TWI238011B - Methods and systems for sub-pixel rendering with gamma adjustment - Google Patents

Abstract

The present invention discloses methods and system for sub-pixel rendering with gamma adjustment. The gamma adjustment allows the luminance for the sub-pixel arrangement to match the non-linear gamma response of luminance channels of the human eyes. The chrominance can match the linear response of the chrominance channels of the human eyes. The gamma correction allows the algorithms to operate independently of the actual gamma value of a display device. The sub-pixel rendering with gamma adjustment disclosed can be optimized for the gamma value of a display device to improve the response time, dot inversion balance, and contrast because the gamma correction and compensation of the sub-pixel rendering algorithm provide the required gamma value through sub-pixel rendering. These techniques can be attached to any specified gamma transfer curve.

Description

psrnm

说明 Description of Invention Method and system for displaying sub-pixels with gamma adjustment on a display. BACKGROUND OF THE INVENTION The latest color single-plane imaging matrices for flat panel displays use RGB color triples, or a single color with vertical bars, as shown in the prior art of FIG. This system uses the Von Bezold color mixing effect (explained further below), which separates three colors and places equal spatial frequency weights on each color. However, these panels do not match well with human vision. Graphics rendering technology has been developed to improve the image quality of previous technology panels. U.S. Patent No. 5,341,153 by Benzschawel et al. Discloses how to reduce a larger image size to a smaller panel. To this end, Benzschawel et al. Have disclosed how to use the "sub-pixel rendering" technology now known in the art to improve image quality. More recently, U.S. Patent No. 6,188,385 to Hill et al. Discloses how to use very similar sub-pixel rendering techniques to lower the actual image of the text one character at a time. The previous techniques described above have inappropriate knowledge of the functioning of human vision. The image reconstructed by the display device in the prior art does not match well with human vision.

The main model used to sample or generate, and then store the images of these displays is RGB pixels (or tri-color pixel elements), where the red, green and blue values are on an orthogonal spatial resolution grid, and coincide. One of the results of using this image format is that it does not match the actual image reconstruction panel, its non-coincided color emitters, and human vision. This would essentially cause redundancy or wasted information in the image.

Martinez-Uriegas et al. U.S. Patent No. 5,398,066 and Peters et al. U.S. Patent No. 5,541,653 disclose a technique for converting and storing images from the RGB pixel format to a camera very similar to that proposed by Bayer in U.S. Patent No. 3,971,065 A color filter array of the imaging device. The advantage of the format of Martinez-Uriegas et al. Is that it simultaneously captures and stores the separated color component data at a similar spatial sampling frequency, just like human vision. However, its first disadvantage is that the format of Martinez-Uriegas et al. Does not match well with the color display panel of the world. For this reason, Martinez-Uriegas et al. Also revealed how to convert the image back to the RGB pixel format. Another disadvantage of the format of Martinez-Uriegas et al. Is that one of the color components is not sampled regularly, in this case red. There will be missing samples in the array, which reduces the structural accuracy of the image during display. Complete color perception is produced in the eye by three forms of color receptor nerve cells called pyramids. The three forms are sensitive to different wavelengths of light: long, medium, and short (red, green, and blue, respectively. The relative densities of the three wavelengths are significantly different from each other. The acceptors for red are slightly more than those for green. Compared to Red or green receptors, blue receptors are very few. In addition to the color terminal I23S011, ochre receptors, there are relatively wavelength-insensitive receptors, called rods, which will help monochrome night vision. The human vision system processes the information detected by the eyes in several perceptual channels: illumination, chromatic aberration, and motion. Motion is more important for imaging system designers only for the threshold of flicker. This illumination channel uses only red and green sensors. The input of the body is: color blindness. It processes information in a similar way and enhances the edge contrast. The color difference channel does not have edge contrast enhancement. Because the illuminance channel uses and enhances each red and green receptor The resolution of the illuminance channel is several times higher than the color difference channel. The contribution of the blue receiver hip to the illuminance perception is negligible. Therefore, by decreasing one octave The error caused by the blue resolution is almost unnoticeable to most perceptual viewers, as it is completely absent. It can be seen in experiments by Xerox and NASA, Ames Research Center (see R. Martin, J. Gille, J ·

Marimer, Detectability of Reduced Blue Pixel Count in Projection Displays, SID Digest 1993). Color perception is affected by what is called, assimilation, or the Von Bezold color mixing effect. This allows a display's separated color pixels (or sub-pixels or emitters) to be perceived as mixed colors. This mixed effect occurs within a given angular distance in the viewing range. Due to the relatively scarce blue receptor, this mixing will occur at a greater angle to blue than red or green. This distance is approximately 0.25 for blue. , And for red or color recording is about 0.12. . At a viewing distance of 12 inches, 0.25. Opposite 50 mils (1,270 μ) on a display. Therefore, if the blue sub-pixel pitch is less than half (625 μ) of this blending pitch, the colors will be blended without loss of image quality. In its most simplified implementation, the sub-pixel rendering operates as a sub-pixel with approximately equal brightness pixels as perceived by the illuminance channel. This allows the sub-pixel to be used as a sampled image reconstruction point, as opposed to using the combined sub-pixel as part of a "real" pixel. By using sub-pixel rendering, the spatial sampling can be increased and decreased This phase error. If the color of the image is to be ignored, each sub-pixel can be considered equal to each other, as if it were a monochrome pixel. However, because color is almost always important (this is why people use a color Monitor), the color balance of a given image is important at every position. Therefore, the sub-pixel rendering algorithm must ensure that the spatial frequency information in the illuminance component of the image to be rendered will not be renamed the color Subpixels cause color errors to maintain color balance. US Patent No. 5,341,153 by Benzchawel et al. And US Patent No. 6,188,385 by Hill et al. Are similar to a commonly used anti-aliasing technique, and their application replaces the elimination Filter to each separate color component of a higher resolution virtual image. This ensures that the illuminance information is not in each color channel If the sub-pixel configuration is optimized for sub-pixel rendering, the sub-pixel rendering will provide spatial addressability that simultaneously increases lower phase errors, and increases the high spatial frequency of the modulation transfer function (MTF) on both axes Resolution. The conventional RGB bar display in inspection view 1. The sub-pixel rendering can only be applied on the horizontal axis. The blue sub-pixel is not perceived by the human illumination channel, so it has no effect on the sub-pixel rendering. Because Only red and green pixels can be used for sub-pixel rendering, and the effective increase in addressability is two crossings on the horizontal axis-10-. The vertical black and white lines must have two main sub-pixels in each column (ie Each black or white line is red and green). This is the same number as the image used for non-subpixel rendering. The MTF, which has the ability to display a given number of lines and spaces at the same time, is Pixel rendering has been improved. Therefore, the conventional RGB stripe sub-pixel configuration shown in Figure 1 is not optimized for sub-pixel rendering. The prior art three-color pixel device configuration is At the same time, the matching of human vision and the general technology of the sub-pixel rendering is very poor. Similarly, the previous image formats and conventional methods are poorly matched for human vision and the actual color emitter configuration. Sub-pixel rendering is another One complexity is dealing with non-linear responses (such as a gamma curve) to the human eye and display device ’s brightness or illuminance, such as a cathode ray tube (CRT) device or a liquid crystal display (LCD). However, sub-pixel compensation The rendered gamma value is not a cumbersome process. That is, it is difficult to provide a high contrast and correct color balance of the sub-pixel rendered image. Furthermore: the prior art sub-pixel rendering systems did not properly provide accurate Gamma control to provide high-quality images. SUMMARY OF THE INVENTION A method for processing data to a display is disclosed. The display includes pixels with color sub-pixels. It receives the pixel data and applies a gamma adjustment to the pixel data converted to the data presented by a pixel. This conversion can produce data for that sub-pixel representation of a single-pixel configuration. The sub-pixel arrangement includes alternating red and green sub-pixels on at least one of a horizontal axis and a vertical axis. The data presented by this sub-pixel is output to

The display. The invention discloses a display system with a plurality of pixels. The pixel may have a primary pixel configuration including alternating red and green sub-pixels on at least one of a horizontal axis and a vertical axis. The system also includes a controller coupled to the display and processes pixel data. The controller also applies a gamma adjustment to convert from the pixel data to the data presented by the sub-pixels. The conversion can generate data presented by the sub-pixels of the sub-pixel configuration. The controller outputs the data presented by the sub-pixel on the display. Other features and benefits of the present invention will be understood from the following detailed description. Brief Description of the Drawings The accompanying drawings are incorporated by reference to form a part of this application to illustrate the invention, and are provided with explanations to explain the principles of the invention. In the drawings, FIG. 1 shows the prior art RGB stripe arrangement of three color pixel elements in an array for a display device on a single plane; FIG. 2 shows the prior art RGB stripe of FIG. 1 The effective sub-pixel configuration presents sampling points; Figs. 3, 4 and 5 do not represent the effective sub-pixel for each color plane of the sampling points of the RGB stripe configuration of the prior art of Fig. 1; Shown is the arrangement of three-color pixel elements in a single plane and an array for a display device; Figure 7 shows the effective sub-pixel rendering sampling points of the configurations of Figures 6 and 27; Figures 8 and 9 are diagrams Another effective sub-pixel of the blue plane sampling points arranged at 6 and 27 presents the sampling area; -12- 修 1238011, ·: 々 · 卜 ... Figure 10 shows a display device on a single plane, an array Another configuration of the three-color pixel element in Figure; Figure 11 shows the effective sub-pixel rendering sampling points of the configuration of Figure 10; Figure 12 shows the effective sub-pixel sampling points of the blue plane configuration of Figure 10 Pixels present sampling areas; Figures 13 and 14 show Figure 6 In the configuration of 10, the effective sub-pixels of the red and green planes present sampling areas. Figure 15 shows an array of identical points and the effective sample areas of the pixel data format of the prior art, where the red, green, and blue The values are in an equal spatial resolution grid and coincide; FIG. 16 shows the array of the sample points of FIG. 15 of the prior art overlaid on the sample points of the sub-pixel presentation of FIG. 11, where the sample points of FIG. 15 are related to The red and green “inspection board” arrays in FIG. 11 have the same spatial resolution grid and overlap; FIG. 17 shows an array of sample points, and its prior art. The effective sample area of FIG. 15 covers the blue plane of FIG. 12 In the sampling area, the sample point of the previous technique in FIG. 15 has the same spatial resolution grid as the red and green “check board” array of FIG. 11 and coincides; FIG. 18 shows the array of sample points and its previous technique. The valid sample area of FIG. 15 is overlaid on the red plane sampling area of FIG. 13. The sample points of the prior art in FIG. 15 have the same spatial solution as the red and green “check board” array of FIG. 11. Degree grids are coincident; Figures 19 and 20 show the array of sample points and their previous techniques. The valid sample area of Figure 15 covers the blue plane sampling area of Figures 8 and 9, where -13-fMWF-front The sample points of Figure 15 are the same spatial resolution grids as the red and green "check board" arrays of Figure 7; they are coincident; Figure 21 shows an array of the same points and the pixel data of the previous technology. The valid sample area of the format, where the red, green and blue values are in an equal spatial resolution grid, and coincide; Figure 22 shows the array of sample points, and its prior art. The valid sample area of Figure 21 covers In the sampling area of the red plane in FIG. 13, the sample points in FIG. 21 do not overlap with the red and green “check board” arrays in FIG. 11 and have the same spatial resolution grid; FIG. 23 shows the sample points. The array and its previous technique The effective sample area of FIG. 21 covers the blue plane sampling area of FIG. 12, where the sample points of FIG. 21 of the prior art are not the same as the red and green “check board” array of FIG. Have the same space The resolution grids also do not overlap; Figure 24 shows the array of sample points and their prior art. The effective sample area of Figure 21 covers the blue flat sampling area of Figure 8. Among them, the sample points of Figure 21 of the previous technology. It does not overlap with the red and green colors in FIG. 7, and the checkerboard array has the same spatial resolution grid; FIG. 25 shows that the valid sample area on the red plane of FIG. 3 covers the sampling area on the red plane of FIG. 13. Fig. 26 shows that the effective sample area of the blue plane of Fig. 5 is overlaid on the sample area of the blue plane of Fig. 8; Fig. 27 shows the three-color pixels in an array of three panels for a display device Another configuration of the components; Figures 28, 29 and 30 show the configuration of the blue 5? 2 color, green and red emitters on each separate panel of the device of Figure 27; Figure 31 shows the diagram The output sample configuration 200 of 11 is overlaid on top of the input sample configuration 70 of FIG. 15 in a special case. When the adjustment ratio is crossed for each two (one red and one green) output sub-pixel as one input pixel; FIG. 32 shows A single repeating unit 202 to convert a 650x480 VGA format image to a PenTile matrix, which has 800x600 total red and green sub-pixels; Figure 33 shows the symmetry in the coefficients of a three-color pixel element below the size of the repeating unit is odd; FIG. 35 shows an example when the size of the repeating unit is even; FIG. 35 shows the sub-pixel 218 from FIG. 33 defined by a presentation area 246, which covers the surrounding 6 input pixel sample areas 248; 36 shows that the sub-pixel 232 and its rendering area 250 from FIG. 33 cover 5 sample areas 252; FIG. 37 shows that the sub-pixel 234 and its rendering area 254 from FIG. 33 cover the sample area 256; FIG. 38 shows The sub-pixel 228 and its presentation area 258 from FIG. 33 cover the sample area 260; FIG. 39 shows that the sub-pixel 236 and its presentation area 264 from FIG. 33 cover the sample area 262; and FIG. 40 shows that it is used to generate blue filtering. The square sampling area of the processor core; FIG. 41 shows the hexagonal sampling area 123 of FIG. 8 with respect to the square sampling area 276; FIG. 42A shows a re--15-

An example of the sampling area indicates the sample area, and FIG. 42B shows an exemplary configuration of three-color sub-pixels on a display device; FIG. 43 shows an exemplary input sine wave; An exemplary image of the output when the input image is subjected to sub-pixel rendering without gamma adjustment; FIG. 45 shows an exemplary display function diagram describing the color error, which can occur using sub-pixel rendering without gamma adjustment; FIG. 46 shows a flowchart of a method for applying a pre-adjusted gamma value before sub-pixel rendering; FIG. 47 shows an exemplary graphic of the output when the input image of FIG. 43 is subjected to gamma-adjusted sub-pixel rendering Figure 48 shows the calculation of the partial coverage of the sample area indicated in Figure 42A; Figure 49 shows a flowchart of a method for sub-pixel rendering for gamma adjustment; Figure 50 is not the input of Figure 43 An exemplary graphic of the output when the image is presented with a gamma-adjusted sub-pixel with an omega function; FIG. 51 shows the flow of a method with the omega-adjusted gamma-adjusted sub-pixel rendering 52A and 52B show an exemplary system to apply the method of FIG. 46 to apply a pre-adjusted gamma value before sub-pixel rendering; FIGS. 53A and 53B show the method of FIG. 49 to perform gamma adjustment Exemplary system presented; Figures 54A and 54B show the implementation of the method of Figure 51 to have an omega-16-

mmM uu.

Exemplary system for gamma-adjusted sub-pixel rendering of the I function; Figures 55 to 60 show exemplary circuits that can be used in the processing blocks of Figures 52A, 53A, and 54A; Figure 61 shows timing during sub-pixel rendering Flow chart of a method of black pixels at the edges; FIGS. 62 to 66 show exemplary block diagrams of a system for improving the color resolution of an image on a display; FIGS. 67 to 70 show a function evaluator that performs high-speed mathematical calculations Exemplary specific embodiments; FIG. 71 shows a flowchart of a sub-presentation process with a gamma adjustment method implemented by software; and FIG. 72 shows a software process of implementation of FIGS. 46, 49, and 51 and / or FIG. 71 The internal block diagram of an exemplary computer system. Detailed Description of the Invention Reference will now be made in detail to the implementation and specific embodiments of the present invention illustrated in the accompanying drawings. Wherever possible, the same reference numbers will be used throughout the drawings and the description below to refer to the same or like parts. A real world image is captured and stored in a memory device. The stored images are generated with some known data configurations. The stored image can be presented on a display device using an array, which can provide improved resolution of a color display. The array contains a plurality of three-color pixel elements, which have at least a blue emitter (or sub-pixel), a red emitter, and a green emitter, which can be mixed to produce all visible to the human eye when illuminated. Other colors.

In order to determine the value of each emitter, a conversion equation in the form of a filter core must first be generated. The core of the filter is generated by determining the relative area of the original data set sample area and the target display sample area overlap. The overlap ratio determines the coefficient values to be used in the filter core array. In order to present the stored image on the display device, the reconstruction point is determined in each of the three-color pixel elements. The center of each reconstruction point can also be the source of the sample points used to reconstruct the stored image. Similarly, the sample points of the image data set can be determined. Each reconstruction point is located at the center of the radiator (for example, when the center of a red radiator is placed at the center of the radiator, a grid of boundary lines is equal to the center of the reconstruction point, etc. Distance from the ground to generate a sample area (where the sample point is located in the center of the grid to produce a brick pattern! The shapes available in this brick pattern can include, but are not limited to, squares, intersecting rectangles, triangles, Hexagons, octagons, rhombuses, staggered squares, staggered rectangles, staggered triangles, staggered rhombuses, Penrose tiles, rhombuses, twisted rhombuses, and lines, and combinations comprising at least one of the foregoing shapes. The image data and the sample points and sample areas displayed by the target have been determined, two of which are overlapped. The overlap produces a secondary area, where the output sample area covers several input sample areas. The area ratio of the input to output is checked by inspection Or calculate the decision and store it as a coefficient in the filter core. Its value is used to weight the input value to the output value to determine each Appropriate value of the emitter. When using a sufficiently high adjustment ratio, the sub-pixel configuration disclosed here -18-

And the rendering method provides better image quality, measurement information addressability, and reconstructed image modulation conversion function (MTF) than previous technology displays. In addition, it reveals the sub-pixel rendering with gamma adjustment. Methods and systems. The data can be processed on a display with color sub-pixel pixels. In particular, the pixel data can be received, and a gamma value can be applied to adjust the conversion from the received pixel data to the data presented by the sub-pixels. This transformation can be used for a single pixel configuration to generate the data presented by that sub-pixel. The sub-pixel configuration may alternate red and green sub-pixels on at least one of a horizontal axis and a vertical axis, or any other configuration. The data presented by the sub-pixel can be output to the display. Because the human eye cannot distinguish between absolute brightness or illuminance values, it needs to improve illuminance contrast, especially at high spatial frequencies, to get higher quality images. As explained in detail below, by adding gamma adjustment to sub-pixel presentation, the illuminance or brightness contrast ratio can be improved for the sub-pixel configuration on a display. Therefore, by improving this contrast ratio, a higher quality image can be obtained. This gamma adjustment can be precisely controlled for a given sub-pixel configuration. FIG. 1 shows the RGB stripe configuration of three-color pixel elements in an array for a single plane of a display device in the prior art, and FIG. 2 illustrates the effective times of the RGB stripe configuration of the prior art in FIG. 1 The pixels represent sampling points. 3, 4 and 5 show the effective sub-pixel rendering sampling area for each color plane of the sampling points of the RGB stripe arrangement of the prior art of FIG. Figures 1-5 are discussed further here. | I23 # ails year month day Figure 6 shows a configuration 20 of several tri-color pixel elements according to a specific embodiment. The three-color pixel element 21 is square and is placed at the origin of the X, Y coordinate system, and includes a blue radiator 22, two red radiators 24, and two green radiators 26. The blue radiator 22 is placed at the center, extends vertically along the X axis of the coordinate system, and extends to the first, second, third, and fourth quadrants. The red radiator 24 is disposed in the second and fourth quadrants, and is not occupied by the blue radiator. The green radiator 26 is placed in the first and third quadrants, and is not occupied by the blue radiator. The blue radiator 22 is rectangular, and its sides are aligned with the X and Y axes of the coordinate system. The relative pairing of the red 24 and the green 26 is usually square. The array is repeated on a panel to complete a device with a desired matrix resolution. The repeated three-color pixel element forms a "check plate", which alternates the red 24 and green 26 emitters with the blue emitter 22, which are evenly distributed in the device, but the red 24 and green 26 emitter Half the resolution. Every other line of the blue emitter is staggered or offset by half its length, as represented by emitter 28. To accommodate such a configuration, and because of edge effects, some blue radiators may be half the size of the blue radiator 28 on the edges. FIG. 7 shows a configuration 29 of effective sub-pixel rendering sampling points for the configurations of FIGS. 6 and 27, and FIGS. 8 and 9 show alternate effective sub-pixel rendering of the blue plane sampling points 23 for the configurations of FIGS. 6 and 27. The sampling areas 123 and 124 are arranged 30 and 31. Figures 7, 8 and 9 are discussed further here. Fig. 10 shows another embodiment 38 of the arrangement 38 of a three-color pixel element 39. The three-color pixel element 39 includes a blue radiator 32, two red radiators 34, and two green radiators 36 in a square shape. The three-color pixel element 39 is a square, which is centered on the origin of an X and Y coordinate system. The blue emitter 32 is centered on the origin of the square and extends to the first, second, third, and fourth quadrants of the X, Y coordinate system. A pair of red radiators 34 are placed in the opposite quadrants (that is, the second and fourth quadrants), and a pair of green radiators 36 are placed in the opposite quadrants (that is, the first and third quadrants). The part of the quadrant occupied by the blue radiator 32. As shown in FIG. 10, the blue radiator 32 is diamond-shaped, and its corners are aligned with the X and Y axes of the coordinate system, and the red 34 and green The relative pairing of 36 is usually square, with truncated inwardly facing corners, forming sides parallel to the blue emitter 32. The array is repeated on a panel to complete a desired Device for matrix resolution. The repeated three-color pixels form one, check board, which alternates the red 34 and green 36 emitters and the blue emitter 32, which are evenly distributed in the device, but the red Half of the resolution of the 34 and green 36 emitters. The red emitters 34a and 34b will be discussed further here. The benefit of the three-color pixel element is that it improves the resolution of the color display. This occurs because only the red and Green emitters can clearly contribute to perception High resolution in the illuminance channel. Therefore, reducing the number of blue emitters and replacing some red and green emitters can improve the resolution to more closely match human vision. Distinguish half of the vertical axis Red and green emitters to increase spatial addressability can improve the vertical signal color bars used in previous techniques. An alternating red and green emitter ff inspection board allows high spatial frequencies -21-

H.1 * xi —i_ & [rate resolution to increase the horizontal axis and vertical axis at the same time. In order to reconstruct the image in the first data format to the display in the second data format, it is necessary to isolate the reconstruction point at the geometric center of each radiator and generate a sampling grid. Fig. 11 shows an effective reconstruction point arrangement 40 of the arrangement 38 of the three-color pixel element of Fig. 10. The reconstruction point (as shown in 33, 35, and 37 in FIG. 11) is centered on the geometric position of the emitter (in FIGS. 10, 32, 35, and 36) in the three-color pixel element 39. The red reconstruction points 35 and the green reconstruction points 37 form a red and green "check board" array on the display. The blue reconstruction points 33 are evenly distributed on the device, but are half the resolution of the red 35 and green 37 reconstruction points. For sub-pixel rendering, the three-color reconstruction points are treated as sampling points and are used to construct an effective sampling area for each color plane, which are processed independently. Fig. 12 shows the effective blue sampling points 46 (corresponding to the blue reconstruction points 33 of Fig. 11) and the sampling area 44 of the blue plane 42 of the reconstructed array of Fig. 11. For a square grid of reconstruction points, the minimum boundary perimeter is a square grid. FIG. 13 shows the effective red sampling point 51, which corresponds to the red reconstruction point of FIG. 11 and the red reconstruction point 25 of FIG. 7, and the effective sampling areas 50, 52, and 53 of the red plane 48. And 54. The sampling points 51 form a square grid array and are 45 with the display boundary. . Therefore, within the center array of the sampling grid, the sampling area forms a square grid. Due to the "edge effect", where the square grid will overlap the border of the display, its shape can be adjusted to maintain the same area and minimize the boundary perimeter of each sample (such as 54). Checking the sample area will It is revealed that the sample area 50 will have the same area as the sample area 52, but the sample area 54 will have a slightly larger area.

? OIW

A slightly larger area, while the sample area 53 in the corner is slightly smaller. This pair causes an error in which the data that changes in the sample area 53 will exceed what is represented, and the data that changes in the sample area 54 will be lower than that represented. However, in a display with hundreds of thousands or millions of emitters, the error will be minimal and disappear in the corners of the image. FIG. 14 shows the effective green sampling point 57, which corresponds to the green reconstruction point 37 of FIG. 11, the green reconstruction point 27 of FIG. 7, and the effective sampling areas 55, 56, 58 of the green plane 60 and 5 9. The inspection view 14 will find that it is basically similar to FIG. 13 with the same sample region relationship, but rotated 180 degrees. . The configuration of these emitters and the sample points and areas obtained by them can be used directly by the drawing software to generate high-quality images, and the drawing elements or vectors are converted to offset the color sample plane. It combines the sampling technology of the prior art and the sampling. Points and areas. Complete graphics display systems, such as portable electronic devices, laptop and desktop computers, and television / video systems, preferably use flat panel displays and these data formats. The display form used may include, but is not limited to, a liquid crystal display, a subtraction display, a plasma panel display, a cold light (EL) display, an electrophoretic display, a field emitter display, a discrete light emitting diode display, an organic light emitting diode ( A QLED) display, a projector, a cathode ray tube (CRT) display, and the like, and a combination including at least one of the foregoing displays. However, the drawing foundation and drawing software installed by Mu Wu Duo uses a previous data sample format, which was originally based on the use of a CRT as a reconstructed display. FIG. 15 shows an array of the same point 74 and the valid sample area 72 of the prior art pixel data format 70, where the red, green, and blue values ea are in an equal spatial resolution grid and coincide. In the prior art display system, this data format was reconstructed on a flat panel display by using only data from each color plane on a prior art RGB strip panel of the format shown in FIG. In the figure, the resolution of each color sub-pixel is the same as that of the sample point, and the three sub-pixels are regarded as a column, although it constitutes a single combined and mixed multi-color pixel, and each color sub-pixel is ignored. Pixel's actual reconstruction point position. In the art, this is commonly referred to as the "essential mode" of the display. This will waste the position information of the sub-pixel, especially the red and green. In contrast, the input RGB data of this application are considered as three planes overlapping each other. To convert data from this RGB format, each plane is processed independently. Display information from the original prior art format on the more efficient sub-pixel configuration of this application requires resampling to convert the data format. The data is resampled in such a way that the output of each sample point is a weighted function of the input data. The weighting function can be the same or different at each output sample point based on the spatial frequency of the separated data samples, as described below. FIG. 16 shows the arrangement 76 of the sample points of FIG. 15 over the sample points 33, 35, and 37 of the sub-pixel presentation of FIG. 11, where the sample point 74 of FIG. 15 is in red with the red (red reconstruction point 35) of FIG. ) And green (green reconstruction point 37) "check board" arrays have the same spatial resolution grid and coincide. FIG. 17 shows the sample point 74 of FIG. 15 and the configuration 78 of the effective sample area 72 covering the blue plane sampling point 46 of FIG. 12, where the sample point 74 of FIG. 15 is the same as the red (red Reconstruction point 35) and green (green reconstruction -24- I2J80i l · point 3 7) "inspection board, the arrays have the same spatial resolution grid and coincide. Figure 17 will be further discussed here. Figure 18 Shown as an array 80 of sample points 74, and the red sample area 35, and the effective sample area 72 of FIG. 15 over the red sample areas 50, 52, 53, and 54 of FIG. 13, where the sample of FIG. 15 The point 74 is the same as the red (red reconstructed point 35) and green (green reconstructed point 37) of FIG. 11, the "check board" array has the same spatial resolution grid and coincides. The inner array of the square sample area 52 completely covers the coincident original sampling point 74 and its sampling area 82, and extends to cover a quarter of each of the surrounding sample areas 84 within the sample area 52. In order to determine the algorithm, the proportion of the output sample area 50, 52, 53 or 54 that covers or overlaps the input sample area 72 is recorded, then multiplied by the value corresponding to the sample point 74, and applied to the output sample Area 3 5. In Fig. 18, the area of the square sample area 52 is filled by the central or coincident input sample area 84, which is half of the square sample area 52. Therefore, the value of the corresponding sample point 74 is multiplied by one-half (or 0.5). By inspection, the area of the square sample area 52 filled by each surrounding non-overlapping input area 84 is each one-eighth (or 0.125). Therefore, the value of the corresponding four input sample points 74 is multiplied by one-eighth (or 0.125). These values are then added to the previous value (eg, multiplied by 0.5) to find the final output value for a given sample point 35. For the edge sample point 3 5 and its five side sample areas 50, the coincident input sample area 82 is completely covered, as in the above example, but only three surrounding input sample areas 8 4, 86, and 92 are overlapping. The overlapping input sample -25- one of the regions 84 represents one-eighth of the output sample region 50. Adjacent input sample regions 86 and 92 along the edge represent three-sixteenths (3/16 = 0.1875) of each of the output regions. As mentioned earlier, the weighted value of the input value 74 from the overlapping sample area 72 is added to obtain the value of the sample point 3 5. This corner and the "close" corner are considered the same. Because the corner 53 and the "close", the image area of the corner 54 covers an area different from the central area 52 and the edge area 50. The weighting of the input sample areas 86, 88, 90, 92, 94, 96, and 98 will be proportional to The aforementioned input sample areas 82, 84, 86, and 92 differ. For smaller corner output sample areas 53, the coincident input sample area 94 covers four-sevenths (or about 0.5714) of the output sample area 53. The The adjacent input sample area 96 covers three-fourteenths (or about 0.2143) of the output sample area 53. For this, 'close' to the corner sample area 54, the coincident input sample area 90 covers the output sample area Eighty-seventeenths of 54 (or about 0.4706). The internal adjacent sample region 98 covers two-seventeenths (or about 0.16) of the output sample region 54. The marginal adjacent input sample region 92 covers three-seventeenths of the output sample region 54. (Or about 0.1765). The corner input sample area 88 covers four-seventeenths (or about 0.2353) of the output sample area 54. As mentioned above, the weighted value of the input value 74 from the overlapping sample area 72 is added to obtain the value of the sample point 35. The resampling of the green plane is calculated in a similar manner, but the output sample array is rotated by 180 °. To restate, the values of the red sample point 35 and the green sample point 37 -26- ν_ are calculated as follows: The central area:

Vout (CxRy) = 0.5_Vin (CxRy) +0.125-V ^ CxqRyhO.US—Vin (CxRy + 1) + 0.125—Vin (Cx + 1Ry) + 0.125—VJCxRy]) The lower edge:

Vout (CxRy) = 0.5_Vin (CxRy) + 0.1875 ^ (0,. ^^ 0.1875_Vin (CxRy + 1) + 0.125_Vin (Cx + 1Ry)

Top edge:

VoutiC.RO ^ O.5 ^, (0 ^ 0 + 0.1875 ^ (0,. ^ 0 + 0.125 ^ (CXR2) + 0.1875—VJCx + A) Right edge:

Vout (CxRy) = 0.5 ^ (0 ^) + 0.125 ^ (^^^ 0.1875 ^ (0 ^ 0 + 0.1875 ^ (0 ^ .0) Left edge:

VoutiC ^^ O.5 ^, (0 ^^ + 0.1875 ^ (0 ^^ 0 + 0.125 ^ (C2Ry) + 0.1875— 乂 以. ~) Upper right corner:

Vout (CxRy) = 0.5714_Vin (CxRy) + 0.2143—VJCwRyH 0.2143_Vin (CxRy + 1) Upper left corner:

VoJQR ^ OMH—Vin (C1R1) + 0.2143_Vin (C1R2) + 0.2143_Vin (C2Ri) Bottom left corner: V0Ut (CxRy) Two 0.5714_Vin (CxRy) + 0.2143—Vin (Cx + 1Ry) + -27-

0.2143 ^ (0 ^, ^) Bottom right corner: V〇ut (CxRy) = 0.5714 ^ (0 ^) + 0.2143 ^ (0, ^^) + 0.2143 ^ (0 ^ .0 Above the edge, the left is close to the corner: V 〇ut (C2R1) = 0.4706_Vin (C2R1) + 0.2353_Vin (C1R1) +0.1176

Vin (C2R2) + 0.1765_Vin (C3Ri) Left edge, close to corner above:

Vout (C1R2) = 0.4706_Vin (C1R2) + 0.1765_Vin (C1R3) +0.1176 Vin (C2R2) + 0.2353 ^ (0 ^ 0 Left edge, close to the corner below: V〇ut (C1Ry) = 0.4706_Vin (C1Ry) + 0.2353_Vin (C1Ry + 1) + 0.1176 ^^ 02 ^) + 0.1765 ^ (01 ^ .0 bottom edge, left corner close to corner:

Vout (C2Ry) = 0.4706—Vin (C2Ry) + 0.2353—ViJC ^ RyH 0.1765_Vin (C3Ry) + 〇. 1176 ^ ,, (02 ^ .0 below the edge, the right is close to the corner:

Vout (CxRy) == 0.4706_Vin (CxRy) + 0.1765 ^ (0, ^^) + 0.2353 ^ (0 ^^,) + 0.1176 ^ (0 ^ 0 The right edge, below the corner:

Vout (CxRy) = 0.4706—Vin (CxRy) + 0.1176_Vin (Cx_1Ry) + 0.2353 ^ (0 ^ 0 + 0.1765 ^ (0 ^ .0 right edge, close to corner above:

Vout (CxR2) = 0.4706_Vin (CxRy) + 0.1176_Vin (Cx-1R2) + -28- 丨 complex

Fiber J 0.1765 ^ (0 ^) + 0.2353 ^ (0 ^ 0 Upper edge, right corner close to corner:

VoJCxR ^ OJTi ^ -Vin (CxR1) + 0.1765_Vin (Cx_1R1) + 0.1176 a ^ ((^ 2) + 0.2353- ^) where Vin is only the color difference value of the color of the sub-pixel at CxRy (CXR indicates red 34th and green 36th pixels in the Xth row, and Ry represents red 34 and green 36th pixels in the yth column, so CxRy represents red 34 or green 36th pixels in the Xth and y columns of the display panel The radiator, starting from the upper left corner, as usual. It is important to note that the sum of the coefficient weights in each equation is added up to one at most. Although there are 17 equations to calculate the complete image transformation, due to this symmetry, there are only four sets of coefficients. This can reduce its complexity when implemented. As mentioned above, FIG. 17 shows the sample point 74 of FIG. 15 and the configuration 78 of the effective sample area 72 covering the blue plane sampling point 46 of FIG. The red (red reconstruction point 35) and green (green reconstruction point 37) "check board" arrays of 11 have the same spatial resolution grid and coincide. The blue sample point 46 of Fig. 12 allows the blue sample area 44 to be determined by inspection. In this example, the blue sample area 44 is now a blue resampled area, which is only the arithmetic average of the blue values around the original data sample point 74, which is calculated as the sample point of the resampled image The value of 46. The blue output value V_ of the sample point 46 is calculated as follows: V〇ut (Cx + _Ry +) = 0.25_Vin (CxRy) + 0.25.Viri (CxRy + 1) +

Year and month Ej 0.25_Vin (Cx + 1Ry) + 0.25_Vin (Cx + 1Ry + 1) where Vin is the blue color difference value of the surrounding input sample point 74, Cx represents the X line of sample point 74, and Ry represents the sample point The y-th column of 74 starts from the upper left corner, as usual. For the calculation of blue sub-pixels, the X and Y values must be odd, as if each pair of red and green sub-pixels has only one blue sub-pixel. Again, the coefficient-weighted sum is equal to the value 1. The weighting of the coefficients of the central region equation of the red sample point 35 affects most of the generated images and is applied to the central resampling region 52, which is the processing of binary offset division, where 0.5 is to the "right" Offset by one bit, 0.25 is the direction, right π is offset by two bits, and 0.125 is the direction, 'right ff is offset by three bits. Therefore, the algorithm is relatively simple and fast, it only involves simple offset division and addition. For maximum accuracy and speed, the addition of the surrounding pixels must be completed first, then shifted to the right by a single three bits, and then the central value of the single bit offset is added. However, the latter involves more complex multiplications for the equations of the red and green sample areas at the edges and corners. On a small display (such as a display with a few pixels in total), it requires a more complicated equation to ensure good image quality display. For larger images or displays, where small errors at the edges and corners are irrelevant, they can be simplified. For this simplification, the first equations of the red and green planes are applied to edges and corners, and there are "missing" input data sample points on the edges of the image, so that the input sample point 74 is set equal to the coincident input sample Point 74. In addition, the "missing" value can be set to black. This algorithm can be easily implemented in software, firmware -30-

Or hardware. 19 and 20 show two other configurations 100, 102 of the sample point 74, and the effective sample area 72 of FIG. 15 overlaid on the blue plane sampling area 23 of FIGS. 8 and 9, where the sample point of FIG. 15 The 74 series has the same spatial resolution grid as the red and green "inspection board" array of FIG. 7 and coincides. FIG. 8 shows the effective sub-pixel rendering sampling area 123. For the radiator configuration of FIG. 6, it has The smallest boundary perimeter of the blue plane sampling point 23 shown in Fig. 7. The method of calculating the coefficient is performed as described above. The proportional overlap of the output sample area 123 covering each input sample area 72 of Fig. 19 It is calculated and used as the coefficient of a conversion equation or filter core. These coefficients are multiplied by the sample value 74 in the following conversion equation: Vout (Cx + _RY + _) = 0.015625—Vin (Cx_1Ry) + 0.234375_Vin (CxRy) + 0.234375—Vin (Cx + 1Ry) +0.015625 One Vin (Cx + 2Ry) + 0.015625 ^ ((^ 1 ^) + 0.234375 ^ (^ 1 ^) + 0.234375_Vin (Cx + 1Ry + 1) H-0.015625_Vin (Cx + 2Ry + 1) Those skilled in the art can find ways to perform these calculations quickly. For example, the coefficient is 0.01562 5 is equal to 6 bits offset to the right. The sample point 74 in FIG. 15 has the same space as the red (red reconstruction point 25) and green (green reconstruction point 27) arrays in FIG. In the case of resolution grid and coincidence, this minimum boundary condition region will cause an increased computational burden, and expand the data to 6 sample points 74. Another effective output sample region 124 configuration 31 of Figure 9 can be used in some applications Or occasion. For example, the sample point 74 in FIG. 15 has the same spatial resolution as the red (red reconstruction point 25) and green (green reconstruction point 27) "check board" arrays in Figure 7-31. When the grids are coincident, or the relationship between the input sample area 74 and the output sample area is shown in Figure 20, the calculation is relatively simple. In the even row, the formula for calculating the blue output sample point 23 is the same The formula developed in Figure 17 above. In this odd line, the calculation in Figure 20 is as follows: V〇ut (Cx + _Ry _) = 0.25_Vin (CxRy) + 0.25_Vin (Cx + 1Ry) + 0.25__Vin (CxU + 0.25—Vin (Cx + 1Ry ·. As usual, the calculations in Figures 19 and 20 above are for The general condition of the central sample area 124 is completed. The calculation at the edge will require correction of the conversion formula or assumptions regarding the value of the sample point 74 at the edge of the screen, as described above. Now refer to FIG. 21, The array 104 and the effective sample area 120 of the sample points 122 in the pixel data format of the prior art are described. Figure 21 shows a grid with equal spatial resolution and overlapping red, green, and blue values, but it has a different image size than the image size shown in Figure 15. Fig. 22 shows the array 106 of the same point 122 and the effective sample area 120 of Fig. 21 overlaid on the red plane sampling areas 50, 52, 53 and 54 of Fig. 13. The sample points 122 in FIG. 21 are not the same as the red (red reconstruction points 25, 35) and green (green reconstruction points 27, 37) "inspection boards in Fig. 7 or 11, respectively. The array has the same spatial resolution grid. In the configuration of Figure 22, it is not allowed to perform a single simplified conversion equation calculation for each output sample 35. However, the generalization is used to generate each calculation based on the covered proportional area. All methods are possible and practical. This is -32-

123101MI rj: Because for any given input-to-output image ratio, especially those that are commonly used in the industry to become standard, it can be established, it will have the smallest common denominator ratio, which will cause the image to be converted into a repeated unit pattern. Complexity can be further reduced due to symmetry, as shown above for coincident input and output arrays. When combined, the repeated three-color sample points 122 and symmetry cause the number of combinations of the unique coefficients to be reduced to a more manageable level. For example, this commercial standard display color image format is called "VGA" (it is the standard for video graphics cards, but now it only represents 640x480), which has 640 rows and 480 columns. This format needs to be resampled or adjusted to appear on a panel in the configuration shown in Figure 10. It has 400 red sub-pixels 34, and 400 green sub-pixels 36 (for a total of 800 sub-pixels), and a minimum of 600. A total of 35 and 36 sub-pixels. This can result in an input pixel to output subpixel ratio of 4 to 5. The conversion equation of each red sub-pixel 34 and each green sub-pixel 36 can be calculated from the partial coverage of the input sample area 120 of FIG. 22 to the sample output area 52. This procedure is similar to the development of the conversion equation of Fig. 18, except that the conversion equation seems to be different for each single output sample point 35. Fortunately, if these calculations are performed, a pattern appears for all these conversion equations. The same 5 transformation equations are repeated on one column, and the pattern of the other 5 equations is repeated on each next row. The final result is that in this example, there is only 5x5 or 25 unique equation combinations for a case where the ratio of one pixel to sub-pixels is 4: 5. This reduces this unique calculation to 25 sets of coefficients. Among these coefficients, other symmetrical patterns can be found, which can reduce the total number of coefficient combinations to only 6 unique groups. The same procedure will produce an identical combination of coefficients for the configuration 20 of FIG. The following is an example that describes how to calculate this coefficient using the geometric method described above. Figure 32 shows a single 5x5 repeating unit 202 from the previous example, which converts a 650x480 VGA format image to a Pen Tile matrix with a total of 800x600 red and green sub-pixels. Each square sub-pixel 204 surrounded by a solid line 206 represents the location of a red or green sub-pixel which must have a set of calculated coefficients. If there is no symmetry, this will require the calculation of 25 sets of coefficients. Figure 32 will be discussed in detail later. Figure 33 shows the symmetry of this coefficient. If the coefficient is written in the general matrix form of the filter core used in the industry, the filter core of the sub-pixel 216 will be a mirror image, and the core of the sub-pixel 218 will turn from left to right. This is true for all sub-pixels on the right side of the line of symmetry 220, each of which has a filter core, which is a mirror image of the filter core of the opposite sub-pixel. In addition, the sub-pixel 222 has a filter core, which is a mirror image, which is turned from the filter core of the sub-pixel 218 from up to down. This is also true for all other filter cores below the line of symmetry 224, each of which is a mirror image of a relative sub-pixel filter. Finally, the filter core of the sub-pixel 226 is a mirror image of the filter of the sub-pixel 228, which is flipped on a diagonal. This is true for all sub-pixels above and to the right of the symmetry line 230, and its filter is a diagonal mirror image of the filter relative to the diagonal of the sub-pixel filter. Finally, the core of the filter on this diagonal is internal diagonal symmetry, which has the same coefficient value on the opposite side of the diagonal of the symmetry line 230. The core of a complete filter

The combined example further illustrates all these symmetries in the filter core. The only filters that need to be calculated are the shaded ones, the sub-pixels 218, 228, 232, 234, 236 and 238. In this example, with a repeating unit size of 5, the minimum number of filters required is only 6. The remaining filters can be determined by flipping the six calculated filters on different axes. Whenever the size of an overlapping unit is odd, the minimum number of the filter is determined. The formula is: ρ + ι (χ [Ρ + Γ

Nfilts = 2 ^ ~ ^ · where P is the odd width and height of the repeating unit, and Nfilts is the minimum number of filters required. Figures 3 to 4 show an example where the size of the repeating unit is even. The only filters to be calculated are those with shadows, sub-pixels 240, 242, and 244. In this example, with a repeating unit size of 4, only three filters have to be calculated. Whenever the size of an overlapping unit is even, the general formula for determining the minimum number of filters is:

Where P is the even width and height of the repeating unit, and Neven is the minimum number of filters required. Returning to FIG. 32, the rendering boundary 208 of the central sub-pixel 204 covers a region 210 that covers four original pixel sample regions 212. Each of these overlapping areas is equal, and its coefficient must be increased to 1, so each of them is 1/4 or 0.25. These are the coefficients of the sub-pixel 238 in FIG. 33, and the core of the 2x2 filter in this example will be: -35- L -------------- ^ 一一 ~ & 1/4 1/4 1/4 1/4 The coefficients of the sub-pixel 218 in FIG. 33 are expanded in FIG. 35. This time the pixel 218 is defined by a presentation area 246, which overlaps five surrounding input pixel sample areas 248. Although this time the pixel is in the upper left corner of a repeating unit, it is assumed that for calculation purposes, there is always another overlapping unit area that has an overlapping edge by having an additional sample area 248. These calculations are done for a general example, and the edges of the display will be treated differently than described above. Because the presentation area 246 spans three sample areas 248 horizontally and three vertically, it must maintain a 3x3 filter core for all coefficients. The coefficient is calculated as described above: the area of each input sample area covered by the presentation area 246 is measured and then divided by the total area of the presentation area 246. The presentation area 246 does not overlap the upper left, upper right, lower left, or lower right sample area 248 at all, so its coefficient is zero. The presentation area 246 overlaps the upper and middle left sample area 248 of the total area of the presentation area 246 by 1/8, so its coefficient is 1/8. The presentation area 246 overlaps the central sample area 248 at the largest ratio, which is 11/16. Finally, the presentation area 246 overlaps the middle right and bottom middle sample area 248 with a minimum of 1/32. Placing these in order results in the following coefficient filter cores: 0 1/8 0 1/8 11/16 1/32 -36-

Pi! 3¾. 12 ^ 0 1/32 0

The sub-pixel 232 from FIG. 33 is shown in FIG. 36, and its presentation area 250 overlaps five sample areas 252. As described above, the area of the sample area 250 that overlaps each sample area 252 is calculated and divided by the area of the presentation area 250. In this example, only a 3x2 filter core needs to be maintained for all coefficients, but for consistency, a 3x3 will be used. The filter core of FIG. 36 will be 1/64 17/64 0 7/64 37/64 2/64 0 0 0. The sub-pixel 234 from FIG. 33 is shown in FIG. 37, and its presentation area 254 overlaps the sample area 256. This coefficient calculation will result in the following cores: 4/64 14/64 0 14/64 32/64 0 0 0 0

Sub-pixel 228 from FIG. 33 is shown in FIG. 38, and its presentation area 258 overlaps the sample area 260. This coefficient calculation will result in the following core: 4/64 27/64 1/64 4/64 27/64 1/64 -37- | 2380_ 0 0 0 Finally, the sub-pixel 236 from FIG. 33 is shown in FIG. 39, which The presentation area 2 62 overlaps the sample area 264. The coefficient calculation for this example will result in the following core: 9/64 23/64 0 9/64 23/64 0 0 0 0

This infers all the minimum number of calculations required for an example with a one-to-subpixel ratio of 4: 5. All other 25 coefficient combinations can be constructed by flipping the above 6 filter cores on different axes, as shown in Figure 33.

For adjustment purposes, the filter core must always add up to 1, or it will affect the brightness of the output image. This is true for all 6 filter cores described above. However, if the core is actually used in this form, the coefficient values are all fractions and require floating-point arithmetic. It is common in the industry to multiply all coefficients by a value to convert them all to integers. Integer arithmetic can then be used to multiply the input core value of the filter core coefficients by simply dividing the total by the same value later. Checking the filter core above, it was found that 64 would be a good number multiplied by all the coefficients. This will result in the following filter cores from the sub-pixel 2 1 8 of Figure 35: 0 8 0 8 44 2 0 2 0 -38-

(Divide by 6 4) In this example, all other filter cores are similarly modified to convert them to integers for ease of calculation. This is particularly convenient when the divisor is a power of two, which is the case in this example. Dividing by a power of two can be done quickly in software or hardware by shifting the result to the right. In this example, an offset of 6 bits to the right will be divided by 64. In contrast, a commercial standard display color image format called XGA (which is used to represent an extended graphics interface card, but now only 1024x768) has 1024 rows and 768 columns. This format can be adjusted to be displayed on configuration 38 of FIG. 10, which has 1600x1200 red and green emitters 34 and 36 (plus 800x600 blue emitter 32). This configuration has an adjustment or resampling ratio of 16 to 25, which results in 625 unique coefficient combinations. Using the symmetry of the coefficients reduces this number to a more reasonable 91 group. However, even this small number of filters can be cumbersome to perform by hand, as described above. Instead, a computer program (a machine-readable medium) can use a machine (such as a computer) to automate this task and quickly generate the set of coefficients. In fact, this program uses it only once to generate a table of filter cores for any given ratio. The table is then used by the adjustment / rendering software, or burned into hard ROM (read-only memory), which performs the adjustment and sub-pixel rendering. In the first step, the filter generation program must be completed, which calculates the adjustment ratio and the size of the repeating unit. This is done by dividing the number of input pixels and the number of output sub-pixels by its GCEK greatest common factor). This can also be done in a small double-layered loop. The external loopback -39- ------------------------- ί ΙΙ23 # 04 k Test the two numbers against a series of prime numbers. This loop will revolve until it has tested prime numbers, up to the bisector of the smaller of the two pixels. The actual typical screen size will never need to test prime numbers greater than 41. In contrast, because this algorithm is to generate the filter core "offline" in advance, the external loop can only execute all numbers from 2 to some unreasonably large numbers, prime numbers, and a non-prime number. This can waste CPU time because it will perform more tests than needed, but the code is only performed once for a specific combination of input and output screen sizes. An internal loop tests the current prime number for the two pixel counts. If both counts are divided by the prime number equally, then both can be divided by the prime number, and the internal loop continues until it is impossible to divide one of the two counts by the prime number again. When the external loop is terminated, the remaining small number will be effectively divided by the GCD. The two numbers will be the "adjusted scale" of the two pixel counts. Some typical values: 320: 640 becomes 1: 2 384: 480 becomes 4: 5 512: 640 becomes 4: 5 480: 768 becomes 5: 8 640: 1024 becomes 5: 8 These ratios will represent the pixel to sub-pixel or P: S ratio, where P is the input pixel numerator and S is the sub-pixel denominator of the ratio. The number of filter cores required to span or repeat the next unit is S in these ratios. The total number of cores required is the product of the horizontal and vertical S values. -40- 1 year 1 year 1 In most of the commonly used VGA screen sizes, the horizontal and vertical repeating patterns will be the same, and the number of filters required will be S2. As can be seen from the table above, a 640x480 image is adjusted to a 1024x768 PenTile matrix, which has a P: S ratio of 5: 8, and will require 8x8 or 64 different filter cores (before considering symmetry). In a theoretical environment, a value added to 1 is used in a wave reducer core. In fact, as mentioned above, the filter core is usually calculated as an integer value with a divisor that is applied afterwards to normalize the total back to one. It is important to start by calculating the weighted value to be as accurate as possible, so the presentation area can be calculated in a sufficiently large coordinate system to ensure that all calculations are integers. Experience shows that the correct coordinate system to be used in the image adjustment situation is that the size of its input pixels is equal to the number of output sub-pixels across a repeating unit, which makes the size of an output pixel equal to number. This is essentially a counter and seems to be backward. For example, in an example of adjusting 512 input pixels to 640, which has a 4: 5 P: S ratio, the input pixel can be drawn as a 5x5 square on a drawing paper, and the output pixel above it becomes 4x4 square. This is the smallest ratio that two pixels can draw, and it keeps all numbers as integers. In this coordinate system, the area of the diamond-shaped presentation area at the center of the output sub-pixel is always equal to 2 times the area of an output pixel, or 2 * P2. This is the smallest integer that can be used as the denominator of the weighted value of the filter. Unfortunately, because the rhombus spans several input pixels, it can be split into triangles. The area of a triangle is its width multiplied by its height divided by two, and this again results in non-integer values. Calculating twice the area can solve the problem -41-| 12380 |, so the program can calculate the area multiplied by 2. This makes the smallest denominator available for integer filters equal to 4 * P2. It must then decide how large each filter core must be. In the hand-made example above, some of the filter cores are 2x2, some are 3x2, and others are 3x3. The relative sizes of the input and output pixels, and how the diamond-shaped presentation areas cross each other, determine the required maximum filter core size. When adjusting an image from a source with more than two output sub-pixels across each input pixel (eg 100: 201 or 1: 3), a 2x2 filter core becomes possible. This will require less hardware to implement. Moreover, the image quality is better than the adjustment of the previous technique, because the obtained image captures the "squareness" of the indicated target pixel, and maintains the best spatial frequency as much as possible. Represented by the edge. These spatial frequencies are used by the font, and the graphic designer can improve the appearance resolution and deceive the Nyquist limitation known in the art. Prior art adjustment algorithms can use interpolation to limit the adjusted spatial frequency to the Nyquist limit, or maintain its sharpness, but produce objectified phase errors. When adjusted down, there are more input pixels than output sub-pixels. At any adjustment factor greater than 1: 1 (for example, 101: 100 or 2: 1), the filter size becomes 4x4 or greater. It will be difficult to persuade hardware manufacturers to add more line buffers to implement this architecture. However, staying in the range of 1: 1 and 1: 2 has the advantage that the core size is maintained at a fixed 3x3 filter. Fortunately, most of the examples that must be implemented in hardware fall into this range, which can reasonably write programs to simply generate 3x3 cores. In some special cases of -42- and M12 in 2012, similar to the hand-made example above, some filter cores will be less than 3x3. In other special cases, even if it is theoretically possible that the filter becomes 3x3, the result is only 2x2 per filter. However, it is easier to calculate the core for a general example, and it is easier to implement a hard core with a fixed core size. Finally, the core filter weight is now calculated only for the area (multiplied by 2) of the 3x3 input pixels, which intersects the output diamond at each < asymmetry position in the repeating unit . This is a very straightforward "presentation" task known in the industry. For each filter core, 3x3 or 9 coefficients can be calculated. In order to calculate each coefficient, a vector description of the presenting area of the rhombus is generated. This shape is cropped at the edge of the input pixel area. It uses a polygon cropping algorithm that is well known in the industry. Finally, the area of the cropped polygon is calculated (multiplied by 2). The area obtained is the coefficient of the corresponding unit at the core of the filter. The sample output from this program is shown as follows: Source pixel resolution 1024 Target sub-pixel resolution 1280 Adjustment ratio 4: 5 The number of filters is divided by 256 The minimum filter required (with symmetry >: 6 here Number of filters generated (without symmetry): 25 -43- divisions 123 ¾ f 032 0 4 28 0 16 16 0 28 4 0 0 32 0 32 176 8 68 148 0 108 108 0 148 68 0 8 176 32 0 8 0 0 8 0 4 4 0 8 0 0 8 0 4 68 0 16 56 0 36 36 0 56 16 0 0 68 4 28 148 8 56 128 0 92 92 0 128 56 0 8 148 28 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 16 108 4 36 92 0 64 64 0 92 36 0 4 108 16 16 108 4 36 92 0 64 64 0 92 36 0 4 108 16 0 0 0 0 0 0 0 0 0 0 0 0 0 0_ _0_ ^ _ 28 148 8 56 128 0 92 92 0 128 56 0 8 148 28 4 68 0 16 56 0 36 36 0 56 16 0 0 68 4 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 8 0 0 8 0 4 4 0 8 0 0 8 0 32 176 8 68 148 0 108 108 0 148 68 0 8 176 32 0 32 0 4 28 0 16 16 0 28 4 0 0 32 0

In the sample output above, all 25 reducer cores required for this example are calculated without regard to symmetry. This allows checking the coefficient and visually verifying horizontal, vertical, and diagonal symmetry in the filter cores in these repeating units. As mentioned before, the edges and corners of the image can be uniquely processed, or can be approximated by filling in the "missing" input data sample with other averages, or the most effective single contributor, or black values. Each set of -44-coefficients is used in a filter core, as is well known in the art. Keep track This position and symmetry operator is the job of a software or hardware designer, and it can use modular mathematical techniques, which are also well known in the art. The work of generating this coefficient is a simple matter to calculate a proportional overlap region of the input sample region 120 to the output sample region 52 for each sample corresponding to the output sample point 35, using a manner known in the art . FIG. 23 shows the array 108 of the same point 122 and the effective sample area 120 of FIG. 21 covered on the blue plane sampling area 44 of FIG. 12, where the sample point 122 of FIG. 21 and the red and The green "inspection board" arrays do not have the same spatial resolution grid, nor do they overlap. The calculation method for generating this conversion equation is performed as described above. First, determine the size of the repeated array of three-color pixel elements, then determine the minimum number of unique coefficients, and then determine the proportional coverage of the input sample area 120 to the output sample area 44 for each corresponding output sample point 46. The values of those coefficients. Each of these values is applied to the transformation equation. The array of the repeated three-color pixel elements and the number of obtained coefficients are the same numbers determined for the red and green planes. FIG. 24 shows the array 110 of the same point, and the effective sample area of FIG. 21 covered on the blue plane sampling area 123 of FIG. 8, where the sample point 122 of FIG. 21 and the red (red reconstructed point) of FIG. 11 35) and green (green reconstruction point 37) "check boards, the arrays do not have the same spatial resolution grid, nor do they overlap. The calculation method for generating this conversion equation is performed as described above. First, That is to determine the size of the repeated array of the three-color pixel element. Next, determine the minimum number of unique coefficients, and then determine each corresponding output sample -45- Γ'year wide moon Japan point 23 input sample area 120 to Outputs the values of those coefficients whose proportionality overlaps the sample region 123. Each of these values is applied to the conversion equation. The former has checked the RGB format of the CRT. A conventional RGB flat panel display configuration 10 has three tri-color pixels. The red 4, green 6, and blue 2 emitters in element 8 are as in the previous art of Figure 1. In order to project a formatted image to the three-color pixel element according to this configuration, as shown in Figure 6 or Figure 10 Must decide Structure points. The placement of the red, green, and blue reconstruction points is shown in configuration 12 in Figure 2. The red, green, and blue reconstruction points do not coincide with each other, and they have a horizontal displacement. According to Benzschawel et al. In the prior art disclosed in U.S. Patent No. 5,341,153, and later in Hill et al. U.S. Patent No. 6,188,385, these positions are referred to as sample points 3, 5, and 7 with sample areas, as shown in Figure 3 The red plane 14 shown in the previous art, the blue plane 16 shown in the previous art in FIG. 4, and the green plane 18 shown in the previous art in FIG. 5. A conversion equation can be calculated from FIG. The prior art configuration presented in 4 and 5 is generated by the method disclosed therein. In the method disclosed above, it can be utilized by calculating the coefficients of the conversion equation, or the filter core, for the selected prior art Each output sample point configured. Figure 25 shows the active sample area 125 of the red plane of FIG. 3 covered on the red plane sampling area 52 of FIG. 13, where the configuration of the red emitter 35 in FIG. Figures 6 and 10 Set the same pixel degree (repeated unit resolution). The method of generating the conversion equation is calculated as described above. First, determine the size of the repeated array of the three-color pixel element. Then the minimum number of unique coefficients That is, it is determined by noticing the symmetry (-46- ^ 23 | 0417 in this example:-; a. 9a 12.-〇 '-; :: :: "is 2 >. Then, those coefficients can be determined The values overlap with the proportionality of the input sample area 125 to the output sample area 52 for each corresponding output sample point 35. Each of these values is applied to the conversion equation. As shown in FIG. 4, the resampling The calculation of the green plane is performed in a similar manner, but the output sample array is rotated by 180. And offset the green input sample area 127. FIG. 26 shows the blue-plane sampling area 123 of FIG. 8 overlaid on the blue-plane sampling area 127 of the prior art blue plane of the prior art. FIG. 40 shows an example of blue corresponding to the red and green samples in FIG. The sample area 266 in FIG. 40 is square instead of diamond in the red and green examples. The number of original pixel boundaries 272 is the same, but there are fewer blue output pixel boundaries 274. The coefficient is calculated as described above; the area covered by the presentation area 266 for each input sample area 268 is measured, and then divided by the total area of the presentation area 266. In this example, the blue sampling area 266 equally covers the four original pixel areas 268, resulting in a 2x2 filter core with four coefficients of 1/4. The eight other blue output pixel regions 270 and their geometric intersection with the original pixel region 268 can be seen in the symmetrical relationship of the filter obtained in FIG. 40. The symmetry of the original pixel boundary 274 in each output pixel region 270 Observed in the configuration. In a more complex example, a computer program is used to generate the blue filter core. This program finds a program very similar to the one that generates the red and green filter cores. The distance between the blue sub-pixel sample point 33 and the red and green sample points 35, 37 in FIG. 11 is twice the distance. It is suggested that the blue rendering area will be twice as wide. However, the red and green areas are diamond-shaped, so they are twice as wide as the interval between the sample points. This makes the red, green-47-Ir_… a year, month, day and blue and the blue and blue rendering areas have the same width and height, which results in several convenient numbers; the size of the blue filter core will be The size is equal to red and green. At the same time, the blue repeating unit size will usually be equal to the red and green repeating unit sizes. Because the interval between the blue sub-pixel sample points 33 is doubled, the ratio of P: S (pixel to sub-pixel) is doubled. For example, the ratio of 2: 3 in red becomes 4: 3 in blue. However, the number of S in this ratio determines the size of the repeating unit, and it is not changed by doubling. However, if the denominator happens to be divided by 2, it can complete an additional optimization. In this example, the two numbers in blue are divided by an extra power of two. For example, if the red and green P: S ratio is 3: 4, the blue ratio will be 6: 4, which can be simplified to 3: 2. This means that in these (even) examples, the blue repeating unit size can be cut in half, and the total number of filter cores required will be one quarter of the red and green. Conversely, in order to simplify the algorithm or hardware design, it is possible to make the blue repeating unit size equal to red and green. The resulting combination of filter cores will have a double (in fact, four times), but its work is equivalent to the red and green combination of the filter core. Therefore, the only correction required to use the red and green filter core programs to generate the blue filter core is to double the P: S ratio and change the presentation area to a square, not a rhombus. Now consider the configuration 20 of FIG. 6 and the blue sample area 124 of FIG. 9. This is similar to the previous example where the blue sample area 124 is square. However, its calculations are complicated because each of its spaced rows is intersected up or down at half its height. At first glance, it seems that the repeating unit size -48- ΓΙ2380ΙΓ will be doubled horizontally. However, the following procedures have been found to generate the correct filter core: 1) Generate a repeating unit combination of a filter core, as if the blue sample points were not interleaved, as described above. Remember the rows and columns of the filter table of the repeating unit, the number of which starts from zero and ends at the size of the repeating unit minus 1 2) the even-numbered rows in the output image, the The filter is modified accordingly. The modulus in the repeated unit size of the output γ coordinate is selected to use which column is used in the core combination of the filter, the modulus in the repeated unit size of the X coordinate is selected to one row, and the Y selection column is told to be used That filter. 3) On the odd output line, before taking its modulus, subtract the Y coordinate from the repeated cell size. The X coordinate system is considered to be the same as the even line. This will select a filter core that is correct for the interleaved example of Figure 9. In some examples, it is possible to perform the modulus calculations in advance and interleave the tables of the filter cores in advance. Unfortunately, this only works with an even number of rows in an example of a repeating unit. If the repeating unit has an odd line, the modulo arithmetic selects the even line half the time and the other half is the odd line. Therefore, the calculation to interleave that row must be performed at the time the table is used, not in advance. Finally, consider the configuration 20 of FIG. 6 and the basket color sampling area 123 of FIG. 8. This system is similar to the previous example in that it has additional complexity for a hexagonal sample area. The first step to consider these hexagons is how to draw them correctly -49- | Ι23 # 04Κ · year .. month or month, or create a vector list of them in a computer program. For best accuracy, these hexagons must be the smallest area hexagons, but they will not be normal hexagons. A geometric proof can be easily implemented in Figure 41. The hexagonal sampling areas 123 of Figure 8 are 1/8 wider on each side than the square sampling area 276. At the same time, the upper and lower edges of the hexagonal sampling region 123 are narrower by 1/8 than the upper and lower edges of the square sampling region 276 at each end. Finally, please note that the hexagonal sampling area 123 and the square sampling area 276 have the same height. The filter cores of these hexagonal sampling regions 123 can be generated in the same geometrical manner as previously described, which are rhombic for red and green, or square for blue. The presenting area is a simple hexagon, and the overlapping area of these hexagons covering the surrounding input pixels is measured. Unfortunately, when using the slightly wider hexagonal sampling area 123, the size of the filter core sometimes exceeds a 3x3 filter, even while maintaining the adjustment ratio between 1: 1 and 1: 2. Analysis shows that if the adjustment ratio is between 1: 1 and 4: 5, the core size will be 40. If the adjustment ratio is between 4: 5 and 1: 2, the core size of the filter will be maintained at 3x3. (Note that because the hexagonal sampling area 123 is the same height as the square sampling area 276, and the vertical size of the filter core remains the same). Designing hardware for a wider filter core is not as difficult as constructing hardware to handle higher filter cores, so it is not reasonable for a hardware-based sub-pixel rendering / adjustment system to be made A 4x3 wave suppressor. However, other solutions are possible. When the adjustment ratio is between 1: 1 and 4: 5, the square sampling area 124 of FIG. 9 is used, which results in a -50- 1238 · f U · 12. —g-house 3x3 filter. When the adjustment ratio is between 4: 5 and 1: 2, the more accurate hexagonal sampling area 123 of FIG. 8 is used, and a 3x3 filter is also required. In this way, the hardware is simpler to maintain and can be constructed cheaper. The hardware only needs to be constructed for the size of a filter core, and the algorithm used to construct those filters is the only thing that changes. Similar to the square sampling area of Fig. 9, the hexagonal sampling area of Fig. 8 is staggered in each other row. Analysis shows that the same method of selecting the filter core of FIG. 9 described above will work for the hexagonal sampling area of FIG. Basically, this coefficient representing the core of the filter can be calculated as if the hexagon is not interlaced, even though it is usually interlaced. This makes the calculation simpler and avoids double the size of the table at the core of the filter. In the example of the diamond-shaped presentation area in Figs. 32 to 39, the area is calculated in a coordinate system designed so that all areas are integers and easy to calculate. This can often result in a larger total area, and the filter core must be divided by a larger number in use. Sometimes this results in the filter core not being a power of two, which makes the hardware design more difficult. In the example of Fig. 41, the extra width of the hexagonal rendering area 123 makes it necessary to multiply the coefficients of the filter core by an even larger number to make them all integers. In all these examples, it is preferable to find a way to limit the size of the divisor of the core coefficients of the filter. In order to make the hard hip easier to design, it is preferable to be able to pick the power of the divisor of two. For example, if all filter cores are designed to divide by 256, this division operation can be performed by an 8-bit right shift operation. Selecting 256 also guarantees that all filter core coefficients will be 8-bit values, which will meet the standard "byte width" of only -51-η read memory (ROM). Therefore, the following procedure is used to generate a filter core with a desired divisor. Since the preferred divisor is 256, it will be used in the following procedures. 1) Use floating-point arithmetic to calculate the area of the filter coefficients. Because this calculation is done offline beforehand, this does not increase the hardware cost of using the resulting table. 2) Divide each coefficient by the known total area of the rendering area and multiply by 256. This would result in a total of 256 filters if all arithmetic is done in floating point, but more steps must be made to construct the integer table. 3) Perform a binary search to find a carry point (between 0 · 0 and 1.0), which when converted to an integer makes the total number of filters 256. Binary search is a common algorithm well known in the industry. If this search is successful, it is complete. A binary search will fail to converge, and this can be detected by testing the loop for more than one execution. 4) If the binary search fails, find a reasonably large coefficient in the core of the filter and add or subtract a small number to force the filter's total to 256. 5) For a special example, check that the filter has a single value of 256. This value will not match a table of 8-bit bytes, where the maximum possible number is 255. In this special case, the single value is set to 255 (256-1), and 1 is added to one of the surrounding coefficients to ensure that the filter still adds up to 256. Fig. 31 shows that in this special case, the output sample configuration 40 of Fig. 11 is overlaid on the input sample configuration 70 of Fig. 15 when the adjustment ratio is one input pixel on every two output sub-pixels. In this configuration 200, when the original -52- 11¾¾1 data is no longer the sub-pixel presented, the pairing of the red emitter 35 in the three-color pixel element 39 will be considered as a combination, which is in the three-color The center of the pixel element 39 has a representative reconstruction point 33. Similarly, the two green emitters 37 in the three-color pixel element 39 are regarded as a single reconstruction point 33 at the center of the three-color pixel element 39. The blue emitter 33 is already at the center. Therefore, the 5 emitters can be regarded as if they reconstruct the sample points of the RGB data format if all three color planes are at the center. This can be regarded as the “essential mode” of this pixel configuration. ^ By resampling and rendering through sub-pixels, an image that has been sub-pixel rendered to another display with a different sub-pixel configuration can maintain many of the Improved original image quality. According to a specific embodiment, it is necessary to generate a transition from the image presented by this pixel to the configuration disclosed here. Please refer to FIGS. 1, 2, 3, 4, 5, 25 and 26 The method already disclosed above can be used, by calculating the coefficient of the conversion filter of each output sample point 35, as shown in FIG. 25, which is the target of the red input sample 5 displaced relative to the right of FIG. 3 Display configuration. The blue emitter is processed as described above, and it calculates the coefficients of the conversion filter of each output sample point of the blue input sample 7 by shifting the target display configuration relative to FIG. 4 ^. In the example of the green plane, as shown in FIG. 5, the input data is already presented as a sub-pixel, because the green data is still in the center, and the example for the non-sub-pixel presentation need not be changed. When the application of text rendered using sub-pixels is included on the sides of graphics and photos rendered along non-sub-pixels, it is preferable to detect that the sub-pixels are present at -53- 月 ί-jj and alternate between the above The spatial sampling filter is switched, but for the adjustment ratio, that is, the normal spatial sampling filter switched back to the area presented by non-subpixels, it is also described above. In order to construct this detector, you must first understand the subpixels What the rendered text looks like, what are its detectable features, and what distance is set from the image rendered by the non-subpixel. First, the pixels at the edges of the glyph rendered by the black and white subpixels will not be local Neutral color: R # G. However, the color will be neutral on several pixels. That is R ^ G. For images or text rendered by non-subpixels, neither of these conditions will occur. . Therefore, we have our own detector to test local R # G and R ^ G for several pixels. Because the sub-pixels on a RGB strip panel are rendered in one dimension, along the horizontal axis, one column Next column, The test is one-dimensional. The following is such a test: If Rx_2 + Rx_i + Rx + Rx + 1 + Rx + 2 three Gn + GxU + Gx + Gx + i + Gx · ^ or if Rx-i + Rx + Rx + 1 + Rx + 2 three Gx_2 + Gx_i + Gx + Gx + i apply another spatial filter for sub-pixel rendering input otherwise apply normal spatial filter In the example where the text is colored, in this form RxaGx There will be a relationship between the red and green components, where "a" is a constant. For black and white text, the value of "a" is 1. This test can be extended to detect colored, and black and white text: I238Q1 he 93. 12. If Rxd + Rxd + Rx + Rx + i + Rxu (aiGxj + Gxd + Gx + Gxw + Gxw) or another spatial filter of the input of the sub-pixel is applied, otherwise the normal spatial filter is applied 1 and GXR represent the values of the red and green components at the coordinates of the pixel row of "x". There can be a critical test to determine if R ^ G is close enough. Its value can be adjusted for best results. The length of the item and the spacing of the test can be adjusted to obtain the best results, but generally follow the above form. Fig. 27 shows the arrangement of three-color pixel elements in an array on three planes in a display device according to another embodiment. FIG. 28 shows the arrangement of the blue radiator pixel elements in the array of the device of FIG. 27. FIG. 29 shows the arrangement of the pixel elements of the green radiator in the array of the device of FIG. Fig. 30 shows the arrangement of the red radiator pixel elements in an array of the device of Fig. 27. This configuration and layout can be used for a projector-based display using three panels, each for red, green, and blue, which combines each image to project on a screen. The configuration and shape of the radiator can be fully matched to FIGS. 8, 13 and 14, which are sample regions of the configuration shown in FIG. 6. Therefore, the graphic generation, conversion equation calculation, and data format for the configuration of Fig. 6 disclosed herein will also work for the three-panel configuration of Fig. 27. For adjustment ratios higher than approximately 2: 3 and higher, the resampled data combination presented by the subpixels in the PenTileTM matrix configuration of the subpixel is more efficient in representing the resulting image. If you want to store and / or transmit -55- 嗲 ί2Net · Yellow ... ~ Poetry.Day 丨 • ”― * ·. *** · * ·· 1 ** ._ ·. **-** • See image It is expected to be displayed on a PenTileTM display and the adjustment ratio is 2: 3 or higher, which is preferably performed by resampling before storage and / or transmission to store in memory storage space and / or bandwidth. This resampled image is called "pre-rendering". So this pre-rendering can be used as an effective low-loss compression algorithm. The benefit of the present invention is that it can take any most of the stored images and pre-render them To any achievable color sub-pixel configuration. Other benefits of the present invention are disclosed by way of example in the methods of FIGS. 46, 49, and 51, which can provide gamma compensation or adjustment using the above-mentioned sub-pixel rendering technology. These three methods of sub-pixel gamma adjustment can achieve correct image color balance on a display. The methods of Figures 49 and 51 can further improve the output brightness or illuminance by improving the output contrast ratio. Specifically, Figure 46 shows the sub-pixel rendering Previously applied a method of pre-adjusting gamma; FIG. 49 shows a method of sub-pixel presentation with a gamma adjustment; and FIG. 51 shows a method of sub-pixel presentation with gamma adjustment with an omega function. These methods The benefits are explained below. The methods of Figures 46, 49, and 51 can be applied to hardware, firmware, or software, as detailed in Figures 52A to 72. For example, the sample code included in the appendix It can be used for the method disclosed here. Because the human eye cannot distinguish between absolute brightness or illuminance values, it needs to improve the illuminance contrast ratio, especially at high spatial frequencies. By improving this contrast ratio, higher quality can be obtained Image, and can avoid color errors, as explained in detail below. The method to improve the contrast ratio is the effect of the gamma-adjusted sub-pixel presentation and the gamma-adjusted sub-pixel presentation with an omega function, which is -56-1238011. '「:: xiν」 丨 The maximum (MAX) / minimum (MIN) point of the modulation transfer function (MTF) at the Nyquist limit, as explained in detail in Figures 43, 44, 47, and 50. Clear Ground The gamma-adjusted sub-pixel rendering technology described here can shift the trend of the MAX / MIN point of the MTF downward to provide a high contrast of the output image, especially at high spatial frequencies, while maintaining the correct color balance The arrangement of the sub-pixels on a display, such as shown in FIGS. 6, 10, and 42B, has alternate red (R) or green (G) sub-pixels on one horizontal axis, or vertical axis, or on both axes. The gamma adjustment described here can also be applied to other display forms, which use a one-time pixel rendering function. That is, the technique described here can be applied to a display using the RGB bar format shown in FIG. 1. Figure 43 shows a sine wave of an input image, which has the same amplitude and increases the spatial frequency. FIG. 44 shows an exemplary graph of the output when the input image of FIG. 43 is sub-pixel rendered without gamma adjustment. The graph of the output ("output energy") shows that the amplitude of the output energy decreases as the spatial frequency increases. As shown in Figure 44, 50% of the MTF value means that the output amplitude at the Nyquist limit is half the amplitude of the original input image or signal. The MTF value can be calculated by dividing the energy amplitude of the output by the energy amplitude of the input: (MAXQut-MINQUt) / (MAXin-MINin). This Nyquist is limited to the point of the input signal sampled at frequency (f), which is at least twice as large as the frequency of reconstruction (f / 2). In other words, the Nyquist is limited to the highest point of the spatial frequency, where an input signal can be reconstructed. The Sparrow is limited to the spatial frequency at MTF = 0. Therefore, measurements at this Nyquist limit, such as contrast ratio -57- i and: i! Year and month & l | cases, can be used to determine image quality. The contrast ratio of the output energy at the Nyquist limit in Fig. 44 can be calculated by dividing the output MAX bright energy level by the output MIN black energy level. As shown in Figure 4, the MAX bright energy level is 75% of the maximum output energy level, and the MIN black energy level is 25% of the maximum output energy level. With this, the contrast ratio can be determined by dividing these] VIAX / MIN values, and a contrast ratio of 75% / 25% = 3 is provided. Therefore, at contrast ratio = 3 and high spatial frequency, the corresponding output of the graph of FIG. 44 on a display will describe alternating dark and bright rods, so that the edge of the rod will have lower sharpness and contrast . That is, a black bar from the input image will be displayed as a dark gray bar, and a white bar from the input will be displayed as a light gray bar at a high spatial frequency. By using the method of Figs. 49 and 51, the contrast ratio can be improved by shifting the MAX and MIN points of the MTF downward. In short, the MTF at the Nyquist limit of the gamma-adjusted sub-pixel rendering method of Fig. 49 is shown in Fig. 47. As shown in Figure 47, the MTF can be shifted downward along a flat trend line, so that compared to the MTF in Figure 44, the MAX value is 65% and the MIN value is 12.5%. The contrast ratio of the Nyquist limit in Figure 47 is therefore 63% / 12.5% = 5 (approximately). With this, the contrast ratio can be improved from 3 to 5. The contrast ratio at the Nyquist limit can be further improved using the method of gamma adjustment with an omega function of Figure 51. Figure 50 shows that the MTF can be shifted further down a downward trend line, so that compared to the MTF in Figure 47, the MAX value is 54.7%, and the MIN value is 4.7% ° at the Nyquist limit The comparison ratio is 54.7% / 4.7% = 11.6 (approximately). Therefore, the contrast -58-ττ: π

The ratio has been improved from 5 to 11.6, thereby allowing high-quality images to be displayed. Figure 45 shows an exemplary graph to describe color errors that can occur using sub-pixel rendering without gamma adjustment. A brief discussion of the human eye's response to illuminance is used to specify the color, gamma, effect of the sub-pixels that are presented. As mentioned earlier, the human eye can experience a change in brightness as a percentage change without changing with an absolute radiant energy value. The relationship between brightness (L) and energy (E) is L = E1 / Y. As this brightness increases, the perceived increase in brightness requires a larger absolute increase in radiant energy. Thus, for an equally perceived increase in brightness on a display, each increment must be logarithmicly higher than the previous one. This relationship between L and E is called a π gamma curve π and is represented by g (x) = x1 / Y. A gamma value of approximately 2.2 (γ 丨 can represent the logarithmic needs of the human eye. Conventional displays can compensate for the above-mentioned needs of the human eye by performing a display gamma function shown in FIG. 45. However, the sub-pixel rendering Processing requires a linear illuminance space. That is, for a primary pixel, such as a green subpixel or red subpixel, the illuminance output must have a value that falls on the linear dotted line pattern. Therefore, when there is a very high spatial frequency The image presented by the pixel is displayed on a display with a non-gamma value, because the sub-pixel's illuminance value is not balanced, and a color error will occur. Specifically, as shown in Figure 45, the red and green colors The sub-pixel cannot get a linear relationship. In particular, the green sub-pixel is set to provide 50% illumination, which can represent a white point logical pixel on the display. However, the illumination output of the green sub-pixel falls on 25% of the display function, not 50%. In addition, the illuminance of the four sub-pixels around the white point (for example, red-page sub-pixel) can be set to each Provides 12.5% illumination, but falls at 1.6% of the apparent function, not at 12.5%. The illumination percentage of the white point pixels and the surrounding pixels must be added up to 100%. Therefore, in order to have the correct For color balance, a linear relationship is required among the surrounding sub-pixels. However, the four surrounding sub-pixels only have [6 ^^ 4 = 6.4%, which is much lower than the 25% required by the central sub-pixel. Therefore, in this example, the central color is more important than the surrounding colors, thereby causing a color error, that is, generating a color point to replace the white point. On more complex images, the non-linear display The color error can produce errors for the part with high spatial frequency in the diagonal direction. The following method of Figure 46, 49 and 51 can apply a conversion (gamma correction or adjustment) to the Data, so that the sub-pixel rendering can be located in a correct linear space. As detailed below, the following methods can provide the correct color balance for the sub-pixels presented. The methods of Figures 49 and 51 can further improve the rendering For the sake of explanation, the following method uses the highest resolution of the pixel-to-subpixel ratio (P: S) to be 1: 1. That is, for a pixel to a pixel Resolution, which uses a filter core with 3x3 coefficient terms. However, for example, by using an appropriate number of 3x3 filter cores, other P: S ratios can be sold. For example, the P: S ratio is In the case of 4: 5, it can use the above-mentioned 25 filter cores. In this one-pixel-to-one-pixel rendering, as shown in FIG. 42A, the output value of a resampled area 282 for a red or green sub-pixel ( Vw), which can be calculated using the input value (Vin :) of the nine indicated sample areas 280. -60-; I238QI l · In addition, for the sake of explanation, the following method uses the primary pixel shown in FIG. 42B Configuration to illustrate. However, the following method can be implemented for other sub-pixel configurations, such as FIGS. 6 and 10, by using the following calculations and formulas for the red and green sub-pixels, and performing appropriate corrections for those blue sub-pixels. FIG. 46 shows a flowchart of a method 300 that applies a pre-adjusted gamma before sub-pixel rendering. Initially, the input sampled data (Vin) of 9 indicated sample areas 280 is received (step 302), for example, as shown in FIG. 42A. Next, each ViM value is input to the function g- \ X; ) = Calculation defined by XY (step 304). This calculation is called, pre-adjusted gamma, and can be performed with reference to a pre-adjusted gamma lookup table (LUT). This g-1⑻ function is a function of the inverse of the reaction function of the human eye. Therefore, when the human eye rotates, the data presented by the sub-pixel obtained after the pre-adjusted gamma can match the response function of the eye to obtain the original image using the g · 1⑻ function. After performing the pre-adjusted gamma, the sub-pixel rendering is performed using the aforementioned sub-pixel rendering technique (step 306). As previously explained in detail, for this pixel rendering step, the corresponding one of the filter core coefficient term Ck is multiplied by the value from step 304, and all the multiplied term times are added. The coefficient term Ck is received by a filter core coefficient table (step 308). For example, the red and green sub-pixels can be calculated in step 306 as follows: V〇ut (CxRy) = 0.5xg-1 (Vin (CxRy)) + 0.125xg-1 (Vin (Cx_iRy)) + 0.125xg_1 + (Vin (Cx + 1Ry)) + 0.125xg-1 (Vin (CxRy_1)) + 0A25xg \ Win (CxRy + l)) -61-

After steps 306 and 308, the data presented by the sub-pixel 乂 is to accept a post-gamma correction for a given display gamma function (step 310 >-the display gamma function is represented by f (X), and Represents a typical non-1 gamma function, for example, for a liquid crystal display. In order to achieve the linearity of sub-pixel presentation, the display gamma function is identified and eliminated by a post-gamma correction function f \ x), which can be calculated by f (x) is generated. Post-gamma correction allows the data presented by this sub-pixel to reach the human eye without interference from the display. Then, the post-gamma correction data is output to the display (step 312 > The method of applying pre-adjusted gamma in FIG. 46 before sub-pixel rendering above can provide appropriate color balance for all spatial frequencies. The method can also provide the correct level of brightness or illuminance at least for low spatial frequencies. However, at high spatial frequencies, using the method of FIG. 46 to obtain the appropriate illuminance or brightness value of the sub-pixels presented remains problematic. Clear In other words, at high spatial frequencies, the sub-pixel rendering needs to be calculated linearly, and according to its average brightness, the brightness value will diverge from the expected gamma-adjusted value. Because all values except zero and 100%, the The correct value can be lower than the linear calculation, which will cause the linearly calculated brightness value to be too high. This will cause excessive and too white text on a black background, and black text cannot be completely eliminated on a white background. As mentioned above, For the method of Figure 46, the linear color balance can be obtained by applying a pre-adjusted gamma step of g-1 (X) = Xγ before the linear sub-pixel is rendered. To. Further improvement of image quality at high spatial frequencies can be achieved by implementing a desired non-linear illuminance calculation, as described below. -62- 123 Lai Fu For further improvements in sub-pixel rendering, use Figure 49 And the method of 51 is obtained for the appropriate illuminance or brightness value, which can cause the MAX and MIN points of the MTF at the Nyquist limit to change downward, thereby further improving the contrast ratio at high spatial frequencies. In other words, the following method allows non-linear illuminance calculation while maintaining linear color balance. Figure 49 shows a flowchart of a method 350 for sub-pixel rendering for gamma adjustment. The method 350 can apply or add a gamma correction, Therefore, the non-linear illumination calculation can be performed without causing color error. As shown in FIG. 47, the exemplary output signal presented by the gamma-adjusted sub-pixel in FIG. 49 shows an average energy, and then a flat trend line at 25% (corresponding to 50% brightness), which is shifted downward from 50% (corresponding to 73% brightness) of Fig. 44. For the gamma-adjusted sub-pixel rendering method 350 of Fig. 49, Referring to FIG. 48, the concept of "local average (〇〇") is introduced. The concept of a local average is that the illumination of a pixel must be balanced with the surrounding sub-pixels. For each edge term (VJCyRw), VJCxU, VJCwRw, VJCwRy), Vin (Cx + 1Ry), VJCyRw), Vin (CxRy + 1), Vin (Cx + 1Ry + 1)), the local average is defined as the average value with the central term (Vin (CxRy)) . For the central term, the local average is defined as the average of all the edge term surrounding the central term, and is weighted by the corresponding coefficient term of the filter core. For example, (Vir ^ CuD + VidCxRy :)) · ^ is the local average of V ^ CyRy), and (Vin (Cx_1Ry) + Vin (CxRy + 1) + Vin (Cx + 1Ry) + Vin (CxRy_1) + 4xVin (CxRy; 0 + 8 is the local average of the central term of the filter core below -63-

Official

0 0.125 0 0.125 0.5 0.125 0 0.125 0

Please refer to FIG. 49. Initially, the input data Vin sampled from 9 designated sample areas 280 is received, as shown in FIG. 42 (step 352). Next, a local average of each of the 8 edge terms (0 is calculated using each edge term Vin and the central term Vin (step 354). Based on these local averages, a π pre-gamma "correction is used For example, a pre-gamma LUT is calculated as step 356>. The pre-gamma correction function must be noted that it is used instead of χγ because the gamma-adjusted sub-pixel rendering makes x (Vin in this example) Multiply in later steps 366 and 368. The result of pre-gamma correction for each edge term is multiplied by a corresponding coefficient term CK, which is received by a filter core coefficient table 360 (step 358).

For this central term, at least two calculations can be used to determine g-ka). For a calculation (1), the local average (a) is calculated for the central term using g-'a) based on the central term local average as described above. For a second calculation (2), a gamma-corrected local average ("GA") is calculated for the central term by using the result of step 358 of the surrounding edge term. The method 350 of FIG. 49 uses calculations (2). The "GA" of the central term can be calculated using the results from step 358 instead of steps 3 56 to refer to the edge coefficients if each edge term has a different contribution to the local average of the central term, for example if it has Sharpening of the same color, as described below -64- 123 8W%. The "GA" of the central term is also multiplied by a corresponding coefficient term CK, which is received from a filter core coefficient table (step 364). The two calculations (1) and (2) are as follows: (1) g'ViViniC. ^ R ^ + V ^ CxR ^ O + Vi ^ C ^ tR ^ + Vi, (CxRy.1) + 4xVin (CxRy ) H8) (2) ((g '\ (Vin (Cx.1Ry) + Vin (CxRy))-2) + gV (Vin (CxRy + 1) + Vin (CxRy)) ^ 2) + g-1 ( (Vin (Cx + 1Ry) + Vin (CxRy))-2) + g · 1 ((νίη ((: χυ + νίη ((: χ; ^)) + 2)) +4) CK g from step 358 · 1⑹ value, and the value of CK "GA" from step 364 using the second calculation · (2) is multiplied by a corresponding term Vin (steps 366 and 368). Then, the sum of all multiplied terms It is calculated (step 370) to generate the output sub-pixel rendered data V (> ut. Then, the -post-gamma correction is applied to VQUt and output to the display (steps 372 and 374). To use the calculation (1 ) To calculate V_, the following calculations for the red and green sub-pixels are as follows: V〇ut (CxRy) = Vin (CxRy) x 0.5x (CxRy.1) + 4xVin (CxRy) K8) + Vin (Cx. 1Ry) x〇.l25xg-1 ((Vin (Cx.1Ry) + Vin (CxRy) H2) + Vin (CxRy + 〇x〇.l25xg-1 ((Vin (CxRy + 1) + Vin (CxRy) K2) + Vin (Cx + lRy) x〇.i25xg '((V ^ C. ^ R ^ + Vi ^ C.Ry)) ^) + Vin (CxRy-1) x〇.125xg_1 ((Vin (CxRy.1) + Vin (CxRy)) + 2) The calculation (2) utilizes this The central term is calculated in the same way as the central Γζΐ3 # (ΜΚ! 火 / 93. if. * &Quot; sqi year J:] tl I ...… ..… two one —--------- Local average of terms. This result can eliminate a color error, which can still be introduced if the first calculation (υ) is used. The output from step 370 uses the second calculation (2) for the red and green sub-pixels, As follows:

Vout (CxRy) = Vin (CxRy) x0.5x ((g_1 ((Vin (Cx-1Ry) + Vin (CxRy)) + 2) + g-1 ((Vin (CxRy + 1) +

Vin (CxRy)) + 2) + g \ (Vin (Cx + 1Ry) + Vin (CxRy) K2) Vk (Vin (CxRy-i) +

Vin (CxRy)) + 2)) + 4) + Vin (Cx.1Ry) x 0.125xg-1 ((Vin (Cx.1Ry) + Vin (CxRy)) ^ 2) + Vin (CxRy + 1) x 〇.125xg-1 ((Vin (CxRy + 1) + Vin (CxRy)) + 2) + Vin (Cx + 1Ry) x 0.125xg-1 ((Vin (Cx + 1Ry) + Vin (CxRy)) + 2) + Vin (CxRy_1) x.125xg_1 ((Vin (CxRy.1) + Vin (CxRy) H2) The formula of the second calculation (2) above is obtained numerically and algebraically with the first calculation (1 ) Has the same result in a gamma combination of 2.0. However, for other gamma settings, the two calculations can be converged with the second calculation (2), but can provide the correct value under any gamma setting. Color rendering. The formula for the sub-pixel rendering of the gamma adjustment of the blue sub-pixel in the first calculation (1) is as follows: V〇ut (Cx + l / 2Ry)-+ Vin (CxRy) x0.5x g- 1 ((4xVin (CxRy) + Vin (Cx.1Ry) + Vin (CxRy + 1) + Vin (C ^ RyhV ^ CxUHS) + Vin (Cx + iRy) x0.5x -66-

g-1 ((4xVin (Cx + 1Ry) + Vin (CxRy) + Vin (Cx + 1Ry.1) + Vin (Cx + 1Ry + 1) + Vin (Cx + 2Ry)) + 8) Use a 4x3 filter The formula for the second calculated blue sub-pixel is as follows: V〇ut (Cx + i / 2Ry) = + Vjn (CxRy) x0.5x ((g-1 ((Vin (Cx_1Ry) + Vin (CxRy) ) +2) + g_1 ((Vin (CxRy + 1) +

Vin (CxRy)) + 2) + g-1 ((Vin (Cx + 1Ry) + Vin (CxRy) H2) + g-1 ((Vin (CxRy-1) + Vin (CxRy)) + 2)) + 4) + Vin (Cx + 1Ry) x0.5x ((g'1 ((Vin (Cx + 1Ry) + Vin (CxRy)) ^ 2) + g-1 ((Vin (Cx + 1Ry + 1) + Vin (Cx + 1Ry) H2) + g-1 ((Vin (Cx + 2Ry) + Vin (Cx + 1Ry)) ^ 2) + g-1 ((Vin (Cx + 1Ry_1) + Vin (Cx + 1Ry)) +2)) + 4) The formula of the blue sub-pixel using the second calculation (2 :) of a 3x3 filter as an approximation is as follows: V0Ut (Cx + i / 2Ry) + Vin (CxRy) xO 5 x ((gl ((Vin (CxRy + 〇 + Vin (CxRy)) ^ 2) + gH (Vin (Cx + 1Ry) +

Vin (CxRy)) ^ 2) + gV (Vin (CxRy_1) + Vin (CxRy))-f2))-3) + Vin (Cx + 1Ry) x0.5x ((g-\ (Vin (Cx + 1Ry) + Vin (CxRy)) ^ 2) + g-\ (Vin (Cx + 1Ry + 1) +

Vin (Cx + 1Ry)) ^ 2) + g-1 ((Vin (Cx + 1Ry_1) + Vin (Cx + 1Ry)) ^ 2)) ^ 3)

The gamma-adjusted sub-pixel rendering method 350 simultaneously provides correct color balance and correct illuminance even at a higher spatial frequency. The non-linear illuminance calculation is performed using a function, which is in the form V ^ fVinxCKxa for each term in the filter core. If a = Vin and CK = 1 are set, the function will return a value equal to the value of the gamma Vin, if the gamma is set to 2. To provide a function to return a value adjusted to a gamma of 2.2 or some other desired value, the form of ν ^ = Σ VinxCKXg'a) can be used in the above formula. This function also maintains the desired gamma for all spatial frequencies. As shown in Figure 47, images using this gamma-adjusted sub-pixel rendering algorithm can have high contrast and correct brightness at all spatial frequencies. Another benefit of using the gamma-adjusted sub-pixel rendering method 350 is that the gamma provided by a lookup table can be based on any desired function. Therefore, for the so-called "sRGB" of the display, standard gamma can also be implemented. This standard has a linear region close to black to replace the exponential curve. When it reaches black, its slope approaches 0 to reduce the required The number of bits is reduced, and the noise sensitivity is reduced. The gamma-adjusted sub-pixel rendering algorithm shown in Figure 49 can also perform Gaussian Difference (DOG) sharpening by using the following, a pixel pair The filter core of the "one-pixel" adjustment mode to sharpen the text image: -0.0625 0.125 -0.0625 0.125 0.75 0.125 -0.0625 0.125 -0.0625

For DOG sharpening, the formula of the second calculation (2) is as follows: ) ^ 2) + 2xg-\ (Vin (CxRy + 〇

Vin (CxRy)) + 2) + 2xg_1 ((Vin (Cx + 1Ry) + Vin (CxRy)) + 2) + 2xg_1 ((Vin (CxRy_1) +

Vin (CxRy)) + 2) + g-k (Vin (Cx.1Ry + 〇 + Vin (CxRy))-2) + g-1 ((Vin (Cx + 1Ry + 1) +

Vin (CxRy)) + 2) + g_1 ((Vin (Cx + 1Ry-1) + Vin (CxRy)) + 2) + g-1 ((Vin (Cx_1Ry-1) +

Vin (CxRy)) +.2)) + 12) + Vin (Cx.1Ry) x0.125xg-1 ((Vin (Cx.1Ry) + Vin (CxRy) H2) + Vin (CxRy + 〇x0.125xg-1 ((Vin (CxRy + 1) + Vin (CxRy))-2) + Vin (Cx + 1Ry) x0.125xg-1 ((Vin (Cx + 1Ry) + Vin (CxRy)) + 2) + Vin (CxRy_1 ) x0.125xg_1 ((Vin (CxRy_1) + Vin (CxRy)) + 2) -Vin (Cx.1Ry + 1) x0.0625xg-1 ((Vin (Cx_1Ry + 1) + Vin (CxRy))-2) -Vin (Cx + 1Ry + 1) x0.0625xg-1 ((Vin (Cx + 1Ry + 1) -fVin (CxRy))-2) -Vin (Cx + 1Ry.1) x0.0625xg-l ((Vin (Cx + 1Ry.1) + Vin (CxRy))-2) -Vin (Cx.1Ry.1) x0.0625xg-1 ((Vin (Cx.1Ry.1) + Vin (CxRy))-2) phase Compared to the diagonal term, the coefficient of the ordinal average term is 2 because the ratio in the filter core is 0.125: 0.0625 = 2. This keeps each contribution to the local average equal. This DOG sharpening can provide odd harmonics of this fundamental spatial frequency, which is -69- fmmmk…

Ej is caused by the pixel edges of vertical and horizontal strokes. The dog sharpening filter shown above borrows the same colored energy from the corner and places it at the center, so the DOG sharpened data becomes a small focus point when the human eye rotates. This sharpening is called the same color sharpening. The amount of sharpening can be adjusted by changing the core coefficients of the middle and corner filters. The intermediate coefficient can be changed between 0.5 and 0.75, and the corner coefficient can be changed between 0 and -0.0625, and the sum is 1. In the above exemplary filter core, 0.0625 is taken from each of the four corners, and the sum of these (ie, 0.0625x4 = 0.25) is added to the central term, so it is increased from 0.5 to 0.75. In general, a sharpened core of a waver can be expressed as follows:

cirx C21 C31 " X c12 C22 + 4X c32 c13-x c23 C33-X where (-X) is called a corner sharpening coefficient; (+ 4x :) is called a central sharpening coefficient: and ... ^^ "It is called the rendering coefficient. In order to further increase the image quality, the sharpening coefficients including the four corners and the center can use the opposite color input image values. This sharpening is called cross-color sharpening because The sharpening coefficient uses the input image value, and its color is opposite to the rendering coefficient. The cross-color sharpening can reduce the tendency of the sharpened saturated color lines or text to appear pointy. Even if the opposite color is used to perform Sharpening, not the same color, -70- fl is different from Y years-the total energy does not change in illumination or chromatic aberration, and the color remains the same. This is because the sharpening coefficient causes the energy of the opposite color Move towards the center, but the balance is OGx-x + Ax-xx ^ O). If you use this cross-color sharpening, the previous formula can be simplified by separating the sharpening term from the rendering term. Because the sharpening term Times It does not affect the illuminance or chromatic aberration of the image, but only affects the energy distribution, and the gamma correction using the sharpening coefficient of the opposite color can be omitted. Therefore, the following formula can be used instead of the previous formula: V〇ut (CxRy ) = Vin (CxRy) x0.5x

((g-Wv ^ CdRyHv ^ CxRyDeHg-WvjCxU + Vin (CxRy)) + 2) + g-1 ((Vin (Cx + 1Ry) + Vin (CxRy)) + 2) + g_1 ((Vin (CxRy_1) +

Vin (CxRy)) + 2)) + 4) + Vin (Cx.1Ry) x0.125xg-1 ((Vin (Cx.1Ry) + Vin (CxRy) K2) + Vin (CxRy + 1) x0.125xg_1 ( (Vin (CxRy + 1) + Vin (CxRy) K2) + Vin (Cx + 1Ry) x0.125xg-1 ((Vin (Cx + 1Ry) + Vin (CxRy) K2) + Vin (CxRy_1) x0.125xg- 1 ((Vin (CxRy-1) + Vin (CxRy)) + 2) (where the above Vin is a full red or green value) + Vin (CxRy) x0.125

HiUxO.03125 -Vin (Cx + 1Ry + 1) x〇.03125 -νίη (εχ + 1υχ〇.〇3ΐ25 71 ·

23 ^ (M K

(Where the above vin are all green or red, respectively, and opposite to the vin selected in the above paragraph) The same and crossed color sharpening mix can be shown as follows:

Vout (CxRy) = Vin (CxRy) x0.5x ((g_1 ((Vin (Cx_1Ry) + Vin (CxRy) H2) + g-1 ((Vin (CxRy + 1) +

Vin (CxRy)) + 2) + g-1 ((Vin (Cx + 1Ry) + Vin (CxRy) H2) + g-1 ((Vin (CxRy_1) + Vin (CxRy)) + 2)) + 4) + Vin (Cx_1Ry) x0.125xg_1 ((Vin (Cx_1Ry) + Vin (CxRy)) + 2) + Vin (CxRy + 1) x0.125xg-1 ((Vin (CxRy + 1) + Vin (CxRy) H2) + Vin (Cx + 1Ry) x0.125xg-l ((Vin (Cx + 1Ry) + Vin (CxRy) K2) + Vin (CxRy_1) x0.125xg-1 ((Vin (CxRy_1) + Vin (CxRy)) + 2)

+ Vin (CxRy) x0.0625 -VinCC ^^^ OxO.015625 -Vin (Cx + 1Ry + 1) xO.015625 -VinCC ^^ y.OxO.015625 -VinCC ^^ y.OxO.015625 (where the above Vin is all red or green value) + Vin (CxRy) x0.0625 -VinCC. ^ Ry + OxO.015625 -V in (Cx + iRy + i) xO.015625 -Vi ^ C. ^ Ry.OxO.015625 -72- | ι—Μ January of the year & ί -Vi ^ Cx ^ Ry.OxO.015625 (where the above Vins are all green or red, respectively, as opposed to the Vin selected in the above paragraph) in In these simplified formulas using cross-color sharpening, the coefficient term is half that of the same color sharpening with gamma adjustment. That is, the central sharpening term becomes half of 0.25, which is equal to 0.125, and the angular sharpening term becomes half of 0.625, which is equal to 0.03125. This is because if there is no gamma adjustment, this sharpening has a greater effect. Only the red and green channels have the benefit of sharpening because the human eye cannot perceive the blue details. Therefore, the sharpening of blue is not performed in this specific embodiment. The following method of FIG. 51 is used for the sub-pixel rendering of gamma adjustment with an omega function, which can control the gamma value without causing color errors. In brief, FIG. 50 shows an exemplary output signal of the gamma-adjusted sub-pixel with an omega function in response to the input signal of FIG. 43. The sub-pixel rendering is adjusted based on the gamma without Omega correction, and the rendered gamma value increases for all spatial frequencies, so the contrast ratio of the high spatial frequency increases, as shown in Figure 47. When the gamma value is further increased, for example, the detailed contrast of black text on a white background will be further increased. However, increasing the gamma value of all spatial frequencies will produce unacceptable photos and video images. The gamma-adjusted sub-pixel rendering method with omega correction in FIG. 51 can selectively increase the gamma value. In other words, the gamma value at the high spatial frequency increases when the gamma value of the zero spatial frequency remains at its optimal point -73-• plus. Therefore, as the spatial frequency becomes higher, the average of the output signal wave exhibiting a downward shift adjusted by the gamma is further shifted downward, as shown in FIG. 50. The average energy at zero frequency is 25% (corresponding to 50% brightness), and it is reduced to 9 · 5% (corresponding to 35% brightness) at the Nyquist limit, if ω = 0.5. FIG. 51 illustrates a method 400 including a series of sub-pixel rendering steps with gamma adjustment. Basically, the Omega 55 function, w (x) = x1 / co (step 404), is inserted after receiving the input data Vin (step 402), and before accepting the data for the local average calculation (step 406) The Omega-corrected local average (β), which is output from step 406, receives the inverted Omega function w'xfx% step 408) in the pre-gamma correction. Therefore, step 408 is called "pre-gamma with omega" correction, and the calculation of g ·, · 1 is performed, for example, with reference to a pre-gamma with omega watch in the form of a LUT. This function is an inverted gamma-like function, and w'x :) is a gamma-like function with the same omega value. The term, "Omega" is chosen as it is often used in electronic devices to represent a signal frequency in radians. This function will result in higher spatial frequencies rather than fewer. In other words, the omega and inverse omega functions do not change the output value at lower spatial frequencies, but have a larger effect at higher spatial frequencies. If "Vi" and "V2" are used to represent the two local input values, which are two local values, the local average value (〇〇 and the Omega-corrected local average (β) are as follows: -74---L —Year n \ (ν! + Ν2) / 2 = α; and (XVD + wCV〗)) /] ^. When VfVz, β = \ ν (ο〇. Therefore, at low spatial frequencies, g-V, g_V ^ h ^ gla). However, at high spatial frequencies (Vl # V2), g-iw-i ((5> g-i⑹. At the highest spatial frequency and contrast, g-1w-1 (p) «g_1w-1 (a) In other words, the function form for subpixel rendering with Omega's gamma adjustment is VouteVinxCKXg-VWwC ^ + wC ^: ^), where g'xhx7-1, w (x) = x1 / G)) and. The result of using this function is that a low spatial frequency exhibits a gamma value of g-1, while a high spatial frequency system effectively exhibits a gamma value of g ·, · 1. When the omega value is set below 1, a higher spatial frequency has a higher effective gamma value, which is a higher contrast between black and white. The operation after the pre-gamma with Omega of Fig. 51 is similar to those of Fig. 49. For each edge term, the result of pre-gamma with omega correction is multiplied by a corresponding coefficient term CK, which is read from a filter core coefficient table 412 (step 410). For this central term, there are at least two ways to calculate the corresponding value. The first method calculates the value of the edge term in the same way, and the second method performs the calculation of step 414 in Fig. 51 to add up the result of step 408. The calculation of step 414 can use the result of step 410 instead of step 408 to represent the edge coefficient of the central term, when each edge term can make a different contribution to the central term's local average. The Omega-corrected local average ("GOA") from the central term of step 414 is also multiplied by a corresponding coefficient term CK (step 416). The value from step 410, and -75- from step 416 using the second calculation (2). 93: 12: '~ 8 ^^ value, that is, multiplied by a corresponding term Vin (steps 418 and 420) . Then, the sum of all multiplied terms is calculated (step 422) to output the data V represented by the sub-pixels. ^. Then, a post-gamma correction is applied to and output to the display (; steps 424 and 426). For example, the output avoided from step 422 using the second calculation (2) for red and green sub-pixels is as follows:

Vout (CxRy) = Vin (CxRy) x0.5x ((g-1W1 ((w (Vin (Cx.1Ry)) + w (Vin (CxRy))) ^ 2) + g'1W1 ((w (Vin ( CxRy + 1)) + w (Vin (CxRy))) ^ 2) + g1W1 ((w (Vin (Cx + 1Ry)) + w (Vin (CxRy))) ^ 2) + g_V-1 ((w ( Vin (CxRy.1)) + w (Vin (CxRy)) K2)) ^ 4) + Vin (Cx.1Ry) x0.125xg-1w-1 ((w (Vin (Cx.1Ry)) + w (Vill (CxRy)))-2) + Vin (CxRy + 1) x0.125xg'1w'1 ((w (Vin (CxRy + 1)) + w (Vin (CxRy))) + 2) + Vin (Cx + 1Ry) x.125xg-1w-1 ((w (Vin (Cx + 1Ry)) + w (Vin (CxRy)) H2) + Vin (CxRy_1) x〇.125xg-V1 ((w (Vin (CxRy_1) ) + w (Vin (CxRy))) + 2) This additional exemplary formula for the red and green sub-pixels can use the simplified method described above to improve the previous formula by the cross-color sharpening and corner sharpening coefficients (X). , Which is as follows: V〇ut (CxRy) = Vin (CxRy) x0.5x ((g'1W1 ((w (Vin (Cx.1Ry)) + w (Vin (CxRy)))-2) + g -1w'1 ((w (Vin (CxRy + 1)) + w (Vin (CxRy)))-2) day 丨 + g-1w-1 ((w (Vin (Cx + 1Ry)) + w (Vin (CxRy))) ^ 2) + g-1w-1 ((w (Vin (CxRy.1)) + w (Vin (CxRy))) ^ 2)) ^ 4) + Vin (Cx_1Ry) x0.125xg- 1w-1 ((w (Vin (Cx.1Ry)) + w (Vin (CxRy)) K2) + Vin (CxRy + 1) x0.125xg-1w-1 ((w (Vin (CxRy + 1)) + w (Vin (CxRy))) ^ 2) + Vin (Cx + 1Ry) x0.125xg-1w-1 ((w (Vin (Cx + 1Ry)) + w (Vin (CxRy)) K2) + Vi n (CxRy_1) x0.125xg-1w_1 ((w (Vin (CxRy_1)) + w (Vin (CxRy))) + 2) + Vin (CxRy) x4x -Vin (CX_l Ry + 1) xx -Vin (Cx + 1Ry + i) xx

-VinCCx + ^ y.OxX -VinCCx ^ Ry.Oxx The formula for the sub-pixels with gamma adjustment of the Omega function for the blue sub-pixels is as follows: V〇ut (Cx +. / 2Ry) = · + Vin (CxRy ) xO. 5 x ((g-1w-1 ((w (Vin (Cx.1Ry)) + w (Vin (CxRy))) ^ 2) + g_1 ((w (Vin (CxRy + 1)) + w (Vin (CxRy)) K2) + g-V1 ((w (Vin (Cx + 1Ry)) + w (Vin (CxRy))) + 2) + gl ((w (Vin (CxRy.1)) + w (Vin (CxRy)) ^ 2)) ^ 4) + Vin (Cx + 1Ry) x0.5x ((g-1w-1 ((w (Vin (Cx + 1Ry)) + w (Vin (CxRy)) K2 ) + g_1 ((w (Vin (Cx + 1Ry + 1)) + w (Vin (Cx + 1Ry)) H2) -77- + g_1w-1 ((w (Vin (Cx + 2Ry)) + w (Vin (Cx + 1Ry))) + 2) + g'1 ((w (Vin (Cx + 1Ry.1)) + w (Vin (Cx + 1Ry)) ^ 2)) ^ 4} The adjustment for the super essence ( That is, the general formula for the representation of the gamma adjustment with Omega with cross-color sharpening and an adjustment ratio of 1: 2 or higher can be expressed as follows for the red and green sub-pixels: V0Ut (CcRr) Di Vin (CxRy) xc22x ((g-1W1 ((w (Vin (Cx.1Ry)) + w (Vin (CxRy))) ^ 2) + g-1w-1 ((w (Vin (CxRy + 1)) + w (Vin (CxRy))) ^ 2) + g-1W1 ((w (Vin (Cx + 1Ry)) + w (Vin (CxRy))) ^ 2) + g-1w * \ (w (Vin (CxRy.1) ) + w (Vin (CxRy))) ^ 2)) ^ 4) + Vin (Cx.1Ry) xc12xg-1w-1 ((w (Vin (Cx.1Ry)) + w (Vin (CxRy)) H2) + Vin (CxRy + 1) xC23xg_V 1 ((w (Vin (CxRy + 1)) + w (Vin (CxRy)) H2) + Vin (Cx + 1Ry) xc32xg-1W1 ((w (Vin (Cx + 1Ry)) + w (Vin (CxRy) ) K2) + Vin (CxRy_1) xc21xg-V1 ((w (Vin (CxRy1)) + w (Vin (CxRy))) + 2) + Vin (Cx_1Ry + 1) xCl3xg-1W-1 ((w (Vin ( Cx_1Ry + 1)) + w (Vin (CxRy))) + 2) + Vin (Cx + 1Ry + 1) xc33Xg_lw " 1 ((w (Vin (Cx + 1Ry + 1)) + w (Vin (CxRy)) ) +2) + Vin (Cx + 1Ry_1) xc31xg-V1 ((W (Vin (Cx + 1Ry_1)) + w (Vin (CxRy)) H2) + Vin (Cx_1Ry_1) xc11xg-1W-1 ((w (Vin (Cx_1Ry_1)) + W (Vin (CxRy)) H2) + Vin (CxRy) x4x

-Vin (Cx_i Ry +1) XX -Vin (CX + lRy + i) XX ifmoiK1… J year ... y

Vin (Cx + lRy-l) xx -V ^ CwUxx The corresponding general formula of the blue sub-pixel is as follows: V0Ut (Cc + 1 / 2Rr) =

+ Vin (CxRy) xc22xR

+ Vin (Cx + 1Ry) xc32xR

+ Vin (Cx.iRy) xci2xR

+ Vin (CxRy_i) xc2ixR

+ Vin (Cx + iRy_i) xc31xR

+ Vin (Cx.iRy.1) xc11xR where R = ((g-1w-1 ((w (Vin (Cx) Ry)) + w (Vin (CxRy))) + 2) + g * 1 ((w (Vin (CxRy + i)) + w (Vin (CxRy)))-2) + g-V1 ((w (Vin (Cx + 1Ry)) + w (Vin (CxRy))) + 2) + g- 1 ((w (Vin (CxRy1)) + w (Vin (CxRy))) + 2) + ((g-1w-l (w (Vin (Cx + 1Ry)) + w (Vin (CxRy)) K2) + g'1 ((w (Vin (Cx + 1Ry + 1)) + w (Vin (Cx + 1Ry))) ^ 2) + g-1w-1 (w (Vin (Cx + 2Ry)) + w ( Vin (Cx + 1Ry))) ^ 2) + g-1 ((w (Vin (Cx + 1Ry.1)) + w (Vin (Cx + 1Ry)) K2) K2) K8) Figures 46, 49 and above The method of 51 can be implemented by the following exemplary system. An example of a system for implementing the steps of FIG. 46 for pre-adjusted gamma presentation at sub-pixels is shown in FIGS. 52A and 52B. The exemplary system can use a thin film A transistor (TFT) active matrix liquid crystal display (AMLCD) panel displays images. Other types of display device packages that can be used to implement the above technologies

Including cathode ray tube (CRT) display device. Please refer to FIG. 52A. The system includes a human computing device (PC) 501 coupled to a sub-pixel rendering module 504 having a primary pixel processing unit 500. The PC 501 may include components of the computing system 750 of FIG. 71. The sub-pixel rendering module 504 of FIG. 52A is coupled to a timing controller (TCON) 506 in FIG. 52B, and is used to control a panel output to a display. Other types of devices available for the PC 501 include a portable computer, a palm computing device, a personal data assistant (PDA), or other similar device with a display. The sub-pixel rendering module 504 can implement the above-mentioned adjusted sub-pixel rendering technology, which has the gamma adjustment technology described in FIG. 46 to output data for sub-pixel rendering. The PC 501 may include a graphics controller or interface card, such as a video graphics card (VGA) to provide image data output to a display. Other types of VGA controllers available include UXGA and XGA controllers. The sub-pixel rendering module 504 may be an independent card or board, which may be set as a field programmable gate array (FPGA), which is programmed to perform the steps described in FIG. 46. In addition, the sub-pixel processing unit 500 may be included in the graphics card controller of the PC 501 as an application-specific integrated circuit (ASIC) for performing pre-adjusted gamma before sub-pixel rendering. In another example, the sub-pixel rendering module 504 may be an FPGA or ASIC in the TCON 506 of a display panel. Furthermore, the sub-pixel rendering module 504 may be implemented in one or more devices or units connected between the pc 501 and the TCON 506 to output images on a display. The sub-pixel rendering module 504 also includes a digital visual interface (DVI) input 508 and a low voltage differential signaling (LVDS) output 526. The sub-pixel rendering module 504 can receive input image data through the DVI input 508, such as a standard RGB pixel format of I238Wt, and perform pre-adjusted gamma on the image data before sub-pixel rendering. The sub-pixel rendering module 504 can also transmit the data presented by the sub-pixel to the TCON 506 through the LVDS output 526. The LVDS output 526 may be a panel interface of a display device, such as an AMLCD display device. In this way, a display can be coupled to any kind of graphics controller or card using a DVI output. The sub-pixel rendering module 504 also includes an interface 509 connected to the PC 501. The interface 509 may be an I2C interface, which allows the PC 501 to control or download and update the gamma or coefficient table used by the sub-pixel rendering module 504, and stores it in an extended display identification information (EDID) unit 510. Get information. In this way, the gamma and coefficient values can be adjusted to any desired value. Examples of EDID information include basic information about a display and its capabilities, such as maximum image size, color characteristics, preset timing frequency range limits, or other information. The PC 501, for example, can read the information in the EDID unit 510 when it is turned on, to determine the type of display connected to it, and how to transmit image data to the display. The operation of the sub-pixel processing unit 500 operating in the sub-pixel rendering module 504 to implement the steps of FIG. 46 will now be described. For the sake of explanation, the sub-pixel processing unit 500 includes processing blocks 512 to 524, which are implemented in a large FGPA, having any number of logic components or circuits, and a storage device to store a gamma table and / or a coefficient table. Examples of storage devices that store these tables include read-only memory (ROM), random-access memory (RAM), or other similar memory. At the beginning, PC 501 transmits an input image data Vin via DVI 508 (eg -81-ifii3tS ¥… L.— 一 土. „—. ·]] '… 一,'. ·! Pixel data in standard RGB format) Go to the sub-pixel rendering module 504. In other examples, the PC 501 may transmit the input image data Vin in the pixel format once, as described above. The way in which the PC 501 transmits vin may be based on the information in the EDID unit 510. In one example, A graphics controller in the PC 501 transmits red, green, and blue sub-pixel data to the sub-pixel rendering unit 500. The input latch and automatic detection block 512 detects the image data received by the DVI 508 and latches the Pixel data. The timing buffer and control block 514 provides buffer logic to buffer the pixel data in the sub-pixel processing unit 500. Here, in Fang Gui 514, the timing signal can be sent to the output synchronization generation block 528 to allow receiving input data. Vin, and sends the output data Vout to be synchronized. The pre-adjusted gamma processing block 516 processes the image data from the timing buffer and the control block 514 to perform step 304 of FIG. This function is calculated from the image data Vin, where the function value at a given γ can be obtained from a pre-adjusted gamma table. The pre-adjusted gamma has been applied to the image data Vin, which is stored at line buffer block 518 Line buffer. In one example, three line buffers can be used to store the input images of three lines, as shown in Figure 55. Other examples of storing and processing image data are shown in Figures 56 to 60. The image data stored in the online buffer block 518 is sampled in the 3x3 data sampling block 519. Here, the 9 values including the central value can be sampled in the register or latch of the sub-pixel rendering processing. Coefficient processing Block 530 executes step 308, and multiplier + adder block 520 executes step 306, where the value of "g" (x) for each of the 9 sampled values is multiplied by the value stored in the coefficient -82- Close Table 53 1 The core coefficient value of the filter, then the multiplied term is added to obtain the output image data V () ut presented by the sub-pixel. The post-gamma processing block 522 performs step 310 of FIG. 46 on the VQUt, in which a post-gamma correction of a display is applied. That is, the post-gamma processing square rose 522 refers to a post-gamma table and uses a function f (x) to calculate f ^ VJ of the display. The output latch 524 interlocks the data from the post-gamma processing block 522, and the LVDS output 526 transmits the output image data from the output latch 524 to the TCON 506. The output synchronization generation stage 528 controls the timing of operations in the squares 516, 518, 519, 520, and 530, and in 522 controls when the output data V ^ t is transmitted to the TCON 506. Referring to FIG. 52B, the TC0N 506 includes an input latch 532 to receive output data from the LVDS output 526. The output data from the LVDS output 526 may contain blocks of 8-bit image data. For example, TCON 506 can receive sub-pixel data based on the sub-pixel configuration described above. In one example, the TCON 506 can receive 8-bit rows of data, where the odd-numbered columns (such as RGBRBGRBG) precede the even-numbered columns (GBRGBRGBR). The 8 to 6-bit color mixing cube 534 converts 8-bit data to 6-bit data to a display that requires a 6-bit data format, which is standard for many LCDs. Therefore, in the example of Fig. 52B, the display uses this 6-bit format. Block 534 sends the output data to the display via a data bus 537. TCON 506 includes a reference voltage and video transmission (VCOM) voltage block 536> Fang Gui 536 provides a voltage reference from the DC / DC converter 538, which is controlled by the row driver 539 and the column driver control 539B to be selectively turned on. Row and column transistors in a panel of the display. In one example, the display is a flat-panel display having a matrix of rows and rows of -83--12-percent-year pixels, and the corresponding transistors are driven by a column driver and a row driver. The sub-pixel may have the above-mentioned sub-pixel configuration. An example of a system for performing the steps of FIG. 49 for performing sub-pixel rendering of gamma adjustment is shown in FIGS. 53A and 53B. This exemplary system is similar to the system of FIGS. 52A and 52B, except that the sub-pixel processing unit 500 uses at least a delay logic block 521, a local average processing block 540, and a pre-gamma processing block 542 to perform the sub-pixel rendering of the gamma adjustment. , And the pre-adjusted gamma processing block 516 is omitted. The operation of the processing method of the sub-pixel processing unit 500 of Fig. 53A will now be explained. Referring to FIG. 53A, the PC 501 transmits the input image data Vin (such as pixel data of a standard RGB format) to the sub-pixel rendering module 504 through the DVI 508. In other examples, the PC 501 can transmit the input image data Vin in pixel format once, as described above. The input latch and automatic detection block 512 detects the image data being received by the DVI 508 and latches the pixel data. The timing buffer and control block 514 provide buffering logic to buffer the pixel data in the sub-pixel processing unit 500. Here, in the square rose 514, the timing signal may be transmitted to the output synchronization generating block 528 to allow receiving the input data Vin and transmit the output data V ^ to be synchronized. The image data Vin being buffered in the timing and control block 514 is stored in the line buffer of the line buffer block 518. The line buffer block 518 can store image data in the same manner as in Figure 52A. The input data stored in the online buffer block 518 is sampled in the 3x3 data sampling block 519, which can be performed in the same manner as in Fig. 52A. Here, the 9 values containing the central value can be taken from the register or latch of the sub-rendering process of the gamma adjustment. -84-

kind. Next, the local average processing square 540 of FIG. 49 performs step 354, where the local average (α) is calculated using the central term of each edge term. Based on the local averaging, the pre-gamma processing block 542 executes step 356 of FIG. 49 for the one, pre-gamma correction, and calculates β′οΟΉ, which uses a pre-gamma lookup table (LUT). The LUT can be included in this square or accessed in the sub-pixel rendering module 504. The delay logic block 521 may delay supplying Vin to the multiplier + adder block 520 until the local averaging and pre-gamma calculations are completed. Coefficient processing block 530 and multiplier + adder block 520 use coefficient table 531 to perform steps 358, 360, 362, 364, 366, 368, and 370, as described in FIG. 49. In particular, the values of CKg, o0 from step 358, and the values of CK "GA" from step 364 use, for example, the second calculation (2) described in FIG. 49, which is multiplied by a corresponding Item Vin (steps 366 and 368). Block 520 calculates the sum of all multiplied terms (step 370 > to generate the data V represented by the output sub-pixels.) The post-gamma processing block 522 and the output latch 524 are performed in the same way as in FIG. 52A. The output image data is transmitted to TCON 506. The output synchronization generation stage 528 in FIG. 53A controls the timing of performing operations in blocks 518, 519, 521, 520, 530, and in 522 controls when the output data is transmitted to TCON 506. The TCON 506 in FIG. 53B operates in the same way as in FIG. 52B, except that the output data obtained by using the method in FIG. 49 is used. The sub-pixel rendering for implementing the gamma adjustment with an omega function of FIG. An example of a step system is shown in Figs. 54A and 54B. This exemplary system is similar to the system of Figs. 53A and 53B, except that the sub-pixel processing -85-E! (With Omega) processing block 545 to perform the gamma-adjusted sub-pixel rendering with the Omega function. The operation of the processing cube of the sub-pixel processing unit 500 of FIG. 54A will now be explained Referring to FIG. 54A, processing blocks 512, 514, 518, and 519 operate in the same manner as the processing block of FIG. 53A. The omega function processing block 544 performs step 404 of FIG. 51, where the omega function w (x) = x1 / 0) is applied to the input image data from the 3x3 data sampling block 519. The local averaging processing block 540 executes step 406, where the central term of each edge term is used to calculate the omega-corrected local mean (β). The pre-gamma (with Omega) processing block 545 executes step 408, where the output from the local average processing block 540 accepts the calculation g · 1 # 1, which is implemented as 8-1 (νΛβ) Μβω) γ "to use The pre-gamma of the Omega LUT has the "pre-gamma" correction. The processing blocks 520, 521, 530, 522, and 524 of FIG. 54A operate in the same manner as in FIG. 53A, except that the result of pre-gamma correction with omega is multiplied by a corresponding coefficient term for each edge term. CK. The output synchronizing block 528 of Fig. 54A controls the timing of execution operations in blocks 518, 519, 521, 520, and 530, and 522 controls when the output data is transmitted to TCON 506 for display. TCON 506 in FIG. 54B operates in the same manner as in FIG. 53B, except that the output data obtained by using the method in FIG. 51 is used. It is possible to make other changes to the examples of Figs. 52A, 52B, 53A-53B and 54A-54B above. For example, the components of the above example can be implemented on a single module and selectively controlled to decide which type of processing to perform. For example, -86-

Paste # Q4K

In other words, this module can be provided with a switch, or it can be set to receive commands or instructions to selectively operate the methods of FIGS. 46, 49, and 51. Figures 55 to 60 show exemplary circuits that can be used by processing cells in the exemplary systems shown in Figures 52A, 5 3 A, and 54 A. The above-mentioned sub-pixel rendering method requires a lot of calculations, which involves multiplying the coefficient filter value by the pixel value, and adding many multiplication terms. The following specific examples disclose a circuit to efficiently perform such calculations. Please refer to Fig. 55, which shows the circuit examples of the line buffer square 518, the 3x3 data sampling block 519, the coefficient processing block 530, and the multiplier + adder block 520 (see Figs. 52A, 53A, and 54A). This exemplary circuit can perform the sub-pixel rendering function described above. In this example, the line buffer square frame 518 includes line buffers 554, 556, and 558, which are combined to store the input data (Vin). Input data or pixels can be stored in these line buffers, which allows latching in the 3x3 data sampling block 519. Go to L9 and sample 9 pixel values. By storing 9 pixel values in latches 1 ^ to L9, 9 pixel values can be processed in a single clock cycle. For example, the 9 multipliers called M9 can be there. The pixel value in the L9 latch is multiplied by the appropriate coefficient value (filter value) in the coefficient table 531 to implement the sub-pixel rendering function described above. In another implementation, the multiplier may be replaced with a read-only memory (ROM), and the pixel value and coefficient filter value may be used to generate a bit address to obtain the multiplied term. As shown in Figure 5, multiple multiplications can be performed and added in an efficient manner to perform the sub-pixel rendering function. Figure 56 shows the line buffers Fang 518, I238QI h B. 3x3 data sampling block 519, coefficient processing Fang 530, and multiplier + adder block 520 using two summing buffers to perform sub-pixel rendering. Features. As shown in Figure 56, three latches 1 ^ to L3 store pixel values, which are input to 9 multipliers 吣 to M9. Multipliers 吣 to M3 are multiplied by the pixel values from the lock 1 ^ to L3 and the appropriate coefficient values in the coefficient table 531, and the result is fed to the adder 564, which calculates the sum of the results and stores the sum in the sum buffer器 560。 In the device 560. The multiplier is multiplied by 4 to M6 by the interlock. To the pixel value of L6 and the appropriate coefficient value in the coefficient table 531, and feed the result to the adder 566, which calculates the sum of the output from% to M6 multiplied by the sum buffer 560, and stores the sum in the sum buffer器 中 562。 In 562. Multiplier% to] ^ 9 times the pixel value from the latch to L9 and the appropriate coefficient value in the coefficient table 531, and feeds the result to the adder 568, which calculates from the centimeter 9 times the sum buffer 562 The sum of the outputs is used to calculate the output value V ^ t. This example of FIG. 56 uses two partial sum buffers 560 and 562, which can store 16-bit values. By using two sum buffers, this example of FIG. 56 can provide an improvement on the three line buffer samples so that it uses less buffer memory. Figure 57 shows an example of a circuit that can be used by the processing blocks of Figures 52A, 53A, and 54A to implement sub-pixel rendering functions for red and green pixels. Specifically, this example can be used for this 1: 1P: S ratio resolution during sub-pixel rendering of red and green pixels. This 1: 1 case provides simple sub-pixel rendering calculations. In this example, all values contained in the core of the filter are powers of 0, 1, or 2, as shown above, which reduces the number of multipliers required, as described below. -88-

0 1 0 1 4 1 0 1 0

Please refer to FIG. 57, which shows 9 pixel delay registers & to 119 to store pixel values. The register core to R3 is fed to the line buffer 1 (570), and the output of the line buffer 1 (570) is fed to the register ruler 4. Registers R4 to R7 are fed to line buffer 2 (572>. The output of line buffer 2 (572) is fed to register R7, which is fed to registers 118 and R9. Adder 575 is added Values from ruler 2 and R4. Adder 576 adds values from heart and R8. Adder 578 adds values from adders 575 and 576. Adder 579 adds values from adder 578 and the bucket. The output of the offsetter 573, its execution multiplies the value from R5 by 4. The output of this adder 579 is fed to a bucket offsetter 574, which performs a division by 8.

Because the 1: 1 filter core is zero in 4 positions (as shown above), it does not require a 4 pixel delay register for sub-pixel rendering, because the value of 4 is 1, which makes it added instead of Multiplication is required, as shown in Figure 57. Figure 5 8 shows an example of a circuit that can be used by the processing blocks of Figures 5 2 A, 5 3 A, and 5 4 A to implement sub-pixels in the case of a 1: 1 P: S ratio of blue pixels. Render. For blue pixels, only a 2x2 filter core is required, thereby allowing the necessary circuitry to be more complex. Please refer to Figure 58, which shows the 9-pixel delay register to receive the input pixel value. Register & to ruler 3 is input to line buffer 1 (580), and the line is buffered -89-1238011; the output of punch 1 (580) is fed to register R4. Registers 4 to 7 are fed to line buffer 2 (582). The output of this line buffer 2 (582) is fed to the register R7, which is fed to the registers 118 and R9. The adder 581 adds the values to the registers R4, R5, H7 and H8. The output of this adder is fed to a barrel offsetter 575, which performs a division by four. Because the blue pixel contains only the values in the four registers, and those values are shifted through the pixel delay register & to R9, and appear in the four different red / green output pixel clock cycles, The blue pixel calculation may be performed early in the process. Figure 59 shows an example of a circuit that can be used by the processing blocks of Figures 52A, 53 A, and 54A to implement a subpixel rendering function of a 1: 1P: S ratio of red and green pixels using two sum buffers. . By using a summation buffer, it simplifies the necessary circuits. Please refer to Figure 5-9, which shows three pixel delay registers & to R3 to receive input pixel values. The register & is fed to the adder 591. Register R2 is fed to sum buffer 1 (583), bucket offsetter 590, and adder 592. The register R3 is fed to the adder 591. The output of this sum buffer 1 (5 83) is fed to an adder 591. The adder 591 adds the values from the registers Ri and R3, and the value of R2 from the bucket offsetter 590 is multiplied by two. The output of the adder 591 is fed to the sum buffer 2 (584), which sends its output to the adder 592, which adds this value and the value in the heart to produce the output. Figure 60 shows an example of a circuit that can be used by the processing blocks of Figures 52A, 53A, and 54A to implement a sub-pixel rendering function with a 1: 1P: S ratio using a sum buffer. By using a sum buffer, this necessary circuit can further simplify the blue pixels. Please refer to FIG. 60, there are not two -90- mfn ~. ~ R, ‘day 丨 pixel delay registers to r2 to receive input pixel values. Registers & and r2 are fed to adders 593 and 594. The adder 593 adds the values from 1 ^ and R2, and stores the output in the sum buffer 1 (585). The output of the sum buffer 1 (585) is fed to the adder 594. The adder 594 adds the value from Rp R2 And sum buffer 1 (585) to generate the output. Figure 61 shows a flowchart of a method 600 for timing black pixels at the edges of a display during the sub-pixel rendering process described above. The above Subpixel rendering calculations require a 3x3 matrix of 3x3 filter values to be applied to a matrix of one pixel value. However, for an image with one pixel at the edge of the display, surrounding pixels will not exist around the edge pixels To provide a 3x3 matrix value of the pixel value. The following method can deal with the problem of determining the surrounding pixel value of the edge pixel. The following method assumes that all pixels of an image at the edge of the display are black and have a zero Pixel value. This method can be implemented by the input latch and automatic detection block 512, the timing buffer and control block 514, and the line buffer block 518 of FIGS. 52A, 53A, and 54A. Initially, a line buffer is initialized to zero for a black pixel before timing the first scan during a vertical turn-back period. Step 602 > The first scan line can be stored in a line buffer. Then, a scan line is The second scan line is output during timing. Step 604> This can occur when the calculation for the first scan line is completed, and its package comes from one of the black pixels. An extra zero is counted in each scan line (step 606) before a first black pixel is timed. Then, the pixel can be output when the second actual pixel is in timing (step-91-1_Li 1] 608>. This can occur when the calculation of the first pixel is completed. One black one of the pixels Zero is counted after the last actual pixel on a scan line has been timed (step 610). For this method, a line buffer or a sum buffer, as described above, can be set to store two additional pixel values to store the black pixel, as described above. The two black pixels can be timed during the horizontal retrace. Then, after the last scan line has been counted, one more scan line is counted for all zero (black) pixels from the above steps. This output can be used when the last scan has been calculated. These steps can be completed during the vertical rewind. Therefore, the above method can provide pixel values of a 3x3 matrix of pixel values of edge pixels during sub-pixel rendering. Figures 62 to 66 show exemplary square rosettes of the system to improve the color resolution of the image on a display. The limitations of current imaging systems in increasing color resolution are disclosed in detail in U.S. Provisional Patent Application No. 60 / 311,138, entitled, "Improved Gamma Table," which was filed on August 8, 2001. In short, adding color resolution is expensive and difficult to sell. That is, for example, if a filtering process is performed, the weighted sum is divided by a fixed value so that the hip effect of the filter result is equal to one. The division calculated by the division (as described above) can be a power of one, so that the division operation can be shifted to the right, or only by discarding the least significant bit. For this processing, the least significant bit is usually discarded, shifted, or removed, and it is not used. But these bits can be used to increase color resolution, as described below. Please refer to FIG. 62, an exemplary square rosette display of a system to perform sub-pixel rendering, which uses a wide digital-to-analog converter or LVDS, which improves color-92- ---. ~ -I.-m. * .-. '_,-., ... T', xr color resolution. In this example, it does not provide gamma correction, and the sub-pixel rendering function produces a 1 1 _ bit result. VGA memory 613 stores image data in an 8-bit format. The sub-pixel rendering square receives image data from the VGA memory 613 and performs a sub-pixel rendering function on the image (as described above) to provide results in a 11-bit format. In one example, the sub-pixel rendering block 614 may represent the sub-presentation processing module 504 of FIGS. 52A, 53A, and 54A. The sub-pixel rendering block 614 may transmit extra bits from the division operation during the sub-pixel rendering to make a wide DAC or LVDS output 615 to process, if set to process 11-bit data. The input data can be maintained in the 8-bit data format, which allows existing images, software, and drivers to remain unchanged for the benefit of increased color quality. The display 616 may be configured to receive image data in an 11-bit format to provide additional color information to the image data in an 8-bit format instead. Please refer to FIG. 63, which is an exemplary block diagram of a system that provides sub-pixel rendering using a wide gamma table or look-up table (LUT) with multiple inputs (11 bits) and fewer outputs ( 8-bit). VGA memory 617 stores image data in 8-bit format. The sub-pixel rendering block 618 receives the image data from the VGA memory 617 and performs a sub-pixel rendering function on the image data (as described above), wherein the gamma value from the wide gamma table 619 is used to apply the gamma correction. The gamma table 619 may have an 11-bit input and an 8-bit output. In one example, sub-pixel processing block 618 may be the same as block 614 of FIG. 62. Block 618 may use the bit-wide gamma LUT from the gamma table 619 to perform the sub-pixel rendering function described above to apply gamma adjustment. The additional -93-f I238MM: bits can be stored in the wide gamma LUT, which can have additional registrations above 256. The gamma LUT of the block 619 may have an 8-bit output of the CRTDAC or LVDS LCD block 620 to display display data in an 8-bit format on the display 621. By using this wide gamma LUT, the output value can be avoided. Please refer to FIG. 64, which shows an exemplary square diagram of a system that provides sub-pixel rendering using a wide input wide output gamma table or look-up table (LUT). VGA memory stores image data in 8-bit format . The sub-pixel rendering block 624 receives the image from the VGA memory 623 and performs a sub-pixel rendering function (as described above) on the image data, wherein the gamma value from the gamma table 626 is used to apply the gamma correction. The gamma table 626 may have an 11-bit input and a 14-bit output. In one example, the sub-pixel processing square 624 may be the same as block 618 of FIG. 63. Block 624 may use a 1-bit-wide gamma LUT from a gamma table 619 with a 14-bit output to perform the sub-pixel rendering function described above to apply gamma adjustment. At block 627, a one-width DAC or LVDS can receive the output in 14-bit format to output data on display 628, which can be set to receive data in 14-bit format. The wide gamma LUT of block 626 may have more output bits (i.e., one less in and more out, or FIMO LUT) than the original input data. In this example, by using this LUT, you can have more output colors than the source image provider. Please refer to FIG. 65, which shows an exemplary block diagram of a system that uses the same kind of gamma table as in FIG. 64, and a space-time color mixing block to provide sub-pixel rendering. VGA memory 629 stores image data in 8-bit format. The sub-pixel rendering block 630 receives the image from the VGA memory 629, -94-

A sub-pixel rendering function (as described above) is performed on the image data, in which the gamma correction from the gamma table 631 is applied. The gamma table 631 may have an 11-bit input and a 14-bit output. In one example, the sub-pixel processing block 640 may be the same as block 624 of FIG. 64. The square rose 630 can use the 11-bit wide LUT from the gamma table 631 with a 14-bit output to perform the above-mentioned sub-pixel rendering function to apply the gamma adjustment. The space-time color mixing block 632 receives 14-bit data and outputs 8-bit data to an 8-bit CDLVDS of an LCD display 634. As a result, existing LVDS drivers and LCD displays can be used without the need for expensive redesign of LVDS drivers, timing controllers, or LCD panels, which can provide better benefits than the exemplary system of Figure 63. Please refer to FIG. 66, which shows an exemplary block diagram of a system that uses a pre-compensated lookup table (LUT) to compensate for the non-linear gamma response of the output display to provide sub-pixel rendering to improve image quality. VGA memory 635 stores image data in 8-bit format. The pre-compensation lookup table block 636 can store the values in an inverted gamma correction table, which can compensate the gamma response curve displayed on the output of the image data in the VGA memory 635. The gamma value in the correction table can provide a 26-bit value to provide a necessary gamma correction value equal to, for example, 3.3. The sub-pixel rendering processing block 637 may use the gamma values in Table 636 to provide the pre-compensation described above. In this way, the exemplary system uses sub-pixel rendering in the same "color space" as the output display, rather than in the color space of the input image stored in the VGA memory 635. The sub-pixel processing block 637 may transmit the processed data to a gamma output generation block 638 to perform the above ---------------------------- ------------- Post-gamma correction of 123801 to. This square can accept 29-bit input data and output 14-bit data. The space-time color mixing box 639 can be converted from the gamma output of an 8-bit LVDS square rose 640 to generate the data received by the square rose 638 to output an image on the display 641. Figures 6 7 to 69 show exemplary embodiments of a function evaluator that performs mathematical calculations, such as generating a gamma output value at high speed. The following specific embodiments can generate a small amount of gamma output from a large number of input values. This calculation can use monotonically increasing functions such as square root, power curve, and trigonometric functions. This is particularly useful for generating gamma correction curves. The following embodiments can use a binary search operation, which has multiple stages using a small parameter table. For example, each stage of the binary search can result in one more bit of precision in the output value. In this way, 8 stages can be used in an 8-bit output gamma correction function. The number of stages can be determined according to the data format size of the gamma correction function. Each stage can be done in parallel for a different input value, whereby the following specific embodiments can use a sequence of pipelines to receive a new input value on each clock cycle. The stages of this function evaluator are shown in Figures 69 and 70. Figure 6-7 shows the internal components of a phase of the function evaluator. Each stage can have a similar structure. Please refer to FIG. 67. In this stage, three input values are received, which include an 8-bit input value, a 4-bit approximate value, and a clock signal. The 8-bit input value is fed to a comparator 656 and an input latch 652 > The 4-bit approximate value is fed to the approximate latch 658. The clock signal is coupled to comparator 21, input latch 652, a single-bit result latch 660, approximate latch 658, and reference -96- f ia ^ (M κ-_ — 立-. Yue Yue丨 Number memory 654. The parameter memory can contain an rAM or ROM, and store parameter values, such as the parameter values shown in Figure 68. These parameter values correspond to the function of sqrt (x) and are used as examples. The 8-bit input and 4-bit approximation are exemplary and can have other bit formats. For example, the input can be a 24-bit value, and the approximation can be an 8-bit value. Now The operation of this phase will be explained. At the rising edge of the clock signal, the approximation is used to find one of the parameter values in a parameter memory 654. The output from the parameter memory 654 is provided by a comparator 656 Compare with the 8-bit input value to generate the result bit fed into the result latch 660. In an example, the result bit is 1, if the input value is greater than or equal to the parameter value, and if The input value is less than the parameter value, which is 0. On the trailing edge of the clock signal, the The input value, result bit, and approximate value are latched to latches 652, 660, and 658, respectively, to maintain the value for the next stage. Refer to Figure 68, a parameter table that can be stored in parameter memory 654 to calculate the A function of the square root of an 8-bit value. This function can be used for any kind of gamma correction function, and the resulting value can be rounded. Figure 69 shows one of the four stages (stage 1 to stage 4). Implement a function evaluator. Each of these stages can include the same components of FIG. 67 and can have the same structure. For example, each stage can include a parameter memory that stores the table of FIG. 8 so that the stage manages Will implement a square root function. The operation of the function evaluator will now be explained. An 8-bit input value is provided to stage 1, because the value flows from stage 1 to stage 4, and finally to the output, which has a continuous time Pulse cycle. That is, for each clock, the square root of each 8-bit value is calculated and the output is provided after phase 4. In one example, phase 1 can initialize the approximate value to 1,000 (Binary), and the result bit of stage 1 outputs the correct value of the most significant bit (MSB), which feeds into the MSB of stage 2. At this time, the approximate latch of each stage transmits this MSB until It reaches the output. In a similar manner, phase 2 sets the second MSB to 1 on the input and generates a second MSB of the output. Phase 3 sets the third MSB to 1 and generates the first MSB of the output. Three MSBs. Stage 4 sets the last approximate bit to 1 and produces the last bit of the resulting output. In the example of Figure 69, stages 1-4 can be equal to simplify manufacturing. It can be done for each stage Implement other changes. For example, to avoid the inefficiency of using internal components, the parameter memory can be replaced by a single latch containing the intermediate value, because all input approximate bits are set to a known fixed value. Phase 2 has only one unknown bit in the input approximate value, so only two latches are required to be included in the half value between the middle and end values from the parameter RAM. The third stage 3 only looks at 4 values, and the fourth stage 4 only looks at 8 values. This means that the parameter RAM does not need to have four equal copies. Instead, if each stage is designed to have the minimum amount of parametric RAM it needs, the required amount of storage is only equal to one copy of the parametric RAM. Unfortunately, each stage requires a separate RAM, which has its own address decoding, because each stage will view a parameter value of a different input value on each clock cycle. (This is very simple for the first stage, it has only one numerical value, find it.) 0 -98- ended 123 # vocational year 9a

El Figure 70 shows how the stages of Figure 69 can be optimized for a function estimator. For example, it may omit the unnecessary output latch of stage 1, and the approximate latch may be omitted by stage 1. Therefore, a single latch 672 coupled to the comparator 665 and the latch 669 can be used for Phase 1. In phase 2, the approximate latches 674 need only have one bit, while in phase 3, the approximate latches 676 and 677 need only have two bits. This can continue until one of the bits in phase 4 is implemented, thereby having latches 680, 681, and 682. In some cases, it does not require this least significant bit. Other changes to this configuration include removing the input value 683 latch of stage 4 because it is not connected to another stage. FIG. 71 shows a flowchart of an exemplary software implementation 700 of the method described above. A computer system, such as computer system 750 of Figure 72, can be used to execute this software implementation. Please refer to FIG. 70. Initially, a window application 702 generates an image to be displayed. A window drawing device interface (GDI) 704 sends the image data (Vin) for output to a display. A pixel rendering and gamma correction application 708 intercepts the input image data Vin, which is guided to a window device data interface (DDI) 706. This application 708 can execute the instructions in the appendix below. The Windows DDI 706 stores the received image data to a frame buffer memory 716 through a VGA controller 714, and the VGA controller 714 outputs the stored image data to a display 718 through a DVI cable. The application 708 intercepts the drawing call from the window GDI 704, and guides the system to present the conventional image data to a system memory buffer 71, rather than to the frame buffer 716 of the graphics card. Application 708 converts this custom image -99-! I23 «0i data to sub-pixel rendered data. The data presented by the sub-pixel is written into another system memory buffer 712, where the graphics card is formatted and transferred to the display via the DVI cable. The application 708 can pre-arrange colors in the PenTileTM sub-pixel order. Windows DDI 706 receives the data presented by the sub-pixel from the system memory buffer 712 and works on the received data as if the data were from Windows GDI 704. FIG. 72 shows an internal block diagram of an exemplary computer system 750 for implementing the methods of FIGS. 46, 49, and 51, and / or the software implementation 700 of FIG. 71. The computer system 750 includes several components, all of which are interconnected via a system bus 760. An example of a system bus 760 is a two-way system bus with 32 data and address lines for accessing a memory 765 and for transferring data between the components. In addition, multiplexed data / address lines can be used instead of separate data and address lines. Examples of the memory 765 include a random access memory (RAM), a read-only memory (ROM), a video memory, a flash memory, or other suitable memory devices. Additional memory devices may be included in the computer system 750, such as, for example, fixed and removable media (including magnetic, optical, or magnetic optical storage media). The computer system 750 can be connected to other computing systems via a network interface 785. Examples of the network interface 785 include Ethernet, or a telephone connection. The computer system 200 may also receive input through an input / output (I / O) device 770. Examples of I / O device 770 include a keyboard, pointing device, or other suitable input device. The I / O device 770 may also represent an external storage device or a computing system or subsystem. The computer system 750 includes a central processing unit (CPU) 755, examples of which include -100- .- ~ 年 施] 1τδ E || A microprocessor of the pentium® series manufactured by Intel® Corporation. However, computer system 750 may use any other suitable microprocessor, micro-, mini-, or processor for a mainframe computer. The CPU 755 is also configured to perform the above method according to a program stored in the memory 765, which uses a table of gamma and / or coefficients also stored in the memory 765. The memory 765 may store instructions or code for executing the program, which causes the computer system 750 to execute the method of the software implementation 700 of Figs. 46, 49, and 51, and Fig. 71. Furthermore, the computer system 750 includes a display interface 780 that outputs sub-pixel presentation data to a display, which is generated by the methods of FIGS. 46, 49, and 51. Therefore, the method and system of sub-pixel presentation with gamma adjustment have been described here. Certain specific embodiments of the gamma adjustment described herein allow the illumination of the sub-pixel configuration to match the nonlinear gamma response of the illumination channel of the human eye, and the color difference can match the linearity of the color difference channel of the human eye reaction. Gamma correction in certain embodiments allows the algorithm to independently operate the actual gamma of a display device. The sub-pixel rendering disclosed herein can optimize the gamma value of a display device to improve the response time, point conversion balance, and contrast compared to some specific embodiments with gamma adjustment, because the sub-pixel rendering The algorithm's gamma correction and compensation can provide the desired gamma value through sub-pixel rendering. Certain embodiments of these techniques can be added to any given gamma transfer curve. In the foregoing specification, the invention has been described with reference to specific exemplary embodiments thereof. However, it can be understood that various modifications and changes can be made without departing from the broader spirit and scope of the invention as set forth in the scope of the appended patent application. Accordingly, the description and drawings are to be regarded as illustrative purposes only, and not as restrictive. Appendix The following is an example C code that can be used to implement the method disclosed here. However, the following code can be translated into any other suitable executable programming language to implement the techniques disclosed herein. In addition, the following code is protected by copyright, in which the copyright owner retains all copyrights contained herein. // **** 氺 氺 氺 氺 ** 氺 氺 氺 氺 氺 氺 氺 ** 氺 氺 氺 氺 氺 氺 氺 氺 * 氺 氺 ** 氺 氺 氺 氺 氺 氺 ****** // Sub-pixel rendering program staticlongBlueSum = 0; // from the sum of red and green, stored as blue unsigned char CalcSubP (BITMAPINFOHEADER * ib, int x, int y, int ox, int oy)

long sum = 0, cent, inner = 0, edge, term; long wcent, bwcent, wedge; // Omega-corrected pixel values • · _ mt i, j, // color component from sub-pixel address int color Two ((ox & l) A (oy & l))? GREEN: RED; unsigned short * pre = color = RED? Precomp: precomp + 256; unsigned short * wgm = color = RED? Wtable: wtable + 256; / / Index-to-filter-102-

unsigned char * myf = filts + (((ox% S) + (oy% S) * S)) * RGXsiz e * RGYsize; unsigned long ccoef; // store the central coefficient // return the omega code and the blue sum // Propose the central input pixel and keep it for a while cent = PTX (x + l, y + l, color);

wcent = wgm [cent] »8; // Use only 8 bits before omega bwcent = wtable [512 + PIX (x + l, y + l, BLUE)] > >8; // Find the blue Color Omega central value inner = 0; // Calculate all terms for (j = 0; j < RGY size; j ++) {for (i = 0; i < RGX size; i ++) {

switch (Xi «4X〇 // For all specific cases, the coordinates are mixed together case 0x00: // corner pixel term case 0x20: case 0x02: case 0x22: edge = PIX (x + i, y + j, color); // The input pixel is always -103-MMm κ 8-bit wedge = wgm [edge] »8; // After looking for Omega, // it is 16, return 8 and now it is 8 bits // The average of the corner pixel and the central pixel is still 9 bits term = :( wedge + wcent) / 2; // view the average in the pre-calculation table to make it 16 bits term = pre [term] ; // Multiply by the filter coefficient, the result must be 24 bits. // However, we use the first database of the multiplier to get a 16-bit result // divide internally by 256 (or do not calculate the Lower 8 bits) term = (term * (unsignedlong) (* myf ++)) »8; // then in the second database of the multiplier, we multiply the gamma correction term by // the unsigned Gamma correction input value. This can effectively add 1 // to the index of the gamma term. Term = (term * edge) »8; -104- to be 11MU t // because the term is multiplied // Coefficients add up to 1, this sum will always match 1 6 bits. Sum + = term; break; case 0x10: // orthogonal edge pixel term case 0x01: case 0x21: case 0x12: edge = PIX (x + i, y + j, BLUE); // get the same Look at the blue pixel color omega blue table wedge ^ wtablel ^ lS + edge]> ^; // Implement it blue // table term ^ wedge + bwcent) / ^; // averaging, its central term = precomp [512 + term]; // Then in GinvWinv // table

Blue Sum + = terns // total and save to blue later

Color calculation with edge = PIX (x + i, y + j, color); // The input pixel is always -105-

Is 8-bit wedge = wgm [edge] >8; // After searching in Omega, it is 1 6 // Return to 8, now it is 8-bit // an edge pixel and the center pixel The average is still 9-bit term = (wedge + wcent) / 2; // look up the average in the pre-calculation table to make it 16-bit times and add up to 4 times term = pre [term]; / / The marginal terms are summed to calculate the central term // this will have to be an 18-bit number to keep the inner + two term; // add the margin to calculate the central term later // multiply by The filter coefficient, which must make the result 24 bits // But we use the first database of the multiplier to get a 16 bit result -106- music // internally divided by 256 (or do not calculate the Lower 8 bits) term = (term * (unsigned long) (* myf ++)) »8; // then in the second database of the multiplier, we multiply the gamma correction term by // the Ungamma-corrected input value. This can effectively add 1 // to the index of the gamma term term = :( term * edge) »8; // Because the term is multiplied by // the coefficients add up to 1, this sum can always match at 1 6-bit sum + ^ term; break; caseOxll: // central pixel ccoef = (long) (* myf ++); // to take only the central coefficient break later; // the sum of the 4 inner terms Take 4 to inner »= 2; get the 16-bit average inner = (inner * ccoef)» 8; // then multiply it by the central coefficient -107- IMMU t year month day sum + = (inner * cent) »8 ; // Finally multiply by the central value to complete the external sum of the filter if (sharpen) {if (color = RED) // switch to the cross color color = GREEN; · else color = RED; // now sharpness It is always done by non-gamma-corrected values. // So, apart from not considering accuracy, we can roughly keep // because we changed the number from 8 bits to 1 bit. // Divided by the sharpness coefficient (1 / 8 and 1/3 2) // Central * 256 to get 16 bits, then multiply by 1/8 sum + = PIX (x + l, y + l, color) * 32; // The corner term is * 256, Then / 32, we get * 8 sum = PIX (x, y + 2, color) * 8; sum = PIX (x + 2, y + 25col or) * 8; sum = PIX (x + 2, ycolor) * 8; sum = max (x, y, color) * 8; sum = max (0? sum); // Sharpness can cause negative numbers -108-

123 faces I sum = min (sum, ginmask); // or more than 8 bits // adjust to output table sum = (sum * goutdiv) / (ginmask + l); size if (color = RED) sum = gwmt [sam]; // Use this 11-bit number to find the output gamma

else sum = gamat [sum + goutdiv]; // red uses _ may be different from the form return ((unsignedchar) (sum)); // returns the sub-pixel // the program that calculates the blue value unsigned char Blue Filter (BITMAPINFOHEADER * ib? int x? int y? int ox, int oy) {

long sum = 0; long teml, tem2; // diagnostic variable unsignedchar * myf; int i, j; // return counter myf = bfilts + (((ox% S) + (oy% S) * S)) * BlueXsize * BlueYsize; BlueSum > > = 3; // Take the average of all those blue sums // This makes the blue sum again a 16-bit number -109- IX3 # Q4fe-)) V-k for (j = 0 ; J <BlueYsize; j ++) {for (i = 0; i <BlueXsize; i ++) {teml two PIX (x + i, y + j, BLUE); // Propose the blue pixel tem2 = (teml ** myf ++ ) »8; // 8 * 16 = 16 times the coefficient teml = (tem2 * BlueSum) > >8; // Same as recursive sum value

}}

BlueSum = 0; // Initialize it for the next blue sum = (sum * goutdiv) / (ginmask + l); sum = gamat [sum + goutdiv * 2]; return ((unsigned char) (sum)) ; // Return the blue superpixel value.

-110- 1238011 CRT Cathode Ray Tube LCD Liquid Crystal Display MTF Modulation Transfer Function FPGA Field Program Gate Array VGA Video Graphics Card ASIC Application Specific Integrated Circuit LVDS Low Voltage Differential Signaling EDID Extended Display Identification Information DVI Digital Visual interface ROM Read-only memory RAM Random access memory Hip MSB Most significant bit GDI Drawing device interface LUT lookup table 20 Configuration 21 Three-color pixel element 22 Blue emitter 23 Blue plane sampling point 24 Red emitter 26 Green emission -Ill-

, 1231¾¾ 丨-~ — & ij 28 emitter 30, 31 configuration 32 blue emitter 33 reconstruction point 34 red emitter 35 reconstruction point 36 green emitter 37 reconstruction point 38 configuration 39 tri-color pixel element 40 configuration 42 Blue plane 44 Sampling area 46 Effective blue sampling point 48 Red plane 50 Edge area 51 Effective red sampling point 52 Central area 53 Output sample area 54, 55, 56 Effective sampling area 57 Effective green sampling point

-112- jl238 «ii. Up. Up, one: / ι 58, 59 60 70 72 74 76 78 80 82, 84, 86, 88 92, 94, 96, 98 100, 102 104, 106, 108 120 122 123 、 124 125 127 202 204 206 Effective sampling area Green flat pixel data format Effective sample area Sample point array configuration Configuration array Sample area Sample area configuration Array input sample area Repeated unit square sub-pixel solid line rendering boundary

-113-208 fmmi 210 area 212 raw pixel sample area 216, 218, 220, 222 sub pixels 224, 226, 228 sub pixels 230 symmetry lines 234, 236, 238 sub pixels 240, 242, 244 sub pixels 246, 250, 254 Rendering area 248, 252, 256 Sample area 260, 262 > 268 Sample area 264, 266 Rendering area 270 Pixel area 272 Original pixel boundary 274 Blue output pixel boundary 276 Square sampling area 280 Sample area 500 times Pixel processing unit 501 Personal operation 504 sub-pixel rendering module 506 timing controller 508 DVI input

-114 MMm I El 509 interface 510 Display identification information unit 512 Input latch and automatic detection block 514 Timing buffer and control box 516 Pre-adjusted gamma processing box 518 Line buffer box 519 3x3 data sampling square box 520 Multiplier + Adder block 521 Delay logic block 522 Post-gamma processing block 524 Output latch 526 LVDS output 528 Output synchronization generation stage 530 Coefficient processing block 531 Coefficient table 532 Input latch 534 Mixed color block 536 Reference voltage and video transmission voltage block 537 Data Bus 538 DC / DC converter 539A line driver control

-115-

539B Column driver control 540 Local average processing block 542 Pre-gamma processing block 544 Omega processing block 545 Pre-gamma processing block 547 Bucket offset 554, 556, 558 Line buffers 560, 562 Sum buffers 564, 566, 568 Addition 570, 572 line buffer 573, 574 barrel offset 575, 576, 578, 581 adder 580, 582 line buffer 583, 584, 585 sum buffer 590 barrel offset 591, 592, 593, 594 addition 613 VGA memory 614 sub-pixel rendering box 615 wide DAC or LVDS output 616 display 617, 623, 629, 635 VGA memory -116-

618, 624, 630, 637 sub-pixel rendering squares 619, 626, 631 620, 627 621, 628, 641 632, 639 633 634 636 638 640 652 654 656 658 660 Gamma table CRTDAC or LVDS square display

Space-time color mixing square cube LCD LVDS cube LCD display pre-compensated lookup table cube gamma output generation cube LVDS square cube input latch parameter memory comparator approximate latch single bit result latch 665, 666, 667, 668 comparator 669, 670, 671, 672 latches 673, 674, 675, 676 latches 677, 678, 680, 681 latches 682 latches Windows applications -117- 702 Data interface 708 sub-pixel rendering and gamma correction application 710, 712 system memory buffer 714 VGA controller 716 frame buffer memory 718 display 750 computer system 755 central processing unit 760 system bus 765 memory 770 input / output (I / O) device 780 display interface 785 network interface

-118-

Claims (1)

fmm \ 、 year 9a # — \! i Scope of patent application 1. A method for processing data of a display containing pixels, each pixel having a color sub-pixel, the method includes: receiving pixel data; Applying a gamma adjustment to the conversion from the pixel data to the sub-pixel rendering data, the conversion produces the sub-pixel rendering data of a primary pixel configuration, which includes alternating red colors on at least one of a horizontal axis and a vertical axis And green sub-pixels; and output the data presented by that sub-pixel. 2. The method according to item 1 of the patent application range, wherein the applied gamma adjustment can provide linearity in the color balance of the sub-pixel presentation data. 3. The method according to item 2 of the patent application, wherein the applied gamma adjustment further provides a non-linear calculation of the illuminance of the sub-pixel presentation data. 4. The method according to item 1 of the patent application scope, wherein the applied gamma adjustment includes: performing a gamma correction on the pixel data to generate the gamma corrected data; converting the gamma corrected data to the sub-pixel rendered data. 5. The method according to item 4 of the patent application, wherein the gamma correction is performed as a function of g-1 (x) = xY. 6. The method according to item 1 of the patent application range, wherein a gamma value of the gamma adjustment is maintained at a selected level for all spatial frequencies, and the selected level corresponds to a certain spatial frequency. A desired contrast ratio. 7. The method of claim 1 in the scope of the patent application, wherein the applied gamma is adjusted. I2-3Ϊμ1: ·, ~, t:]! "! J includes calculating a local average based on the pixel data. 8. The method according to item 7 of the patent application scope, wherein the applied gamma adjustment further comprises performing a gamma correction on the local average to generate a gamma corrected local average, and converting the gamma multiplied by the pixel data The modified local average of the pixels becomes the data presented by the sub-pixel. 9. The method of claim 8 in which the gamma correction is performed as a function of g · 1g. 10. The method according to item 1 of the patent application range, wherein a gamma value of the gamma adjustment is selected to increase as the spatial frequency increases. ^ 11. The method according to item 1 of the patent application range, wherein the applied gamma adjustment includes: performing omega correction on the pixel data to generate omega corrected data; and calculating an omega corrected part based on the omega corrected data. average. 12. The method according to item 11 of the scope of patent application, wherein the omega correction is a function of w (x) = x1 / v ^. 13. The method according to item 11 of the scope of patent application, wherein the applied gamma adjustment further comprises: calling on the local average of the omega correction to perform a gamma correction to produce a local local gamma average with the omega correction; and a conversion multiplication The local average of the gamma with the omega correction based on the pixel data becomes the data presented by the sub-pixel. 14. The method according to item 13 of the scope of patent application, wherein the gamma correction is performed as a function of gdw-WMxW) 7.1. 15.—A kind of sample data for converting an image to a MMm 1 ^, + nm that can display sub-pixel presentation using a three-color sub-pixel element. The method includes: receiving sampling data including a plurality of first data values, each first data value representing each data point of each color in the image; for each of the first data in the sampling data Value to perform gamma correction to generate gamma-corrected data; and based on the gamma-corrected data to calculate data represented by sub-pixels containing a plurality of second data values, each second data value corresponding to the display Each sub-pixel element of each color. 16. The method according to item 15 of the scope of patent application, wherein calculating the data presented by the sub-pixel includes calculating the pixel configuration on the display, which includes alternating red on at least one of a horizontal axis and a vertical axis And green sub-pixel elements. 17. The method according to item 15 of the scope of patent application, wherein the data for calculating the sub-pixel presentation includes: referring to a filter core including a plurality of coefficient terms; and modifying the gamma of each of the first data values. The data is multiplied by each corresponding coefficient term in the filter core; and each multiplied term is added to generate each of the second data values. 18. The method of claim 15 in the scope of patent application, wherein the gamma correction compensates the human eye's response function to illuminance. 19. The method according to item 15 of the scope of patent application, further comprising performing post-gamma correction on the data presented by the sub-pixel, the post-gamma correction compensating a gamma function provided by the display; and [U-JL Output the post-gamma-corrected sub-pixel rendering data to the display. 20. The method according to item 15 of the patent application, wherein the gamma correction is performed as g_1 (X) = XY〇 21. The method according to item 15 in the patent application, further comprising determining each color The sample area specified in the sampled data of each data point; and determining the resampling area of each sub-pixel element corresponding to each color, wherein calculating the data presented by the sub-pixel includes using a filter core A plurality of coefficient terms, each coefficient term representing an overlap percentage of a given one of the resampling regions over each of the specified sample regions having the given one of the resampling regions. 22. —A method for converting sampling data of an image to a display capable of displaying sub-pixel rendered data using a three-color sub-pixel element, the method comprising: receiving sampling data including a plurality of first data values, each first The data value represents each data point of each color in the image; for each of the first data values in the sampled data, a gamma-corrected data value is generated; and based on the gamma-corrected data value and the first A product of a data value is used to calculate data presented by a sub-pixel including a plurality of second data values, and each second data value corresponds to each sub-pixel element of each color on the display. 23. The method according to item 22 of the scope of patent application, wherein calculating the data presented by the sub-pixel includes calculating a pixel configuration on the display, which includes the alternation of at least one of 1: JL on a horizontal axis and a vertical axis Red and green sub-pixel elements. 24. The method of claim 22, wherein the gamma-adjusted data value includes: calculating a local average of each of the first data values based on the sampled data; and performing gamma on the local average. Adjustment. 25. The method of claim 24, wherein the gamma adjustment is performed as a function of g-Hxhf1. Lu 26. The method of claim 24 in which the first data value includes an edge term and a central term, and the calculation of the local average includes: using the central term of each edge term to calculate A first average; calculating a second average of the central term based on the first average. 27. The method of claim 24, wherein the first data value includes an edge term and a central term, and calculating the local average includes using the central term of each edge term to calculate an average The generated gamma-adjusted data values further include: · Using the average of the edge-terms of the gamma-adjustment to generate a local average of a gamma-adjustment of the central term. 28. The method according to item 22 of the scope of patent application, wherein the calculation of the data presented by the sub-pixel includes: referring to a filter core that includes a plurality of coefficient terms; adjusting the gamma of each of the first data values Multiplying the data by each corresponding coefficient term in the filter core and each of the first-b, H'j Ell data values; and adding each multiplied term to generate each of the second Data value. 29. The method of claim 22, wherein the product of the gamma-adjusted data value and the first data value can compensate a response function of the contrast of the human eye. 30. The method according to item 22 of the scope of patent application, further comprising: performing post-gamma correction on the data presented by the sub-pixel, the post-gamma correction compensating a gamma function provided by the display; and outputting the post-gamma correction Gamma-corrected sub-pixels present data to the display. 31. The method according to item 22 of the scope of patent application, further comprising: determining a sample area specified in the sampled data of each color point of each color; and determining a value of each sub-pixel element corresponding to each color Re-sampling the area, and wherein calculating the data presented by the sub-pixel includes using a plurality of coefficient terms in a filter core, each coefficient term representing a given one of the re-sampling areas overlapping each having the re-sampling The percentage of overlap of the given sample area of the given one in the sample area. 32. The method of claim 31 in the scope of patent application, wherein the first data value includes a corner term, an edge term except for the corner term, and a central term, and the data for calculating the sub-pixel presentation includes : Less use of the corresponding one of the first data values relative to the one represented by the overlap percentage of each of the corner terms; and relative to the one represented by the overlap percentage of the central term I | Ϊ2 ^ (Μ M…-1 #: month-: '. Day uses the corresponding one of the first data value. 33. The method of claim 22 in the scope of patent application, wherein the first data value includes a corner Term, in addition to the edge term of the corner term, and a central term, and the data calculated for the calculation of the sub-pixel includes: reducing the effect of the corner term; and strengthening the effect of the central item to balance the weakening 34. The method of claim 33, wherein the product uses the first data value in a first color of the edge term and the central term, and the weakening and strengthening uses the Corner item and the central item ® the first data value in a second color. 35. The method of claim 22 in the patent application range, wherein the data value for generating the gamma adjustment includes: calculating each of the first data based on the sampled data A local average of one omega adjustment of the data values; and a gamma adjustment of the local average of the omega adjustment. 36. For the method of item 35 of the patent application scope, wherein calculating the local average of the omega adjustment includes: _ For the Omega adjustment is performed for each of the first data values in the sampled data; and a local average of each of the first data values is determined based on the sampled data for the omega adjustment. 37. If the method of item 36 of the patent application, The omega adjustment is an approximation of the response function of the contrast of a human eye. 38. The method according to item 35 of the scope of patent application, wherein the gamma adjustment is performed by f 12 vis 11 11] as g Iw-1 (X) = (W "(X)) Y_1, where W-1 (X) is the inverse function of w (x) = x1 / w. 39. The method of the 35th item of the patent application, wherein the first data value includes an edge term And one central item, The calculating the local average of the omega adjustment includes: using a central term of each of the edge terms to calculate a first omega adjusted average; and calculating a second omega of the central term based on the first omega adjusted average. Adjusted average. 40. The method of claim 35, wherein the first data value includes a ® edge term and a central term, and calculating the local average of the omega adjustment includes using each edge term The central term to calculate an average of an omega adjustment, the generated gamma-adjusted data value further includes: using the gamma-adjusted average of the edge term to generate a local average of a gamma-adjusted central term . 41. The method of claim 22, wherein the data value for generating the gamma adjustment includes: calling for performing the omega adjustment of the first data value; and performing an inverse omega adjustment to generate the data value for the gamma adjustment such that The omega adjustment and the inverted omega adjustment can more affect the data values of the gamma adjustment when the pan frequency of the image becomes more local. 42 · —A system for processing data of a display containing pixels, each pixel having a color sub-pixel, the system includes: Jt h :. .. i — [.… 月 Η A receiving module for receiving pixels Data; and a processing module for performing a conversion from the pixel data to sub-pixel rendering data, and applying a gamma adjustment to the conversion, the conversion generating the sub-pixel rendering data of a pixel configuration, including Alternating red and green sub-pixels on at least one of the horizontal axis and a vertical axis. 43. The system according to item 42 of the patent application scope, wherein the processing module is to provide linearity in the color balance of the data presented by the sub-pixel. 44. The system according to item 43 of the patent application scope, wherein the processing module is to provide a non-linear calculation of the illuminance of the data presented by the sub-pixel. 45. If the system of claim 42 is applied for, the processing module is to perform gamma correction on the pixel data to generate gamma corrected data, and convert the gamma corrected data to the sub-pixel presentation. data. 46. The system according to item 45 of the patent application, wherein the processing module is to perform a gamma correction using a function such as g ^ (x:) = xY. 47. If the system of claim 42 is applied for, the processing module maintains a gamma value adjusted by the gamma for all spatial frequencies at a selected level, and the selected level corresponds to The desired contrast ratio at a certain spatial frequency. 48. The system according to item 42 of the patent application, wherein the processing module calculates a local average based on the pixel data. 49. The system of claim 48, wherein the processing module performs a gamma correction on the local average to generate a gamma corrected local average, and the processing module converts and multiplies the pixel data. The local average of the gamma correction becomes the data presented by the sub-pixel. 50. The system of claim 49, wherein the processing module is to perform a gamma correction using a function such as. 51. The system according to item 42 of the patent application, wherein a gamma value of the gamma adjustment is selected to increase as the spatial frequency increases. 52. If the system of claim 42 is applied for, the processing module is to perform omega correction on the pixel data to generate omega correction data, and calculate a local average of an omega correction based on the omega correction data. 53. The system of claim 52, wherein the processing module uses the function w (x) = x1 / w to perform Omega correction. 54. The system of claim 52, wherein the processing module performs gamma correction on the local average of the omega correction to generate a gamma local average of the omega correction and converts it to the pixel data. The local average of the gamma with Omega correction becomes the data for sub-pixel presentation. 55. The system as claimed in claim 54 in which the processing module is to perform a gamma correction using a function such as gUw ^ xMxy1. _ 56. —A computing system comprising: a display with a plurality of pixels, wherein at least one of the pixels includes a sub-pixel configuration of red and green sub-pixels alternated in at least one of a horizontal axis and a vertical axis; And a controller coupled to the display, the controller processes the pixel data and applies gamma adjustment to the conversion from the pixel to the sub-pixel data, the conversion generates the data presented by the sub-pixel configured by the sub-pixel, and outputs The data presented by the sub-pixel is displayed on the display. 57. A controller for a display, comprising: a receiving unit for receiving pixel data; and a processing unit for applying gamma adjustment to conversion from the pixel data to sub-pixel presentation data, the The data generated by the sub-pixel configured by the sub-pixel configuration is converted, and the data presented by the sub-pixel is output on the display. 58. —A computer-readable medium of record storage instructions that, when executed by a computing system, enables the computing system to execute a method for processing data on a display containing pixels, each pixel With color sub-pixels, the method includes: receiving pixel data; applying gamma adjustment to the conversion of the pixel data to sub-pixel rendering data, the conversion generating the sub-pixel rendering data in a pixel configuration, which includes a horizontal axis And alternating red and green sub-pixels on at least one of a vertical axis; and outputting data presented by the sub-pixels. · 59. — A computer-readable medium of record storage instructions, which when executed by an operating system, enables the operating system to execute a method for displaying sub-pixel presentation data using a three-color sub-pixel element A method for converting sampling data of an image by a display, the method comprising: receiving sampling data including a plurality of first data values, each first data value representing each data point of each color in the image; for the sampling Gamma repair-11-ll: I23fOI: W-'i 1-8, ι | is performed for each of the first data values in the data to generate gamma corrected data; and calculations are performed based on the gamma corrected data Sub-pixel presentation data including a plurality of second data values, each second data value corresponding to each sub-pixel element of each color on the display. 60. A computer-readable medium of record storage instructions that, when executed by an operating system, enables the operating system to execute a display that uses three-color sub-pixel elements to display sub-pixel presentation data A method for converting sampling data of an image, the method comprising: receiving sampling data including a plurality of first data values, each first data value representing each data point of each color in the image; for the sampling Each of the first data values in the data to generate a gamma-adjusted data value; and based on the product of the gamma-adjusted data value and the first data value, calculating Data, each second data value corresponds to each sub-pixel element of each color on the display. -12-