TW594627B - Conversion of a sub-pixel format data to another sub-pixel data format - Google Patents

Abstract

A method of converting a source pixel data of a first format for a display of a second format having a plurality of three-color pixel elements is disclosed. The method comprises determining implied sample areas for each data point of each color in the source pixel data of the first format. The resample areas for each emitter of each color in the display is also determined. A set of fractions for each resample area is formed. The denominators are a function of the resample area and the numerators are the function of an area of each of the implied sample areas that at least partially overlaps the resample areas. The data values for each implied sample area is multiplied by its respective fraction and all products are added together to obtain luminance values for each resample area.

Description

594627 A7 B7 V. Description of the Invention (Cross-Reference to Related Applications This application claims the benefits of the following provisional patent application at that time: US Provisional Patent Application Serial No. 60 / 29〇86, filed on May 9, 2011, named Conversion of RGB (RGB color model) pixel format data to pentile matrix sub-pixel data format '', US Provisional Patent Application Serial No. 60 / 290.087 filed on May 9, 2001, named, calculates filtering for different quantization modes Core value of the controller '; Quantitative sub-pixel rendering of the US Provisional Patent Application Serial No. 60 / 290.143 filed on May 9, 2001, entitled "PenTil ^ 6 array," and the United States filed on August 16, 2001 Provisional patent application serial number 60/3 13.054, the full name is "RGB stripe sub-image sketch edge detection", and it is fully incorporated herein by reference. BACKGROUND OF THE INVENTION This application describes the conversion of graphic data formats; In particular, this application describes' an improved color pixel arrangement for converting RGB (red, green, blue) graphics into a display. For a flat panel display, the single flat shadow of the color Matrix technology currently uses the RGB tri-color set or a single color in a vertical stripe, as shown in Figure 1, previous technology. The system uses the color mixing effect of von Bezold to separate the three colors And make each color have the same spatial frequency proportion (explained further here). However, these panels are extremely unsuitable for human vision. Graphic drawing techniques have been developed to improve the image quality of previous technology panels. Benzschawel et al. In U.S. Patent No. 5,34 !, 153 teaches me how to reduce a large image to a smaller panel. When implemented, ____- 4-This paper size applies the Chinese National Standard (CNS) A4 specification (210 X 297 mm) 594627 A7 B7 V. Description of the invention (2)

Benzschawel et al. Taught me how to improve the image quality by using a technique known as π sub-pixel rendering in this technique. Recently, HU1 et al. In U.S. Patent No. 6,1 88,385 taught me how to use the aforementioned sub-pixel rendering technique to reduce a virtual text image to improve the quality of the text, one character at a time. The foregoing prior art did not carefully consider the question of "how human vision works." The image created by the display element in this prior art is very unsuitable for human vision. This is important for sampling or generating images, and then storing the images The display model is the RGB pixel (or three primary color pixel elements), where the values of the red, green, and blue are based on an orthogonal spatial parsing grid, and the values are consistent. Use this One of the results of the image format is that it is extremely unsuitable for the new image (spaced, inconsistent, color emitter) of the real image, and human vision. This causes serious information excessive or waste in the image.

Martinez-Uriegas et al. And peterS et al. Teach us in U.S. Patent Nos. 5,398,066 and 5,541,653, respectively, "to convert the image from the 11 (} 3 pixel format to a format that is very similar to what Bayer taught in U.S. Patent No. 3,971,065. Similar formats, and the technique of storing such images, of which Bayer taught the format as a filter array for cameras and their imaging devices. The advantage of Martinez-Uriegas et al.'S format is that it samples spatially similar to human vision Frequency to capture and store the individual color component data. However, the primary disadvantage of the format of Martinez-Uriegas et al. Is that it does not match the actual color display panel. Based on this, Martinez-Uriegas et al. _______ Ί --- this Paper size applies Chinese National Standard (CNS) Λ4 specification (210X297 mm)

Binding Ao 594627 A7 ___ B7 V. Description of the invention () \ 3; People also taught me how to change the image back to the RGC pixel format. Another disadvantage of this format is that one of the color elements (in this case, work color) is sampled on an irregular basis. Because there are missing samples in the array, the accuracy of the newly created image will be reduced when displayed. The sensory organ nerve cell pattern called the three primary colors of the cone produces colored perception in the eye. These three types are very sensitive to different light wavelengths: long wave, medium wave, and short wave (respectively, red "," green ", and" blue ~ "°. The relative densities of these three wavelengths are significantly different from each other. Red The sensory organs are slightly more than the green sensory organs. Compared with the red or green sensory ears, the blue sensory organs are few. In addition to these color sensory organs, there are also called retinal rods, which are insensitive to relative wavelengths The sensory organs provide monochrome night vision. The human visual system uses several perceptual channels to process the conditions, illuminance, chroma and movement detected by the eyes. The movement is only the flicker valve for the programmer of the imaging system Important. The illuminance channel only accepts input from these red and green sensory organs. The illuminance channel is color-blind. It processes the information in a contrast-enhanced manner. The chrominance channel does not enhance edge contrast. Because the illuminance channel uses and Enhances each red and green sensory organ, so sometimes the resolution of the illuminance channel is higher than that of the chrominance channel. The blue sensory organ control The benefit of perception is very small. Therefore, only the most perceptive observer will notice the error (if the error really exists) caused when the resolution of beneficial color is reduced by one octave. For example, Xerox (full record) And nasa (National Aeronautics and Television Corporation) and Ames Research Center (R Martin. J. Gille · J.

The detectability of Lari met's "reduced total number of blue pixels in the projected display, ----------6-_ __________ This paper size applies the Chinese National Standard (CNS) A4 specification (210X297 public dream) '. ^ 594627

1 () 93 years S〗 〖D Abstract) experimentally demonstrated. Hong is affected by a process called "Assimilation" or the Von Bezold color mixing effect. This allows pixels (or sub-pixels or emitters) with different colors in a display to be treated as mixed colors. The mixing effect is In this field of view, a specific: distance: above occurs. Because the blue sensory organs are quite scarce, the mixing: the angle of occurrence of blue is greater than that of red or green. For blue, the $ off is about 0.25 In terms of red or green, the distance is about 〇12. The next display on a 12-inch visual distance is, 〇25. Yes, miles (1.270 magic). If so, the blue When the sub-pixel pitch is less than two-half of the mixed pitch (625 V), the colors will be mixed without degrading the graphics quality. &Amp; Under the simplest implementation, the sub-pixel drawing uses these sub-pixels as Pixels seen by this illuminance channel operate at nearly equal brightness. This allows these sub-pixels to serve as new points for sampling d-images compared to using sub-pixels such as 涊 combined as part of a "real" pixel. By using The elementary rendering increases the spatial sampling and thereby reduces the phase error. When the color of the image is to be ignored, each sub-pixel can be used as if it were a monochrome pixel, where each sub-pixel is equal. However, = Because color is almost always important (why does another person use a color display?), The color balance of a particular image at each position is important. If so, the general description of the subpixel must be ensured by The frequency information in the illuminance element of the image to be described does not use these color subpixels as alias files, which causes color errors to maintain color balance. Benzschavvel et al. And Hill et al. Each have US Patent No. 5,341, 153 ______- 7. This paper size applies to Chinese National Standard (CNS) A4 Gong (210 X 297 mm) 594627 A7

And the method adopted in 6, m, 385 is a common anti-offset technique. It seems that ‘Yu Ji’ applies an alternative decimal filter to the one-degree imaginary image and each element of its different color. This ensures that the illuminance message has no alias file in each color channel. & When the configuration of the pre-pixels is optimized for sub-pixel rendering, the sub-pixel material will increase the spatial addressability, reduce the phase error, and increase the modulation shift of the two axes. Function (MTF) High Spatial Frequency Unfolding Check the well-known RGB stripe display in Figure 1, where the pre-pixel rendering is only available on the horizontal axis. The human sub-pixel is not visible in the illuminance channel, so the blue sub-pixel cannot play a role in the sub-pixel rendering. Since only red and green pixels are useful in the sub-picture sketch margin, it effectively doubles this addressability on the horizontal axis. The vertical white and black lines must have two dominant sub-pixels in each column (ie, each black or white line has red and green). This is the same as the number in the image used for non-subpixel rendering. The sub-picture sketch edge is not stubborn about the MTF, where the MTF is the ability to display a specific stop line or a specific space at the same time m. The well-known RGB stripe sub-pixel configuration shown in Figure 1+ is not optimized for sub-pixel rendering. . The three primary color pixel element configurations of these previous techniques also show that they are extremely unsuitable for human vision and the generalized sub-pixel rendering technique. Similarly, the image formats and conversion methods of these first-class technologies are extremely unsuitable for human vision and actual color emitter configuration. Summary of the Invention By converting RGB pixel format data into penTne τ M matrix sub-pixel data

Binding This paper scale is applicable to the Chinese national standard x} 5. Description of the invention (6) The material format overcomes the shortcomings and disadvantages of previous techniques. A method of converting original pixel data in a first format and displaying the original pixel data in a second format having a plurality of three primary color pixel elements is disclosed. The method includes determining an implicit sampling area for each data point of each color of the original pixel data in the first format. It also determines the resampling area of each emitter for each of its colors in the display. A score is formed for each resampling area. The denominators are a function of the resampling area π; and the numerators are a function of each of the fe-containing sampling areas at least partially overlapping the resampling areas. Each resampling area's respective score is multiplied by the data value of each implicit sampling area, and all the products are added to obtain the illuminance value of each resampling area. It also discloses a method of "determining an implicit sampling area of each piece of original pixel data of a first format and each data point, so that the original pixel data is displayed in a second format having a plurality of three primary color pixel elements" The method includes determining a geometric center of each emitter of each of the three primary color pixel elements of the first format to define a sampling point. Then, each of the implicit sampling areas is defined by a line, and using The lines form a grid point, which is formed at an equal distance between "the geometric center of the emitter of one of the three primary color pixels and the geometric center of the emitter of one of the three primary color pixels and the same color." Etc. Reveal a method to limit the filter core divisor in a filter core to a value used to simplify hardware execution. The method includes using floating-point arithmetic to calculate differences; the area of filter coefficients, and Next, the paper of each controller is listed in the National Standard (C-NS) A4 specification (21〇X297 = issue) 5. Description of the invention (f number divided by one) The total area of the marginal area to obtain -S-results. Secondly, the first result is multiplied by a divisor to produce a total sum of the filter into a single search, to get the sum of the filter. A rounding point; and the total of the filter is converted to an integer. For a brief description of the diagram, please refer to Figure 4 and other diagrams, where the same elements are numbered with the same number: 、 Figure 1 μ Description-a display element "乂A previous, rgb, and arranging technique is to arrange two primary color pixel elements in an array in a single plane; Figure 2 illustrates an example of the previous technique of bright picture i, f RGB stripe configuration f, is effective Sub-picture sketch edge sampling points; Figures 3, 4 and 5 are illustrations of the previous techniques], R (} B stripe configuration, each of its sampling points is a valid sub-image sketch edge sampling area ; 〃 FIG. 6 illustrates a configuration in which a three-primary color pixel element is arranged in an array in a single plane in a display element; FIG. 7 illustrates effective sub-pixel drawing sampling of such configurations in FIGS. 6 and 27 Points; Figures 8 and 9 The illustration illustrates the alternative effective sub-pixels of the blue plane sampling points of the configurations shown in Fig. 6 and Fig. 27 depicting the sampling area. Fig. 10 illustrates an arrangement of three primary color pixel elements in a single plane in a display element. Another configuration in the array; Figure Π, which illustrates Figure 10. The effective sub-pixels of this configuration depict sampling points; Figure 丨 2, which illustrates Figure 10. The effective sub-image sketch of this configuration's blue plane sampling points, will sample Area; This paper standard Tong Cai a® Home Standard (CNS) field grid (21Gχ297 | ^) 594627 A7 B7 ^ 11-V. Description of the invention (Figure 13 and Figure 14 'Examples are shown in Figure 6 and Figure 1 〇 The valid sub-pixels of the red and green planes of these configurations ride the sampling area; Figure 15 also illustrates a sample point array of a prior art pixel data format, and the temple J is the same, the effective sampling area of the sample point, where The values of red, green, and blue are based on an equal spatial resolution grid, and these values are consistent; Figure 16 and Example 4 show the previous art map. Figure 5 overlays its subpixels. Drawing points Array of sampling points, # 15 The sampling points in Figure 15 are based on the same spatial analysis grid points and conform to the red and green "checkerboard" arrays in Figure 11; Figure 17 illustrates the previous technique. The configuration of the "sampling points" and "effective sampling areas of these sampling points," which are overlaid on the blue plane sampling area of Fig. 12, is based on the same spatial resolution grid of the sampling points of Fig. 15 Points' and match the red and green colors of figure n, checkerboard "array; Figure 18, illustrating the prior art Figure 15, which is superimposed on the red plane sampling area 5 of Figure 13, "Sampling points" and, which An array of effective sampling areas of equal sampling points, in which the sampling points of FIG. 15 in the prior art are based on the same spatial analysis grid and conform to the red and green “checkerboard” array of FIG. 11; FIG. 19 and FIG. 20, exemplifying the array of "sampling points" and "effective sampling areas of these sampling points" overlaid on the color plane sampling areas of Figs. 8 and 9 in the prior art. Point system The same spatial analysis grid points, and the red and green, checkerboard, array of Figure 7; Figure 21, illustrates the pixel data format of a prior art, and its "sampling point wood paper scale applies Chinese national standards ( CNS) A4 size (210 X 297 mm) binding 594627 A7 B7 V. Description of the invention (9) " ~ ^ ~ ^ ~ • and " An array of effective sampling areas of these sampling points ", where the The values of equal red, green, and blue are based on a uniform spatial analysis grid, and the values are consistent; Figure 22 'Illustrates the prior art Figure 2 1 superimposed on Figure 1 3 its red plane sampling An array of “,” sampling points, and “effective sampling areas of these sampling points”, where the sampling points of FIG. 21 are not planted based on the same spatial resolution grids and do not conform to the red of Figure Π And green "checkerboard" array; Figure 23 'illustrates an array of "sampling points" and "effective sampling areas of these sampling points" superimposed on the blue plane sampling area of Figure 12 in the prior art; Where first The sampling points of the previous technique in Figure 21 are not planted based on the same spatial analysis grid, nor do they match the red and green "checkerboard," array of Figure 11; Figure 24, illustrating the previous technique superimposed on Figure 21 Fig. 8 shows an array of "-π sampling points" and "effective sampling areas of these sampling points" on the blue plane sampling area, wherein the sampling points of the previous technique in Figure 21 are not based on the same spatial resolution grid points. Also does not conform to the red and green "checkerboard" array of Fig. 7; Fig. 25 illustrates the effective sampling area superimposed on the red plane sampling area of Fig. 13 in the red plane of Fig. 3; Fig. 26 illustrates the diagram 5 The effective sampling area overlaid on the blue plane sampling area in FIG. 8 in its blue plane; FIG. 27 illustrates one display element and another in which three primary color pixel elements are arranged in an array of three panels Configurations; Figures 28, 29, and 30, which illustrate the configuration of the blue, green, and red emitters on each individual panel of the device of Figure 27; _____-12-This paper size applies the Chinese National Standard (CNS) Α4 specification (210 X 297 (public expense) 594627 A7 B7

Fig. 31 illustrates the special case when the horizontal quantization ratio of an input pixel and an output sub-pixel (one red and one green) is 1: 2; The output sampling configuration is 200; Figure 32 illustrates an example of converting a 64x48VGA (video graphics interface) format image into a single image with a total of 800x600 red and green pixels? 611 is a single repeating grid of the matrix 202; Figure 33 illustrates the symmetry between the coefficients of a three-primary-color pixel element when the repeating grid size is odd; Figure 34 An example when the number is even; Figure 3 5 illustrates the sub-pixel 2 1 8 ′ delimited by a drawing area 246 in FIG. 33, where the drawing area overlaps with these six surrounding input pixel sampling areas 248; FIG. 36, The illustration illustrates the sub-pixel 232 in FIG. 33, where the drawing area 250 of the sub-pixel 232 overlaps with the five sampling areas 252; FIG. 37 illustrates the sub-pixels 2 3 4 in FIG. 3, where the sub-pixels 2 3 The drawing area 254 of 4 overlaps with the sampling area 256. FIG. 38 illustrates the sub-pixel 228 in FIG. 33, where the drawing area 258 of the sub-pixel 228 overlaps with the sampling area 260. FIG. 39 illustrates the sub-pixel in FIG. 33. Pixel 236, in which the drawing area 262 of sub-pixel 236 overlaps with sampling area 264; Figure 40 'illustrates the square sampling areas used to generate the blue filter core; Figure 41, illustrating the hexagons in Figure 8 Sampling area 1 2 3 with Correlation square sample area is 276. _-13-The size of this paper is applicable. National Standard (CNS) Α4 (210 X 297 mm)

Hold

594627 A7 _________ B7 V. Description of the invention () Details of the invention Generally, those skilled in the art will understand that the following description of the present invention is only used as an example to explain the use of the moon, and the meaning of the limitation. Other embodiments of the present invention will quickly and naturally appear in the minds of those skilled in the art. * Capture an image of a solid area and store the image in a memory to device. Where the image is generated from a known data configuration, the stored heart image can be depicted on a display element using an array that provides an improved color display resolution. The array is composed of a plurality of three primary color pixel elements, where the three primary color pixel elements include at least one monitor color emitter (or sub-pixel), a red emitter, and a green emitter, and when illuminated, the Emitters produce all other colors to the human eye. To determine the value of each of these emitters, one must first generate a conversion equation in the form of a filter core. The filter cores are generated by judging the relative area overlap between the sample area of the original data set and the target display sampling area. The overlap ratio determines the coefficient values used in the core array of the filter. In order to trace the stored image on the display element, a new point is determined in each of the three primary color pixel elements. The center of each new point will also be the source of the sampling point used to create the stored image. Similarly, the sampling point of the image data set is determined. Each new point is located at the center of the emitter (for example, at the center of a red emitter). When the newly-created points are placed at the center of the emitter, a boundary line grid point is formed at an equal distance from the centers of the newly-created points, thereby generating a sampling area (where these sampling points are _ ___ 14 二 ________ This paper size applies the Chinese National Standard (CMS) A4 specification (210X297 mm) 594627 A7 B7 5. Description of the invention (12 digits in the middle ~). The formed grid point produces a tile pattern. It can be used in the tile pattern The shapes include (but are not limited to) squares, rectangles, triangles, /, squares, octagons, rhombuses, cross squares, interleaved rectangles, interleaved triangles, X cross diamonds, penrose tiles, oblique squares, and distorted Oblique, ..., etc., and a combination including at least one of the foregoing shapes. It has been determined that the sampling points and sampling areas in the image lean material and the target display are heavy. The overlap produces sub-areas, where the outputs The sampling area of is overlapped with several input sampling areas. The ratio of the input area to the output area is determined by checking or calculating, and the ratio is stored as a coefficient in the core of the filter. The ratio The value is used to shift the proportion of the input value to the output value to determine the appropriate value of each emitter. When the quantization ratio to be used is sufficiently high, the sub-pixel configuration and rendering method disclosed here will be more than the previous The technology display provides better image quality, which uses information addressability and the new ed image modulation transfer function (mtf) to measure the image quality. Figure 1 illustrates a display element with a previous " RGB stripe The configuration "Technology" arranges the three primary color pixel elements in an array in a single plane, and Fig. 2 illustrates the prior art of Fig. 1 "Rgb stripe configuration. The effective sub-pixels depict sampling points. Figs. 3, 4 and 5 The example illustrates the prior art " RGB stripe configuration of Fig. 1 where the effective sub-pixels of each color plane of the sampling points delineate the sampling area. Fig. 5 will be further discussed here. Fig. 6 illustrates a specific implementation For example, the arrangement of several three primary color pixel elements is 20. The primary color pixel element 21 is a square, which is arranged at the origin of an X, Y coordinate system, and includes a Blue emitters 2 2, two -15-This paper wave size applies to Chinese National Standard (CNS) A4 specifications (2〗 0 X 297 male f) Binding t 594627

A red bus emitter 2 4 and a rain green emitter 2 6. The blue emitter 22 is arranged at the center ', which extends vertically along the X axis of the coordinate system to the first, second, third, and fourth quadrants. The red emitters 24 are arranged in the second and fourth quadrants which are not occupied by the blue emitters. The green emitters 2 6 are arranged in the first and third quadrants which are not occupied by the blue emitters. The monitor color emitter 22 is rectangular, and its edges are aligned along the x-axis and the γ-axis of the coordinate system, and the pair of opposite red and green emitters 24 and 26 on each side is usually square. The array is repeated over an entire panel to complete a device with a suitable matrix resolution. The repeated three primary color pixel elements form one, a checkerboard, in which the red emitter 24 / green emitter 26 and the blue emitter are staggered and displayed on the entire device, but the blue emitter The resolution is half the resolution of the red emitter 2 4 / green emitter 2 6. The blue emitters in the second row are staggered with the blue emitters in the previous row, or shifted by half the length of the blue emitter, as represented by the emitter 28. In order to fit this and because edges

Edge effect, so the blue emitters on some edges are only half the length of the blue emitters28. I FIG. 7 illustrates a configuration 2 9 in which the effective sub-pixels of the configurations in FIGS. 6 and 27 depict sampling points; and FIGS. 8 and 9 illustrate the configuration in FIG. 6 and FIG. At points 2 3, the alternative effective sub-pixels depict the configurations 30, 31 of the sampling areas 123, 124. Will diagrams be discussed further here? , 8 and 9. Fig. 10 illustrates an alternative exemplary embodiment of the configuration of the three primary color pixel elements (39) ...- Configuration 38. The three primary color pixel elements 39 are composed of a square ___- 16 'This paper size applies to the Chinese National Standard (CNS) A4 specification (210 X 297)

Equipment t 594627 A7 B7 14 V. Description of the invention (one of the blue emitters 3 2, two red emitters 3 4 and two green emitters 3 6. The three primary color pixel elements 39 are square, which are concentrated in an X , Y coordinate system origin. The blue emitter 32 is concentrated on the origin of the square and extends to the first, second, third and fourth of the X, γ coordinate system. In the quadrant, a pair of red emitters 34 are arranged in the opposite quadrants (that is, the second and fourth quadrants are not occupied by the blue emitter 32); and a pair of green emitters 36 are arranged opposite In the quadrant (that is, the parts in the first and third quadrants that are not occupied by the blue emitter 32). As shown in Fig. 丨, the blue emitter 32 is diamond-shaped, and its angle is the same as that of the coordinate system. The χ-axis and the ¥ -axis are aligned, and the pair of opposite red emitters 34 and the opposite green emitters 36 are usually fitted as squares to cut off the corners toward the center, where the sides forming are parallel to the sides of the blue emitters 3 2 T Repeat the array over an entire panel to complete a device with a suitable matrix resolution. The The repeated three primary color pixels form a "checkerboard", in which the red emitter 34 / green emitter 36 and the blue emitter 3 2 are staggered and displayed on the entire device, but the resolution of the blue emitter is Half of the resolution of the red emitter 34 / green emitter 36. The red emitters 34a and 34b will be further discussed here. One of the advantages of the three primary color pixel element array is that it improves the resolution of the color display. This is because Only the red and green emitters can significantly benefit the perception of radon resolution in this frequency range. If so, reduce the number of blue emitters and replace some blue emitters with red and green emitters, making them more relevant to humans. The vision is more matched, and the resolution is improved. Both are divided by the red and green emitters on the vertical axis in half to increase the space 594627

Addressability, thereby improving the vertical single-color ir and red / green emitter `` checkerboard '' known in the prior art, allowing high spatial frequency resolution to increase both the horizontal and vertical axis Space addressability.隔离 Isolate the newly created points in the geometric center of each emitter and generate a sampling grid to create a new image of the first data format on the display of the second data format. FIG. 丨 illustrates a configuration 40 of the effective newly created point of the three primary color pixel element configuration 38 of FIG. 10. The newly created points (for example, 33, 35, and 37 in FIG. Η) are concentrated on the geometric positions of the emitters in the three primary color pixel elements 39 (for example, 32, 34, and 36 in FIG. 10, respectively). Red new points 35 and green new points 37 form a red and green n checkerboard array on the display. The blue new points 3 3 are evenly distributed throughout the entire device, but the resolution is half that of the red new points 35 / green new points 37. In terms of sub-pixel rendering, these color newly-created points are regarded as sampling points, and are used to create effective sampling areas for each color plane, where the color newly-created points are processed individually. Figure 2 illustrates the new array with its effective blue sampling point 46 (corresponding to the blue new point 3 3 in Figure n) and the sampling area 44 in the blue plane 42 of the new array. For a new square grid point, the smallest perimeter is a square grid point. Figure 1 3 illustrates the effective red sampling points 5 on the red new points 3 5 and 7 on the red new points 2 5 in Figure 11 and the effective sampling areas 50, 52, 53 and 5 on the red plane 4 8 54 ^ Sampling points 51 form a square grid array in a direction where the display boundary is 4 50. If so, the sampling areas form a square grid in the center array of the sampling grid. Because '' Edge effect '' causes the square grid to overlap the display boundary, _ -18- This paper size applies to China National Standard (CNS) A4 (210X 297 mm)

Binding

594627 A7 —__ —_____ B7 V. Description of the invention () 16, J1] k 4 'f shape to maintain the same area and minimize the perimeter of the boundary of each sample (eg 5 4). Examination of these sampling areas will show that the sampling area 50 has the same area as the sampling area 52. However, the area of the sampling area 54 is slightly larger; the area of the sampling area 53 in the corners is slightly smaller. This does cause an error, because the description of the shell material in the sampling area 5 3 is too exaggerated; and the shell material changed in the ample area 5 4 is insufficiently described. However, in a display with hundreds of thousands to millions of emitters, the error will be minimal and disappear into the corners of the image. Fig. 14 illustrates the effective green sampling points 57 corresponding to the green new point 37 in Fig. N and the green new point 27 in Fig. 7, and the effective sampling areas 55, 56, 58 and 59 of the green plane 60. Examining FIG. 14 will show that FIG. 14 is basically similar to FIG. 13 in that it has the same correlation of the sampling regions, but its rotation angle is 180. . It is best to directly use the configuration of the emitters and the sampling points / sampling areas generated by the graphics software to produce high-quality images, and convert the graphics primitives or inwards to compensate for the color sampling plane, combining the sampling techniques of previous techniques With such sampling points and sampling areas. Ideal graphics display systems (such as portable electronics, laptop and desktop computers, and TV / video systems) will benefit from the use of flat panel displays and such data formats. These display types used may include, but are not limited to, liquid crystal displays, subtractive displays, plasma panel displays, cold light (EL) displays, electrophoretic displays, field-emitter displays, discrete light-emitting displays, and organic light-emitting diodes (OLED). Display, projector, cathode ray i (CRT) # 员 示 State, ..., etc., and at least one of the aforementioned displays --------------------- 19-This paper size applies to Chinese National Standard (CNS) Λ4 specification (210X 297 public goods) ^ 627 A7

in order? And together. Figure 13 illustrates a prior art pixel data format 70 of a sampling point (74) array and a sampling area 72 of the sampling point 74. The values of red, green, and monitored colors are based on an equal spatial resolution grid. , And these values are consistent. In the display system of the prior art, only the data of each color plane on the RGB stripe panel of the previous art of the 7F foot type shown in FIG. 1 is used to create the data format on a flat panel display. In FIG. 1, the resolution of each color sub-pixel is the same as that of the sampling points, and the three sub-pixels in a column are regarded as they are. A single combination and multi-color pixels are ignored, while each of these is ignored. The position of a color sub-pixel for the actual new point. In this technique, this usually means the display, the default mode, ·. This wastes the position information of these sub-pixels, especially the red and green sub-pixels. In contrast, the incoming RGB data in this application is regarded as three planes overlapping each other. In order to convert the data in the RGB format, each plane is processed individually. When information with this original prior art format is displayed on its more effective sub-pixel configuration in this application, it will be required that the data format will be converted by resampling. Resample the data in this way, "the output of each sampling point is a weighted function of the input data". The weighting function may be the same or different, depending on the spatial frequency of each of these data samples, as explained below. FIG. 16 illustrates the sampling point arrangement 76 of FIG. 15 superimposed on the sampling points 33, 35, and 37 of the sub-picture sketch edge of FIG. 11, where the sampling point 74 of FIG. 15 is based on the same spatial analysis lattice Point, and red in accordance with Figure 1 (red new ____- 20-This paper size applies to China National Standard (CNS) A4 specifications (210 X 297 mm) 594627 A7 ____B7 V. Description of the invention (~ 建 点 3 5) And green (green checkpoint 3 7) " checkerboard " array. Figure 17 illustrates the `` sampling point 74π '' and `` sampling points '' of Figure 15 overlaid on the blue plane sampling point 46 of Figure 15 The configuration 78 of the effective sampling area 72 " of 74, wherein the sampling point 74 in FIG. 15 is based on the same spatial analysis grid point and conforms to the red (red new point 3 5) and green (green new point 3) of FIG. 1 7), checkerboard "array. Figure 7 will be discussed further here. Figure 18 illustrates the overlapping of the red plane sampling point 35 and the red sampling areas 50, 52, 53 and 54 in Figure 15 in Figure 15 "Sampling point 74," and "Sampling point 74 effective sampling area 72" configuration 80, of which Figure 1 5 Sampling point 74 is based on the same spatial analysis lattice points, and conforms to the red (red new point 35) and green (green new point 37) "checkerboard" arrays of Figure n. The array in the square sampling area (52) will be The matching original sampling point 74 and its sampling area 82 are completely covered and extended to cover one quarter of each of the surrounding sampling areas 84 located inside the sampling area 52. In order to determine this rule, the output sampling area 50, 52, 53, or 54 which covers or overlaps the portion (fraction) on the input sampling area 72, and then multiplies the value of the corresponding sampling point 74 by the score 'and applies the product to the output sampling area 35. In FIG. 18, the area filled by the central or matching input sampling area 84 in the square sampling area 52 is half of the square sampling area 52. If yes, multiply the corresponding sampling by 丨 / 2 (or 0.5) The value of point 74. It is found by inspection that each of the square sampling areas 52 filled by the input areas that do not match each other is one-eighth (or 0.125) of the square sampling areas 52. ). If so, multiply the corresponding four sampling points by 丨 / 8 (or 0.125), and then 'add these values to the previous value (for example, multiply by I--; _______ ___________ -21-This paper rule & Common Chinese National Standard (CNS) Λ4 specification (210X297 mm) B7 19 V. Description of the invention (value of 0.5) to get the final output value of a specific sampling point 35. As far as the edge sampling point 35 and its five side sampling areas 50 are concerned, the private sampling area 8 of this phase is completely covered as in the above example, but it is only related to the three surrounding areas. The input sampling areas 84, 86, and 92 overlap. The input sampling area 84 of one of the parents represents one-eighth of the output sampling area 50. The adjacent input sampling areas 86 and 92 along the edge each represent three-sixteenths of the output area (that is, 0.1 875). As before, the weights of the input values 74 in the overlapped sampling area 72 are added to produce a value for the sampling point 35. The corners and "close" corners are treated in the same way. Because the corners 53 and “approach” in this image, the area covered by corners 54 is different from the central area 52 and the edge area 50. Therefore, the weighting of the input sampling area%, μ, 90, 92, 94, 96, and 98 is weighted. Will be different from the aforementioned input sampling areas 82, 84, 86, and 92. For smaller corner output sampling areas 53, this matching input sampling area 94 covers four-sevenths (or about 0.5 714). These adjacent input sampling areas 96 cover three-fourteenths (or approximately 0.2143) of the output sampling area 53. For a 'close,' corner sampling area 54, the matching input sampling area 54 90 covers eighty-seventeenth (or about 0.4706) of the output sampling area 54. The inner adjacent sampling area 98 covers two-seventeenths (or about 0 · η 76) of the output sampling area 54. The edges along the The adjacent input sampling area 92 covers three-seventeenths (or about 0.1765) of the output sampling area 54. The corner input sampling area 88 covers four-seventeenths (or about 0.2353) of the output sampling area 54. As before, In the overlapping sampling area 72 The weighted values of these input values 74 are added to generate sampling points __- 22-This paper size applies the Chinese National Standard (CNS) A4 specification (210 X 297 mm) 594627 A7 _ B7 V. Description of the invention (2〇 ) The value of 3 5. The resampling of the green plane is calculated in a similar manner, but the output sampling array is rotated by 180 °. The values ν_ for red sampling point 35 and green sampling point 37 are calculated as follows: Central area : V ^ CC,! ^} = 0.5 ^^ (0 ^) + 0.125 ^ Vm (CK., Ry) ^ +

〇 ″ 25 一 my + 〇K (C, M lower edge: V plus (c, Ry) = aSUCJEg + 0.1875_Vh (C,., Ry) Ben 0.1875_Vm (C ^ fg 4.0.125 ^ (0, ^ ,) Top edge: V ,,; rc, R. ≫ = 0.5..V.JCR,) + 0.1875_v..rc,, R.) 4-0.125.ynrcR :) φ 0.1875-HR,) right edge : V ^ Cry, VjC, Ry) = 0, 125'VJC ^ Ry), 0.t875_V, n (C, R, '·), 0.1875 into (d,) Left edge: ^ (C ^, ) = OS ^ V ^ fCtR,) ^ 0.1 + OMS ^ yjC ^) + 0-1875 ^ (0,1 ^.,) Upper right corner: V ^ QR,) = 0-57ia_V {<. (CfRy) 0.2 l43_V; n (Ct., Ry) 0.2143 ^ (0 ^,) _____- 23-This paper size applies to Chinese National Standard (CNS) A. 彳 size (210 x 297 mm) A7 B7 594 like 7 Ming (2 mouth Upper corner of friend: V ^ C ^ R.) = 0.5714 ^ ViH (CtR |) ^ 〇 2143 ^ Vm (C, R2) 4- On ^ JVjQ ^) Lower left corner: V ^ iCR,) = O.57l4yin ( CKk ^ ο 2ΐ43 ^ νΐΛ (〇 ^^ γ) 4- 0.2143 ^, ^, ^,) Bottom right corner: V0Uf (C, Ry) = 0.5714_V, rt (C, Ry) 4-0.2143 ^^^,) + 02WJVln (Ctk ^) Edge, left corner 靠:

Vn «» (C2R,) = 0.4706-v JC2R, U 0.2353- 乂 "((: Van Bo 1176-Vm (QR2) + 〇-1765 ^ Vlfl (C? R,) :

ΆΚ2); 〇47〇6, VjC, R7) + 〇, 1765J ^ C, fc3) + α 1176-V, "(CR2) + 〇 · 2353 乂 (C, RJ Left edge, bottom corner 隅: V ^ fCiRy) = 0.4706 ^ Vln (CtRy) + 0.2353. ^ ,, (0, ^,) 0.1J76J / JC? Kr) ^ 〔H (C, RT ·,) bottom edge, left side corner 隅: V ^ C ^) ^ 0.4706 ^ ,, (0 ^) + 0-2353 ^, ^, ^,) 4. 〇. 1765 ^ Vffl (CjRy) ^

〇.mvjc2M lower edge, right side corner 隅: V 细 (QRy) = 0.4706- ^ 1 ^ 4 0.1765 一 Ά ,, βν) ▲ 〇 2353_V, " (C ^ Rt) aimd ·)

This paper size applies the Chinese National Standard (CMS) A4 specification (21.0 X 297 mm) 594627 A7 B7 V. Description of the invention (22) Right edge, bottom corner 隅: V ^ tCCRy) = OAyoe ^ yjCnRy) ^ Ol l76_V, n (C „.iRy) 0 + Ο.Π65_ν, ΗΠ) Right edge, top angle 隅: = 〇.47〇6 ^ Vrt (C.R2) ^ 0.14-0.1765 ^ V; M (CtR〇 + 〇- 2353 ^ V, n (CxRf) Top edge, right side corner 隅: UQR ·) = CU706-VJQRO + 0 · 17W-VJC, ·,! ^) + 〇. 117d (C, R2) + 0.2353-V, H ) Where Vin is only the chromaticity value of the color of the sub-pixel on CxRy (the red sub-pixel 34 / green sub-pixel 36 in row X of the CxR table; and Ry represents the red sub-pixel 34 / green sub-pixel 36 in column y If so, CxRy represents the red sub-pixel emitter 34 or the green sub-pixel emitter 36 in the Xth and yth columns on the display panel, starting from the top left corner (as usual). Note 'Each of these It is important that the sum of the coefficients in the equation equal to 1. It is important that although there are seventeen equations to calculate the conversion of the entire image, there are only four sets of coefficients due to symmetry. This reduces the complexity during execution. As mentioned earlier, FIG. 17 illustrates the configuration 78 of the “sampling point 74” and the effective sampling area 72 of the “sampling point 74” of FIG. 15 superimposed on the blue plane sampling point 46 of FIG. The sampling point 74 of 5 is based on the same red space (red newly created point 35) and green (green newly created point 3J) φ checkerboard, array, which is in accordance with the same space with noon grid. The blue sampling point 仏 in Fig. I 2 allows the blue sampling area 44 to be determined by inspection. In this case, the blue sample area

Paper > 〇㈣rt Coffee House Fresh (CNS) Weaving Knowledge ---- J 594627

AT B7 V. Description of the Invention Before (23), there is a blue resampling area, which is purely the raw data sampling point 74. The arithmetic mean of the blue values around it, where the arithmetic mean is calculated as the sampling point of the resampled image The value of 46. Calculate the blue output value of sampling point 46 as follows: R ^) = 0.25 ^ Vle (QRy) 4- 0.25.V4- 0.25_VJC, .tfty) 4. 〇.25J / Uy + _) where Vin is the surrounding input sampling point The blue chromaticity value of 74; cx represents the sampling point 74 in the x-th row; and Ry represents the sampling point 74 in the y-th column (starting from the upper left corner as usual). In terms of calculating blue subpixels, since there is only one blue subpixel in each red subpixel / green subpixel pair, x * y must be an odd number. The total of these coefficients is equal to 1. Weight the coefficients of the central region equation of the red sampling point 3 5 (which has the greatest impact on the sense image) and apply it to the central re-sampling region 52 for the binary shift removal number, where 0.5 is shifted one bit to the right; 〇25 is two bits shifted to the right; and 0.1 25 is three bits shifted to the right. If so, the general rule is very simple and fast, it only involves simple removal and addition. In order to achieve the highest accuracy and speed, the surrounding pixels should be added first, then shift a single 3-bit to the right, and then increase the shifted single-bit central value. However, the following equations for the red / green Sampe regions on these edges and corners involve more complex multiplication operations. On a small display (for example, a display with only a few pixels), a more complex equation may be required to ensure a good image quality L ------ 26 This paper scale applies Chinese national standards ( CNS) A4 specification (at 0 × 297 mm) 594627 A7 B7 5. Description of the invention (Quality. For large images or displays, a small error on such edges and corners may not matter, it can be simplified For the sake of simplicity, with the “missing” on the edge of the image, the input data sampling point applies the first square of these red / green planes to the edges and corners, resulting in rotation sampling. The point 74 is set equal to " in accordance with the input sampling point 74, .. Or, the missing value such as 泫 can be set to black. The general rule can be easily implemented in software, carcass or hardware. Linear addition is important, which means that the sub-pixel mapping must be completed before gamma (r) correction. The output of the aforementioned general rule can be fed into a children's gamma correction table. For example, perform gamma before sub-pixel mapping. Fork timing, then Unexpected chrominance errors can occur. Figures 19 and 20 illustrate the overlapping of the blue plane sampling area 23 of Figures 8 and 9 in Figure 15 with "Sampling point 74", and "Sampling point is effective" The two alternative configurations of the sampling area 72 are 100 and 102. The sampling points in FIG. 5 are based on the same two analytical grid points and conform to the red and green checkerboard array of FIG. 7. Fig. 8 illustrates the configuration of the emitter in Fig. 6 whose blue plane sampling point 23 (not shown in Fig. 7) is sufficient to have a valid self-pixel delineating sampling area 1 2 3 with the smallest boundary perimeter. The method will proceed as described above. Calculate the overlap ratio of the output sampling area 123 and each of its input sampling areas 72 in Figure 9 and use the X-fold ratio as a conversion equation or coefficient in the core of the filter. Below Multiply these sampling values by 74 in the conversion equation of the following equation: _ -_1 ^ 7- This paper size is applicable to China. National Standard (CNS) A4 specification (210 X 297)

Binding

Line 594627 A7 B7 V. Description of the invention (25) V〇 «(C ^^) = 0.015625 ^ VjCi (Ry) 4. 〇.234375 ^ Vtn (CtRy) ^ 0.2i4375 ^ Vlw (Cx., Ry) + 〇 '015625-Ά ▲ 九） + 〇. 015625— Vm (Cx_, RY', Bu 0-2 such as 3 W (C, ') + 0 · 234375 a ν · "(") Ben 〇〇1562mR—) One Those skilled in the art can find ways to perform these calculations quickly. For example, a coefficient of 0.0015625 is equivalent to shifting six bits to the right. When the sampling point 74 of Fig. 15 is based on the same spatial analysis grid and conforms to the red (red newly-created point 25) and green (green newly-created point 27) "checkerboard" arrays of Fig. 7, the minimum Boundary condition regions can result in additional computational load, and, spreading the data across six sampling points 74. The alternative effective output sampling region (24) of Figure 9 can be configured with 3 1 for some applications or In the case, for example, when the sampling point 74 of Fig. 5 is based on the same two analytical grid points and conforms to the red (red newly-created point 25) and green (green new-built point 27) of Fig. 7 ”Array; or when the correlation between the input sampling area 74 and the output sampling area is as shown in FIG. 20, then these calculations will be simpler. In the even rows, this is used to calculate the blue The formula for the color output sampling point 2 3 is exactly the same as the formula developed above for Figure 17. In these odd rows, the calculation in Figure 20 is as follows: = 0 · 25—V; " (C, Ry) + 0 · 25-V; I1 (C, -Ry) 4 «CK25" Vlrt (Qi? Y 丨) 4 * 0.25-m,) as usual FIG central calculation executing sampling region 124 of FIG. 19 and the 2〇. When calculating such edges, the conversion formula needs to be modified or the value of the sampling point 74 deviating from the edge of the screen needs to be assumed, as described above. ______- 28-National Standard (CNS) A4 specification (210 X 297 mm) 594627 A7 B7 V. Description of the invention (26) Now turn to Figure 2i, which illustrates the pixel data format of a prior art. · Sampling point 122 & quot And "Sampling point 122 is a valid sampling area 120," a configuration of 04. Figure 21 illustrates that these red, green, and blue values are based on an equal spatial resolution grid, and that these values are consistent. However, the image size is different from the image size illustrated in Figure 15. FIG. 22 illustrates an array 10 of the effective sampling area 1 2 0π of “sampling point 122” and “sampling point 122” superimposed on the red plane sampling areas 50, 52, 53 and 54 of FIG. 6. Among them, the sampling point 12 2 in FIG. 21 is not planted based on the same spatial analysis lattice, and it does not meet the red (red new points 25, 35) and green (new green points 27, 37) of FIG. 7 or FIG. 11, respectively. "Chessboard" array. In the configuration of Fig. 22, it is not permissible to calculate each output sample 35 with a single oversimplified conversion equation. However, it is possible and feasible to generalize the method used to generate each of these calculations based on the proportion of the covered area. If the ratio of the input image to the output image is any particular ratio (especially those that are common and standard in the industry), there is indeed a least common denominator ratio that causes the image to be converted into a repeating grid pattern. The complexity is further reduced due to symmetry. As demonstrated above, these input and output arrays are consistent with 4 cases. When combined, the repeated three primary color sampling points 1 2 2 and the symmetry will reduce the total number of unique coefficient groups to a more easily controlled level. For example, the commercial standard display color called "VGA" (used as a standard for video graphics interface cards, but now only means 640x480) has _ rows and 480 columns. The format needs to be resampled or quantized To show in a match

594627 A7 B7

As shown in FIG. 10, the panel has 400 red sub-pixels M and 400 green sub-pixels 36 (800 sub-pixels in total) in the horizontal direction, and 600 sub-pixels 34/36 in total in the downward direction. This results in a ratio of input pixels to output subpixels of 4 to 5. The transfer equation of each of the red sub-pixels 34 and each of the green sub-pixels 36 can be calculated according to the portion of the sampie output area 52 overlay and its input sampling area 120. This procedure is similar to the developed transfer equations for the figure "1. The transfer equations for each of the single-output sampling points 35 seem to be quite different. Fortunately, when starting to calculate all these transfer equations , A pattern will emerge. The same five transfer equations are repeated over and over on one column; and the five million equations in the other pattern are repeated on each row. The ratio of one pixel to subpixel is 4 For example, the result is only 5 meanings, 5 or 25 sets of unique 10,000 formulas. This reduces these unique calculations to 25 sets of coefficients, and other symmetry modes can be found between these coefficients, causing the total The number of coefficient sets is reduced to only six unique sets of coefficients. The same program will produce a set of identical coefficients for the configuration 20 of Figure 6. The following is a general example describing how to calculate these coefficients using the geometric method described above. Figure 32 illustrates an example of the above, converting a 64 × 48VgA format image into a PenTile matrix with a total of 800 × 600 red / green subpixels ” A single 5x5 repeating grid in 002. Each of these square subpixels 205 bounded by a solid line 206 indicates the position of a red or green subpixel that must have a set of calculated coefficients. If there is no symmetry, 25 sets of coefficients are required to be calculated. Figure 32 will be discussed in more detail later. Figure 3 j illustrates the symmetry of these coefficients. If it is used in this industry ------------ ___- JU-this Paper size applies Chinese National Standard (CNS) A4 specification (210 X 297 public capital)

Binding

594627 A7 _B7 V. The common matrix form of the filter core in the description of the invention (28) When these coefficients are written down, the sub-pixel 2 1 6 will pass; the core of the preparation will be a reflection image --- Yu pixel 7 1 The core of 8 flips from left to right. This is true for all sub-pixels located on the right side of the symmetry line 22, where each of these sub-pixels has a filter core, where the filter core is the sub-pixel's relative sub-pixel Image of the core of the device. In addition, the sub-pixel 222 also has a filter core that reflects the scene> the core of the sub-pixel 2 1 8 is flipped from top to bottom. This is also true for all filter cores located below the line of symmetry 224, where each of these filter cores is a reflection image of a relative sub-pixel filter. After taking the 'sub-pixel 2 2 6's filter, the core of the filter is an image reflecting the ___ sub-pixel 228's filter is flipped on a corner. This is true for all sub-pixels located at the upper right of the symmetry line 230, where the filter of each of these sub-pixels reflects the image as a diagonal line of the sub-pixel filter. Finally, the diagonal of the filter core system on the diagonal line is symmetrical to the ground diagonally, and the coefficient values on the diagonal line of the symmetrical line 230 are exactly the same. Here we further provide a complete set of filter core examples to demonstrate all of these symmetric filter cores. The filters to be calculated are only the sub-pixels 218, 228, 232, 234, 236, and 238 of the shadow portion. In this case, when the size of a repeating cell is 5, the required filter is only at least 6. The remaining filters can be determined by flipping the six calculated filters on the different axes. Whenever the size of a repeated cell is an odd number, the formula for determining the minimum number of filters is as follows: ____-31-This paper is sized to the Sichuan National Standard (CNLS) Λ4 size (210 X 297 mm) 594627 A7 B7 V. Description of Invention (29

NfiUs =

Where P is the width and height of the odd repeating grid; and Nfihs is the minimum required number of filters. FIG. 34 illustrates an example in which the repeated cell size is an even number. These need to be considered differently, and the filter only has the sub-pixels 24, 242, and 244 in the shadow portion. In this case, when a repeating cell size is 4, only 3 filters need to be calculated. Whenever the size of the repeated grid is even, the formula used to determine the minimum number of filters is as follows: where P is the width and height of the even repeating grid; and Neven is the minimum required filter number. Turning back to FIG. 32, the drawing boundary 208 of the central sub-pixel 204 surrounds an area 21 which overlaps with the four original pixel sampling areas 2 丨 2. Each of these overlapping areas is equal, and their coefficients must add up to equal 丨, so that each of these overlapping areas is 1/4 or 025. These coefficients are the coefficients of the sub-pixels 23 8 in Figure D. In this case, the 2 × 2 filter core will be: 1/4 Γ ------ _1 / 4 1/4 _m ___ —-32- This paper scale applies the Chinese National Standard (CNS) A4 specification (210 X 297 mm) 594627 A7 _______B7 V. Description of the invention () In Figure 3 :), the coefficients of the sub-pixels 2 丨 8 in Figure 33 have been developed. A sub-pixel 2 1 8 is delimited by a drawing area 2 * 6 that overlaps with the surrounding private pixel sampling area 2 4 $ of these ′ and five. Although the sub-pixel is in the upper left corner of a repeating grid, it is assumed for the sake of "difference" that there is always another repeating grid passing through and overlapping the sampling area 248 outside the grid. Complete these calculations for a general case, and process the displayed edge and mouth drawing area 246 in a way different from that described above, so that three horizontal sampling areas 248 and three vertical = sampling areas 248 intersect, Therefore, a 3 × 3 filter core is needed to accommodate all these coefficients. Calculate the coefficients as described previously: measure the area of the drawing area 246 covering each of these input sampling areas; and then divide each of these faces by the total area of the drawing area 246. Since the drawing area 2 does not overlap with the upper left, upper right, lower left, or lower right sampling areas 248, their coefficients are all zero. The sampling area 246 is the sampling area above the center and the left or 2 4 8 parent $. The size of the drawing area is 1/8 of the total area of the drawing edge area 2 4 6. Therefore, the f is 1/8. The area where the drawing area 246 overlaps with the center sampling area 248 is a maximum of 1 1/16. Finally, the area of the drawing area 246 and the sampling area 248 of the middle and central Shirazaki on the right is the smallest, ie, m2. Placing all these coefficients in order will result in the following coefficient filter cores: 0 1/8 0 1/8 11/16 1/32 0 1/32 0 矣 Figure 3 illustrates the example in Figure 3 3 Pixels 2 3 2 in which the delineated area is 2 5 0 ___—______. 33-This paper 'Ut applies Chinese National Standard (CNS) Λ · 1 specifications (210 X 297 male sharp)' W4627 A7

Entered into five sampling areas 252. 4? '4. — uf. ^ — 1 If the' counter nose drawing area 250 and the mother sample area 252 intersect each other, and divide each of these areas by 4 Tiansu area 2 5 0 area. # 人 今 吉 „丨, ten thousand; in the case of the example, only a 3x2 filter 'core is needed to accommodate all these coefficients. However, for consistency, a 3x3 filter core is used. Figure% Fuser core ... 1/64 17/64 0 7/64 37/64 2/64 0 0 0 The sub-pixel 234 of FIG. 33 is illustrated in FIG. 37, where the drawing area 254 overlaps with the sampling area 256. Calculation The coefficient of this case will lead to the following core:

4/64 14/64 0 14/64 32/64 0 0 0 0 Order

The lines exemplify the sub-pixels 2 2 8 of FIG. 3 in FIG. 38, where the drawing area 2 5 8 overlaps with the sampling area 260. Calculating the coefficients for this case will result in the following core: 4/64 27/64 ---- Π 1/64 4/64 27/64 ---- 1/64 0 0 0 ____- 34-This paper scale applies to China National Standard (CNS) A4 (210 x 297 mm) 594627 A7 B7

V. Description of the Invention (Finally, the sub-pixel 230 of FIG. 33 is illustrated in FIG. 39, in which the drawing area 2 62 overlaps with the sampling area 204. Calculating the coefficient of this case will result in 4/64 —___________ — 2 7 / 64 1/64 4/64 27/64 1/64 0 0 0 The total number of all minimum calculations required to terminate this instance with "pixels: subpixels = 4: 5" can be reversed by flipping the six on these different axes. Each filter core constructs the remaining 25 sets of coefficients, as described in Figure 33. For quantification, the sum of the filter cores must always equal 丨, or the filter cores will cause the output image to Brightness. This is true for all the filter cores above. However, if the cores are actually used in this form, the coefficient values will all be fractions and require floating-point arithmetic. It is often seen in the industry to multiply all the coefficients by a certain value to convert all the coefficients into integers. Then, as long as the total is divided by the same value in the future, integer filtering can be used to filter these coefficients. The core coefficient is multiplied by the input sample value. Checking the filter core above, it appears that M is a value suitable for multiplying all these coefficients. This will result in the filter core of the sub-pixel 2 1 8 in Figure 35 as follows: 0 8 0 8 44 2 0 2 0 (Divided by 64) ___-35 This paper size applies the Chinese National Standard (CNS) A4 specification (2 public funds of 7 × 7) 594627 A7 B7 V. Description of the invention (33 In this case, the same can be modified All other filter cores 'to convert them to integers' for easy calculation. In this case, it is particularly convenient when the divisor is a power of two (in this case, the divisor is 2 6th power). You can move the result to the right to quickly divide the element in software or hardware by a power of 2 ". In this case, one shifts 6 bits to the right The shift will be divided by 64. In contrast, a commercial standard display color image format called "XG A" (used as a standard for extended graphics cards, but now only means 1024x768) has 1024 rows and 768 columns. This format can be quantified to have a "1600 \ 1200 red emitter 34" in Figure 10. And green emitter 36 (plus 80 (^ 600) blue emitter 32) "configuration 38 is shown. The architecture's quantization or resampling ratio is 16 to 25, which results in 625 sets of unique coefficients. Use this The symmetry between the equal coefficients reduces this total coefficient group to a more appropriate number, which is 91. However, even if the total number of filters is small, it still feels tedious when done by hand, as described above. However, one A computer program (a machine can read the media), but a machine (such as a computer) can be used to automate the task, and these coefficient sets can be generated quickly. In fact, the program is used once to generate a filter core table for any particular ratio. Then, use the table by quantization / tracing software; or burn the table into the read-only memory (ROM) of the hardware that performs quantization and sub-image sketching. The first step that must be completed when the filter generates a program is to calculate the quantization ratio and the size of the repeating grid. This is done by dividing the total number of input pixels and the total number of output sub-pixels by their greatest common denominator (GCD). This can also be done with a small two-way nested loop. This external circuit is compared with a set of quality ------- 36-This paper size is applicable to China National Standard (CNS) A4 specifications (210 X: 297 public funds) 594627 A7 B7

V. Description of the invention (34 Number test these two values, 1 until the square root of the smaller number of the number of two pixels, such as the prime and 涊, is equal. In fact, 'the typical screen In terms of size, there should be no need to test prime numbers larger than 41. On the contrary, because the general rule intends to produce the filter core offline in advance, the external circuit can only The execution of prime and non-prime numbers between the range of values. This may be ^ CPU time because it performs some unnecessary tests, however, it only performs this encoding once for a particular input and output screen size combination. An internal loop The total number of the two image numbers is tested against the current prime number. If both total numbers can be divided by the prime number, the two total numbers are divided by the prime number, and the internal loop continues until one of the values does not May be divided by the prime number again. When the external loop is terminated, the remaining smaller value will effectively be divided by the GCD. The two values will be the "quantity" of the total number of the two pixels Conversion ratio '·. Some typical values: 320: 640 becomes 1: 2 384: 480 becomes 4: 5 5 12: 640 becomes 4: 5 480: 768 becomes 5: 8 640: 1024 becomes 5: 8 The ratio means '' the ratio of the pixel to the subpixel 'or' P · S ratio ", where P is the input pixel numerator of the ratio; and S is the subpixel denominator of the ratio. A repeating cell has a horizontal direction. The total number of filter cores required in the downward direction is s in these ratios. The total number of cores required is "the level s value X the ___, 37-this paper size ) 210X 297 * ^ 17 594627 A7 B7

Vertical S value ". Nearly all of the common VGA-derived screen sizes' the size of the horizontal and vertical repeating graphics results are exactly the same, and the total number of filters required is S 2. According to the table above A 640x480 image quantized into a 1024x768 PenTile matrix has a p: s ratio equal to 5: 8 'and would require 8x8 or 64 different filter cores (before considering symmetry).

Hold

In a theoretical context, the fractional values that add up to 丨 are used in a filter core. In fact, as mentioned above, the filter core is usually calculated as an integer value, and then a divisor is applied to normalize the total back to 1. It is important to start by calculating these specific gravity values as accurately as possible, so the drawing areas can be calculated in a sufficiently large coordinate system to ensure that all calculation results for children are integers. Experience has shown that using the correct coordinate system under image quantization is a reason for an input pixel size equal to the total number of output subpixels across a repeating cell, which makes an output pixel equal to the size of the crossing a repeated cell. The total number of input pixels. i is counter-intuitive and seems to be unfavorable. For example, when 512 input pixels are quantized to 640 with a 4: 5 p ·· s ratio, the input pixels can be drawn as 5x5 squares on graph paper and the 5x5 squares can be The output pixels are drawn as 4x4 squares. This is the smallest size that can describe two pixels of a book, while keeping all these values at whole numbers. In the coordinate system, the area of the diamond-shaped trajectories concentrated on the output sub-pixels will always be equal to twice the area of an output pixel or 2 * p2. This is the smallest integer value that can be used as the denominator for these filter specific gravity values. Unfortunately, when the diamond falls on several input pixels, then -38-

594627 V. Description of the invention (36 diamond cut into triangles. The area of this triangle is A. t. Younu, Si Yukuan X X / 2 "'This can lead to non-positive numbers again. Calculate twice the area to solve This problem is different, so the formula is different. 2 X area. This makes — the human denominator equal to 4V. ^ Take a small effective integer and then, it is necessary to determine each filter must be. In the example of hand 60%, some of the filter cores are, ^ spoon, some are 3χ2, and others are filtered: the core is 3χ3. 如何 The relatively large input and output pixels ^, and the shape of the riding area each other Intersect to determine the maximum filter core size that is needed. When you click on the private image, you say, Tian Ru ..., 屌 i mouth ~ like an image, and each of the original images has its input pixels When there are more than two output sub-pixels on it (for example, 100: 2001 or 1: 3), then a 2 × 2 filter core will be possible. This will require less hardware implementation. , Because the generated image captures the square hidden target pixels, it is as much as possible 'Preserve spatial frequency' is represented by the ㈣ edge of many flat-panel displays, so the image quality is better than the quantification of previous techniques. Font and icon programmers use spatial frequency to improve the resolution of the appearance and cleverly escape this technique. The Nyquist limitation is already well known to me. The quantization rules of previous techniques either use the intersection method to limit the quantized spatial frequencies to the limit, or keep the sharp and clear, but produce annoying phase errors. When the ratio is reduced, the input pixels are more than the output sub-pixels. When any quantization factor is greater than 1: 1 (such as 10: 1: 100 or 2: 1), the filter size becomes 4X 4 or greater It is difficult to persuade hardware manufacturers to add more line buffers to implement this system. However, the advantage of staying in the range of 1: 1 and 1: 2 is that the core size is kept on a fixed 3x3 filter. Fortunately National Standards for Paper Tissues (CNS) Λ Frame (210 X 297 Public Goods 594627 A7 ------ B7) V. Description of the Invention (37) is that these must be performed in hardware falling within this range Most cases' and edit It is appropriate to write a program that only produces 3 x 3 cores. In some special cases (such as the example executed by hand above), then some filter cores will be smaller than 3x3. In other special cases, Even though the filter may theoretically become 3x3 ', the result is that each filter is only 2x2. However, it is easier to calculate the cores and implement a hard system with a fixed core size for this general case. Finally, the difference The weight values of the core filters are only an item at this moment to calculate the area (multiplied by 2) of the 3 × 3 input pixels of the output diamond-shaped fathers at each unique (asymmetric) position in the repeating grid. This is A very simple, easy-to-do, "describe," work that is well known to me in this industry. Calculate 3 × 3 or 9 coefficients for each filter core. In order to calculate each of these coefficients, a vector describing the diamond-shaped drawing area is described. The shape is clipped against the edges of the pixel areas of these losses. Take advantage of the polygon trimming rules already known to me in this industry. Finally, calculate the $ product of the clipped multilateral two (multiply by 2). The resulting area is the number of corresponding grids of the filter core. An example output from this program is shown below: #. Original pixel resolution 1024 mesh sub-pixel resolution 1280 quantization ratio is 4: 5 All filter values are divided by 2 5 6 The minimum required filter (with symmetry ): 6 filters (no symmetry) generated here ·· 25 ~ 4-0 ~

This paper size applies to the Chinese National Standard (CNS A4 specification, < 297) to approximate the edges and corners of the image. Applying each set of coefficients to a filter core 'is well known to us in this art. Recording these positions and symmetric operators is a job for the software or hardware programmer to use the mathematical function of the modular function ’, which is also familiar to me in this technology 594627 A7

Fang, Shang spoon 蚣 蚣 出 出 中 'without considering symmetry, such as all 25 _ cores required for this example, in order to allow inspection = equal coefficients and repeated check box There is a horizontal, vertical, and diagonal symmetry between the cores of the death. If #, you can process the edges and corners of the image independently; or you can fill in the other averages, the highest effective contributing factor, or black in the "missing, input data sampling ___ ___ 41->, Paper-scale currency_ $ 料 (CNS) :, 4 specifications (210 > < Canada)

594627

Known. The work of generating a third-level coefficient is a matter of calculating the input sampling area using a device known in the art, and the output sampling point corresponding to each sample J 5 and the ratio of the area where the output sampling area 5 2 overlaps. FIG. 23 illustrates an array 108 of “sampling point 122” and `` effective sampling area 122 of sampling point 122 '' superimposed on the blue plane sampling area 44 of FIG. 2 in FIG. Sampling point 122 is not based on the same spatial analysis lattice points, nor does it conform to the red and green, checkerboard "array of Figure 11. The method of generating these conversion equations is calculated as described earlier. First, determine The size of the repeated three primary color pixel element arrays. Next, determine the minimum total number of unique coefficients, and then determine the dependencies, and input the sampling area 120 · 'and the output samples corresponding to each of the output sampling points 46 Area 44, the value of the coefficient of the parent stack ratio. Each of these values is applied to the conversion equation. The repeated three primary color pixel elements and the resulting total number of coefficient arrays are the same as those determined for the red and green planes FIG. 24 illustrates one of the `` sampling points '' and `` effective sampling areas of these sampling points '' superimposed on the blue plane sampling area in FIG. 8 Array 1 10, of which sampling point 122 in FIG. 21 is not planted based on the same spatial analysis lattice, and does not meet the red (red new point 35) and green (green new point 3 7) "checkerboard" of Figure 11 Array. The calculation method for generating these conversion equations is performed as described above. First, determine the size of the repeated three primary color pixel element arrays. Next, determine the minimum total number of unique coefficients, and then determine the input sampling area according to丨 20 ”and“ Each output sampling area 1 2 3 ”corresponding to each of the output sampling points 2 3” and the value of the coefficient of its overlap ratio. Each of these values is applied to the conversion equation. ____- 42-This paper size is in accordance with Chinese National Standard (CNS) A4 specifications (: 2〗 0 X 297 male |) 594627 A7 _B7 V. Description of the invention (40) The facets have been RGB format of 4 CRT 'a well-known RGB The flat-panel display configuration 10 has 4 red emitters, 6 green emitters, and 2 blue emitters arranged in a two-element bar pixel element 8. As in the previous art, Figure 1, in order to project an image formatted according to this configuration onto the three primary color pixel elements illustrated in Figure 6 or Figure 10, these new points must be determined. The configuration presented in Figure 2 of the previous art illustrates the placement of these red, green, and monitored color new points. The red, green and blue new construction points are not consistent with each other, they have a horizontal displacement. According to the prior art disclosed by Benzschawel et al. In U.S. Patent No. 5,341,153, and later Hill et al. In U.S. Patent No. 6,18,3,85, these locations are used for sampling with sampling areas Points 3, 5 and 7. The red plane 14 is shown in FIG. 3 of the prior art, the blue plane 16 is shown in FIG. 4 of the prior art, and the green plane 18 is shown in FIG. 5 of the prior art. A conversion equation can be generated according to the prior art configuration presented in Figures 3, 4 and 5 and the methods disclosed herein. The method outlined above can be utilized by calculating the conversion equations or filter core coefficients for each output sampling point of the selected prior art configuration. FIG. 25 illustrates the effective sampling area 1 25 overlaid on the red plane sampling area 52 in FIG. 13 in the red plane of FIG. 3, wherein the configuration of the red emitter in FIG. 25 and the configuration in FIG. 6 / FIG. Same pixel level (repeat unit) resolution. The method of generating these conversion equations is performed as described above. First, the size of the repeated two primary color pixel element array is determined. Then, the smallest total number of unique coefficients is determined by paying attention to the symmetry (2 in this case). Then, by taking the input sampling area of 3 5 for each corresponding output sampling point, 25 pairs of output are taken. The paper size is suitable for financial standards (CMS) Λ grid (21 () X 297 公 #) 594627 A7 ___ B7 5 2. Description of the invention (41) ~ The ratio of the sample area 5 2 overlaps, and determines the values of those coefficients. Each of these values is applied to the conversion equation. The resampling of the green plane is calculated in a similar manner (as exemplified in Figure 4), however, the output sampling array is rotated by 180 °, and the green input sampling area 127 is compensated. Fig. 2 illustrates an example of the first cutting technique. Fig. 4 shows an effective sampling area 1 2 7 superimposed on the blue plane sampling area 1 2 3 in the blue plane of the figure. Fig. 40 illustrates a monitor color example corresponding to the red and green examples in Fig. 32. The sampling area 266 in FIG. 40 is a square, not a crying shape, as in the magenta and green examples. The total number of original pixel boundaries 272 is the same, but the total number of blue output pixel boundaries 274 is smaller. The coefficients are calculated as previously described: the area of each of the input sampling areas 268 covered by the drawing area 266 is measured; and then each of these areas is divided by the total area of the drawing area 266. In this example, the blue sampling area 266 evenly overlaps these four original pixel areas 268, resulting in a filter core with 4 coefficients, each of which is 1/4. The other eight blue output pixel regions 27 and their geometric intersection with the original pixel region 268 can be seen in FIG. 40. The relative correlation of the resulting filters can be observed in a symmetrical configuration of the original pixel boundary 274 in each output pixel region 27. In more complicated cases, a computer program is used to generate the blue filter core ~. The private result is very similar to the red and green filter heart program. The distance between the blue sub-pixel sampling points 33 in Figure η is twice the distance between the red and green sampling points 35, 37, which implies that these blue drawing areas will be twice as wide. However, these red and green ones are: ---------44-This paper Phantom Financial ® National Standard (CNS) A4 specification (21〇x2fn public funds) 594627 A7 B7

The drawing area is diamond-shaped, and if it is twice the width between the sampling points. This makes the red and green painted areas have the same width and height as the blue painted areas, resulting in several suitable values; the blue color; the size of the core and the material red and green "The size of the core is all the same. Same as Xiao ', the size of the blue repeating grid is usually exactly the same as the size of the red and green repeating grids. Since the blue pre-pixel sampling points are separated by 33, the P: s (the ratio of pixels to sub-pixels) becomes twice. For example, a 2: 3 ratio of red becomes blue 4: 3. However, the size of the repeated cell is determined by the s value in the ratio, and it is not affected by Doubling and changing. However, if the denominator can be divided by 2, another optimization can be achieved. If so, the two blue values can be divided by a power of 2. For example, if the red and When the green P: S ratio is equal to 3: 4, then the blue ratio will be 6: 4, which can be reduced to 3.2. This thinking refers to the case of 涊 and other (even) cases where the blue Repeat before. The size is cut in half, and the total number of required filter cores will be red / green. 1/4 of the total number of state cores. Conversely, to simplify the general rule or hardware design, it is possible to make the blue repeating cell size exactly the same as the red / green repeating cell size. The resulting filter core group will be exactly the same The same (actually four times), but it has exactly the same function as the red and green filter core group. Therefore, the red and green filter core program is adopted and the most necessary to produce the blue bar filter core A suitable modification is to double the numerator of the p: s ratio and change the drawing area to a square instead of a rhombus. Now consider the configuration 20 of Fig. 6 and the blue sampling area 24 of Fig. 9. This is similar In the previous example, because the blue sampling areas 24 are squares:-__ _ __ _-45-tm ^ (CNS) A x 297

Hold

594627 A7 _____ B7 V. Description of the Invention (43) However, because the blue sampling area 1 24 is staggered up or down every other line, that is, half of its height, these calculations are complicated. At first glance, the size of the repeating cell seems to double in the horizontal direction. However, it has been found that the following procedure produces these correct filter cores: (1) Generate a duplicate filter core grid group as described above, as if the blue sampling points were not staggered. Paste the rows and columns in the filter table of the repeating box with numbers, where the starting number is zero; the ending number is "the size of the repeating box minus 1 丨 '. (2) In the even rows of the output image, the The filter in the repeated cell is equally correct. The output γ coordinate selects which column of the core group of the modulo function in the repeated cell size. The χ coordinate selects a row and reports the modulo function in the repeated cell size. Which filter in the selected column is used for the γ coordinate 2. (3) In the odd-numbered output lines such as Sun, before taking the modulo function of the γ coordinate (in the repeated cell size), first Subtract 1 from the Υ-coordinate. Treat the χ-coordinate as if it were an even-numbered row. This will select a correct filter core for the staggered cases in Figure 9. In some cases, perform the modulo functions in advance Calculation and staggering in advance> The core table of the filter is possible. Unfortunately, this is only a repetition for an even cell with an even number =. If the repetitive cell is an odd number of rows, the modulo function arithmetic has Choose those even numbers half the time And the odd number of rows are selected at the other half of the time. Therefore, when using the table, it is necessary to calculate which row should be staggered instead of being calculated in advance. Finally, consider the configuration 20 of FIG. 6 and the blue sampling area 123 of FIG. 8. This Class 丨 X 297 male > 594627

Similar to this previous case,... '' '' A more complex a-sided sampling area. The first step 7% of these τς · edges is how to draw them correctly or how to produce a vector table of f, a, and μ 77 polygons in a computer program. To be more precise, the ', edge opening " are hexagons with the smallest area, but they will not be regular hexagons ^, which can easily be 70 into a geometric proof, as illustrated in Figure 41 Fig. 8. Each side of the spoon and ear opening region 1 23 is wider than the square sampling region 276 by one-eighth. The top and bottom edges of the rectangular sampling region 1 23 are larger than the top and bottom edges of the square sampling region 276, respectively. The bottom edge is narrowed by 1/8. Finally, note that the hexagonal sampling area 123 is the same height as the square sampling area 276.: The core of the filter that produces the hexagonal sampling area 123 in the same geometrical manner as described above, where red and green The filter core is diamond $, or the blue filter core is square. The painted areas are purely hexagons, and the areas where the hexagons overlap with the surrounding input image f are measured. Unfortunately, When using a slightly wider hexagonal sampling area 23, the size of these filter cores sometimes exceeds a 3x3 filter, even when kept within a range of 1: 1 and 1: 2 quantization ratios This is also the case. Analysis shows that if the quantization ratio is between 1.1 and 4: 5 In time, the core size will be 4 × 3. When the quantization ratio is between 4: 5 and 1: 2, the size of the filter core will still be 3x3. Note that because of the hexagonal sampling area 123 and The height of the square sampling area 276 is the same 'so the vertical size of the filter cores remains the same). Designing hardware to give a wider filter core is not as difficult as building a higher filter core, so in When the filter is 4 × 3, it is unreasonable to require a hard system based on the sub-image sketch clearing / quantization system. However, another solution is possible. When the quantization ratio is between 丨: 丨 and ______- 47- The size of the paper and paper is applicable to the Chinese national standard CNS) Α. 丨 Specifications (such as χ 297 gong) 594627 A7 ———--- —__ B7 ___ V. Description of the invention (45) 4: > "Interval, use The square sampling area 124 in FIG. 9 results in a 3x3 filter. When the quantization ratio is between 4: 5 and 丨: 2, then the more accurate hexagonal sampling areas 1 23 in FIG. 8 are used, and A 3x 3 filter is also required. In this way, the hardware is still simpler And it costs less to set up. It is only necessary to build the hardware for a filter core of one size, and the only thing that changes is the general rule for establishing such filters. Like the square sampling area in Figure 9, the figure The hexagonal sampling area of 8 is staggered every other line. The analysis shows that the method of selecting the filter cores described above with respect to FIG. 9 can also be applied to the hexagonal sampling area of FIG. 8. Basically, this This means that the coefficients of the filter cores can be calculated as if the hexagons are not staggered (even if they are always staggered). This makes the calculation of 1¾ etc. simpler and prevents the filter core table from becoming twice as large. In the case of these diamond-shaped areas depicted in Figures 32 to 39, these areas are calculated in a coordinate system designed to make all areas integers for easy calculation. This occasionally results in large The total area and the filter core must be divided by a large value. Sometimes this results in the filter core not being a power of two, which makes the design of the hardware more difficult. In the case of Figure 41, the hexagonal outline of the edge area The extra width of 1 23 would make it necessary to multiply the coefficients of the filter cores by a larger value so that all of these areas are integers. In all of these cases, find a way to restrict the filter cores The divisor of the coefficient is more appropriate. In order to make the design of the hardware easier, it would be advantageous if a divisor of '' 2 can be selected. For example, if all of these When the filter core is designed to be divided by 256, the division can be performed by an 8-bit right shift operation. Choosing 256 also guarantees that all of these fimer cores ________-48-This paper standard is applicable to the country of China 檫Standard (CNS) A4 Specification (Sin) / 297 male hair) '' ~: ------

Hold

594627 -49-V. Description of the invention (Heart coefficients are all in line with 丨, p; ~ points, · a σ wide 8-bit read-only memory (ROM) standard 8-bit value. Therefore, South The 々, 々 program produces a divisor with a suitable divisor; consider the moxibustion nucleus, ° Because the divisor of the larger 卩 乂 乂 Bi Bijia is 256 ', it is used in the following program: (1) Use The i-point operation calculates the area of these filter coefficients. Because the operation is completed in advance offline, it is necessary to add the cost of the hardware using these result tables. (2) Each- The coefficient is divided by the total area of the known delineation area, and then multiplied by 25 ^. If all arithmetic is done with floating-point operations, the filter sum will equal I 256. However, 'more needs to be done Steps to build a table of integers. (3) Perform a dichotomy search to 'when converted to an integer, find the filter that makes the filter, and the rounding point equal to 256 (between 〇 · _ 〇). A dichotomy The law search is a general rule that is well known to me in this industry. If the search is successful, it ends. One two The method search sometimes converges with beneficial methods, which can be detected by testing the circuit with too many executions. (4) If the dichotomy search fails, find a considerable coefficient in the filter core, and Add or subtract a decimal value to force the total sum of the converter to be 2 5 6. (5) In this particular case, a single-value 256 "check the filter. This value is not suitable for an 8-bit bit Tuple table, where the maximum possible value is 2 55. 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 sum is still Equal to 256. Figure 3 1 illustrates the horizontal quantification of an input pixel and an output sub-pixel. The paper's sense of meaning applies to the Chinese National Standard (CNS) Λ4 specification (21〇X 297 public goods). Gutter 594627 A7

At the moment, the S sub-image sketch is detected. The above-mentioned alternative space will be detected and opened. _-50-This paper size applies the Chinese National Standard (CNS) A4 specification (210X: 297 mm>

48594627 A7 B7 V. Description of the invention

The fader is advantageous, but if it is at this quantization ratio, it is switched back to a regular spatial sampling filter for areas that are not described by non-subpixels (also described above). In order to establish this checker, we must first understand what the subpixel-depicted text τ looks like, what its identifiable characteristics are, and what non-subpixel-depicted image sets separate it from. First of all, the color of the pixels on the edges of the black and white sub-sketch sketches is partially unnatural ... that is, R is G. However, the colors on several pixels are natural ... that is, R30. In the case of images or text not depicted by sub-pixels, these two situations will not occur at the same time. If so, let the verifier test the local R off G and R three G 0 on several pixels. Because the sub-pixels depicted on an RGE stripe panel are one-dimensional, they are row by row along the horizontal axis. The test is also one-dimensional. The test is described as follows:

If R, G, and 4 * Gt.i + Gx + G ^ t 4- 〇% Ml

If Rq + U R, +1 + RT < triple Or

If R * “本 ++ + Cloud C3t ·: + G * .i Ben G * + then apply the subspace rendering input instead of spatial filter. Otherwise apply the general spatial filter. When the text is colored, then these R x = There will be a correlation between the red and green elements in the form of a G x, where na " is a constant. In the case of black and white text, T '' 'a " is equal to 1. This test can be extended to detect Colored and accented text: -51-This paper size is applicable to the Chinese family standard (CMS): \ 4 specifications (21 × x 7 public love) 594627 A7 ______B7 5. Description of the invention (~) ^ ~ '~'

If Rt Φ aGt and tf K, 4 · H + R "2 a a (G ,, G '丨 + G〆 G, — G,‘ 2)

Or

If + R〆R “t this RXA2 sa (Gm this G Xia Gx +) and then apply the subspace drawing to input the alternative spatial filter. Otherwise apply the general spatial filters RX and G x to represent the red and green components. ,, Χ,, The value on the pixel row coordinates. There may be a valve test to determine whether R Ξ G is tight enough. You can adjust this value to achieve the best result. You can adjust the length of these terms and the magnitude of the test to achieve The best result, but it usually follows the above form. Figure 27 illustrates an arrangement in which a two-primary-color pixel element is arranged in an array of three panels in a display element according to another embodiment. 28 illustrates the arrangement of the blue emitter pixel elements in an array of the device of FIG. 27. FIG. 29 illustrates the arrangement of the green emitter pixel elements in an array of the device of FIG. 27. FIG. 30 illustrates the arrangement of FIG. 27 The arrangement of the red emitter pixel elements in an array of the device. The arrangement and arrangement are beneficial to a projector based on the display using three panels, of which two The plates each serve as the three primary colors of red, green and blue, in which the images of each of these panels are combined to be projected on a screen. The emitter configurations and shapes are well matched to those shown in Figures 8, 13, and 14, It is the sampling area of the configuration shown in Figure 6. If so, the graphic generation, conversion equation calculation, and data format disclosed here for the configuration of Figure 6 will also work on the three-panel configuration of Figure 27. ___- 52- This paper size applies the Chinese National Standard (CNS) A4 specification (210 X 297 mm) 594627 A7 B7 V. Description of the invention (50. For the quantization ratio of approximately 2: 3 above and the higher quantization ratio, the Pe The resampled data set of the sub-pixel rendering in the η T i I τ M matrix sub-pixel configuration is more effective in representing the generated image. If you want to display an image to be stored and / or transmitted in one? ^ 丁 丨41 ^ on display, and if the quantization ratio is 2: 3 or higher, it is advantageous to perform the resampling to save in memory storage space and / or bandwidth before storing and / or transmitting the image The one that has been resampled The image is called "pre-painted." If so, the pre-painting serves as an effective lossless compression rule. And the advantage of the present invention is that it can obtain most of the stored image image on any actual color sub-pixel configuration. When the present invention has been described in terms of a representative specific embodiment, it should be understood that various changes can be made to the present invention and equivalent elements can be substituted for the elements of the present invention without departing from the scope of the present invention. AA ^ ^ _ In addition, the present invention can be made many, P, seven, L, people, and overnight U under the basic Qiantian Temple that does not depart from the month of the present month, so as to make a special situation or to anticipate the temple's teachings. Therefore, ^ V, ^ / 4. It is hoped that the present invention is limited to the specific embodiment for carrying out the present invention, and the present invention is intended to include all of the specific embodiments. ^ The scope of the patent is within the scope of this paper. The paper size is applicable to the Chinese National Standard (CNS.) Λ4 specification (21〇;

Claims (1)

〇9110% 77 Patent Application Application Patent Range Replacement (April, 1993) The scope of the patent application is more than 1 type, which converts the original pixel data of the first format with a plurality of three primary color pixel elements. A method for displaying the original pixel data in a second format with a plurality of three primary color pixel elements in a second format, including determining a value of each data point of each color of the original pixel data in the first format Implied sampling areas; the resampling area of each emitter of each color of the declension family; forming a set of scores for each of the resampling areas, where the denominator of the scores is the resampling area _ And the numerator of the scores are a function of an area in each of the implicit sampling areas that at least partially overlaps with the resampling areas; multiplying each of the resampling areas with their respective scores by the each- Imply the data values of the sampling area to generate a-product; and add each of the products to obtain the illuminance value of each of the resampled areasIncluding the first format, including the first format, including the first format, and driving with the illuminance values 2. The number of components of the method such as the application of the scope of patent application No. 1 In the display hardware 3. If the method of applying for the scope of patent application item 1 is stored in the display software 4. If the method of applying for the scope of patent application item 1 is stored in the display body. 5 · The method as claimed in item 1 of the patent scope This display. 6. The method of patent model, if requested, 'includes the storage of these illuminances I. The paper size applies the Chinese National Standard (CNS) A4 specification (210X 297 public directors)' '* ----- 594627 VI. Application scope value. It also includes the method of transmitting these illuminances such as the scope of patent application item 1. s. According to the item in the scope of the patent application, the number of the three primary color pixel elements in the horizontal and vertical formats of the format W is two-half of this display format. 9. The method according to page ^ of the scope of patent application, wherein the display is selected from the group consisting of a group LCD display, a subtraction display, an organic photoelectrode device, an electrophoretic display, a field emitter display, and a discrete photoelectrode display. For electric panel display, cold light (EL) display, projector, and cathode ray tube (CRT) display. 10 · kinds of data-a method of converting the original pixel data of the first format to display the original pixel data in a second format having a plurality of three primary color pixel elements, including determining the original of the first format Implied sampling area of each data point of each color in pixel data; determination of resampling area of each emitter of each color of the display; determination of each resampling area and the implicit sampling area A percentage of overlapping regions to obtain data values; multiply the parent® region percentage by the data values of each of the implicit sampling regions to produce a product; and add each of the products to obtain the each Illumination value of a resampled area. Including the first format, including the first format, including the first format, driving with the illuminance values, including storing the illuminance, including transmitting the illuminance for 1 h. In the method of item 10, the number of elements is stored in the display hardware. 12. According to the method of claim 10, the number of children is stored in the display software. 13. According to the method of claim 10, the number of children is stored in the display carcass. 14. If you want to apply for the method in item 10 of the patent scope, move the display. 15 'as the method value of item 10 of the patent scope. 16. If the value of the method of the scope of the patent application is 0, the value is 17. 17. The method of the scope of the 10th category, in which the first format is used. Two exchanges: two primary pixels in the two directions of water and vertical direction . The number is one-half of the two display formats of the child. Team: The method of applying for item 10 of the patent scope, wherein the display is selected from the group consisting of-!.: LCD, subtraction display, electric panel display 7 =, EL display, organic LED (0LED) ) Monitors, headlights, field emitter displays, discrete photopole displays, projectors, I cathode-ray tube (CRT) displays. 19. · A method to determine each of the original pixel data in a -first format, the hidden sampling area of each data point of the color, so that the original pixel data is displayed in a second format with a plurality of three primary color pixel elements The method includes: -3-1 Yindi 4¾ ,, ... 〇, the two primary color pixel elements, a geometric center of each emitter 'to define a sampling point; each of the eight is defined by a line ^ ^ ^ ^ Φ S3 sampling area, which is located in the same color: the geometric center of the emitter of the Xin-4 element and the adjacent three original elements:; = its: the use of equal distances between the geometric centers of the emitters of the same color Wait for the lines to form a grid. 2 0. If the scope of the patent application is No.9, No.9, the grid points of 3 lines are arranged in a tile pattern. 21. t = The method of item 20 of the patent scope, wherein the tile pattern is selected from the group consisting of square, rectangle, triangle, hexagon, octagon, rhombus, staggered square, staggered rectangle, staggered triangle , Qing_ tile, oblique, distorted oblique, and staggered rhombus. 22. A method to determine each color of a first format of the original pixel data "resampled area of each data point, so that the original pixel data is displayed in a second format with a plurality of primary color pixel elements The method includes determining a geometry of each emitter of each of the three primary color pixel elements, and defining each resampling area by a line, wherein the geometric center of the emitter of a three primary color pixel element and the An adjacent three primary color pixel element has the same distance between the geometric centers of the emitters of the same color; and 61 uses these lines to form a grid point. 594627 AB c D tV 4-Application scope 23 The method of item 22, wherein a grid of the lines is placed with a tile. A 24. The method of item 23 of the scope of patent application, wherein the tile pattern is selected from the group consisting of: square, rectangular , Triangle, hexagon, octagon, rhombus, staggered square, staggered rectangle, staggered triangle, penrose tile, oblique square, distorted Oblique squares and staggered rhombuses 25. — A method to limit the filter core divisor in a filter core to a value that is used to simplify hardware execution, including: Calculating filter coefficients using floating-point arithmetic Area; divide each filter coefficient by the total area of a drawing area to obtain a first result; multiply a first divisor by the first result to generate a filter sum; complete a dichotomy search To obtain a rounding point for the total of the filter; and convert the total of the filter to an integer. 26. For the method of applying for a scope item of the patent, the divisor is ^ 27. The method of 25 items still includes using a coefficient in the core of the filter; and adding a value to force the total of the filter equal to the divisor. 28. The method of item 25 of the patent application also includes using the filter A coefficient in the core of the filter; and subtracting a value to force the total of the filter to equal the divisor. 29.-A method of operating a spatial sampling filter, including: detecting sub-pixels The painted area; -5-- present paper suitable silver China National Standard Scale (CNS) A4 size (21〇Χ297 the public) 594627 Open a first set of spatial sampling filters to the areas depicted by the sub-pixels in response to the areas depicted by the detected sub-pixels; detect areas not depicted by the sub-pixels; and open a second set of spatial sampling The filter maps the non-subpixel regions' in response to the detected non-subpixel regions. 30. The method according to item 29 of the patent application, wherein the areas drawn by the detected sub-pixels include and υυΐΙχπχ + ι + Ruler ... = + + +; and R is a red value, and G is a The green value, and χ is a pixel position in a column of pixels. 31. The method according to item 29 of the scope of patent application, wherein the areas drawn by the detected sub-pixels include Rx # 1 and Rx • + Rx + Rx + l + R ^ p I ″; and R is A red value, G is a green value, and X is a pixel position in a column of pixels. 32. The method according to item 29 of the scope of patent application, wherein the areas depicted by the detected sub-pixels include Rx ^ (^ and! ^ _ 2 + 1 ^ _1 +] ^ + 1 ^ ”+ 1102 sVGxj + Gx-i + Gx + Gx + i + Gx + J; and R is a red value, G is a green value, a is a constant, and χ is a pixel position in a column of pixels. Method, where the detected sub-picture sketch edges include Rx # aGx and Rx_i + Rx + rx + i + Rx + 2 three a (Gj ^ + Gx-j + Gx + Gx + i), and Where R is a red value, g is a green value, a is a constant, and x is a pixel position in a column of pixels. -6-This paper size applies to China National Standard (CNS) A4 (210 X 297 mm) ) A8 B8 34. The kind of brain can be described as a recording medium, which is specifically compiled into a set of instructions that can be executed by a machine, wherein the set of instructions is used to perform a conversion of original pixel data in a first format with a plurality of three primary color pixel elements, A method for causing the original pixel data to be displayed in a second format having a plurality of three primary color pixel elements, the method comprising: determining the data value of each data point of each color of the original pixel data in the first format Implicit sampling area of the form; determine the resampling area of each emitter of each color in the display, and form a set of scores for each resampling area, where the denominator of the score is the resampling area A function, and the numerator of the scores are a function of the regions in each of the implicit sampling regions that at least partially overlap the resampling regions; multiplying each of the resampling regions with their respective scores Imply the data value of the sampling area to produce a product; and add each product to obtain each Resample the area's illuminance value. 35. If the computer-readable recording medium in item 34 of the scope of patent application includes the storage of the component number of the first format in the display hardware. 36. If the computer-readable green media of item 34 in the scope of patent application includes the storage of the component number of the first format in the display software. 37. If the computer-readable recording medium in item 34 of the scope of patent application includes the storage of the component number of the first format in the display body. 38. If the computer-readable recording medium in item 34 of the scope of patent application, including this paper size, the Chinese National Standard (CNS) A4 specification (210 X 297 mm) is applicable 594627 The display is driven with these illuminance values. 39. If the computer-readable recording medium in item 34 of the scope of patent application includes the storage of these illuminance values. 40. If the computer-readable recording medium in item 34 of the scope of patent application includes the transmission of these illuminance values. 41. For example, the computer-readable recording medium of item 34 of the scope of patent application, wherein the number of elements of the three primary colors of the data in the first format in the horizontal direction and the vertical direction is bisected by the second display format. one. 42. If the computer-readable recording medium of item 34 of the scope of patent application, the display is selected from the group consisting of a liquid crystal display, a subtraction display, an organic light emitting diode (OLED) display, an electrophoretic display, a field Emitter displays, discrete photopole displays, plasma panel displays, cold light (EL) displays, projectors, and cathode ray tube (CRT) displays. -8-This paper size applies to Chinese National Standard (CNS) A4 (210 X 297 mm)