KR100328585B1 - Method for Digital Image Enlargement Using Delta-Shape Pseudomedian Filter - Google Patents

Abstract

In the digital image magnification method, the first intermediate value is a median value of three peripheral pixels located at a vertex of a triangle and a vertex of an inverted triangle among six peripheral pixels masked by a main mask. And obtaining a second intermediate value, obtaining an average value of pixel values located at the bottom center point of the triangle and pixel values located at the bottom center point of the inverted triangle, and then applying the first intermediate value to the second intermediate value and the average value. The present invention relates to a digital image magnification method using a delta-type pseudo median filter for determining a median value of an intermediate pixel as an interpolation value of an empty pixel point.

According to the present invention, by setting the sub-window of the pseudomedian filter for digital image enlargement to the delta type, excellent pixel reproducibility can be provided by effectively reflecting the pixel distribution characteristics of the vertical, horizontal, and diagonal directions of the surrounding pixels. There is an advantage to this.

Description

Method for Digital Image Enlargement Using Delta-Shape Pseudomedian Filter

The present invention relates to a digital image magnification method using a delta-type pseudo median filter, and more particularly, to provide excellent pixel reproducibility by setting a sub-window of a pseudo median filter for digital image expansion to a delta type. A digital image enlargement method using a delta pseudomedian filter.

Image enlargement techniques can be broadly classified into optical image enlargement and digital image enlargement. Both optical and digital image enlargement are intended to enlarge the image while maintaining the characteristics and shape of the original image as much as possible. The biggest difference is the use of optical lens and computer calculation. Optical magnification technology is relatively poor in practicality because it requires the use of multiple lens combinations to provide the desired magnification, but provides excellent image quality even at high magnifications. On the other hand, digital image magnification technology has the advantage of easy implementation and practicality by using digital signal processing techniques, while providing a relatively low quality image at high magnification.

In general, digital image magnification can be described by dividing into two stages: image extension and image interpolation. First, an image expansion process is performed in which the pixels of the original image are spaced apart by an interval to be enlarged, and then an image interpolation process is performed in which the pixel values of the grid points that do not exist in the grid of the extended image are filled with appropriate values step by step. As such, the image that has undergone the process of image expansion and image interpolation is represented by the number of pixels spatially increased compared to the original image, which means that the image is enlarged. However, in view of the preservation of the amount of information, the magnification technique using only the information of spatially adjacent pixels in one still image can increase the size of the image, but it is known that the contribution to the substantial resolution improvement is insufficient. This is because the amount of information that can be used to enlarge the image, that is, the frequency band of the original image information is limited in the original image.

Representative prior applications related to the image magnification method of a display device according to the prior art are Korean Patent Application No. 1993-011678, '12' Normal Display Mode 15 'Full Screen Display Mode Switching Circuit of Monitor', and Korean Patent Application No. 1997-019423 U.S., 'Automatic Vertical Extension Circuit for Wide TV', U.S. Patent Publication No. US04816929, 'Dual access frame store for field or frame playback in a video disk player', U.S. Patent Publication No. US05907640, 'Functional' interpolating transformation system for image processing ', US Patent Publication No. US05844617,' Method and apparatus for enhancing the vertical resolution of a television signal having degraded vertical chrominance transitions', US Patent Publication No. US05835029, 'Digital convergence apparatus', US Patent Publication US05349385, 'Adaptive scan converter', US Patent Publication US05299300, 'Interpolationprocessing' of digital map imagery data ', US Patent Publication No. US04858026,' Image display 'and the like.

Digital image magnification technology is typically implemented by image interpolation technology. In the TV field, interpolation technology has been introduced for the purpose of converting interlaced scanning of TV signals into sequential scanning. In the case of interlaced scanning, aliasing occurs due to the incorporation of components on the time-vertical axis, and thus, the problem can be minimized by converting the interlaced scan into a sequential scan. The interpolation technique is used to reproduce the omitted scan lines in the frame in interlaced scanning.

Conventional image interpolation methods include linear interpolation, such as zero order interpolation (ZOI), first order interpolation (FOI), spatially-weighted adaptive interpolation (SWAI), and nonlinear interpolation. Median method, Edge based Line Average (ELA), Pseudomedian Filter (PMED) are typical.

Hereinafter, each known interpolation scheme will be briefly described with reference to the accompanying drawings in order to help understanding of the present invention.

First of all, the ZOI method is also called a line repetition method. The ZOI method is a method of repeating a line above an omitted line as it is, but the implementation is simple but has limited applicability. FIG. 1A shows that the pixel value of the previous line is transferred directly to the pixel value located on the missing scanning line. This method is suitable for still images and is partially used for still regions in actual three-dimensional processing.

The FOI method is also referred to as a line averaging method. As shown in FIG. 1B, a scanning line missing from each interlaced field is interpolated by averaging pixel values of adjacent up and down scanning lines, and is suitable for moving pictures. Applied to localized animation processing mode in dimension processing. In addition, because the structure is simple and satisfactory performance, it is the method used when considering the price performance ratio.

The SWAI method calculates the difference between the vertical and diagonal directions in a filter that averages the pixels above and below the field to be interpolated in the field, and a filter that averages the pixels in four diagonal directions. This method calculates the value of the interpolated signal by multiplying. This approach is difficult to apply to systems that require real-time processing because of the large amount of computation.

The median method is one of the most common nonlinear techniques used in signal processing, which is conceptually simple but requires a large amount of computation. The median filter used in this way is an operator that is easily computed by the window function passing the signal at each point, and the output of the filter takes the median inside the window. In order to apply such a median filter to an interpolation technique, as shown in FIG. 1C, when the pixel to be interpolated is g, since there are six neighboring pixels used, it is difficult to find an intermediate value. One solution is to use a modified median filter, in which the average of pixels located above and below g in FIG. 1C is added to an input pixel. The output of the median filter by this method is expressed by Equation (1).

The ELA method is used for TV signal processing and is excellent in removing impulse noise, but has a disadvantage of blurring the outline of an image because it is based on FOI. The method of obtaining the pixel x (i, j) to be interpolated is as shown in FIG. 1D. This is briefly described in C-program form as follows.

a = ABS {x (i-1, j-1)-x (i + 1, j + 1)} b = ABS {x (i-1, j)-x (i + 1, j)} c = ABS {x (i-1, j + 1)-x (i + 1, j-1)} if MIN {a, b, c} = bx (i, j) = 0.5 {x (i-1, j ) + x (i + 1, j)} elseif MIN {a, b, c} = ax (i, j) = 0.5 {x (i-1, j-1) + x (i + 1, j + 1 )} elsex (i, j) = 0.5 {x (i-1, j + 1) + x (i + 1, j-1)}

The median filter operates on discontinuous signals and has a window of 2N + 1 sample width centered on the signal to be processed, and the median value of the 2N + 1 samples included in this window becomes the output of the filter. Pseudo-median filters also operate on discrete signals and have a sample width of 2N + 1, but unlike median filters, 2N + 1 samples are masked with N + 1 sized windows to average the minimum of the maximums and the maximum of the minimums. Calculate the output of the pseudomedian filter. For example, the pseudomedian filter may be expressed as Equation 1 and Equation 2 for N = 1 and N = 2, respectively.

If is the window value ego, If is the window value to be. here Means 'pseudomedian filter'. Pseudomedian filters, unlike medians, Since only the sub-window of is used, it has a characteristic of adding a stronger weight to the center point than the median filter. Accordingly, the response characteristics of edge and impulse are evaluated to be superior to the median filter.

An H-shped PMED (H-PMED) is used to convert an applied interlaced signal into a sequential scan type, and FIG. 2A illustrates pixels used for PMED calculation. a, b, and c are Pixels located on the first scan line, and d, e, and f are Are the pixels located in the first position. In this case, g denotes a pixel to be interpolated, and at this time, a line on which g is located is a scan line to be interpolated.

Each sub window for the entire window of FIG. 2A is {a, b, c}, {d, e, f}, {b, e}. At this time, the pixel value g interpolated by this method is expressed by Equation 4.

As represented by Equation 4, the interpolated pixel value g is calculated by linear and nonlinear operations of adjacent pixels in the horizontal and vertical directions.

On the other hand, any pixel in the image is highly correlated not only horizontally and vertically with the peripheral pixel but also in the diagonal direction. Whereas H-PMED is a method in which horizontal and vertical directions are considered, an asterisk-shaped PMED (hereinafter referred to as A-PMED) shown in FIG. 2B is a method in consideration of a diagonal direction and a vertical direction. This method uses the interpolated pixel g and the pixel groups {a, f}, {c, d} located in the diagonal direction and the pixel group {b, e} located in the vertical direction with g as sub-windows. Therefore, the output by the average of the minimum of the maximum value of such a sub-window and the maximum of the minimum value is represented by (5).

H-type PMED is widely evaluated to provide more outstanding performance than ZOI method, FOI method, median method, ELA method, and SWAI method, and is known to provide excellent pixel reproducibility on average compared to similar star-shaped PMED method. .

However, since the H-type PMED is intended to improve the contour reproducibility of the empty pixel point in consideration of the vertical and horizontal orientations of neighboring pixels, there is a problem in that the diagonal direction is not considered. In particular, such a problem is a slight performance degradation at low magnification, but at a high magnification, the weak effect is accumulated, thereby providing a somewhat unnatural pixel reproducibility even in the texture area.

Accordingly, the present invention has been made to solve such a problem, and in the digital image magnification method, vertical and horizontal and horizontal and horizontal pixels of the peripheral pixels are set by setting the sub-window of the pseudomedian filter for digital image magnification to delta type. The present invention relates to a digital image enlargement method using a delta-type pseudo-median filter which provides excellent pixel reproducibility by effectively reflecting pixel distribution characteristics in a diagonal direction.

1A, 1B, 1C, and 1D are pixel layout diagrams for explaining a zero order interpolation method (ZOI), a first order interpolation method (FOI), a median method, and an ELA method, respectively;

2A and 2B are exemplary diagrams illustrating a pixel arrangement and a sub-window shape for explaining the H-type PMED scheme and the star-shaped PMED scheme;

3 is a flow chart showing a preferred embodiment of the method of the digital image magnification method using a delta-type pseudo median filter according to the present invention;

4 is an exemplary diagram illustrating a pixel arrangement and a sub-window shape for explaining a delta-type pseudo median filter method;

5A to 5C are block diagrams illustrating a digital image enlargement apparatus using a delta-type pseudo median filter according to the present invention.

In order to achieve the above object, the present invention provides a method for enlarging a digital image using a delta-type pseudo median filter, which includes a median value of three peripheral pixels located at a vertex of a triangle among six peripheral pixels masked by a main mask. After obtaining the first intermediate value, which is the median value, and obtaining the second intermediate value, which is the median value for the three peripheral pixels located at the vertex of the inverted triangle, among the six peripheral pixels masked in the main mask, After obtaining the average value of the pixel value located at the bottom center point of the triangle and the pixel value located at the bottom center point of the inverted triangle, the median value of the first intermediate value, the second intermediate value, and the average value is calculated. By determining the interpolation value of the empty pixel point, it is possible to effectively reflect the pixel distribution characteristics in the vertical, horizontal, and diagonal directions of the surrounding pixels when the image is enlarged. It is characterized by a lock.

Hereinafter, a preferred embodiment of the method of the digital image enlargement method using the delta-type pseudo median filter according to the present invention will be described with reference to the accompanying drawings.

Figure 3 is a flow chart showing a preferred embodiment of the method for digital image enlargement method using a delta-type pseudo median filter according to the present invention.

As shown in FIG. 3, the preferred embodiment of the method of the digital image magnification method using the delta-type pseudo median filter according to the present invention, in the digital image magnification method,

Placing a pixel point of the original image to be enlarged by an interval to be enlarged (S10);

Masking six peripheral pixels adjacent to the empty pixel point to be interpolated with a main window (S20);

Obtaining a first intermediate value which is a median value of three peripheral pixels located at a vertex of a triangle among the peripheral pixels masked by the main mask (S30);

Obtaining a second intermediate value which is a median value of three peripheral pixels located at a vertex of an inverted triangle among the peripheral pixels masked by the main mask (S40);

Obtaining an average value of a pixel value located at the bottom center point of the triangle and a pixel value located at the bottom center point of the inverted triangle (S50); And

And determining a median value for the first intermediate value, the second intermediate value, and the average value as an interpolation value of the empty pixel point (S60).

The operation of the preferred embodiment of the digital image enlargement method using the delta-type pseudo median filter according to the present invention configured as described above will be described in detail with reference to the accompanying drawings.

First, in step S10, an image extension process is performed in which a pixel point of an original image to be enlarged is spaced by an interval to be enlarged.

Subsequently, in step S20, six peripheral pixels adjacent to the empty pixel point to be interpolated are masked with the main window.

In operation S30, a first intermediate value, which is a median value of three peripheral pixels located at a vertex of a triangle, is obtained from the peripheral pixels masked in the main mask, and in operation S40, the peripheral pixels masked in the main mask. Among them, a second intermediate value, which is a median value for three peripheral pixels located at the vertex of the inverse triangle, is obtained. In operation S50, an average value of the pixel value located at the bottom center point of the triangle and the pixel value located at the bottom center point of the inverted triangle is obtained.

Finally, in step S60, the above process of determining the median value for the first intermediate value, the second intermediate value, and the average value as the interpolation value of the empty pixel point is repeatedly performed for all the empty pixel points. As a result, the original image is enlarged.

4 is an exemplary diagram illustrating a pixel arrangement and a sub-window shape for explaining a delta-type pseudo median filter method.

The median in the triangular subwindow {a, c, e} and the inverse triangular subwindow {b, d, f} and the '|' shaped subwindow {b, e} The intermediate value of the average value is obtained to calculate the value of the pixel to be interpolated. This can be expressed as an equation.

5A to 5C are block diagrams illustrating a digital image enlargement apparatus using a delta-type pseudo median filter according to the present invention.

As shown in Fig. 5A, the delta-type pseudo median filter is composed of nine comparators, nine multiplexers (MUXs), one adder, and one multiplier. If you take 8 bits from the next least significant bit (LSB) of the output from the adder to the carry bit, you do not need a multiplier.

The MUX of SORT2 (two-input aligner) shown in Fig. 5B uses the output of the comparator as the selection control signal and outputs a high value for a large number and a low value for a small number. SORT2 is used to configure SORT3 (3-Input Sorter) in FIG. 5C. In addition, FIG. 5D calculates an average value of pixel values located at the bottom center point of the triangle and the pixel values located at the bottom center point of the inverted triangle through the component AVRG.

Overall, SORT3, AVRG, adder, and the like are used to form an overall Telta-type pseudo median filter as shown in FIG. 5A.

Terminologies used herein are terms defined in consideration of functions in the present invention, which may vary according to the intention or practice of a person skilled in the art, and the definitions should be made based on the contents throughout the present application. will be.

Although specific embodiments of the invention have been described and illustrated, it will be apparent that the invention may be embodied in various modifications by those skilled in the art.

In addition, since the present invention has been described through the preferred embodiment of the present invention, in view of the technical difficulty aspects of the present invention, those having ordinary skill in the art can easily be different from another embodiment of the present invention. As modifications may be made, it is obvious that both the embodiments and modifications cited in the above description belong to the appended claims.

As described in detail above, in the digital image magnification method, a median value for each of three peripheral pixels located at a vertex of a triangle and a vertex of an inverted triangle among six peripheral pixels masked by a main mask Calculate a first intermediate value and a second intermediate value, obtain an average value of pixel values located at the bottom center point of the triangle and pixel values located at the bottom center point of the inverted triangle, and then calculate the first intermediate value and the second intermediate value. According to the present invention which determines the median value of the mean value and the mean value as the interpolation value of the empty pixel point, the peripheral pixel is set by setting the sub-window of the pseudomedian filter for digital image enlargement to the delta type. It is advantageous to provide excellent pixel reproducibility by effectively reflecting the pixel distribution characteristics in the vertical, horizontal and diagonal directions.

Claims (1)

In the digital image magnification method using a delta pseudomedian filter: Spacing pixel points of the original image to be enlarged by an interval to be enlarged; Masking six peripheral pixels adjacent to the empty pixel point to be interpolated into a main window; Obtaining a first intermediate value, which is a median value of three peripheral pixels located at a vertex of a triangle, among the six peripheral pixels masked by the main window; Obtaining a second intermediate value which is a median value of three peripheral pixels located at a vertex of an inverted triangle among the six peripheral pixels masked by the main window; Obtaining an average value of a pixel value located at the bottom center point of the triangle and a pixel value located at the bottom center point of the inverted triangle; And And determining a median value for the first intermediate value, the second intermediate value, and the average value as an interpolation value of the empty pixel point.