KR101329134B1 - System and method for converting image using sub-pixel rendering - Google Patents

Abstract

Disclosed are an image conversion system and method using subpixel rendering. An image conversion system using sub-pixel rendering uses at least one filter set for each output sub-pixel by using a sharing amount of each of the at least one output sub-pixel for each input sub-pixel for the input pixel. And a sub-pixel rendering unit configured to select a filter optimized for an input image from the generated filter set generation unit and convert the value of the input sub-pixel into a value of the output sub-pixel through the selected filter.

Rendering, contribution, pattern detection, input structure, output structure, filter mask

Description

Image conversion system and method using subpixel rendering {SYSTEM AND METHOD FOR CONVERTING IMAGE USING SUB-PIXEL RENDERING}

The present invention relates to an image conversion system and method using subpixel rendering. In particular, the present invention relates to a method for generating a filter set using a normalized contribution amount to an output subpixel, and to select a filter according to characteristics of an input image among the generated filter sets. The present invention relates to an image conversion system and method using subpixel rendering to filter an input image.

A subpixel rendering method is an existing technique of converting an RGB input image having a subpixel structure. One of the subpixel rendering methods is to construct a rendering filter for several new subpixel structures, and define a suitable resampling area for the structure. The resampling areas should be adjacent to each other without overlapping areas, and there should be no spaces missing for resampling in the image. The filter value is determined based on the area ratio of the overlapping area between the resampling area having the corresponding condition and the pixel area in the existing pixel structure.

Resampling Areas can be proposed in a variety of ways, but having the right shape of Resampling Areas in place is entirely a matter of design, making it difficult to respond to patterns or adjust sharpness in the conversion process.

In addition, although the actual primary display has the same primary color set, the subpixels may have different sizes. In addition, there may be a case where the primary color set is different from the RGB structure and the RGBW structure. In this case, a process of converting a color expressed in RGB into a color expressed in RGBW is required.

As the size of the subpixel increases, the number of subpixels occupying the same area decreases, thereby reducing the cost in terms of a driving circuit for controlling the pixel. And, there is an advantage that the brightness is increased by increasing the aperture ratio. However, in the display structure, when the number of subpixels is small, a resolution required to express a corresponding video signal may not be satisfied.

Therefore, there is an urgent need for an invention that can easily create a filter set having various sizes and characteristics by using a distribution function and parameters, and adaptively filter according to the characteristics of the image to preserve the characteristics of the image to be converted. do.

The present invention provides an image conversion system and method using sub-pixel rendering that can minimize the visual difference between an existing image and a reconstructed image by generating a filter set using a distribution function.

The present invention provides an image conversion system and method using sub-pixel rendering that can increase the degree of freedom in filter design by changing the distribution function or the parameters of the distribution function.

The present invention provides an image conversion system and method using sub-pixel rendering that can improve the quality of an output image by detecting a pattern of the input image and adaptively filtering the input image through a filter advantageous to the pattern.

The present invention provides an image conversion system using subpixel rendering capable of preserving the amount of color to be shared by sharing subpixel values of an input image with subpixels of an output image corresponding to a periphery reflecting visual characteristics. Provide a method.

An image conversion system using subpixel rendering according to an embodiment of the present invention uses at least one output subpixel by using a sharing amount of at least one output subpixel with respect to a value of each input subpixel for an input pixel. A filter set generation unit for generating one filter set and a sub-pixel that selects a filter optimized for an input image from the generated filter set and converts the value of the input sub-pixel into a value of the output sub-pixel through the selected filter It may include a renderer.

In this case, the filter set generation unit calculates a sharing amount of each of the at least one output subpixel representing the same color as the input subpixel by using a distribution function for the value of each input subpixel for the input pixel. Allotment calculation part to say,

A contribution normalization unit for normalizing the contribution amount to the input subpixel using the area ratio between the input subpixel and the output subpixel, and a filter corresponding to each output subpixel using the normalized contribution amount It may include a filter configuration.

According to an embodiment of the present invention, an image conversion method using subpixel rendering uses at least one output subpixel by sharing a value of each input subpixel with respect to an input pixel. Generating a filter set and converting a value of the input subpixel to a value of the output subpixel through the selected filter by selecting a filter optimized for an input image in the generated filter set It may include.

The generating of at least one filter set for each output subpixel may include generating at least one output color representing the same color as the input subpixel using a distribution function of the value of each input subpixel for the input pixel. Calculating an amount of contribution each subpixel shares, normalizing the contribution amount to the input subpixel using an area ratio of the input subpixel and the output subpixel, and using the normalized contribution amount The method may include configuring a filter corresponding to each output subpixel.

According to the present invention, there is provided an image conversion system and method using subpixel rendering capable of minimizing the visual difference between an existing image and a reconstructed image by generating a filter set using a distribution function.

According to the present invention, there is provided an image conversion system and method using sub-pixel rendering that can increase the degree of freedom in filter design by changing the distribution function or the parameters of the distribution function.

According to the present invention, there is provided an image conversion system and method using sub-pixel rendering that can improve the quality of an output image by detecting a pattern of the input image and adaptively filtering the input image through a filter advantageous to the pattern.

According to the present invention, by subdividing subpixel values of an input image into subpixels of an output image corresponding to a periphery reflecting visual characteristics, image conversion using subpixel rendering capable of preserving the amount of color to be shared Systems and methods are provided.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. The image conversion method using subpixel rendering according to an embodiment of the present invention may be performed by an image conversion system using subpixel rendering.

1 is a block diagram illustrating an image conversion system using subpixel rendering according to an embodiment of the present invention.

Referring to FIG. 1, an image conversion system using subpixel rendering may include a filter set generator 101 and a subpixel renderer 102. In this case, the filter set generating unit 101 may include a sharing amount calculating unit 103, a sharing amount normalizing unit 104, and a filter constitution unit 105. In addition, the subpixel renderer 102 may include a pattern detector 106, a filter selector 107, and an image filter 108.

The filter set generating unit 101 may generate at least one filter set for each output subpixel by using a sharing amount of each of the at least one output subpixel sharing the value of each input subpixel with respect to the input pixel.

In this case, as shown in FIG. 1, the filter set generating unit 101 may receive information about each of the input structure representing the input image and the output structure representing the output image. The input structure includes a plurality of input pixels, and each input pixel includes at least one input subpixel unit. The output structure also includes a plurality of output pixels, and each output pixel may be configured by at least one output subpixel unit.

For example, the filter set generator 101 may have different sizes of subpixels with respect to a structure representing an image. When the size of the subpixels representing the same image is different but the subpixels representing the color components are different, a filter set that can be converted between the images can be generated in consideration of the subpixels of each structure.

Hereinafter, the filter set generating unit 101 will be described in detail through the sharing amount calculating unit 103, the sharing amount normalizing unit 104, and the filter constitution unit 105.

The sharing amount calculator 103 may calculate a sharing amount of each of the at least one output subpixel representing the same color as the input subpixel by using a distribution function of the value of each input subpixel for the input pixel. have.

Because of the subpixel size differences, the input and output structures may not match each other. In this case, when the output subpixels share the value of the input pixel, a problem that may occur due to the difference in size between the subpixels may be overcome by using the sharing amount shared by each output subpixel. At this time, even if each of the output subpixels shares the value of the input pixel, the sum of the sharing amount shared by each of the input pixel value and the output subpixel before sharing can be kept the same.

In this case, the occupancy calculation unit 103 may mask an output subpixel around the input pixel through an N * N filter mask to calculate a contribution amount according to at least one output subpixel included in the filter mask. have.

That is, in order to express the value of one input pixel, the mask may be masked to the center of the input pixel through an N * N filter mask. However, the value of N is not limited and may be changed at any time according to the setting, and thus the masking area may be enlarged or reduced.

The contribution calculator 103 may calculate an allocation amount allocated to each output subpixel so that the output subpixels in the filter mask express the color represented by each input subpixel. When the output subpixels in the filter mask have a value corresponding to the allocated contribution amount, the output subpixels can be expressed almost the same as the input image.

The burden calculator 103 may calculate the burden shared by each output subpixel using a distribution function. In one example, the distribution function may include a Gaussian function that takes an input pixel and an output subpixel as parameters. A filter with various sizes and characteristics can be created using a distribution function and its parameters. In this case, the sharing amount may correspond to a value obtained by integrating each output subpixel through a Gaussian function.

The burden calculator 103 will be described in detail with reference to FIG. 3.

For example, the sharing amount normalizing unit 104 may normalize the sharing amount with respect to the input subpixel using the area ratio between the input subpixel and the output subpixel.

Since the digital values of the input subpixels and the output subpixels are the same, the color values of the input subpixels represent different amounts of colors due to the difference in area, and thus a normalization process may be required to correct the difference in the areas.

For example, in an input structure composed of three input subpixels per input pixel, the number of input pixels is four, and in the output structure composed of four output subpixels per output pixel, the number of output pixels is five. Suppose The area of four input pixels and the area of five output pixels are the same, but the area of the input subpixel and the output subpixel is assumed to be different.

In one example, the area ratio may be determined by the area of the input subpixel of the input structure relative to the area of the output subpixel of the output structure. Thus, in the above example, the area ratio is (3/12) / (5/12) = 3/5. This area ratio may vary depending on the input structure and the output structure.

The sharing amount normalizing unit 104 may normalize the sharing amount according to Equation 1 using the area ratio.

는 입력 부화소와 출력 부화소의 면적비,

는 분포 함수

를 이용한 분담량,

는 입력 화소,

는 출력 부화소,

는 필터 마스크에서 입력 화소의 위치에 대한 인덱스)(

Is the area ratio of the input subpixel to the output subpixel,

Is a distribution function

Share using

Is the input pixel,

Is the output subpixel,

Is the index of the position of the input pixel in the filter mask)

The filter configuration unit 105 may configure a filter corresponding to each of the output subpixels using the normalized sharing amount.

Specifically, the filter configuration unit 105 may collect the normalized sharing amount according to the output subpixels to construct a filter indicating a position where each output subpixel and each of the input pixels overlap.

The configured filter may be set to digitalize the normalized burden amount so as to be easily calculated digitally, and to sum to 256 in the case of an 8-bit image. If you use these filters, you can obtain the value by dividing the filtered result by the sum of each filter coefficient. The size of the filter may be determined according to the filter mask. The characteristics of the constructed filter can be changed in various ways by adjusting the distribution function and the parameters for the function.

The filter set is defined and collected for all the output subpixels of the output structure.

The subpixel rendering unit 102 may select a filter optimized for the input image from the generated filter set and convert the input subpixel value into the output subpixel value through the selected filter.

That is, it means filtering by selecting a filter favorable to the input image from the generated filter set. For example, a filter that is advantageous to the input image may be selected according to the overall characteristics of the input image (eg, sharpness or blur) and the local characteristics of the input image (such as a pattern direction of a specific region). The subpixel renderer 102 will be described in detail through the pattern detector 106, the filter selector 107, and the image filter 108.

The pattern detector 106 may determine edge components in the horizontal and vertical directions with respect to the input image and detect local patterns of the input image according to the edge components.

In this case, the pattern detecting unit 106 may be excluded in the present invention. That is, when the overall characteristics of the image are used rather than the local characteristics of the input image, the local pattern of the input image may not be detected by the pattern detector 106.

For example, the pattern detector 106 may determine an edge direction through a ratio of edge components, and detect a pattern of an input image according to an angle formed by the edge direction with a horizontal line or a vertical line.

In addition, the pattern detector 106 may detect a local pattern of the input image using a pattern detection filter. In detail, the pattern detector 106 may determine an edge component of the input image according to Equation 2 below.

는 가로 방향에 대한 엣지 성분,

는 세로 방향에 대한 엣지 성분,

는

에 위치한 입력 화소의 값,

는 가로 방향의 패턴 검출 필터의 계수값,

는 세로 방향의 패턴 검출 필터의 계수값)(

Is the edge component for the horizontal direction,

Is the edge component for the longitudinal direction,

The

The value of the input pixel located at,

Is the coefficient value of the pattern detection filter in the horizontal direction,

Is the coefficient value of the vertical pattern detection filter)

In the present invention, a sobel filter may be used as the pattern detection filter in detecting a local pattern of the input image. However, in this case, the Sobel filter is only an example and may be replaced with another pattern capable of detecting a local pattern of the input image.

A detailed process of detecting a local pattern of the input image using the pattern detection filter will be described in detail with reference to FIG. 4.

The filter selector 107 may select a filter optimized for the input image from the generated filter set.

For example, a filter optimized for the input image may be selected from the filter set generated by using metadata about the input image. That is, the filter optimized for the input image may be selected using metadata such as text information and program information related to the input image.

Here, the optimization may mean that the characteristics of the input image are improved or the output image of the improved form is derived when the subpixel structure is changed. In this case, the case of using the metadata may refer to the case of not passing through the pattern detector 106.

If the local pattern of the input image is detected through the pattern detector 106, the filter selector 107 may select a filter optimized for the input image according to the detected pattern.

According to the present invention, an output image that is a result of filtering by selecting an optimized filter in consideration of the input image and adaptively filtering the input image using the selected filter may well reflect the characteristics of the input image.

The image filtering unit 108 may filter each input subpixel of the input image to each output subpixel using the selected filter. In other words, it may mean that the input image represented by the input structure is converted into the output image represented by the output structure using a filter.

For example, the image filtering unit 108 may multiply the values of the filter for the output subpixel and the color component of the input subpixel by elements, and add up the result values multiplied by each element. Through this process, the image filtering unit 108 may determine the value of the output subpixel representing the same color as the input subpixel of the input image.

FIG. 2 is a diagram illustrating a display having different subpixel structures and masking each subpixel through a filter mask in an embodiment of the present invention.

Display 201 shows a portion of an input structure consisting of indexed input pixels.

Each input pixel in the display 201 may be configured of at least one input subpixel.

Display 202 shows a portion of an output structure composed of indexed output subpixels. As can be seen in FIG. The size of the five input pixels corresponds to the size of the twelve output subpixels.

The filter mask 203 shows an example of masking the input pixel indexed by g using a 3 * 3 filter mask. In order to represent the red color of the input pixel g as the output subpixel through subpixel rendering, an output subpixel representing the red color in the filter mask 203 may be used. For example, R1, R2, R4, R5, R1, and R2 present in the filter mask 203 may be used to represent the red component of the input pixel g.

The size and type of the index, the input pixel, and the output subpixel shown in FIG. 2 are only examples, and may be changed according to the settings of the system.

3 is a diagram illustrating a process of dividing an input pixel by each output subpixel and configuring a filter for the output subpixel according to an embodiment of the present invention.

Reference numeral 301 denotes a process in which each output subpixel in the filter mask shares a red component of the input pixel g.

As can be seen at 301, each output subpixel in the filter mask can be represented by a distribution function f. In one example, f may be a Gaussian function.

At this time, R1 is f (g, R1,9), R2 is f (g, R2,8), R4 is f (g, R4,5), and R5 is f It can be seen that it is represented by (g, R5,4). Further, R1 at the bottom of the filter mask may be f (g, R1,3), and R2 at the bottom of the filter mask may be expressed as f (g, R2,2). In this case, the contribution amount may be expressed as a distribution function such as an integral value of a Gaussian function.

In this case, each parameter of the function f may mean an input number to be shared, an output subpixel to be shared, and an identification number for the input pixel. The identification number is arbitrary, and it can be seen that the filter mask uses the following identification number.

9 8 7 6 5 4 3 2 One

Reference numeral 302 illustrates a process of configuring a filter for the output subpixel R1 using the normalized contribution amount.

As shown by reference numeral 302, the shared contribution amount may be collected from the peripheral input pixels through the process of reference numeral 301, and a filter for the output subpixel R1 may be configured. The filter for the output subpixel R1 is as follows.

Here, the function f _N may mean a normalized sharing amount, and each parameter may mean a relative identification number based on the referenced input pixel, the output subpixel that is the object of the filter configuration, and the output subpixel. Here, the relative identification number may use the aforementioned identification number.

For example, f _N (j, R1,1) means that the output subpixel R1 refers to the input pixel j and is at the position of the identification number 1 when j is located at the center. f _N may mean a normalized contribution amount.

As a result, according to the present invention, each color component of the input pixel is assigned to the surrounding output subpixels by the reference numeral 301 and the reference numeral 302, and the assigned sharing amount is normalized for each output subpixel. It collects and comprises a filter.

4 is a diagram illustrating an example of a pattern detection filter capable of detecting a local pattern of an input image according to one embodiment of the present invention.

The pattern of the local part may be detected while scanning the entire input image through the pattern detection filter. Rather than using the same filter for the entire input image, it is possible to filter by applying different filters for each region of the input image according to the detected pattern. In such a case, an appropriate filtering result for each area of the input image may lead to an output result having improved visibility than the existing input image.

As already mentioned above, the pattern detection filter may include a sobel filter. However, the Sobel filter is only one example of detecting a local pattern of the input image, and may be replaced with another filter capable of exhibiting the same function and effect.

The pattern detection filter 401 represents an example in which the vertical pattern of the input image can be detected. The pattern detection filter 402 shows an example in which a horizontal pattern of the input image can be detected. The value expressed in each of the pattern detection filters 401 and 402 is merely an example, and may be replaced with another value capable of detecting a vertical pattern or a horizontal pattern of the input image.

The pattern detection filters 401 and 402 may detect a local pattern of a corresponding input image while scanning the entire input image.

Suppose that a specific area of the input image is as follows.

0 One 2 0 One 2 0 One 2

The pattern detection filters 401 and 402 may be applied through Equation 2 below. That is, the sum of the sum of the pattern detection filters 401 and 402 and the corresponding input image may be calculated by multiplying each element. According to a result of applying the pattern detection filters 401 and 402, a pattern of a corresponding region of the input image may be detected.

When the pattern detection filter 401 is applied, the filtering result is 2 + 4 + 2 = 8. When the pattern detection filter 402 is applied, it becomes zero. That is, when the specific region of the input image is as described above, the local pattern of the input image may be detected as a vertical pattern.

)로 결정될 수 있다.If it is difficult to intuitively detect a local pattern of the input image through the result of applying the pattern detection filters 401 and 402, the edge directionality may be checked as the ratio of each element value. At this time, a specific area of the input image is a horizontal pattern when the angle formed by the edge direction and the horizontal line is in a preset reference range, and a vertical pattern when the angle formed by the edge direction and the vertical line is in the preset reference range. Can be detected. At this time, the angle between the edge direction and the vertical or horizontal line is tan ^-1 (

). ≪ / RTI >

FIG. 5 illustrates an example of a distribution function of a filter for filtering an input image according to an embodiment of the present invention.

The function 501 represents an example of a horizontal and vertical distribution function for general filtering. In one example, the distribution function may be expressed as a Gaussian function. The function 501 means a flat representation of the Gaussian function form. The function 501 may be used both when the input image represents a horizontal pattern or a vertical pattern.

Function 502 represents an example of a distribution function for a horizontal filter that can be advantageously applied when localized portions of the input image are detected in a horizontal pattern. The vertical direction is spread evenly over the entire range, but the horizontal direction shows a large value by focusing on a specific part.

Thus, if a localized portion of the input image is detected as a landscape pattern, it may be desirable to filter the input image using a filter representing the function 502. For example, if someone is wearing horizontal stripes in the input image. For the horizontal stripe cloth, using a filter representing the function 502 can improve the quality of the output image.

That is, the visibility of the pattern in the horizontal direction may be improved by using a filter that adaptively represents the function 502 in a portion of the horizontal image representing the horizontal pattern with respect to the input image.

Function 503 illustrates an example of a distribution function for a filter that can be advantageously applied when localized portions of an input image are detected in a vertical pattern. The vertical direction indicates a large value by focusing on a specific portion, and the horizontal direction is evenly spread over the entire range.

Thus, if a localized portion of the input image is detected as a vertical pattern, it may be said to filter the input image using a filter representing the function 503. For example, if there is a vertical cage in the input image, using a filter representing the function 503 for that portion may yield more effective filtering results.

That is, there is an advantage in that visibility of the pattern in the vertical direction may be improved by using a filter in which the function 503 is adaptively applied to the portion in which the pattern in the vertical direction is displayed locally.

In other words, the function 502 and the function 503 mean a filter that is advantageous to each corresponding pattern in consideration of a pattern for a specific region of the input image.

FIG. 6 is a flowchart illustrating an image conversion method using subpixel rendering according to an embodiment of the present invention. An image conversion method using subpixel rendering may correspond to an image conversion system using subpixel rendering.

Referring to FIG. 6, in the image conversion method using subpixel rendering, at least one output subpixel using at least one output subpixel sharing a value of each input subpixel with respect to an input pixel. Generating a filter set (S601 to S603) and selecting a filter optimized for an input image in the generated filter set to convert the value of the input subpixel to the value of the output subpixel through the selected filter. The conversion may include the steps S604 to S606.

In an image conversion method using subpixel rendering according to an embodiment of the present invention, at least one output subpixel expressing the same color as the input subpixel using a distribution function of a value of each input subpixel for an input pixel. Calculate the amount of contribution each share (S601).

In this case, in the calculating of the sharing amount (S601), the output subpixel may be masked around the input pixel through an N * N filter mask to calculate the sharing amount through the at least one output subpixel included in the filter mask. Can be.

As a preferred example, the distribution function may include a Gaussian function that takes an input pixel and an output subpixel as parameters. In this case, the sharing amount may correspond to a value obtained by integrating each output subpixel through a Gaussian function.

 In the image conversion method using sub-pixel rendering according to an embodiment of the present invention, the sharing amount is normalized with respect to the input sub-pixel using the area ratio of the input sub-pixel and the output sub-pixel (S602).

In step S602 of normalizing the burden amount, the burden amount may be normalized with respect to the input subpixel according to Equation 3 below.

는 입력 부화소와 출력 부화소의 면적비,

는 분포 함수

를 이용한 분담량,

는 입력 화소,

는 출력 부화소,

는 필터 마스크에서 입력 화소의 위치에 대한 인덱스)(

Is the area ratio of the input subpixel to the output subpixel,

Is a distribution function

Share using

Is the input pixel,

Is the output subpixel,

Is the index of the position of the input pixel in the filter mask)

In the image conversion method using subpixel rendering according to an embodiment of the present invention, a filter corresponding to each of the output subpixels is configured using the normalized sharing amount (S603).

In step S603 of configuring a filter corresponding to each output subpixel, the normalized sharing amount may be collected according to the output subpixel to configure a filter indicating a position where each output subpixel and each input pixel overlap. .

According to an embodiment of the present invention, an image conversion method using subpixel rendering determines edge components in a horizontal direction and a vertical direction with respect to an input image, and detects a local pattern of the input image according to the edge components. S604).

The detecting of the local pattern of the input image according to the edge component (S604) may not be essential to the image conversion method.

At this time, the step of detecting a local pattern of the input image according to the edge component (S604) determines the direction of the edge through the ratio of the edge component to take the pattern of the input image in consideration of the angle of the edge direction with the horizontal line or the vertical line Can be detected.

In this case, in the detecting of the local pattern of the input image according to the edge component (S604), the local pattern of the input image may be detected using a pattern detection filter.

Here, in the detecting of the local pattern of the input image according to the edge component (S604), the edge component may be determined according to Equation 4 below.

는 가로 방향에 대한 엣지 성분,

는 세로 방향에 대한 엣지 성분,

는

에 위치한 입력 화소의 값,

는 가로 방향의 패턴 검출 필터의 계수값,

는 세로 방향의 패턴 검출 필터의 계수값)(

Is the edge component for the horizontal direction,

Is the edge component for the longitudinal direction,

The

The value of the input pixel located at,

Is the coefficient value of the pattern detection filter in the horizontal direction,

Is the coefficient value of the vertical pattern detection filter)

The image conversion method using subpixel rendering according to an embodiment of the present invention selects a filter optimized for the input image from the generated filter set (S605).

In this case, in operation S605 of selecting a filter optimized for the input image, the filter optimized for the input image may be selected using metadata of the input image.

In this case, in operation S605 of selecting a filter optimized for the input image, the filter optimized for the input image may be selected according to the detected pattern.

The image conversion method using subpixel rendering according to an embodiment of the present invention filters each input subpixel of the input image to each output subpixel using the selected filter (S606).

A part not described with respect to an image conversion method using subpixel rendering according to an embodiment of the present invention may refer to FIGS. 1 to 5.

In addition, the image conversion method using sub-pixel rendering according to an embodiment of the present invention includes a computer readable medium containing program instructions for performing various computer-implemented operations. The computer readable medium may include program instructions, data files, data structures, and the like, alone or in combination. The media may be program instructions that are specially designed and constructed for the present invention or may be available to those skilled in the art of computer software. Examples of computer-readable media include magnetic media such as hard disks, floppy disks and magnetic tape; optical media such as CD-ROMs and DVDs; magnetic media such as floppy disks; Magneto-optical media, and hardware devices specifically configured to store and execute program instructions such as ROM, RAM, flash memory, and the like. The medium may be a transmission medium such as an optical or metal line, a wave guide, or the like, including a carrier wave for transmitting a signal designating a program command, a data structure, or the like. Examples of program instructions include machine language code such as those produced by a compiler, as well as high-level language code that can be executed by a computer using an interpreter or the like.

While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it is to be understood that the invention is not limited to the disclosed exemplary embodiments, but, on the contrary, Modification is possible. Accordingly, the spirit of the present invention should be understood only in accordance with the following claims, and all equivalents or equivalent variations thereof are included in the scope of the present invention.

1 is a block diagram illustrating an image conversion system using subpixel rendering according to an embodiment of the present invention.

FIG. 2 is a diagram illustrating a display having different subpixel structures and masking each subpixel through a filter mask in an embodiment of the present invention.

3 is a diagram illustrating a process of dividing an input pixel by each output subpixel and configuring a filter for the output subpixel according to an embodiment of the present invention.

4 is a diagram illustrating an example of a pattern detection filter capable of detecting a local pattern of an input image according to one embodiment of the present invention.

FIG. 5 illustrates an example of a distribution function of a filter for filtering an input image according to an embodiment of the present invention.

FIG. 6 is a flowchart illustrating an image conversion method using subpixel rendering according to an embodiment of the present invention.

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

101: filter set generator 102: sub-pixel rendering unit

103: contribution amount calculation unit 104: contribution amount normalization unit

105: filter configuration section 106: pattern detection section

107: filter selection unit 108: image filtering unit

Claims (25)

A filter set generator configured to generate at least one filter set for each output subpixel by using a sharing amount of each of the at least one output subpixel with respect to a value of each input subpixel for an input pixel; And A subpixel rendering unit which selects a filter optimized for an input image from the generated filter set and converts the value of the input subpixel into a value of the output subpixel through the selected filter Image conversion system comprising a. The method of claim 1, The filter set generator, A contribution amount for calculating a contribution amount of each of at least one output subpixel representing the same color as the input subpixel using a distribution function for the value of each input subpixel for the input pixel A calculator; A burden normalization unit for normalizing the burden amount with respect to the input subpixel using an area ratio between the input subpixel and the output subpixel; And A filter constructing unit for constructing a filter corresponding to each of the output subpixels by using the normalized sharing amount Image conversion system comprising a. 3. The method of claim 2, The contribution calculation unit, And convert the contribution amount according to at least one output subpixel included in the filter mask by masking the output subpixel around the input pixel through an N * N filter mask. system. 3. The method of claim 2, The distribution function is A Gaussian function using the input pixel and the output subpixel as parameters; The sharing amount which each output subpixel shares, And an output value obtained by integrating the output subpixels through the Gaussian function. 3. The method of claim 2, The burden normalization unit, And the contribution amount is normalized with respect to the input subpixel according to Equation 5 below.

(

Is the normalized contribution,

Is the area ratio of the input subpixel to the output subpixel,

Is a distribution function

Share using

Is the input pixel,

Is the output subpixel,

Is the index of the position of the input pixel in the filter mask) 3. The method of claim 2, The filter component, And a filter indicating a position at which each of the output subpixels and each of the input pixels overlap each other by collecting the normalized sharing amount according to the output subpixels. The method of claim 1, The subpixel rendering unit, A filter selector for selecting a filter optimized for an input image from the generated filter set; And An image filtering unit filtering each of the input sub-pixels of the input image to each of the output sub-pixels using the selected filter Image conversion system comprising a. The method of claim 7, wherein The filter selection unit, And a filter optimized for an input image from the filter set generated by using metadata about the input image. The method of claim 7, wherein The subpixel rendering unit, A pattern detector which determines edge components in the horizontal and vertical directions with respect to the input image, and detects a local pattern of the input image according to the edge components. More, The filter selection unit, And a filter optimized for the input image according to the detected pattern. 10. The method of claim 9, The pattern detection unit, And determining the direction of an edge through a ratio of the edge components, and detecting a pattern of the input image according to an angle formed by the edge direction with a horizontal line or a vertical line. 10. The method of claim 9, The pattern detection unit, And a local pattern of the input image is detected using a pattern detection filter. 12. The method of claim 11, The pattern detection unit, And an edge component of the input image is determined based on the value of the input pixel and the coefficient value of the pattern detection filter. Generating at least one filter set for each of the output subpixels by using a sharing amount of each of the at least one output subpixel with respect to a value of each input subpixel for an input pixel; And Selecting a filter optimized for an input image from the generated filter set and converting a value of the input subpixel to a value of the output subpixel through the selected filter Image conversion method comprising a. 14. The method of claim 13, The generating of at least one filter set for each output subpixel may include: Calculating a share of each of at least one output subpixel representing the same color as the input subpixel by using a distribution function of a value of each input subpixel for the input pixel; Normalizing the contribution amount with respect to the input subpixel using an area ratio between the input subpixel and the output subpixel; And Constructing a filter corresponding to each of the output subpixels using the normalized contribution amount Image conversion method comprising a. The method of claim 14, The calculating of the sharing amount of each of the at least one output subpixel representing the same color as the input subpixel may include: And masking the output subpixel with respect to the input pixel through an N * N filter mask to calculate the contribution amount according to at least one output subpixel included in the filter mask. The method of claim 14, The distribution function is A Gaussian function whose parameters are the input pixel and the output subpixel, The sharing amount which each output subpixel shares, And the output subpixel corresponding to a value obtained by integrating the output subpixels through the Gaussian function. The method of claim 14, The step of normalizing the contribution amount to the input subpixel, And the contribution amount is normalized with respect to the input subpixel according to Equation 6 below.

(

Is the normalized contribution,

Is the area ratio of the input subpixel to the output subpixel,

Is a distribution function

Share using

Is the input pixel,

Is the output subpixel,

Is the index of the position of the input pixel in the filter mask) The method of claim 14, The step of configuring a filter corresponding to each output subpixel, And a filter indicating a position at which each of the output subpixels and each of the input pixels overlap each other by collecting the normalized sharing amount according to the output subpixels. 14. The method of claim 13, The step of converting a value of an input subpixel to the value of the output subpixel through the selected filter, Selecting a filter optimized for an input image from the generated filter set; And Filtering each of the input subpixels for the input image to each of the output subpixels using the selected filter. Image conversion method comprising a. 20. The method of claim 19, The step of selecting a filter optimized for the input image from the generated filter set, And selecting a filter optimized for the input image from the filter set generated using the metadata of the input image. 20. The method of claim 19, The step of converting a value of an input subpixel to the value of the output subpixel through the selected filter, Determining an edge component in a horizontal direction and a vertical direction with respect to an input image, and detecting a local pattern of the input image according to the edge component; More, The step of selecting a filter optimized for the input image from the generated filter set, And selecting a filter optimized for the input image according to the detected pattern. 22. The method of claim 21, The step of detecting a local pattern of the input image according to the edge component, And determining a direction of an edge through a ratio of the edge components, and detecting a pattern of the input image according to an angle formed by the edge direction with a horizontal line or a vertical line. 22. The method of claim 21, The step of detecting a local pattern of the input image according to the edge component, And a local pattern of the input image is detected using a pattern detection filter. 24. The method of claim 23, The step of detecting a local pattern of the input image according to the edge component, And an edge component of the input image is determined based on the value of the input pixel and the coefficient value of the pattern detection filter. A computer-readable recording medium in which a program for executing the method of any one of claims 13 to 24 is recorded.

Images