KR20060104817A - Interpolation filter and interpolation filtering method for video deinterlacing - Google Patents

Abstract

The present invention relates to an interpolation filter and an interpolation filtering method for deinterlacing an image signal.

In the motion adaptive interpolation based deinterlacing apparatus according to the present invention, an image deinterlacing interpolation filter includes a pixel storage unit 110 for receiving pixel data of upper and lower lines adjacent to each other and storing pixel data of a predetermined region, and the stored pixels. A correlation calculation unit 120 that receives data and calculates correlations between pixels positioned in diagonal directions of the upper and lower lines, and interpolates the pixel region detected as the maximum correlation by comparing the calculated correlation values. An interpolation direction determiner 130 for determining a direction, an interpolation direction storage unit 140 for storing the determined interpolation directions, and an interpolation direction for determining a final interpolation direction by receiving an interpolation direction of a current pixel and a previous pixel interpolation direction. After determining a candidate pixel to be interpolated from the position correction unit 150 and the corresponding position pixel of the pixel storage unit 110 according to the final interpolation direction. It is configured to include the pixel determination unit 160, the final interpolation (170) for performing a final interpolation output to the determined candidate pixels.

Image Conversion, De-Interlacing, Interlacing, Progressive, Interpolation Filter

Description

Image deinterlacing interpolation filtering device and interpolation filtering method {INTERPOLATION FILTER AND INTERPOLATION FILTERING METHOD FOR VIDEO DEINTERLACING}

1 shows an outline of an interlaced signal;

2 shows an outline of deinterlacing.

3 shows an outline of a line repetition method for deinterlacing.

4 is a diagram illustrating an outline of a conventional interfield interpolation method without motion compensation.

5 is a diagram illustrating an outline of a conventional interfield interpolation method.

6 is a diagram showing the configuration of an image deinterlacing interpolation filter of the present invention.

7 is a diagram showing an example of window area setting in the present invention;

8 is a view showing the outline of the correlation with the sub-window area in the present invention

9 is a view showing the outline of the correlation with the sub-window area in the present invention

10 is a flowchart of a method for determining an interpolation direction in the present invention.

11 is a view showing an outline of a candidate pixel determination method in the present invention.

12 shows an outline of a candidate pixel determination method in the present invention.

<Explanation of symbols for the main parts of the drawings>

100: interpolation filter 110: pixel storage unit

120: pixel correlation calculation unit 130: interpolation direction determination unit

140: interpolation direction storage unit 150: interpolation direction correction unit

160: candidate pixel determiner 170: final interpolator

The present invention relates to an interpolation filter and an interpolation filtering method for deinterlacing an image signal, and in particular, to improve the performance of an intrafield interpolation filter in a deinterlacing device using a motion adaptive interpolation method. An interpolation filter and its filtering method.

A deinterlacing apparatus is used as a means for converting an interlace screen input in real time into a progressive screen.

For the interlaced video signal, if NTSC is taken as an example, two frames are divided into 1/60 seconds by dividing a screen of one frame at 1/30 second intervals into an odd field and an even field as shown in FIG. It is a signal type that shows alternately by time interval. That is, in displaying one image frame F (n), an odd field image f (n_odd) composed of odd lines of the image frame F (n) and an even field image f (n_even) composed of even lines The two fields are alternately shown at 1/60 seconds (field frequency). In the same way, in displaying the next image frame F (n + 1), an odd field image f (n + 1_odd) consisting of odd-numbered lines of the image frame F (n + 1) and an even-numbered line The two fields are alternately displayed at 1/60 seconds (field frequency) by dividing each of the even field images f (n + 1_even).

 The interlaced signal is converted into a progressive video signal as shown in FIG. 2 through a separate conversion device such as a deinterlacer. That is, the interlaced odd field image f_odd and the even field image f_even are converted into progressive image frames F (n), F (n + 1), ... using a deinterlacer 10. .

As such, converting the interlaced video signal into a progressive video signal through a deinterlacer depends on the display method and characteristics of the video display apparatus. That is, according to the image display apparatus, it is required for an apparatus for implementing a screen by processing a progressive image without processing an interlaced image. For example, in the case of a computer monitor, an image is displayed using a progressive method. In this case, a means for deinterlacing an interlaced video signal and converting the video signal to a progressive video signal is required.

In particular, a deinterlacing apparatus for converting an interlaced video signal into a progressive video signal in such a display device in order to process the video signal in a display device that receives an interlaced video signal and processes it as a progressive video signal includes: It is necessary.

A variety of techniques for converting interlaced video signals into progressive video signals have been proposed. Techniques for converting interlaced video signals to progressive video signals include line repetition, inter-field interpolation without motion compensation, and intra-field interpolation. ).

3 schematically shows a line repetition method. The line repetition method is a method of constructing a frame by simply repeating line information of a current field. As shown in FIG. 3, the method corresponds to a method of constructing a frame image of a total of 2N lines by repeating each line once more with respect to an image of N lines. However, the line repetition method has advantages such as simplicity and convenience of implementation. On the other hand, since a new frame image is formed by simple repetition of the same line, the image quality drops sharply after interpolation.

4 schematically shows a method for interfield interpolation without motion. The interfield interpolation method without motion corresponds to a method of making a progressive frame image by simply inserting image data obtained by dividing the previous field and the next field line into two lines between the current field lines. That is, as shown in Fig. 4, for each of the field images successive to the even field Even1, the odd field Odd1, the even field Evne2, and so on, the previous field (Odd1) is referred to. The line corresponding to 1/2 of the sum of the nth line of Even1) and the nth line of the next field Even2, that is, Interpolation Line (m) = [Previous Field Line (n) + Post Field Line (n)] / A frame image is constructed by inserting a value of 2 into an interpolation line (m) for interpolation between lines of the current frame image.

However, the motion-free interpolation method is easy to implement and has a simple advantage, whereas the interpolation of images with inter-field motions causes a sharp deterioration of image quality as in the line repetition method.

Fig. 5 shows schematically an intrafield interpolation method. In-field interpolation is a method of completing one image frame by using an interpolation value of data obtained by dividing data between two lines in an area between two lines. That is, as shown in Fig. 5, a frame image composed of 2N lines is constructed by using 1/2 value of the sum of the adjacent lines for the N-line field image as the interpolation line value. For example, a method of constructing a progressive video frame by setting (1Line + 2Line) / 2, which is 1/2 of the sum of the first line and the second line, to the interpolation line value between the first line and the second line. to be.

However, in-field interpolation is also easy to implement and has a clearer image quality than line repetition, whereas in case of a tilted image, the step-edge phenomenon is prominent after interpolation. In the case of an image, the flicker phenomenon that the screen shakes after interpolation is also worsened, resulting in deterioration of image quality.

Each of the deinterlacing methods described so far has advantages and disadvantages, and a technique for performing adaptive interpolation by compensating them is also proposed. Motion-compensated interpolation corresponds to this, and also has been proposed an adaptive motion-compensated interpolation to improve the disadvantage of the motion-compensated interpolation.

The former method, that is, motion compensation interpolation, is a method of dividing a screen into units of blocks, estimating a motion vector for each block, and implementing one frame with reference to the estimated motion vector. However, while the motion compensation interpolation method improves image quality after interpolation, the motion compensation interpolation method has a disadvantage in that it has a considerable level of complexity for using the block division and the block vector motion vector estimation. In addition, when the motion vector is incorrectly obtained, the influence on the deterioration of image quality is very large, and thus, an error in interpolation and a decrease in reliability thereof become a significant problem.

The latter method, that is, the motion adaptive interpolation method, adaptively selects the inter-field interpolation method and the intra-field interpolation method, which interpolates the field data before and after the current field into two values according to the presence or absence of the motion of the image. That's how. In such a motion adaptive interpolation method, in particular, the performance and reliability of the intrafield interpolation filter for applying the intrafield interpolation method are important factors.

Therefore, there is a need for an optimized design of an intrafield interpolation filter for using a motion adaptive interpolation method, a filtering method that can minimize errors in image quality, and is easy to implement.

The performance of the conventional interpolation filters in several fields shows that the stepped noise phenomenon is prominent due to simple vertical interpolation at edges with gentle slopes or edges with large movements. Because it is too sensitive to the noise of, it has the problem of interpolating the remaining wrong edges.

In the deinterlacing method used to convert an interlaced screen input in real time into a progressive screen, an object of the present invention is to prevent image quality deterioration and reduce sensitivity to noise of an input pixel based on a motion adaptive interpolation method. The present invention provides an image deinterlacing interpolation filter and an interpolation filtering method for improving the quality of a converted image.

It is still another object of the present invention to provide an interlaced screen that is input in real time to a progressive screen, which is one of interpolation methods in a field. The present invention provides an image deinterlacing interpolation filter and an interpolation filtering method that can reduce sensitivity to noise and improve the quality of a converted image by performing interpolation using correlation between two line data.

Another object of the present invention is the most important in the de-interlacing apparatus using a motion adaptive interpolation method which is not sensitive to input noise and improves image quality after interpolation even for a moderately sloped image by using correlation of up and down line data. This is to improve the performance of inter-field interpolation filter.

Another object of the present invention is less susceptible to noise of input pixels than conventional line repetition methods or in-field interpolation methods without motion compensation, and interpolates diagonally instead of vertically even at edges with high gradients. The present invention provides an image deinterlacing interpolation filter and an interpolation filtering method that can improve the image quality of a converted image by preventing a sudden change of slope.

An image deinterlacing interpolation filtering apparatus of the present invention for achieving the above object is a deinterlacing apparatus for converting an interlaced video signal into a progressive video signal, the diagonal inter-correlation between the pixel data of the adjacent upper and lower lines And interpolation direction determining means for determining an interpolation direction using the extracted correlation values, and interpolation value output means for outputting an interpolation value using pixel data of a position corresponding to the determined interpolation direction. Characterized in that made.

In the image deinterlacing interpolation filtering apparatus of the present invention, the correlation extracting means may include pixel storage means for storing pixel values corresponding to a predetermined window area, and diagonal lines between adjacent upper and lower lines based on the window area. And correlation calculation means for calculating the direction correlation.

In the image deinterlacing interpolation filtering apparatus of the present invention, the interpolation direction determining unit may determine a direction having a maximum correlation value in a diagonal direction between adjacent upper and lower lines as an interpolation direction.

In the image deinterlacing interpolation filtering apparatus of the present invention, the interpolation direction determining means includes interpolation direction storage means for storing the interpolation direction, and interpolation direction correction means for correcting the interpolation direction by using the previous interpolation direction and the current interpolation direction. It is characterized by.

In the image de-interlacing interpolation filtering apparatus of the present invention, the interpolation value output means may include candidate pixel determination means for determining a corresponding pixel in each of the upper and lower lines of the determined interpolation direction as candidate pixels, and the determined candidate pixel values. And final interpolation means for generating an interpolation pixel value by use.

In addition, the video deinterlacing interpolation filtering method of the present invention for achieving the above object, in a deinterlacing method for converting an interlaced video signal into a progressive video signal, a window area of a predetermined size is set on each of the adjacent upper and lower lines A correlation calculation step of obtaining correlations between pixels positioned diagonally to each other based on the window area; An interpolation direction determining step of determining the window area having the largest correlation as the interpolation direction; Outputting a final interpolation value using the pixel value using a pixel in the window area determined in the interpolation direction as a candidate pixel; Characterized in that comprises a.

In the image deinterlacing interpolation filtering method of the present invention, the window area has a size of 5 pixels.

In the image deinterlacing interpolation filtering method of the present invention, a pixel located in the center of the window region is selected as a candidate pixel.

Also, in the image deinterlacing interpolation filtering method of the present invention, the correlation is a correlation value of a sum of differences of pixel values in a relative window region positioned diagonally to each other while moving the set window region in a right direction and a left direction. It is characterized by calculating.

In the image deinterlacing interpolation filtering method of the present invention, a direction corresponding to a window area having the largest correlation among the correlation values between pixels positioned in the diagonal direction is determined as the interpolation direction.

In the image deinterlacing interpolation filtering method of the present invention, a plurality of pixels positioned before and after the temporal direction are selected with respect to the determined interpolation direction, and the interpolation directions of the pixels are monitored to correct for a sudden change in the interpolation direction. Characterized in that it further comprises the step.

In the image deinterlacing interpolation filtering method of the present invention, the interpolation value is determined as a sum of two minutes of the sum of the candidate pixels.

Hereinafter, with reference to the accompanying drawings will be described an embodiment of the present invention.

A feature of the present invention is a method of interpolating a pixel value located on a diagonal line between the upper and lower sides of a maximum correlation by receiving the upper and lower adjacent line data of the current field and calculating the correlation between the pixels located in a predetermined region. to be.

In the motion adaptive interpolation based deinterlacing apparatus according to the present invention, the configuration of the intrafield interpolation filter 100 is shown in FIG. As illustrated in FIG. 6, the image deinterlacing interpolation filter according to the present invention receives a pixel data of upper and lower lines adjacent to each other and stores pixel data of a predetermined region, and inputs the stored pixel data. A correlation calculation unit 120 that calculates a correlation between pixels positioned in diagonal directions of the upper and lower lines with each other, and compares the calculated correlation value to determine an interpolation direction with the pixel region detected as the maximum correlation. An interpolation direction determiner 130, an interpolation direction storage unit 140 storing the determined interpolation directions, and an interpolation direction correction unit configured to determine a final interpolation direction by receiving an interpolation direction of a current pixel and a previous pixel interpolation direction. A candidate pixel determination unit 160 that determines candidate pixels to be interpolated from corresponding position pixels of the pixel storage unit 110 according to the final interpolation direction, and And a final interpolator 170 for performing final interpolation on the determined candidate pixels.

The interpolation filtering operation by the interpolation filter configured as described above is performed as follows.

First, an appropriately sized window area is set around the pixel at the current position. In this case, the window area may be arbitrarily set, and the number of pixels to enter the window area is set to an odd number. 7 shows an example of the window area. The window area of size 17 (pixel) is shown for the upper line (I-1 LINE) and the lower line (I + 1 LINE). The pixel storage unit 110 stores the upper and lower line data as much as the size of the window area. The pixel data stored in the pixel storage unit 110 is used by the pixel correlation calculator 120, the interpolation direction corrector 150, and the candidate pixel determiner 160.

The pixel correlation calculator 120 calculates a correlation value between pixels in the set window area. The correlation value between pixels calculates the correlation between pixels located in diagonal directions of the upper and lower line data, and designates a sub window area of an appropriate size within 17 pixels, which is the total window size of each other. The correlation is obtained by moving this subwindow diagonally between the upper and lower lines. For example, the size of the sub window may be set to 5 (pixels), and FIGS. 8 and 9 illustrate correlations between pixels by moving the sub window area and the sub window area diagonally between the upper and lower lines. It shows how.

That is, to obtain the correlation in the right direction, the sub-window of the upper line (I-1 LINE) moves to the right, and the sub-window of the lower line (I + 1 LINE) moves to the left, and the pixels placed in diagonal directions with each other. Calculate the correlation.

In general, the correlation between pixels is obtained by the sum of products of correlated values. The larger the value, the greater the correlation between the pixels. The present invention is based on this, but considering the convenience of hardware implementation, the correlation is calculated by the sum of the difference of the pixel values corresponding to each other in the diagonal direction between the upper and lower lines, the smaller the value is the corresponding It is determined that the correlation between the pixels is large.

To this end, a difference between pixels positioned diagonally to each other in the sub window areas of the upper and lower lines is obtained, and the difference values are summed to obtain a correlation corresponding to each sub window area.

8 shows a method of obtaining correlation of the first sub-window region in the right direction. A sub-window of size 5 (pixel) is applied to the pixel position J as a reference, and the first sub-window region in the right direction from the upper line (I-1 LINE) and the lower line (I + 1 LINE) corresponding thereto is applied. The difference between the pixels in the diagonal direction of the first sub-window regions in the left direction is obtained and the sum of the difference values is shown.

If this is expressed as an equation, it is expressed as Equation 1 below.

That is, considering the size of the sub window area, when K = 0 to 4 (5 pixels), and when K = 0, the pixel p (I-1, J-1) of the upper line I-1 and the corresponding diagonal The difference value with the pixel P (I + 1, J-3) of the lower line I + 1 placed in the directional position is obtained, and the pixel p (I-1, J) of the upper line I-1 when K = 1. And the difference value between the pixel P (I + 1, J-2) of the lower line I + 1 at the diagonal position corresponding to this, and the pixel p (of the upper line I-1) when K = 2. The difference between I-1, J + 1 and the pixel P (I + 1, J-1) of the lower line (I + 1) at a diagonal position corresponding thereto is obtained, and when K = 3, the upper line ( The difference value between the pixel p (I-1, J + 2) of I-1 and the pixel P (I + 1, J) of the lower line (I + 1) at a diagonal position corresponding thereto is obtained, and K = When 4, the pixel p (I-1, J + 3) of the upper line I-1 and the pixel P (I + 1, J + 1) of the lower line I + 1 placed at diagonal positions corresponding thereto. And the difference between all the difference values It is to obtain the correlation between the first one of the sub-area to the right direction.

9 shows a method for obtaining correlations in the sixth sub-window region in the right direction. A sub-window of size 5 (pixel) is applied to the reference pixel position J, and the sixth sub-window area in the right direction from the upper line (I-1 LINE) and the lower line (I + 1 LINE) corresponding thereto. It shows how to calculate the difference between the pixels placed in the diagonal direction of the sub-window area six times in the left direction and sum the difference values.

If this is expressed as an equation, Equation 2 is obtained.

That is, considering the size of the sub window area, when K = 0 to 4 (5 pixels), and when K = 0, the pixel p (I-1, J + 4) of the upper line I-1 and the corresponding diagonal The difference value with the pixel P (I + 1, J-8) of the lower line I + 1 placed in the directional position is obtained. When K = 1, the pixel p (I-1, J + of the upper line I-1 is obtained. 5) and the difference value between the pixel P (I + 1, J-7) of the lower line I + 1 at the diagonal position corresponding to this, and the pixel of the upper line I-1 when K = 2. Find the difference between p (I-1, J + 6) and the pixel P (I + 1, J-6) of the lower line (I + 1) at a diagonal position corresponding to this, and obtain the upper side when K = 3. The difference value between the pixel p (I-1, J + 7) of the line I-1 and the pixel P (I + 1, J-5) of the lower line I + 1 placed at a diagonal position corresponding thereto The pixel p (I-1, J + 8) of the upper line I-1 and the pixel P (I + 1) of the lower line I + 1 placed at diagonal positions corresponding to the pixel p (I-1, J + 8) when K = 4. , J-4) and all the difference values To obtain the sum as correlated 6 of the six right second sub-region.

The pixel correlation calculator 120 can obtain correlations as many as the number of sub window regions in one direction. In the present invention, the pixel correlation calculation unit 120 includes the six sub window regions in the right direction described above based on the pixel position J to be interpolated. In consideration of all six subwindow regions in the left direction, a total of 12 correlations in six directions are obtained. Here, the correlation of the sub-window region in the left direction is to move the sub-window region having a size of 5 pixels in the left direction with respect to J in the upper line (I-1) and correspond to J in the lower line (I + 1) correspondingly. The sub window area having a size of 5 pixels is moved to the right direction to obtain the sum of the differences between the pixels placed in the diagonal direction.

Meanwhile, along with the correlation between the pixels in the diagonal direction, the correlation in the vertical direction is also obtained. The correlation in the vertical direction is obtained by the sum of the differences between adjacent pixels in the vertical direction of the upper line (I-1) and the lower line (I + 1), which is a pixel in a subpixel region of 5 pixels in size relative to J. Do it for them.

The vertical correlation is expressed by the following equation (3).

That is, considering the size of the sub-window area, when K = 0 to 4 (5 pixels), and when K = 0, the pixel p (I-1, J-2) of the upper line (I-1) and the corresponding vertical The difference value with the pixel P (I + 1, J-2) of the lower line I + 1 placed in the directional position is obtained, and when K = 1, the pixel p (I-1, J- of the upper line I-1 1) and the difference value between the pixels P (I + 1, J-1) of the lower line I + 1 placed at the vertical position corresponding thereto, and when K = 2, the pixels of the upper line I-1. The difference between p (I-1, J) and the pixel P (I + 1, J) of the lower line (I + 1) placed at the vertical position corresponding thereto is obtained, and when K = 3, the upper line (I−) The difference value between pixel p (I-1, J + 1) of 1) and pixel P (I + 1, J + 1) of the lower line I + 1 placed at the vertical position corresponding thereto is obtained, and K = When 4, the pixel p (I-1, J + 2) of the upper line I-1 and the pixel P (I + 1, J + 2) of the lower line I + 1 placed at a vertical position corresponding thereto. And the difference between all the difference values A will obtain a vertical correlation of the sub-window area.

The interpolation direction determiner 130 determines the sub window area having the largest correlation among the right direction, left direction, and vertical direction by using the correlation calculated by the correlation calculation unit 120 of the pixel. That is, as described above with reference to FIGS. 8 and 9, the correlation in the diagonal direction is 6 right and 6 left directions based on J for the upper and lower lines when the 5 pixel sub window area is used as a reference. Correlations were obtained between 12 subwindows placed in a diagonal direction, and a correlation between the subwindows placed in a direction perpendicular to each other based on J was obtained. The smallest value is retrieved and the corresponding direction is determined as the interpolation direction.

Looking at the method of determining the interpolation direction in more detail as follows.

Step 1: Correlation between the six subwindow regions in the left direction (the upper line is left with respect to J and the corresponding diagonal subwindow with the lower line is the right), left_cor1, left_cor2, left_cor3, The sub window area having the smallest value among left_cor4, left_cor5, and left_cor6 is determined as the leftward correlation Left_cor.

This is expressed as follows.

'Left_cor = MIN (left_cor1, left_cor2, left_cor3, left_cor4, left_cor5, left_cor6)'.

Step 2: Correlation between the six subwindow regions in the right direction (the upper line is to the right with respect to J, and the corresponding diagonal subwindow of the lower line is to the left), right_cor1, right_cor2, right_cor3, A subwindow region having the smallest value among right_cor4, right_cor5, and right_cor6 is defined as a rightward correlation Right_cor.

This is expressed as follows.

'Right_cor = MIN (right_cor1, right_cor2, right_cor3, right_cor4, right_cor5, right_cor6)'.

Third Step: When Left_cor is small by comparing Left_cor and Right_cor, the interpolation direction DIR = left is determined.

This is expressed as follows.

'If min_cor = Left_cor then DIR = left'.

Fourth step: comparing Left_cor and Right_cor, if the value of Right_cor is small, determines that interpolation direction DIR = right (right).

This is expressed as follows.

'If min_cor = Right_cor then DIR = right'.

Fifth step: If neither of the third and fourth steps is determined, determine the interpolation direction DIR = mean (vertical).

The overall process so far is expressed as follows.

Left_cor = MIN (left_cor1, left_cor2, left_cor3, left_cor4, left_cor5, left_cor6)

Right_cor = MIN (right_cor1, right_cor2, right_cor3, right_cor4, right_cor5, right_cor6)

If min_cor = Left_cor then

           DIR = left

If min_cor = Right_cor then

           DIR = right

Else

           DIR = Mean

The interpolation direction determined by the interpolation direction determiner 130 is stored in the interpolation direction storage 140 and is also input to the interpolation direction corrector 150. The interpolation direction corrector 150 monitors the interpolation directions of several pixels in front and behind in time when the interpolation direction is classified into left (0 degrees to 90 degrees), vertical, and right (90 degrees to 180 degrees). . Therefore, if the change of the angle to be interpolated is sharply severe, this is prevented, and even if the angle to be interpolated is changed to maintain the angle to some extent, thereby preventing deterioration of image quality.

In addition, since the method of determining the interpolation direction may vary depending on the input pixel value itself, the sensitivity to input noise may be reduced by the above correction operation.

10 is a flowchart showing the operation principle of the interpolation direction correction unit. The first step S10 is a step of monitoring the interpolation direction Dj of 2 to 3 pixels in front and rear. The second step (S20) is a step of determining whether the angle of the interpolation is abruptly changed by searching the monitored result. If the condition is met, the second step (S20) proceeds to the third step (S30) and corrects the interpolation direction Dj = Da. The flow proceeds to the fourth step S40. The fourth step (S40) is a step for determining whether the angle of the interpolation is abruptly changed by searching the monitored result. If the condition corresponds to the fifth step (S50), the interpolation direction Dj = D90 is corrected. The sixth step S60 of comparing the correlations Cor_a and Cor_b is then performed. If Cor_a <Cor_b is satisfied in the sixth step S60, the process proceeds to the seventh step S70 to correct the interpolation direction Dj = Da, and if not, the process moves to the eighth step S80 to correct the interpolation direction Dj = Db. .

The interpolation direction information corrected by the interpolation direction correction unit 150 is stored in the interpolation direction storage unit 140 as previous interpolation direction information.

The candidate pixel determiner 160 determines a candidate pixel to be used for interpolation in the final direction determined by the interpolation direction corrector 150. In this case, the candidate pixel selects one pixel in the sub-window area having the largest correlation among the correlations of the directions described above by the interpolation direction determiner 130. Decide This is shown in FIGS. 11 and 12.

11 shows that when the first sub window area in the right direction in the upper line I-1 LINE has the greatest correlation with the left first sub window area in the lower line I + 1 LINE corresponding to this diagonally, A case in which a pixel positioned in the center of each sub window area is determined as a candidate pixel is illustrated. That is, the candidate pixels Up_candidate_pixel = P (I-1, j + 1) of the upper line (I-1 LINE) and the candidate pixels Dn_candidate_pixel = P (I + 1, J-1) of the lower line (I + 1 LINE). It is shown that pixels in each sub window area are determined as candidate pixels.

12 shows that when the fourth sub-window region in the left direction on the upper line I-1 LINE has the greatest correlation with the right fourth sub-window region in the lower line I + 1 LINE corresponding to this diagonally, A case in which a pixel positioned in the center of each sub window area is determined as a candidate pixel is illustrated. That is, the candidate pixels Up_candidat_pixel = P (I-1, j-4) of the upper line (I-1 LINE) and the candidate pixels Dn_candidate_pixel = P (I + 1, J + 4) of the lower line (I + 1 LINE). It is shown that a pixel in each sub window area is determined as a candidate pixel.

The final interpolator 170 receives a pixel value to be used for interpolation determined by the candidate pixel determiner 160, and generates and outputs a pixel value to be finally interpolated from this value. For example, 1/2 of the sum of the corresponding candidate pixel values of the upper and lower lines is output as the final interpolation value.

That is, interpolation value = [Up_candidat_pixel + Dn_candidate_pixel] / 2.

As described above, in the image de-interface interpolation filter according to the present invention, a window area having a predetermined size is set in each adjacent line, and the correlation between pixels located in diagonal directions with each other based on the window area is obtained. In-field interpolation filtering is performed by determining the window area having the largest correlation and outputting the final interpolation value (interpolation pixel data) using this pixel value using the center pixel of the determined window area as a candidate pixel. Doing. In addition, as described above, the interpolation directions of some pixels are monitored before and after the interpolation direction determined from the correlation of the pixels so that there is no sudden change, and the interpolation direction change of an appropriate level is induced according to the monitoring result. It was possible to prevent the deterioration of the image quality and to make the noise sensitivity dull.

In-field interpolation apparatus according to the present invention can reduce the sensitivity to the noise of the input pixel than the conventional line repetition method or in-field interpolation method that does not compensate for motion, and even in the direction of a large slope, diagonal direction rather than vertical direction It is possible to interpolate and to prevent the sudden change of slope, thereby preventing deterioration of image quality and improving interpolation image quality.

In particular, the present invention is easy to implement a deinterlacing apparatus by applying to a motion adaptive interpolation method, and improves the performance in deinterlacing, and also converts an interlaced video signal into a progressive video signal. There is an effect of enabling high quality image implementation in.

Claims (12)

A deinterlacing apparatus for converting an interlaced video signal into a progressive video signal, comprising: correlation extracting means for obtaining a diagonal correlation between pixel data of adjacent upper and lower lines, and an interpolation direction using the extracted correlation values And interpolation value determining means for outputting an interpolation value using pixel data of a position corresponding to the determined interpolation direction. The method of claim 1, wherein the correlation extracting means comprises: pixel storage means for storing pixel values corresponding to a set window area having a predetermined size, and calculating diagonal correlations between adjacent upper and lower lines based on the window area; An image deinterlacing interpolation filtering device comprising correlation calculation means. The apparatus of claim 1, wherein the interpolation direction determining unit determines a direction having a maximum correlation value in a diagonal direction between adjacent upper and lower lines as an interpolation direction. The image according to claim 1, wherein the interpolation direction determining means includes interpolation direction storage means for storing the interpolation direction, and interpolation direction correction means for correcting the interpolation direction using a previous interpolation direction and a current interpolation direction. Deinterlacing interpolation filtering device. 2. The interpolation value output means according to claim 1, wherein the interpolation value output means comprises: candidate pixel determination means for determining a corresponding pixel in each of the upper and lower lines of the determined interpolation direction as candidate pixels, and using the determined candidate pixel values. And a final interpolation means for generating the deinterlacing interpolation filtering device. A deinterlacing method for converting an interlaced video signal into a progressive video signal, comprising: setting a window area having a predetermined size in each of adjacent upper and lower lines, and arranged in diagonal directions with each other based on the window area A correlation calculation step of obtaining correlations therebetween; An interpolation direction determining step of determining the window area having the largest correlation as the interpolation direction; Outputting a final interpolation value using the pixel value using a pixel in the window area determined in the interpolation direction as a candidate pixel; Image deinterlacing interpolation filtering method comprising a. 7. The method of claim 6, wherein the window area has a size of 5 pixels. 7. The method of claim 6, wherein the pixel located in the center of the window region is selected as a candidate pixel. The method of claim 6, wherein the correlation calculates the sum of the difference between pixel values in the relative window regions located in diagonal directions while moving the set window region in a right direction and a left direction as a correlation value. Image deinterlacing interpolation filtering method. 7. The method of claim 6, wherein the direction corresponding to the window area having the largest correlation among the correlation values between the pixels in the diagonal direction is determined as the interpolation direction. The method of claim 6, further comprising: selecting a plurality of pixels positioned before and after the temporal interpolation direction by monitoring the interpolation directions of the pixels and performing correction on a sudden change in the interpolation direction. An image deinterlacing interpolation filtering method, characterized in that. 7. The method of claim 6, wherein the interpolation value is determined as a value of two minutes of the sum of the candidate pixels.

Images