, RAM, CD, DVD, magnetic tape and the like. The computer readable recording medium can also be distributed over network coupled computer systems so that the computer readable code is executed.

While the present invention has been described in connection with what is presently considered to be practical exemplary embodiments, it is to be understood that the invention is not limited to the disclosed embodiments, but, on the contrary, You will understand. It is therefore to be understood that the above-described embodiments are illustrative in all aspects and not restrictive.

100: deinterlacing device 110: input unit
120: masking application unit 130: edge detection unit
140: edge determination unit 150: interpolation method selection unit
160: interpolator

Claims (13)

A masking applying unit which calculates a magnitude of the gradient vector with respect to the reference pixel of the input interlaced image;
An edge detector for detecting an edge of the reference pixel by comparing the magnitude of the calculated gradient vector with a preset value;
An edge determination unit that determines whether the interpolation target pixel is edged according to the edge of the reference pixel;
An interpolation method selector which selects an interpolation method to be applied to the interpolation target pixel according to an edge of the interpolation target pixel; And
And an interpolation unit configured to interpolate the interpolation target pixel by using the selected interpolation method. The method of claim 1,
The masking applying unit, the de-interlacing apparatus for performing a calculation by applying a 3x3 Sobel mask. The method of claim 1,
And the edge detector is configured to determine the reference pixel as an edge when the magnitude of the gradient vector is larger than the preset value. The method of claim 1,
The reference pixel is positioned in a vertical upper scan line or a vertical lower scan line of the interpolation target pixel;
And the edge determining unit determines the interpolation target pixel as an edge when the reference pixel is an edge, and determines that the interpolation target pixel is not an edge when the reference pixel is not an edge. The method of claim 1,
The interpolation method selector selects an interpolation method using geometric duality when the interpolation target pixel is determined as an edge, and uses line averaging when it is determined that the interpolation target pixel is not an edge. De-interlacing device to select an interpolation method. 6. The method of claim 5,
The interpolation method using the geometric duality is a deinterlacing apparatus for interpolating interpolation target pixels according to the following equations (a) and (b):
[α] = ([C] ^T · [C]) ^-1 ([C] ^T · [y]) (a)

(b)
Here, [α] is a weight vector for each direction of the reference pixel which is a reference for interpolation of the interpolation target pixel, [y] is a data vector, and [C] is [{2n X (2n + 1)} X 6 ] Or [{2n X (2n + 1)} X 8], where n is any natural number, N is a pixel value of a plurality of pixels adjacent to the interpolation target pixel, and Y is 6 Pixel value of the interpolation target pixel. Calculating the magnitude of the gradient vector for the reference pixel of the input interlaced image;
Detecting an edge of the reference pixel by comparing the magnitude of the calculated gradient vector with a preset value;
Determining whether an interpolation target pixel is edged according to whether the reference pixel is edged;
Selecting an interpolation method to be applied to the interpolation target pixel according to an edge of the interpolation target pixel; And
And interpolating the interpolation target pixel using the selected interpolation method. delete delete delete delete delete delete

Images