TW558899B - Methods and systems for improving display resolution in images using sub-pixel sampling and visual error filtering - Google Patents

Abstract

The present invention provides a method and an apparatus for converting a color image of a higher-resolution to an image of a lower-resolution. With the method of the present invention, based on characteristics of spatial frequency responses of the human visual system, a lower-resolution image (116) is formed by separating a higher-resolution image (72) into a luminance channel (76) and chrominance channels (98 and 100) so as to carry out suitable sampling and filtering for each of the luminance channel (76) and the chrominance channels (98 and 100), then by combing the luminance channel (86) and the chrominance channels (111 and 114) after the sampling. As to the chrominance channels (98 and 100), traditional sub-sampling (101 and 103) is carried out to avoid chromatic aliasing, while sub-pixel sampling (80) is performed for the luminance channel (76) to improve resolution of luminance component.

Description

558899 A7 B7 V. Description of the invention (1) Field of the invention The specific embodiments of the present invention relate to the field of displaying high-resolution images using lower resolutions, where the display uses three primary color arrangements to display R, G, and B or other components of the image (Component). For example, this three-primary-color arrangement is found in intuitive LCD displays' and in this arrangement, a single pixel is composed of three side-by-side sub-pixels. Each sub-pixel controls only one of the three primary colors (ie, red (R), green (G), and blue (B)), and is then controlled by only the three primary colors represented by the digital image. High-resolution images can be obtained in memory or directly from an algorithm (vector graphics, certain glyph designs, and computer graphics). The subject matter of this patent application is related to the U.S. patent application title "Methods and Systems for Improving Display Resolution using Sub-Pixel Sampling and Visual Error Compensation" invented by Scott Daly on December 12, 2000, and is granted a U.S. patent application number 09 / 735,454. This U.S. patent application is incorporated herein by reference. The subject matter of this patent application is also related to the US patent application titled "Methods and Systems for Improving Display Resolution in achromatic Images using Sub-Pixel Sampling and Visual Error Filtering" invented by Rajesh Reddy K Kovvuri and Scott Daly on December 12, 2000, And granted U.S. Patent Application No. 09 / 735,425. This U.S. patent application is incorporated herein by reference. BACKGROUND OF THE INVENTION The method most commonly used in the field of displaying high-resolution images on lower-resolution displays is to down-sample pixel 2 of the high-resolution image 4 to a low resolution. -5- This paper scale applies the Chinese National Standard (CNS ) A4 size (210 X 297 mm) binding

558899 A7 _ B7 V. Description of the invention (2) The resolution of the resolution display 6 is shown in FIG. 1. Then, the R, G, and B values of each down-sampled color pixel 8 are mapped to the R, G, and B elements 10, 12, and 14 of each display pixel 16. The R, G, and B elements 10, 12, and 14 of a display pixel are also called sub-pixels. Because the display device does not allow overlapping color elements, the sub-pixel can only represent one of the three colors of R, G, or B. However, the amplitude of the color can be changed throughout the grayscale range (for example, 0 to 255). The sub-pixels usually have a 1: 1 aspect ratio (width: height), so the resulting pixel 16 is a square. Sub-sampling / mapping techniques do not take into account the fact that the R, 0, and B sub-pixels of the display are spatially shifted; in fact, these techniques assume that these sub-pixels are overlapped in the same way as in the high-resolution image shown in Figure 1. . This type of sampling can be referred to as subsampling or traditional subsampling. The pixels of the high-resolution image 4 are presented as three slightly offset stacked squares 8 to indicate the correlation of rgb values at the same spatial location (ie, pixels). A display pixel 16 is composed of R, G, and B sub-pixels 10, 12, and 14, and a black line is used as a part of the lower-resolution tri-primary display 6 in FIG. The other display pixels are indicated by light-colored dotted lines. In this example, the high-resolution image has a resolution (horizontal and vertical dimensions) that is more than 3 times higher than that of the display. Since this direct subsampling technique can cause aliasing artifacts, various methods are used, such as averaging adjacent unsampled pixels and sampled pixels. Note that the common technique of averaging adjacent elements when subsampling is mathematically equivalent to using a rectangular (rect) filter in advance; consider wave high-resolution images. Also, please note that selecting pixels other than the leftmost pixel (as shown) can be considered as pre-filtering that only affects the phase. Therefore, most of the treatments related to the prevention of aliasing are handled by -6 _ This paper size applies the Chinese National Standard (CNS) A4 specification (210X 297 mm) " 558899 A7 B7 V. Description of the invention (3) For filtering operations on high-resolution images, even if the kernel is only applicable to the sampled pixel positions. As defined in this specification and the scope of patent applications, achromatic images have no visible color change. Achromatic conditions occur when the image contains only one layer or color channel, or when the image has multiple layers or color channels, but each color layer is exactly the same to produce a single color image. It should be understood that the aforementioned techniques do not take advantage of possible display resolutions. For background information in this area, see R. Fiegenbiatt (1989) "Full color imaging on amplitude color mosaic displays" Proc. SPIE V · 1075, 199-205; and J. Kranz and L. Silverstein (1990) "Color matrix display image quality: The effects of luminance and spatial sampling ", these documents (29-32) are incorporated herein by reference. For example, in the display shown in FIG. 1, when the resolution of the display pixel 16 is 1/3 of the high-resolution image (source image) 4, the resolution of the sub-pixels 10, 12 and 14 is equal to the resolution of the source image. Degree (horizontal dimension). If the display is intended for color blind people only, the spatial position of the sub-pixels may be used. Figure 2 shows this approach, in which the r, G * B sub-pixels 10, 12 and 14 of the display are obtained from the corresponding colors of the different pixels Π, 13 and 15 of the high-resolution image. This allows the final horizontal resolution to become sub-pixel resolution, which is three times the display pixel resolution. But what about 疋 ’for display viewers who are not colorblind? That is, most viewers. Fortunately for display designers, even this paper size is applicable to the gg 豕 standard (CNS) μ specification (⑽X tear public love) A7 B7 V. Description of the invention (4) Viewers with ideal color 5 vision are the highest space Color blindness in frequency. As shown in Figure 3, the figure shows the idealized spatial frequency response of the human vision system. As shown in the figure 7JT, the degree 1 7 represents the achromatic contrast of the viewing image, and the chromaticity 19 represents the I in the color. The vision system processes it to equal brightness from red to green and blue to yellow. ) Modulation. The R-G and B-Y of the video color difference signal are approximately about these modulations. For most viewers, I, the chrominance frequency response bandwidth is 1/2 of the luminance frequency response bandwidth. Sometimes the bandwidth of the blue-to-yellow adjustment response is lower than about 1/3 of the brightness. As shown in FIG. 2, the sampling including mapping the color elements from different image pixels to the sub-pixels of the two primary colors of the display pixel can be referred to as sub-pixel sampling. Please refer to FIG. 4 'In the horizontal direction of the display, the frequency range is between Nyquist (display pixel = three primary color pixels of display pixel 16), assuming that three primary color Nyquist is 0.5 cycles and three sub-pixels Nyquist frequencies of 10, 12, and 14 (0.5 cycles per pixel = ι · 5 cycles / triad pixels). The rectangular area 20 shown in Fig. 4 indicates this frequency range. The light-dotted dotted curve 22 in the figure indicates the sine function generated by convolving a high-resolution image using a rect function with a width equal to the display sampling space. When the display is an LCD, this is the most common approach used to model a display MIF (modulation transfer function). The dashed curve 24 in the figure indicates a sine function generated by using a rect convolving high-resolution source image equal to sub-pixel space, which has a higher frequency. This is because the display considers the sub-pixel to be a rect in a one-dimensional dimension.

Due to the limit. As shown in the rectangular area 20, the sub-pixels can display brightness information, but do not display color information. In fact, any chroma information in this block is obfuscated. Because Λ, in this block, by allowing chrominance obfuscation ': a higher frequency luminance signal than that allowed by the three primary color (i.e., display) pixels can be achieved. This is a "benefit" block provided by using sub-pixel sampling. For fusiform display applications, black & white glyphs are usually processed in advance, as shown in Figure 5. Standard preprocessing includes micro-compression, which means that the glyph strokes centered on the pixel are concentrated, that is, the phase shift specific to the glyph strokes. I is usually followed by low-pass filtering, also known as greyscale antialiasing. Binding

Figure 3 shows the idealized visual frequency response (CSF). In fact, the visual frequency response has a limited decreasing slope, as shown in Figure 6A. The brightness CSF 30 has been mapped from the cy / deg unit to the display pixel domain (assuming a viewing distance of 1280 pixels). As shown in the figure, the solid line 30 has a maximum frequency close to 1.5 cycles / pixel (Cy / pixei; display pixel), and the passband shape has a peak close to 0.2 cycle / pixel three primary colors (cy / pixehriad). The dotted line indicates R: G CSF 32, which is a low frequency band with a maximum frequency close to 0.5 cycles / pixel (cy / pixel). The dotted virtual lpf curve indicates β: Y modulation CSF 34, whose maximum frequency is similar to r: g CSF, but with a lower maximum response. The range between the chrominance CSF 32 and 34 and the cutoff frequency of the luminance CSF 30 is a block that allows chrominance confusion to improve the luminance bandwidth. Figure 6A also shows the idealized image power spectrum 36 as a 1 / f function, as shown by a straight line with a slope of -1 (due to the use of the log axis in the figure). This frequency repeats at the sampling frequency. The pixels 38 and sub-pixels 40 which are repeatedly present in the horizontal direction are taken as -9.-This paper size applies the Chinese National Standard (CNS) A4 specification (210 X 297 mm) 558899 A7 B7. 5. Description of the invention (6 sample rate. The repetitive spectrum that occurs at low frequency 38 is due to pixel sampling, and the repetitive spectrum that occurs at high frequency 40 is due to sub-pixel sampling. Please note that the shape changes due to plotting on the log frequency axis. These The frequency of the repetitive spectrum extending below the lower frequency of Nyquist is called aliasing. The leftmost is the chromaticity confusion caused by the pixel sampling rate 38 'and the confusion 40 and the higher subpixel sampling rate Relevant, so will occur at higher frequencies. In Figure 6 A, the source spectrum has not been pre-filtered. Therefore, the aliasing caused by pixel sampling (ie, chromaticity aliasing) extends to a very low frequency 35. Therefore, Even if the chromaticity CSF has the bandwidth of the low-luminance CSF, the color epigenesis is still visible (depending on the noise and contrast of the display). In Figure 6B, the source power spectrum has been pre-filtered (equal to Source image pixel rect function), as shown by the dotted line 22 in Figure 4, and it can be found that it will affect the base-band spectrum 42 through 0_5 cycles / pixel (cy / pixei), resulting in it has a steeper than -1 The slope of is as indicated by 44. Repeating also shows the effect of this pre-filtering. Even using this filtering, we can find that some kind of chromaticity confusion occurs at the frequency 46 below the cut-off frequency of chromaticity CSF 32a and 34a (located in the more Low-frequency repetitive spectrum) ^ Therefore, it can be found that the simple type of luminance pre-filtering has a difficult time that is difficult to remove chrominance confusion 'and cannot be removed by 0 · 5 cycles / pixel (cy / pixel) (that is, the "advantage" area Block) of all brightness frequencies. Since we rely on the visual system bandwidth difference as a function of brightness or chrominance to provide brightness bandwidth improvement in the "Benefit Block" 20, one possibility is to pre-filter based on the visual system model design Please refer to c -10- for relevant instructions. This paper size applies Chinese National Standard (CNS) A4 specifications (210 X 297 mm) 558899 A7 B7 V. Description of invention (7)

Betrisey et al. (2000) "Displaced filtering for patterned displays" SID Symposium digest 296-299, which is incorporated herein by reference, as shown in Figure 7. This technique conceptually depends on which color layer and which color sub-pixel of the image to sample to use different pre-filters. Therefore, there are 9 filters. They have been designed using a human parallax model. For a description, please refer to the work by X · Zhang and B · Wandell (1996) "A spatial extension of CIELab for digital color image reproduction" SID Symp. Digest 73 1-734, which is mentioned by reference Incorporated herein, as shown in Figure 7. This is done offline, assuming the image must be black & white. In the final implementation, the rect function is used to store the calculation results instead of using the resulting filter. In addition, some residual chrominance errors still occur because chrominance aliasing extends to frequencies below the chrominance CSF cutoff frequency (as shown in Figure 6B). However, the visual model used does not take into account the occlusion characteristics of the visual system, which can cause chromaticity to be obscured by luminance when the luminance is at a medium or high contrast level. Therefore, in larger glyphs, the chroma epigenesis at the edges of the glyphs is obscured by the brightness contrast of the glyphs. However, as the size of the glyph is reduced, the brightness of the glyph decreases, and then the epigenetic phenomenon of the same chromaticity becomes very obvious (for example, in very small glyphs, the black / white part of the glyph disappears, leaving only partial color Point (color speckle)). Summary of invention

The design of the present invention is based on the aforementioned specificity of the spatial frequency response of the human visual system, in other words, the brightness CSF is higher than the chromaticity CSF -11-This paper standard is applicable to the Chinese National Standard (CNS) A4 specification (210 X 297 mm) 558899 A7

= Stop Frequency Basis. The method of forming the low-resolution image of the present invention is to divide the high-resolution image into _ luminance data and a chroma data to extract and perform sampling and filtering suitable for the luminance data and the chroma data, ... Combing the luma data and the chroma data after the foot ° As far as the chroma data is concerned, 'the traditional subsampling will be performed chromatic chromatic auasing' and the subpixel sampling will be performed on the luma data at the same time ' To improve the brightness component resolution. In addition, high-pass filtering is performed on the luminance data sampled by the sub-pixel to remove the low-frequency epigenetic phenomenon that occurs during the luminance data sampled by the sub-pixel. Hereinafter, the concept of the present invention will be explained. In Figures 丨 and 2, the R, G, and B values of the subpixels 10, 12, and 14 of the R'G * B subpixels 10, 12, and 14 shown in Fig. 1 (subsampling) reflect the high resolution image 4 The respective R, G, and B values of color pixel 8 (ie, u) "However, the respective R, G, and B values of sub-pixels 10, 12, and 14 are not exactly the same as the respective R, 6 of color pixel 8 (11) And 6. For the luminance component, the sub-pixel sampling shown in FIG. 2 is performed so that the r, g, and B values of the sub-pixels 10, 12, and 14 reflect the luminance components of the color pixels 11, 13, and 15, respectively. The specific embodiments of the present invention include methods and systems that are less dependent on filtering, and are assumed to be linear and capable of processing input color images. These specific embodiments can directly remove low-frequency chroma epigenetics caused by sub-pixel sampling. These are generated by The LPF version of the image chroma content is completed by adding to the luma and chroma obfuscation version. This is to remove the color epigenesis caused by sub-pixel sampling by using a color gamut other than the additive color gamut (ie, RGB). Phenomenal way is done. In fact, only lower frequency colors need to be eliminated -12- This paper applies the scale of Chinese National Standard (CNS) A4 size (210X 297 public attack :)

Binding

558899 A7 B7 V. Description of the invention (9) This phenomenon is due to the low bandwidth of the chromaticity CSF, so no high frequency chromaticity epigenesis can be seen, as shown in Figure 6A. The method and system of the present invention can be used to obtain a higher-resolution luminance signal without significant chromaticity confusion, which is suitable for viewing the display at a distance farther than the design specification. These techniques do not require the source image to be text or the image to be achromatic. A specific embodiment of the present invention converts a higher-resolution image into a lower-resolution image, and reduces errors caused by the sub-sampling processing program. When higher resolution color images are not in a format that allows separation of luma and chroma data, the image is converted to such a format. Many contrast color domains can be divided into beta contrast color domains, thereby separating the luminance channel from the chrominance channel to allow separate processing. Then, the brightness channel is converted into an additive color domain (ACD), such as RGB, and the ACD brightness image is sampled by sub-pixels to save the brightness data when the resolution is reduced. After sub-pixel sampling, the image after sub-pixel sampling (SPS) is converted back to the contrast color domain (OCD), and the separated luminance and chrominance channels are again segmented. Then, the SPS chrominance channel generated by this segmentation process is subjected to high-pass filtering to remove the low-frequency chrominance epigenesis generated during the sub-pixel sampling period. The SPS brightness channel is usually not modified to save the original brightness data. Then, the chroma channel from the original image is low-pass filtered, and then sub-sampled to provide chroma data for the lower resolution image. Then, the low-pass filtered chroma channels are combined with the high-pass filtered, sub-pixel sampled chroma channels established from the original luminance channel. -13- These paper sizes are also applicable to China National Standard (CNS) A4 (210 X 297 mm) 558899 A7

Chroma is usually combined with the SpS luminance channel to form a reduced-error, lower-resolution image, usually a contrasting color gamut. The reduced error, lower resolution shirt image can then be converted to add color gamut or other color gamut compatible with the desired application. Schematic illustration

In order to obtain the methods listed above and other advantages and objectives of the present invention, a more detailed description of the present invention briefly described above will be presented with reference to specific embodiments of the present invention as shown in the accompanying drawings. It should be understood that these drawings depict only typical specific embodiments of the present invention, and therefore should not be considered as limiting the present invention. Additional details and details of the present invention will be described and illustrated with the accompanying drawings, in which: Figure 1 Display traditional image sampling for display with three primary color pixel configurations, order

Figure 2 shows sub-pixel image sampling suitable for a display with three primary color pixel configurations; Figure 3 shows the idealized cSi ^ 々 pattern mapped to a digital frequency plot; Figure 4 shows the Nyquist analysis of pixels and the sub-pixels that mark the merits Nyquist block; Figure 5 shows a typical pre-processing technique; Figure 6 A shows a diagram using analysis of a 1 / f power spectrum repeated at pixel sampling and sub-pixel sampling frequency; Figure 6B shows a modification using pre-processing Schematic diagram of 1 / f power spectrum analysis of pixel sampling and sub-pixel sampling frequency repetition; Figure 7 shows a block diagram of known uses of visual models; Figure 8 shows a flowchart of a specific embodiment of the present invention; -14- this paper scale Applicable to China National Standard (CNS) A4 (210 X 297 mm) 558899

Fig. 9 shows a flowchart of a specific embodiment of the invention; and Fig. 10 shows a diagram of signals retained by a specific embodiment of the invention. Modes for Carrying Out the Invention By referring to the drawings, one can better understand the presently preferred embodiments of the present invention, wherein similar parts throughout the drawings are indicated by similar numbers. The illustrations listed earlier are explicitly incorporated into this detailed description. As broadly described herein and illustrated in the drawings, it will be readily apparent that the components of the present invention can be arranged and designed in a variety of different configurations. Therefore, the following more detailed description of the specific embodiments of the method and system of the present invention is not intended to limit the scope of the present invention, but merely represents the presently preferred specific embodiments of the present invention. Hardware, firmware, or software may be used to embody the elements of a particular embodiment of the invention. In addition, software can be stored in a computer-readable medium, in this way, software (program) embodying functions of specific embodiments of the present invention is provided. In addition to recording media such as information storage devices (semiconductor memory, floppy disks, hard drives, etc.) and optical storage devices (CD-ROM, DVD, etc.), examples of such media include computer networks (Broadband network (Lan), Internet, and other wide area network (WAN) and wireless communication networks) Communication media (optical fiber, wireless communication lines, etc.) used in the system The function of the specific embodiment is realized as a computer signal embodied in electronic transmission. Although the exemplary embodiments disclosed herein only illustrate one form of these forms', those skilled in the art should know that it is possible to implement these elements in any of these forms, which are all examples of the present invention. -15- This paper size applies Chinese National Standard (CNS) A4 specification (210X 297 mm) 558899 A7 B7 V. Description of the invention (12) The term "achromatic" will be used to explain and propose specific embodiments of the present invention. When used with the images in this specification and the drawings, the term "achromatic" means an image that has no visible color change. An achromatic image is an image that includes only one layer or color channel, or an image that has multiple layers or color channels, but each color layer is exactly the same to produce a single color image. Specific embodiments of the present invention will be explained and proposed with reference to the "RGB" image or color gamut, or "additive color domain" or "additive color image" of the art. In the scope of this specification and the patents of related applications, these terms may represent any form of multi-component image domain, which includes, but is not limited to, various RGB color gamuts and CMYK color gamuts. Reference will also be made to the "YCrCb" image or color gamut, "opp0nent color" color gamut, image, or channel, or "color difference" color gamut, image, or channel to explain and propose specific embodiments of the present invention . In the scope of this specification and related patent applications, these terms can represent any form of multi-component image domains with different luminance channels and chrominance channels including, but not limited to, YCrCb, LAB, YUV, and YIQ color gamuts. The specific embodiment of the present invention can be used to convert a higher-resolution image into a lower-resolution image, and there are fewer visible errors in the converted interface. Although these specific embodiments are often used with display devices to down-convert an image with a resolution higher than that of the display to a resolution available to the display, it is also applicable to other applications. The images converted by the specific embodiments of the present invention may exist in various formats. -16- This paper size applies to China National Standard (CNS) A4 (210 X 297 mm) binding

Line 558899 A7 B7 V. Description of the invention (13) When these formats are not compatible with the processing procedure of the specific embodiment of the present invention, the image is converted into a compatible format before processing, and after processing, as needed Convert the image back to the original format. A specific embodiment of the present invention can be explained with reference to FIG. 8, which shows a diagram outlining a specific embodiment. This processing program 70 starts with an image existing in an contrast color gamut (OCD), such as YCrCb, LAB, YUV, YIQ, or a similar color gamut. When an image exists in an "additive color domain" (ACD) such as the RGB or CMYK color gamut or some other color space, the image can be converted into a contrast color gamut before using the invention. The embodiment processes the converted image. Some embodiments include steps to convert the image to a compatible format before processing. Once the image belongs to the contrast color gamut 72 and has separate luminance and chrominance channels, the image will be "segmented" (step 74) to separately process the luminance and chrominance channels. "Segmentation" (step 74) may include a method of sampling or filtering the original OCD image 72, or isolating the luminance data or chrominance data from the original image 72. Segmentation can also include image conversion. After segmentation, the initial luminance channel 76 is converted (step 78) into an ACD luminance image, such as an RGB image. This is to enable the luminance image to be sampled in the format or color gamut where the luminance image is ultimately displayed. Once the luminance image has been converted (step 78), sub-pixel sampling is performed on the image (step 80) to improve the resolution of the lower-resolution image produced. In this method, luminance data from each successive pixel in the original higher-resolution image is assigned to each corresponding sub-pixel in the lower-resolution image. -17- This paper size applies the Chinese National Standard (CNS) A4 specification (210 X 297 mm) 558899 A7 B7 V. Description of the invention (14) When sub-pixel sampling (step 80) is completed, sub-pixel sampling (SPS) ) Brightness image. This SPS brightness image is converted (step 82) into an OCD image, which can be referred to as an SPS-OCD brightness image. This conversion is performed to allow further segmentation (step 84) of the SPS luminance image into separate luminance and chrominance channels. The SPS brightness channel 86 is usually maintained as it is, and the other channels are directly combined (step 88). However, the SPS chrominance channels 90 & 92 are filtered before being further combined. These SPS chroma channels 90 & 92 can be divided into a red-to-green channel 90 and a blue-to-yellow channel 92. These channels usually include the Cr and Cb channels of the YCrCb image, the "a" and "b" channels of the LAB image, the U and V channels of the YUV image, the I and Q channels of the YIQ image, or similar channels in the color gamut of other color spaces. Then, the chroma channel 90 & 92 is high-pass filtered (step 94 & 96) to remove the low-frequency epigenetic phenomenon occurring during the sub-pixel sampling. In some embodiments of the present invention, an unsharp mask method is used to perform high-pass filtering (steps 94 & 96). The halo mask can use a low-pass core. The low-pass core that produces the low-pass version of the image is usually used to process the original image. The original unfiltered image is then subtracted from this low-pass version while the average value of the image is saved. Successful embodiments have used Gaussian low-pass cores, where the sigma is about 0.3 pixels to 0.8 pixels. A sigma value of 0.6 pixels is considered particularly successful and produces a cut-off frequency in the frequency domain of about 0.168 cycles / pixel. This provides a very ideal halo mask filter. Instructions for deriving the Gaussian core are provided below. -18- This paper size applies to China National Standard (CNS) A4 (210 X 297 mm) binding

558899 A7 B7 V. Description of the invention (15 In some specific embodiments, the one-dimensional Gaussian function shown below is used .... = -4- ^ 2 / 2σ2 μ = 0 (1) Fourier transform of this function It is shown as follows: = (2) It can be found here that σ (pixel unit) in the spatial domain is equivalent to 1 / π2σ (Jewel / pixel unit) in the frequency domain. This relationship helps to know The cutoff frequency of the filter is determined in the case of ， and conversely, the space σ of the halo mask is determined under the known frequency, which can be guided by the CSF model. In some specific embodiments, the following two are used Gaussian function: Body loading (X, less) = —1 c 12σχ ΐσΐ) 2πσ, less σ

(3) Since the Gaussian function is a separable Cartesian, when considering the importance of σ, the frequency response of the two-dimensional Gaussian function is similar to equation (2). That is, ~ in the time domain is 1 / π2σχ in the frequency domain, and ^ in the time domain is 1 / π2σγ in the frequency domain. A successful embodiment of the present invention has been implemented using a 3 × 3 core size. Gaussian halo mask filter. The selected value is 0 · 6 ', which causes the cut-off frequency of the low-pass / filter to be about 0.2 cycles / pixel. Order

k -19- This paper size applies the Chinese National Standard (CMS) A4 specification (210X 297 public attack :) 558899

Other embodiments of the present invention may use a high-pass filter, which is equivalent to the inverse CSF of the respective contrast color channel. These CSFs can be mapped from the cy / deg domain (modeled) to the loop / pixel (0 ^ /]? D) digital domain. The actual mapping handler takes into account the viewing distance and allows customization of 'the use of special display resolutions in pixels / mm (pixel / mm)' for different applications and different expected or projected viewing distances. Due to the method of the present invention, when the display is viewed at a distance farther than the designed viewing distance, the chroma epigenesis phenomenon will not be seen. However, the brightness resolution can be improved. Filtering can be performed on all chrominance channels 90 & 92 or selected channels based on the total amount or intensity of epigenetic phenomena caused by a particular sampling process (steps 94 & 96). The low-pass filtering of the original OCD chrominance channel 98 & 100 (step 102 & 104) can occur simultaneously with the processing of the luminance channel 105, or at other times. A low-pass filtering of the OCD chrominance channel 98 & 100 is performed (step 102 and 104) to remove the substantial chrominance frequency above the Nyquist frequency of the display pixel. So 'the channels can be sampled (steps ioi & i03) in a conventional manner with a factor of 1: 3' without generating chromaticity confusion in the chrominance pathway 11o. Once the filtering operation is completed, the isolated channels can be combined. The chrominance channel combination will vary depending on the color gamut used. In this exemplary embodiment, the high-pass filtering, sub-pixel sampling blue to yellow (HPFSPS-B / Y) chroma channel 97 is combined (step 106), and the low-pass filtering, traditional prime sampling blue Turn the yellow (LPFSS-B / Y) chrominance channel 10 to form a single high-low filter (HLF) B / Y chrominance channel 111. It will also combine high-pass filtering and sub-pixel sampling red-to-green (HPFSPS-R / G) channel 95 combination (step 108) low-pass filtering -20- This paper size applies to China National Standard (CNS) A4 specifications (210 X 297 male (Centi) 558899 A7 B7 5. Description of the invention (17), the traditional prime sampling red to green LPFSS-R / G) channel 107, to form a single high-low filter (HLF) R / G channel 114. Please note that the method of the specific embodiment of the present invention is applicable to other color spaces and color gamuts, which may include other color channels and other color channel quantities, and other changes in the brightness or brightness channels. The combined 111 ^ chroma channels 111 & 114 can be further combined (step 88) SPS luminance channel 86 to form a lower resolution OCD image 116. Then, the lower resolution OCD image 116 can be converted into other image formats or fields according to the needs of various applications. The methods and systems of these embodiments provide lower resolution images with less visible chrominance. In addition, the foregoing specific embodiments may be modified in various ways. For example, in some cases, the low-pass filtering of the chroma channel 98 & 100 may be ignored (step 102 & 104). Alternatively, the segmented (step 74) luminance channel 76 may be used alone to construct the lower resolution image 116. In other words, each step of the luminance path 105 related to the luminance channel 76 can be performed, while each step of the chrominance path 110 and steps 106 and 108 related to the chrominance channel 98 & 100 are ignored, in order to combine the SPS luminance channel 86 and HPFSPS-R / G channel 95 and HPFSPS-B / Y channel 97. In this method, a lower-resolution image 116 can be constructed.

Referring now to FIG. 9, a specific exemplary embodiment of the present invention is explained. This particular embodiment can be used to process higher-resolution RGB images for displaying images on lower-resolution display devices. Can be pre-processed (step 122) according to the specific needs of the user or application (step 122) Higher resolution RGB -21-This paper size applies the Chinese National Standard (CNS) A4 specification (210X297 mm) Binding

558899 A7 B7 V. Description of the invention (18) Image 120. The pre-processing (step 122) may include hinting, low-pass filtering, or other types of processing techniques. The pre-processing (step 122) can also be completely skipped. After pre-processing (step 122), the RGB conversion (step 124) can be performed into an opponent color domain, such as LAB, YCrCb, YIQ, YUV, or other image color gamuts. In this example, the LAB image color gamut is used. Once converted into this color gamut, the image can be segmented (step 126) into L, a, and b color gamut channels to separate the processing channels. In this method, the chrominance and luma channels can be processed separately. The "L" channel 127 is then converted (step 128) back to the RGB color gamut so that it can be sampled in the final display format. This transformation can include copying the L layer or channel directly to three identical R, G, and B layers. However, a single layer can also be used, and the actual conversion method depends on the color conversion selected. Sub-pixel sampling is then performed on the RGB luminance image (step 130) to preserve the horizontal luminance resolution of the original RGB image 120. After the sub-pixel sampling, the sampled image is converted again (step 132) into an contrast color domain (OCD), such as LAB. The sampled LAB image is segmented (step 134) to isolate the luminance channel and chrominance channel for further processing of the chrominance channel. Here, the SPS brightness channel is usually not processed to save the original brightness data. However, the chroma channel 150 & 152 and the segmented image after sub-pixel sampling are all subjected to high-pass filtering (step 146 & 148) to remove the low-frequency epigenetic confusion that occurs during the sub-pixel sampling. As with the specific embodiment explained previously, a high-pass filtering may be performed using a halo mask filter using a Gaussian low-pass core. When using this -22- this paper size applies Chinese National Standard (CNS) A4 (210 X 297 mm) binding

558899 A7 B7 V. Description of the Invention (19) In the specific embodiment of the method, the chroma channel is filtered to generate a low-pass filtered chroma image. This is a chroma image subtracted from the SPS-RGB chroma image to form " High-pass filtering (HPF) SPS chroma image or channel i47 & 149. High-pass filtering is usually performed on the “a” and “b” channels (step i46 & 148), but when conditions permit, high-pass filtering can be performed on only one of the channels. The low-pass filtering (steps 138 & 140) of the original "a" and "b" chrominance channels 154 & 156 can occur simultaneously while processing the "L" channel, or at other times. Low-pass filtering of the "a" and "b" chroma channels is performed (step 138 & 14) to remove the substantial chroma frequency above the Nyquist frequency of the display pixel. After low-pass filtering (steps 138 & 140), these channels can be sampled (step 142 & 144) in a conventional manner with a factor of 1: 3 without chrominance confusion. When the channels have been filtered and sampled, the channels are combined to form a lower resolution image with fewer errors. The high-pass filtered "a" luminance channel 147 is combined (step 160) through the sub-sampled and low-pass filtered "a" channel 143 to form a processed ra "channel 164. The high-pass filtered "b" luminance channel 149 is combined (step 158) and the sub-sampled and low-pass filtered "b" channel 145 is combined to form a processed "b" channel 162. Then, the chrominance channel 162 & 164 is combined (step 166), and the SPS luminance channel 136 'is used to form a lower-resolution, lower-resolution lab image 168. The error-reduced images can be converted (step 70) to RGB color gamut to produce error-reduced, lower-resolution RGB images 172, which can be output to a display or other device. The function of the processing procedure of the specific embodiment of the present invention will now be described with reference to FIG. 10, which shows the reserved signals related to the luminance CSF 180 and chrominance CSF 182. -23- This paper standard applies to China National Standard (CNS) A4 specification (210 X (297 mm)

Claims (1)

A B c D 558899 Patent Application No. 090130880 Chinese Application for Patent Scope Replacement (June 1992) 々, Patent Application Consumption 1. A method for converting a higher resolution image into a lower resolution image and reducing visible errors The method includes the following actions: segmenting a higher resolution contrast color domain (OCD) image into an initial luminance channel and one or more initial chrominance channels; performing a second time on the initial luminance channel Pixel sampling, thereby creating an additive color domain (ACD) brightness image; converting the ACD brightness image into a 0CD brightness image; segmenting the OCD brightness image into one pixel sampling (SPS) brightness channel and one Or one or more SPS chrominance channels; high-pass filtering the one or more SPS chrominance channels; and combining the one or more high-pass filtered SPS chrominance channels with the SPS luminance channel to form a lower error reduction Resolution image. 2. A device for converting a higher resolution image into a lower resolution image and reducing visible errors, the device includes: a segmentation device for converting a higher resolution contrast color gamut (OCD) ) An initial luminance channel and one or more initial chrominance channels separated by image segmentation; a sub-pixel sampling device for performing sub-pixel sampling on the initial luminance channel, thereby establishing an additive color domain ACD) brightness image; a conversion device for converting the ACD brightness image into an OCD brightness image; a segmentation device for splitting the 0CD brightness image into one O: \ 75 \ 75526-920603 DOC \ 7-1 -This paper size applies to China National Standard (CNS) A4 (210 X 297 mm) 558899 A8 i said · ΰ · B8:! Year G C8] D8 L______, patent application pixel sampling (SPS) luminance channel and one or more SPS chrominance channels; high-pass filter for high-pass filtering one or more of these An SPS chrominance channel; and a combination device for combining the one or more high-pass filtered SPS chrominance channels with the SPS luminance channel to form a lower resolution image with reduced errors. A computer-readable medium storing a method for converting a higher-resolution image into a lower-resolution image and reducing visible errors, the method includes the following actions: a higher-resolution contrast color gamut (opponent color domain; An OCD) image segmentation consists of an initial luminance channel and one or more initial chrominance channels; performing sub-pixel sampling on the initial luminance channel, thereby establishing an additive colo rdomain (ACD) luminance image; Converting the ACD luminance image into an OCD luminance image; segmenting the OCD luminance image into a pixel sampling (SPS) luminance channel and one or more SPS chrominance channels; high-pass filtering the one or more SPS chrominance channels ; And combining the one or more high-pass filtered SPS chrominance channels with the SPS luminance channel to form a lower resolution image with reduced errors. A method for converting a higher resolution image into a lower resolution image and reducing visible errors. The method includes the following actions: Segmenting a higher resolution contrast color domain (OCD) image into separate components. One starting brightness channel and several starting colors-2- O: \ 75 \ 75526-920603 DOC \ 7 This paper size is applicable to China National Standard (CNS) A4 specification (210 X 297 mm) 558899 A BCD Positive charge 6. Apply for a patent range channel; perform sub-pixel sampling on the initial brightness channel to establish an additive color domain (ACD) brightness image; convert the ACD brightness image into an OCD brightness image; OCD luminance image segmentation consists of one pixel sampling (SPS) luminance channel and several SPS chrominance channels; high-pass filtering of these SPS chrominance channels; low-pass filtering of these initial chrominance channels; sub-sampling and filtering Starting chrominance channel; combining the sub-sampling, low-pass filtering chrominance channel and the high-pass filtering SPS chrominance channel; and combining the combined chrominance channel and the SPS luma channel to form a lower error reduction Resolution image. 5. If the method of the fourth item of the patent application is applied, wherein the higher resolution image is an additive color gamut image, the image will be converted into a contrast color gamut image before the segmentation of a higher resolution OCD image. . 6. The method according to item 4 of the patent application, wherein the color gamut image is an RGB image. 7. The method according to item 4 of the patent application, wherein the contrast color gamut image is a YCrCb image. 8. If the method of claim 4 is applied for, the contrast color gamut image is a LAB image. 9. The method according to item 4 of the patent application, wherein the high-pass filtering includes unsharp-mask filtering. O: \ 75 \ 75526-920603 DOC \ 7 * 3-This paper size applies to China National Standard (CNS) A4 (210 X 297 mm) 558899 A8 B8 C8 D8 6. The scope of patent application 10. The method according to item 4 of the patent scope, wherein the high-pass filtering includes the following actions: filtering the SPS chroma through a halo mask filter having a Gaussian low-pass core. Channel to generate a low-pass SPS chrominance channel: and subtract the SPS chrominance channels from the SPS chrominance channels to generate a high-pass filtered SPS chrominance image. 11. The method according to item 4 of the patent application, wherein the chrominance channels include a red-to-green channel and a blue-to-yellow channel. 12. The method according to item 4 of the patent application, wherein the chrominance channels include Cr and Cb channels of a YCrCb image. 13_ The method of claim 4 in the scope of patent application, wherein the chrominance channels include the "a" and "b" channels of a CIELab image. 14. The method of claim 4, wherein the method further comprises converting the lower resolution image with reduced errors into an RGB image. 15. The method of claim 4, wherein the performing sub-pixel sampling step includes converting the starting luminance channel into an additive color gamut (ACD) luminance image, and sampling the ACD luminance image. 16. A method for displaying a higher resolution image at a lower resolution that reduces visible errors, the method including the following actions: converting a higher resolution RGB image into a higher resolution contrast color gamut ( OCD) image; segment the higher resolution contrast gamut (0CD) image into _ starting luminance channels and several starting chrominance channels; O: \ 75 \ 75526-920603 DOC \ 7-4 This paper size is applicable _ Jiascale (⑽) A4 specifications _ χ Norwegian public love) 558899 A8 B8 C8 D8 6. The scope of the patent application converts the starting brightness image into an RGB brightness image; performs sub-pixel sampling on the RGB brightness image; converts the sub-pixel sampling (SPS) RGB brightness image into an SPS-OCD brightness image ; An SPS luminance channel and several SPS chrominance channels that divide the SPS-OCD luminance image into components; high-pass filtering the SPS chrominance channels; low-pass filtering the initial chrominance of the higher-resolution OCD image Channel; sub-sampling the filtered starting chrominance channels; combining the sub-sampling, low-pass filtering chrominance channels and the high-pass filtering SPS chrominance channels; combining the combined chrominance channels and the SPS luma channel, Forming an OCD lower resolution image with reduced errors; and converting the OCD lower resolution image with reduced errors into a lower resolution RGB image with reduced errors. 17. A method for converting a higher-resolution image into a lower-resolution image and reducing visible errors, the method includes the following steps: a higher-resolution contrast color domain (OCD) image Segment a starting luminance channel and several starting chrominance channels; perform sub-pixel sampling on the starting luminance channel to establish an additive color domain (ACD) luminance image; and convert the ACD luminance image Converted into an OCD brightness image; One pixel sampling (SPS) of this OCD brightness image is divided into O: \ 75 \ 75526-920603 DOC \ 7-5-This paper standard is applicable to China National Standard (CNS) A4 specification (210 X (297 mm) 558899 A8 B8 C8 D8 々 、 Patent application scope Luminance channel and several SPS chrominance channels; High pass filtering of these SPS chrominance channels; Low pass filtering of these starting chrominance channels; Sub-sampling of these filtered starting chrominance channels ; Combining the sub-sampling, low-pass filtered chrominance channels and the high-pass filtered SPS chrominance channels; and combining the combined chrominance channels and the SPS luminance channel to form a lower-resolution image with reduced errors. 18. A system for converting a higher-resolution image into a lower-resolution image and reducing visible errors, the system comprising: a first divider for comparing a higher-resolution contrast color gamut (opponent color domain; OCD) an initial luminance channel and several initial chrominance channels separated by a segment; a primary pixel sampler for performing sub-pixel sampling on the initial luminance channel, thereby establishing an additive color domain (domain; ACD) luminance image; a converter for converting the ACD luminance image into an OCD luminance image; a second divider for dividing the OCD luminance image into a one-time pixel sampling (SPS) luminance channel and Several SPS chrominance channels; a high-pass filter for high-pass filtering the SPS chrominance channels; a low-pass filter for low-pass filtering the initial chrominance channels; a sampler for sub-sampling The filtered initial chrominance channels; a first combiner for combining the sub-sampled, low-pass filtered chrominances O: \ 75 \ 75526-920603 DOC \ 7-6 _ This paper size is applicable to the Chinese national standard Standard (CNS) A4 (210 X 297 mm) 558899 A BCD Positive charging, patent-applied channels and the high-pass filtered SPS chrominance channels; and a second combiner for combining the combined chrominance channels and the SPS luminance channel to form a lower resolution that reduces errors image. 19. A computer-readable medium including instructions for converting a higher-resolution image into a lower-resolution image and reducing visible errors, the instruction including the following actions: comparing a higher-resolution contrast color gamut (opponent color (domain; OCD) image segmentation into an initial luminance channel and several initial chrominance channels; performing sub-pixel sampling on the initial luminance channel, thereby establishing an additive color domain (ACD) luminance image; Convert the ACD luminance image into an OCD luminance image; divide the OCD luminance image into one pixel sampling (SPS) luminance channel and several SPS chrominance channels; high-pass filter the SPS chrominance channels; low-pass filter the Starting chrominance channels; sub-sampling the filtered starting chrominance channels; combining the sub-sampling, low-pass filtering chrominance channels and the high-pass filtering SPS chrominance channels; and combining the combined chrominance channels and The SPS luminance channel constitutes a lower resolution image with reduced errors. O: \ 75 \ 75526-920603 DOC \ 7 This paper size applies to China National Standard (CNS) A4 (210 X 297 mm)