KR100422575B1 - An Efficient Spatial and Temporal Interpolation system for De-interlacing and its method - Google Patents

Abstract

The spatial / time-axis interpolation system and method for deinterlacing of the present invention improves the processing capability of edges with a gentle slope to provide excellent image quality, removes false motion, and provides excellent image quality, and motion missing. (Motion Missing) Eliminate errors and provide excellent image quality, and classify them into high-motion, normal, low-motion, and still images according to the degree of motion of the input image. The purpose is to provide excellent image quality by applying different thresholds. To this end, in the spatial and time-axis interpolation system for de-interlacing, a motion value is determined by using a difference of pixel values, and a plurality of reference fields are used to classify an input image into a plurality of motion regions according to the magnitude of the motion value. Motion detection means for receiving a pixel value; An edge-based median filter which receives a pixel value from a plurality of reference fields to change a pixel selection according to whether an edge exists and to determine a spatial interpolation pixel value Z according to the selected pixel; A minimum absolute pixel difference filter for receiving a pixel value from a plurality of reference fields to obtain a difference between a plurality of candidate pixels located in a field before and after temporally, and to determine a minimum interpolation pixel value by comparing with a threshold value; And a pixel value including a pixel value output from the edge-based median filter, a pixel value output from the minimum absolute pixel filter, and a pixel value output from the edge-based median filter and minimum absolute pixel filter. It characterized in that it comprises a soft switch for determining whether to use.

Description

An Efficient Spatial and Temporal Interpolation system for De-interlacing and its method

The present invention relates to a spatial / time-axis interpolation system and method for de-interlacing. Specifically, the present invention relates to a progressive scan method in an interlaced scan method that supports various standard formats adopted in digital television and is used for improving image quality. A spatial axis / temporal axis interpolation system and method in an interlaced scanning video to pregressive scanning video conversion (IPC) conversion system.

In general, NTSC (National Television System Committee) television systems play screens using interlaced scanning, while video monitors (VGAs) on computer monitors or digital televisions are screened by progressive scan. Play it. Here, the interlaced scan method is divided into two fields (Field: odd rows, even rows) after dividing one frame (Frame: 30 frames per second in an NTSC color TV screen) and sequentially scanning them. One frame is created, and the progressive scan method creates one frame and scans it on the screen at once. Deinterlacing is the process of converting an interlaced image divided into two fields into one frame before outputting it to the screen. It is easy to convert a progressive scan image into an interlaced scan image, but it is difficult to convert an interlaced scan image into a progressive scan image. However, current digital television receivers support not only interlaced images but also progressive images, so that the monitor needs to be deinterlaced in order to support progressive scanning or convert the interlaced images into other formats.

Various methods of deinterlacing have been proposed so far, and there are largely a motion compensation method and a no motion compensation method. The motion compensation method has a disadvantage in that a large amount of computation, a complicated structure, and an error are transmitted when compensating for an incorrect motion, and the motion non-compensation method are adaptively changed according to an approach and a motion using various spatial filters. Adaptive) method.

Edge-based line average (ELA) method, which is a widely used method to implement de-interlacing, uses the edge-based line average method to extract the edge-to-line average. There are disadvantages such as using sensitive and incorrect edge information.

Motion adaptive is a method that detects the motion of the input image and uses the space- and time-base filters according to the motion. Motion missing and false motion due to faulty motion detection are performed. False Motion) Error occurs. Here, motion missing is an error generated by considering a moving area as a non-moving area, which has a fatal effect on image quality, and a false motion is a motion in an area where there is no motion. Degradation of the image quality caused by considering the area as is caused.

Another important factor in determining the performance of the deinterlacer is the processing of moving diagonal edges, which is a function of smoothly changing edges in the case of the edge-based line averaging method, which is mainly used as a spatial filter. There is a disadvantage in that the picture quality is reduced.

In order to solve the above problems, the present invention provides a spatial axis and time axis interpolation system and method for de-interlacing that can provide excellent image quality by improving the processing capacity of the edge having a gentle slope.

In addition, the present invention provides a spatial axis and time axis interpolation system and method for de-interlacing that can provide excellent image quality by removing the false motion (False Motion).

In addition, the present invention provides a spatial axis and time axis interpolation system and method for de-interlacing that can provide excellent image quality by eliminating motion missing (error).

In addition, according to the present invention, the motion image is classified into a high motion image, a normal image, a low motion image, and a still image according to the degree of motion of the input image. The present invention provides a spatial axis and time axis interpolation system and method for deinterlacing that can provide excellent image quality.

1 is a block diagram of a deinterlacer according to the present invention;

2 is an explanatory diagram of motion extraction from four fields existing in the time axis according to the present invention;

3 is a flowchart for classifying and determining a threshold value of an input image according to the present invention;

4 is a flowchart of subdividing an image obtained by FIG. 3 into a motion area, a background area, and a boundary area;

5 is a flowchart of searching for a three-step optimal pixel value according to the present invention;

FIG. 6 is an exemplary view of selecting a candidate pixel in a three step search process according to the present invention; FIG.

7 is a flowchart for finally extracting the direction of the edge according to the present invention;

8 is an exemplary diagram for determining a candidate value according to the present invention.

* Description of the symbols for the main parts of the drawings *

110: motion detection means 111: motion detection unit

113: maximum detection unit 115: median filter unit

117: threshold determination unit 120: EBMED filter

130: AMPD filter 140: soft switch

In order to achieve the above object, the present invention is to determine the motion value using the difference of the pixel value in the spatial axis and time axis interpolation system for de-interlacing, and to convert the input image into a plurality of motion regions according to the magnitude of the motion value Motion detection means for receiving pixel values from a plurality of reference fields to distinguish them; An edge-based median filter which receives a pixel value from a plurality of reference fields to change a pixel selection according to whether an edge exists and to determine a spatial interpolation pixel value Z according to the selected pixel; A minimum absolute pixel difference filter for receiving a pixel value from a plurality of reference fields to obtain a difference between a plurality of candidate pixels located in a field before and after temporally, and to determine a minimum interpolation pixel value by comparing with a threshold value; And a pixel value including a pixel value output from the edge-based median filter, a pixel value output from the minimum absolute pixel filter, and a pixel value output from the edge-based median filter and minimum absolute pixel filter. It characterized in that it comprises a soft switch for determining whether to use.

In addition, the motion detection means of the present invention, according to the motion value, the input image is divided into four images of a high motion image, a normal image, a low motion image and a still image, respectively, the threshold value of the upper limit and the lower limit, respectively; And a threshold value determination unit for setting.

In addition, the threshold value of the present invention is characterized by being used in a minimum absolute pixel filter and a soft switch.

In addition, the four images of the threshold determination unit of the present invention may be further subdivided into a motion region, a background region, and a boundary region according to the magnitude of the motion value.

In addition, when the pixel value is divided into 256 steps, the threshold value of the motion area is 90 to 110, the threshold value of the background area is 15 to 25, and the threshold value of the boundary area is 3 to 10.

In addition, the motion detecting means of the present invention, in order to detect the motion value of the pixel at a predetermined position. It characterized in that it comprises a motion detector using a.

In addition, the edge-based median filter of the present invention is characterized by using a 7 (row) x 3 (column) window to extract the direction of the edge.

In addition, the edge-based median filter of the present invention, when the edge is present, the average value of the pixel values corresponding to the edge in the current field, the average value of the pixel values located inside the edge in the current field, located outside the edge in the current field The candidate pixel value is determined by using an average value of pixel values, an average value of pixel values located inside and outside an edge in a current field, and an average value of pixel values in two front and back fields adjacent to the current field.

Also, the edge-based median filter of the present invention, when no edge exists, the average value between the upper and lower pixels adjacent to the interpolation target pixel, the upper right adjacent to the interpolation target pixel, the average value between the lower left pixels, and the upper left adjacent to the interpolation target pixel. The candidate pixel value is determined by using an average value between the lower right pixels, an average value between upper right, left upper, lower left and lower right pixels adjacent to the interpolation target pixel, and an average value of pixel values in the front and rear two fields adjacent to the current field.

In addition, the soft switch of the present invention, in order to find the pixel value z to be finally interpolated

- ,

Where α is Im-

It characterized by using.

In addition, in the spatial and temporal axis interpolation method for the deinterlacing of the present invention, a plurality of motions are determined in order to determine a motion value by using a difference of pixel values, and to divide the input image into a plurality of motion areas according to the magnitude of the motion value. A motion detection step of receiving a pixel value from a reference field; An edge-based median filtering step of selecting pixels according to whether edges are present and receiving pixel values from a plurality of reference fields to determine the interpolated pixel value Z according to the selected pixel; A minimum absolute pixel difference filtering step of obtaining a difference between a plurality of candidate pixels located in a field before and after temporally and receiving pixel values from a plurality of reference fields to determine a minimum interpolation minimum pixel value by comparing with a threshold value; And a pixel value of the pixel value output from the edge-based median filter, the pixel value output from the minimum absolute pixel filter, and the pixel value mixed with the pixel value output from the edge-based median filter and the minimum absolute pixel filter. It characterized in that it comprises a soft switching step of determining whether or not to use.

The edge-based median filtering step of the present invention may include extracting edge maps in each direction by applying an operator in a horizontal direction and a vertical direction; And performing median filtering by extracting an edge direction using a 7 (row) x 3 (column) window based on the edge map.

In addition, the minimum absolute pixel filtering step of the present invention may include a three-step searching process, wherein the first step is a step of selecting a minimum pixel by obtaining a difference value between five candidate pixels located in a field before and after in time. In the step, the minimum pixel is calculated for the surrounding 4 candidates based on the minimum pixel obtained in step 1 and compared with a threshold value. When the value is less than the threshold value, the minimum value is determined. If the difference between the pixel values in both fields before the search is smaller than the threshold, the median is applied to the average value of the pixel values of the front and back fields and the upper and lower pixel values in the current field without performing the three-step search. Wanted to be filtered.

In addition, in the spatial and time-axis interpolation system for de-interlacing, a motion value is determined by using a difference of pixel values, and a plurality of motion regions are used to classify the input image into a plurality of motion regions according to the magnitude of the motion value. A motion detection function for receiving a pixel value from a reference field; An edge-based median filtering function that receives a pixel value from a plurality of reference fields to change the pixel selection depending on whether an edge exists and to determine the pixel value Z interpolated according to the selected pixel; A minimum absolute pixel difference filtering function for obtaining a difference between a plurality of candidate pixels located in a field before and after temporally and receiving pixel values from a plurality of reference fields to determine a time interpolation minimum pixel value by comparing with a threshold value; And a pixel value of the pixel value output from the edge-based median filter, the pixel value output from the minimum absolute pixel filter, and the pixel value mixed with the pixel value output from the edge-based median filter and the minimum absolute pixel filter. And a computer readable recording medium having recorded thereon a program for realizing a soft switching function for determining whether or not to use the "

The above objects, features and advantages will become more apparent from the following detailed description taken in conjunction with the accompanying drawings. Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings.

1 is a block diagram of a deinterlacer according to the present invention.

The deinterlacer according to the present invention detects motion and obtains motion information, motion detection means 110, an EBMED filter 120 for interpolating motion information in a spatial domain, an AMPD filter 130 for interpolating motion information in a time domain, and Software for determining whether to use the interpolated value using the EBMED filter 120, the interpolated value using the AMPD filter 130, or to use the two values mixed with the pixel values obtained through the above means. Switch 140.

When the motion detector 111 obtains a difference between each pixel value to determine a motion region using pixel values input from three reference fields, the maximum detector 113 uses a difference of three candidate pixels. The maximum pixel difference is set as a motion value to determine a motion region, and the median filter 115 is passed to remove noise.

Here, the threshold determination unit 117 first grasps the degree of motion of the input image and classifies it into four images by using the obtained motion value. Then, according to the degree of motion, the absolute minimum pixel difference (AMPD) is determined. ) A threshold to be used in filter 130 and a threshold to be used in soft switch 140. In addition, the threshold value determination unit 117 divides the input image into a moving image, a normal image, a moving image, and a still image to set respective upper and lower limits.

The soft switch 140 uses this value to use the pixel value passed through the edge based median (EBMED) filter 120, the pixel value passed through the AMPD filter 130, or the EBMED filter. It is determined whether the pixel value passed through 120 and the pixel value passed through AMPD filter 130 are mixed.

EBMED filter is used for efficient interpolation of moving diagonal edge, and AMPD filter is used to determine the pixel value by considering surrounding pixel values in three steps to eliminate motion missing. In this case, since the motion detection unit 111 uses a brightness difference, false motion does not occur.

Each block will be described in more detail as follows.

Motion detection means (110)

One of the most important factors affecting the performance of deinterlacing in the motion non-compensation method in which the motion detection unit 110 determines the threshold value through motion detection is accurate motion detection. Here, the motion vector is obtained using the brightness difference of the pixels from the three reference fields around the time axis.

2 is an explanatory diagram for extracting motion from four fields existing in the time axis according to the present invention.

Equation 1 is used to obtain a motion value of the pixel at the 'z' position.

Equation 1 has a disadvantage in that minute motion in a specific portion is not detected when a motion value is obtained using a brightness difference. However, in this case, the motion missing occurring in this specific part is compensated by the AMPD filter 130, so that the degradation of the image quality does not occur.

Using the obtained motion information, the input image is classified into four types of a large motion image, a normal image, a low motion image, and a still image, and the image thus obtained is again divided into regions with motions, background regions, and boundary regions. Subdivide into.

3 is a flowchart for classifying an input image and determining a threshold according to the present invention.

When the maximum brightness difference is calculated (301) and the motion value is determined (303), it is determined whether the current motion value is larger than the reference value (305). If the current movement value is less than the reference value, If it is larger (309), if it is small, it is recognized as a still image (317). If it is large, the current motion value is It is determined whether it is greater (311). At this time, the current movement value If larger, it is recognized as a normal image (313), if smaller, it is recognized as an image with less movement (315).

FIG. 4 is a flowchart for subdividing the image obtained by FIG. 3 into a motion region, a background region, and a boundary region.

When motion information of the image obtained by FIG. 3 is input (401), a region with movement and a region without movement are determined (403), and the region with movement 405 is set to a threshold value (100), Peripheral pixels 409 in the motion area have a threshold value of 5 (411), and the motionless area 413 has a threshold value of 20 (415).

The threshold value used here is a value obtained through simulation, and when the pixel value is divided into 256 levels, the threshold value of the motion area is 90 to 110, the threshold value of the background area is 15 to 25, and the threshold value of the boundary area. The value is preferably 3 to 10, but the value used in FIG. 4 is the best value.

In the case of moving images, more information about the pixel values to be interpolated exists in the spatial domain, so the interpolation focuses on the spatial filter, and in the case of still images, more pixel information exists in the time domain. Interpolate with emphasis.

In other words, the input image is divided into a motion region, a background region, and a boundary region, and the threshold value used for each of them is changed to increase the computational efficiency. Since the motion area uses a spatial filter, there is no need to perform a three-step search, so simple calculations are given with high thresholds, and still areas also have relatively high thresholds because there is little chance of motion missing. Determine the pixel value. However, in the case of the boundary part, motion missing can occur, which gives a low threshold.

AMPD Filter 130 for Temporal Interpolation

If accurate motion information is extracted, inter-field interpolation is used in the still area of the image, so even if the average of the front and back fields in the same position as the pixel to be interpolated can obtain a good result, When not present, motion missing errors will occur, resulting in severe degradation of picture quality. The motion detector 111 extracts motion by using the brightness difference from four fields including three reference fields and one current candidate field. In this case, fine motion in a specific portion cannot be detected. In order to overcome this problem, a motion detection method using a hardness difference has been proposed, but this method has a disadvantage of being sensitive to noise. Therefore, the present invention goes through a three-step search process to eliminate such motion missing errors.

FIG. 5 is a flowchart of searching for a three-step optimal pixel value according to the present invention, and will be described with reference to an embodiment of selecting a candidate pixel in the three-step search process according to the present invention of FIG. 6.

The shaded pixel in the i-th field of Fig. 6A is the pixel to be interpolated, and the absolute value of the pixel value of the previous field ((i-1) th field) and the next field ((i + 1) th field) at the same position. Is less than the threshold (501), the median filter is applied to the obtained three values (3 taps), such as the previous field pixel value, the next field pixel value, and the average value of the upper and lower pixel values in the current field, without performing a three-step search. (503).

On the other hand, if the absolute value of the pixel value of the previous field ((i-1) th field) and the next field ((i + 1) th field) at the same position is larger than the threshold value, the previous field ((i-1) of FIG. The difference value (five) is obtained between the second field) and the next field (the (i + 1) th field), and a rough search for selecting the pixel having the smallest value is performed (505). As shown in FIG. 6C, when the leftmost pixel in the previous field ((i-1) th field) is selected as the smallest value (507), four pixels close to the position and the next field ((i + 1) are selected. ) And the smallest pixel among the (th) field) and the difference value is selected (509).

If the selected pixel value is smaller than the threshold value (511), the value is selected (513). If the selected pixel value is larger than the threshold value, the difference between the four pixels around the minimum absolute pixel of FIG. The smallest value is selected (515).

That is, the minimum value is selected by obtaining a difference value for five candidate pixels located in the front and back fields in time (step 1). The minimum pixel difference value is obtained for the surrounding 4 candidates based on the pixel position obtained in step 1, and the minimum pixel value is selected when the value is smaller than the threshold value (step 2). If it is larger than the threshold value, the search is performed again to select the minimum pixel value (step 3).

EBMED Filter 120 for Spacial Interpolation

One of the major issues that affect the performance of deinterlacing algorithms is the handling of moving diagonal edges.

This is divided into two phases: first, using the 3 * 3 sobel operator to extract the edge map, and then applying the direction rules to improve the deinterlacing of the slow-moving diagonal edges. Select the 7 * 3 window.

7 is a flowchart for finally extracting the direction of the edge according to the present invention.

In order to find the directionality, the operator is applied in the horizontal direction and the vertical direction, respectively (701) to extract the horizontal edge map and the vertical edge map (703). This means that a typical 3 * 3 window can extract edge maps in the 45, 0, -45 degree directions, while using 7 * 3 windows in the 45, 30, 15, 0, -15, -30, -45 degree directions. Excellent performance against gentle edges.

The median filter is taken based on the obtained edge map (705). First, the directionality is detected within the 7 * 3 window based on the edge map. The division of directionality is classified into seven of 45, 30, 15, 0, -15, -30, and -45 degrees. If the directionality is not detected or zero degrees, it is replaced by the average value of the upper and lower pixels in the field. However, in the remaining six directions, five candidate values are calculated and passed through the 5-tap median filter. Through this process, the direction is determined (707).

FIG. 8 is an exemplary diagram for determining a candidate value according to the present invention, and FIG. 8 (A) shows a method for determining a candidate value X1 in the presence of a 45 degree edge and in the presence of a 30 degree edge when no edge exists. 8 (B) shows a method for determining a candidate value X2 in the absence of an edge and in the presence of a 45 degree edge and in the presence of a 30 degree edge, and FIG. 8 (C) shows a method for determining an edge value when the edge does not exist. Fig. 8 (D) shows a method for determining candidate value X4 when there is no edge, when there is no edge, and when there is a 45 degree edge and when there is a 30 degree edge. 8 (E) shows a method for determining the candidate value X5 in the absence of an edge, in the presence of a 45 degree edge and in the presence of a 30 degree edge.

First, the five candidate values are described as an example in which an edge exists. First, it is an average value of pixel values corresponding to the edge in the current field (Fig. 8 (A)), and second, it is an average value of pixel values located inside the edge in the current field (Fig. 8 (B)). The average value of the pixel values located outside the edge (Fig. 8 (C)), and the fourth, the average value of the pixel values located inside and outside the edge in the current field (Fig. 8 (D)). It is an average value of pixel values in two fields (Fig. 8 (E)).

Even when no edge exists, the candidate value is determined as shown on the left side of FIG. First, the average value is taken between the upper and lower pixels adjacent to the interpolation target pixel (Fig. 8 (A)), and second, the average value is taken between the upper right and lower left pixels adjacent to the interpolation target pixel (Fig. 8 (B)). The average value is taken between the upper left and right lower pixels adjacent to the target pixel (Fig. 8 (C)), and the fourth is the average value between the upper right, upper left, lower left and lower right pixels adjacent to the interpolation target pixel (Fig. 8 (D)). Is an average value of pixel values in the front and rear two fields adjacent to the current field (Fig. 8 (E)).

As the 5-tap median filter is applied to the selected pixel, the pixel value of 'z' interpolated on the spatial axis is determined by Equation 2 below.

Here, FIR stands for Finite Impulse Resonse filter.

Soft switch (140)

The soft switch 140 uses the interpolated pixel value using the EBMED filter 120 or the interpolated value using the AMPD filter 130 as the final pixel value for the pixel values obtained through the above methods. Or decide whether to mix the two values.

Equation 3 below is a coefficient for determining this Equation 4 is an equation for determining the 'z' pixel value to be finally interpolated according to the present invention.

Here, "Med" indicates a value passed through the median filter unit 115 to remove noise with respect to the maximum motion value obtained by the maximum value detector 113, and "H" and "L" indicate an upper limit value and a lower limit value, respectively.

The upper and lower threshold values used herein depend on the type of image. As described above in the threshold determining means 110, the input image is divided into a moving image, a normal image, a moving image, and a still image, and a different threshold value is determined for each to improve performance.

The present invention described above is capable of various substitutions, modifications, and changes without departing from the spirit and scope of the present invention to those skilled in the art to which the present invention pertains. It is not limited.

According to the configuration of the present invention as described above, the direction component of the edge can be accurately obtained, and it shows good performance even with a diagonal edge with false motion and movement. In addition, the present invention can remove the motion missing error by dividing the input image into a moving region, a background region, and a boundary region and interpolating an accurate pixel value in three steps. Finally, the present invention can finally improve the image quality by classifying the input image into a moving image, a normal image, a moving image, and a still image and applying different threshold values.

Claims (22)

In the spatial and time-axis interpolation system for de-interlacing, Motion detection means for determining a motion value by using a difference of pixel values, and receiving pixel values from a plurality of reference fields in order to classify the input image into a plurality of motion regions according to the magnitude of the motion value; An edge-based median filter that varies pixel selection depending on whether an edge exists and receives pixel values from a plurality of reference fields to determine a spatial interpolation pixel value Z for the selected pixel; A minimum absolute pixel difference filter for receiving a pixel value from a plurality of reference fields to obtain a difference between a plurality of candidate pixels located in a field before and after temporally, and to determine a minimum interpolation pixel value by comparing with a threshold value; And Any pixel value is used which is a mixed pixel value output from the edge-based median filter, a pixel value output from the minimum absolute pixel filter, and a pixel value output from the edge-based median filter and minimum absolute pixel filter. Includes a soft switch for determining whether to The motion detection unit determines a threshold value for setting an upper limit and a lower limit threshold value by dividing the input image into four images of a high motion image, a normal image, a low motion image, and a still image according to the motion value. Includes wealth, The four images of the threshold value determination unit are subdivided into motion regions, background regions, and boundary regions according to the magnitude of the motion values, respectively. delete The method of claim 1, wherein the threshold value, Spatial and temporal axis interpolation system for deinterlacing characterized in that it is used in the minimum absolute pixel filter and soft switch. delete The method of claim 3, When dividing the pixel value into 256 steps, the spatial axis for the deinterlacing, characterized in that the threshold of the motion region is 90 to 110, the threshold of the background region is 15 to 25, and the threshold of the boundary region is 3 to 10. Time base interpolation system. The method of claim 1, wherein the motion detecting means, Equation for detecting a motion value of a pixel at a predetermined position - - Space and time axis interpolation system for a de-interlacing, characterized in that it comprises a motion detection unit using. delete The edge-based median filter of claim 1, wherein the edge-based median filter uses a 7 (row) x 3 (column) window to extract an edge direction. If there is an edge, the average value of the pixel values corresponding to the edge in the current field, the average value of the pixel values located inside the edge in the current field, the average value of the pixel values outside the edge in the current field, the inside edge of the current field and 12. A spatial and temporal axis interpolation system for deinterlacing, wherein the candidate pixel value is determined using an average value of pixel values located outside and an average value of pixel values in two front and rear fields adjacent to the current field. The edge-based median filter of claim 1, wherein the edge-based median filter uses a 7 (row) x 3 (column) window to extract an edge direction. If no edge exists, the average value between the upper and lower pixels adjacent to the interpolation target pixel, the upper right adjacent to the interpolation target pixel, the average value between the lower left pixel, the upper left adjacent to the interpolation target pixel, the average value between the lower right pixel, and the upper right adjacent to the interpolation target pixel And a candidate pixel value is determined using an average value between upper left, lower left and right lower pixels, and an average value of pixel values in front and rear two fields adjacent to the current field. The method of claim 1, wherein the soft switch, To obtain the final interpolated pixel value z - , Where α is Im- Space- and time-axis interpolation system for de-interlacing, characterized in that using. In the spatial axis and time axis interpolation method for deinterlacing, A motion detection step of determining a motion value by using a difference of pixel values, and receiving pixel values from a plurality of reference fields in order to classify the input image into a plurality of motion regions according to the magnitude of the motion value; An edge-based median filtering step of selecting pixels according to whether edges are present and receiving pixel values from a plurality of reference fields to determine the interpolated pixel value Z according to the selected pixel; A minimum absolute pixel difference filtering step of obtaining a difference between a plurality of candidate pixels located in a field before and after temporally and receiving pixel values from a plurality of reference fields to determine a minimum interpolation pixel value by comparing with a threshold value; And The pixel value of the pixel value output from the edge-based median filter, the pixel value output from the minimum absolute pixel filtering filter, and the pixel value output from the edge-based median filter and the minimum absolute pixel filtering filter are mixed. A soft switching step of determining whether to use, The motion detecting step may include setting a threshold value by dividing the input image into four images of a high motion image, a normal image, a low motion image, and a still image according to the motion value. The four images are subdivided into motion regions, background regions, and boundary regions according to the magnitudes of the motion values, respectively. delete The method of claim 11, wherein the threshold value, Space- and time-axis interpolation method for de-interlacing characterized in that it is used in the least absolute pixel filtering step and the soft switching step. delete The method of claim 13, When dividing the pixel value into 256 steps, the spatial axis for the deinterlacing, characterized in that the threshold of the motion region is 90 to 110, the threshold of the background region is 15 to 25, and the threshold of the boundary region is 3 to 10. Time base interpolation method. The method of claim 11, wherein the detecting of the motion comprises: Equation for detecting a motion value of a pixel at a predetermined position - - A spatial axis and time axis interpolation method for de-interlacing, characterized in that using. delete The method of claim 11, The edge-based median filtering step includes extracting edge maps in each direction by applying an operator in a horizontal direction and a vertical direction, and using a 7 (row) x 3 (column) window based on the edge map. Extracting a direction to perform median filtering, In the edge-based median filtering step, when an edge exists, an average value of pixel values corresponding to an edge in the current field, an average value of pixel values located inside an edge in the current field, and a value of pixel values outside an edge in the current field The spatial and temporal axis interpolation for deinterlacing comprises determining a candidate pixel value using an average value, an average value of pixel values located inside and outside an edge in a current field, and an average value of pixel values in two front and back fields adjacent to the current field. Way. The method of claim 11, The edge-based median filtering step includes extracting edge maps in each direction by applying an operator in a horizontal direction and a vertical direction, and using a 7 (row) x 3 (column) window based on the edge map. Extracting a direction to perform median filtering, In the edge-based median filtering step, when no edge exists, an average value between upper and lower pixels adjacent to an interpolation target pixel, an upper right adjacent to an interpolation target pixel, an average value between lower left pixels, and an upper left and lower right pixel adjacent to an interpolation target pixel A spatial axis and a time axis for deinterlacing, wherein the candidate pixel value is determined using an average value, an average value between upper right, left upper, lower left, and lower right pixels adjacent to the interpolation target pixel, and an average value of pixel values in the front and rear two fields adjacent to the current field. Interpolation method. The method of claim 11, wherein the soft switching step, Equation - , Where α is Im- The spatial and temporal axis interpolation method for deinterlacing, wherein the pixel value z is finally interpolated. The method of claim 11, wherein the minimum absolute pixel filtering step is Three-stage search process-The first step is to select the minimum pixel by obtaining the difference value between five candidate pixels located in the before and after fields temporally, and the second step is based on the surrounding pixels based on the minimum pixel obtained in the first step. Obtaining the minimum pixel for the candidate and comparing it with the threshold value, and determining the smallest minimum value when less than the threshold value, the third step is performed when the value is larger than the threshold value, the pixels present in both fields before searching The median filtering is performed on the average value of the pixel values of the front and rear fields and the upper and lower pixel values of the current field without performing a three-step search when the difference between the values is smaller than the threshold value. Space and time axis interpolation method. delete