TW201243808A - System and method for improving color and brightness uniformity of backlit LCD displays - Google Patents

Abstract

Systems and methods for improving color and brightness uniformity of an image displayed on a backlit LCD are disclosed. In one example, a correction map is computed and applied to the LCD pixel values. In another example, the voltage settings of the backlight source components are also corrected in addition to the LCD pixel values. For efficient hardware implementation, corrections are applied using function representation of a grid data transformation relating measured values to corrected values. In one particular exemplary embodiment, the backlight source is provided by a plurality of LEDs. In another exemplary embodiment, the display consists of a plurality of OLEDs wherein the light source and the display panels coincide.

Description

201243808 VI. Description of the Invention: [Technical Field] The embodiments described herein relate generally to electronic image and video processing, and more particularly to correction of color and brightness non-uniformity of a backlit LCD panel. [Prior Art] Light-emitting diodes (LEDs) are made of a special material based on the spontaneous composite radiation of visible electrons and holes in the p-n junction, visible and invisible (covering the infrared to ultraviolet range of the electromagnetic spectrum). A forward bias voltage is typically applied to the P-n junction to accelerate electron-hole recombination and produce sufficient brightness. The wavelength (and color) of the emitted light depends on the energy of the semiconductor bandgap. Early LEDs emit low-intensity red light. New semiconductor materials with large energy band gaps to cause the LEDs to emit green light have recently appeared, followed by new semiconductor materials with large energy band gaps to cause the LEDs to emit blue light. In addition, advances in LED brightness and energy efficiency have led to the invention of white LEDs. Liquid crystal displays (LCDs) are commonly used in TV panels and computer monitors, using the light modulation characteristics of liquid crystal (LC). The LC is a transmissive dement. They can only transmit but not directly. Therefore, the LCD panel itself cannot produce light and requires an external illumination mechanism to be visible. Conventionally, a cold cathode fluorescent lamp (CCFL) is placed behind the LCD panel to provide illumination. Recently, with the advancement of HD TV and high-frequency video content, LED backlit LCD panels have emerged in the television industry to replace CCFL backlight LCDs. There are two LED backlighting technologies, white LED backlighting and red/green/blue (RGB) LED backlighting. White LEDs (used extensively in notebook and laptop screens) are actually blue LEDs combined with yellow phosphors to provide white light perception. In this case, the spectral curve has a large interruption in the green and red portions. The RGD LED consists of red, green and blue LEDs. The RGD LED can be controlled to produce different whites of temperature. RGB LEDs provide a huge color gamut to the screen. A backlight from three different LEDs produces a color spectrum that closely matches the color filter in the LCD pixel itself. In this way, the bandpass of the LCD color filter can be narrowed such that each color component allows only a very narrow band to pass through the LCD. This can improve the efficacy of the display because a very small amount of light is blocked when white is displayed. And the actual red, green and blue dots can be further visualized to enable the display to reproduce more realistic colors. Both types of LED backlights can be arranged in an array to illuminate the screen. 201243808 LEDs exhibit many advantages over incandescent sources, including greater color gamut, higher luminous efficiency, deeper black levels (higher contrast), lower power consumption (reduced light waste) ), longer life, improved robustness, smaller size, faster switching response and better durability and reliability. Considering lower cost and energy, lower environmental impact (green) and thinner displays, the industry has aggressively improved backlighting through the rapid transition from old CCFL backlights to more efficient and flexible LED backlighting. An example of such a system in which an RGB LED array or bank is driven by a particular circuit to provide light to each pixel of the display is disclosed in U.S. Patent No. 6,888,529. The intensity and color content can be adjusted from the source by controlling the individual color components directly through the drive circuit. However, LED sources still have some drawbacks. The main drawback is that since the LED is a discrete light source, it increases the unevenness of color and brightness. Due to the difference in LED manufacturing process and LED aging (different LEDs aging at different rates), the uniformity is significantly lower compared to CCFL backlights. Broadband TVs require a large number of (100th order) LEDs for display compared to a small number (10th order) CCFL tubes, each with a different brightness level. Even if these sources are sorted and graded, these sources can still have a brightness difference of up to +/- 10% between the device and the device. In addition, the use of three separate red, green, and blue light sources means that the white point of the display can move at different rates as the LED ages. White LEDs also age, and white LED aging is accompanied by a change in color temperature of a few hundred K. White LEDs also face a higher temperature blue offset. As a result, they require more accurate current and thermal management than traditional light sources, so they are more expensive to build. In fact, if a certain level of uniformity is not achieved, the display produced will be waste, which will cause damage to the manufacturer. In recent years, organic light-emitting diodes (OLEDs) have replaced LCDs for use in TV screens and other displays. Unlike LC cells, OLEDs are active cells in which an emissive electroluminescent layer of organic semiconductor compound that responsive to current luminescence is present. This layer is a film between two electrodes, one of which is typically transparent. The organic compound is a small molecule or polymer that allows the OLED to be used directly as a display pixel. In this way, the OLED display does not require backlight operation. In other words, the backlight and modulator panel are identical. Typically, an image from a display, LCD, OLED, or other device is a spatial varying pattern of color and brightness that is intended to match the pattern of the input signal. If the input signal is spatially constant, then it is expected that the display will be rendered constant in color and brightness for 201243808. This is referred to as color and brightness uniformity requirements, which are important requirements for accurate color reproduction of the display. The LCD panel has several other components, including a light guide, diffuser for guiding the light toward the front and distributing the light evenly. While these components help to improve uniformity, as the thickness of the panels tends to decrease, their designs become more complex and will result in reduced efficiency. Alternative methods that require color and brightness adjustment overcome the above deficiencies in a more efficient and cost effective manner. Some prior art solutions have been used primarily to improve backlight quality by controlling the source voltage after sensing or visualizing the output signal. These solutions focus on the uniformity of the backlight panel rather than the LC panel in which the image is actually observed, so the effect is limited. For example, US 2007/0200513 discloses a device that controls LED driving in response to temperature and voltage variations. US 2006/0007097 discloses a backlight adjustment method for an LED-backlit LCD device. The luminance measuring sensor is disposed on the substrate together with the thin film device as a pixel on the panel and integrated with the LCD panel. These prior art techniques are not dedicated to addressing the non-uniformities that the observer will observe. In addition, there is no solution to adjust the chromaticity of the signal. It is an object of one or more aspects of the present invention to provide an electronic unit for improving the color and brightness uniformity of an LCD that is directed to simultaneously adjusting the backlight and the LC light modulator. All aspects of the invention relate to modulators that do not consider the application of the light source. In one embodiment, the LED provides backlighting. The present invention not only addresses the non-uniformity due to LED aging, but also helps manufacturers save costs by having fewer cells and tolerating units that do not meet the threshold and will be discarded. This teaching can also be applied to a conventional CCFL backlight consisting of multiple CCPL tubes. In another embodiment, a laser diode (LD) is used as the backlight. When used in backlit LCD panels, the LD operates in a similar manner to LEDs. The main difference is the generation of light. Here, the recombination of electrons and holes is generated by excitation rather than spontaneously, which is of course a necessary condition for laser light emission. The spectrum of the LD is much narrower than that of the LED, resulting in a more clearly defined color. The direct light LED backlight system as an exemplary embodiment is mainly shown in this light source part and related drawings. However, as described, the present invention can be applied to edge-lit LEDs, LDs and CCFLs, and OLEDs (LEDs are special cases of direct LEDs) as long as the displayed light is relevant. SUMMARY OF THE INVENTION 201243808 Embodiments described herein provide, in one aspect, improved backlighting by examining measurable physical output from a display and efficiently creating a correction map to correct pixel defects of the display. A method of color and brightness uniformity of a liquid crystal display (LCD). In addition, the voltage control of the components of the light source can be modified in the same scheme. Embodiments described herein further provide in a further aspect a system for improving the color and brightness uniformity of a backlit liquid crystal display (LCD). The system includes an image generator for displaying a plurality of reference input images on a display; an image capture device for measuring uniformity of the displayed images, the uniformity of the displayed images being physically Measured quantity characterization; a processor for generating a global display corresponding function from the measurement data and creating a corrected grid data map such that the response function becomes a constant 横跨 across the display, and for correcting the correction The grid data map is converted into a functional form and the associated correction function is applied to the second processor of the input signal. The backlights can be generated using different technologies, such as LEDs, CCFLs, and laser diode sets. The embodiments described herein further provide, in another aspect, improved color and brightness uniformity of organic LEDs (0LCDs). Sexual systems and methods. This is an example of a special LED in which the backlight and the display are essentially an integrated unit that makes the brightness and color non-uniformity problems even more relevant. [Embodiment] It will be appreciated that a number of specific details have been described in order to provide a thorough understanding of the exemplary embodiments described herein. However, those skilled in the art will appreciate that the embodiments and domain implementations described herein can be implemented without these specific details. In other instances, well-known methods, procedures, and components are not described in detail to avoid obscuring the embodiments and/or implementations described herein. In addition, the description is not to be considered as limiting the scope of the embodiments described herein, but rather to the structures and operations of the various embodiments and domain embodiments described herein.

Fig. 1 shows a typical LCD system 'which is at a high level' in terms of color and brightness and can be regarded as a light source (also referred to as a backlight panel) 1 or 11 and a light modulator 16 display. The backlight panel, as shown in Figure 1-A, consists of multiple red, green, and blue (RGB) LEDs 12 'This panel is called an LED panel. In a low-cost consumer display, 3 RGB 201243808 LEDs are usually single Replace with white LED. Or the light source may be composed of a plurality of ccFL tubes 13 or laser diodes (not shown) as shown in FIG. i_b. The light emitted by the light source has an intensity distribution 14, which depends on the type of the light source and the number of units. Etc. and usually provided by the manufacturer. The light modulator is an LCD panel 16 which is a liquid crystal layer. When an RGB filter layer is added in front of it, the liquid crystal layer is composed of an LC array forming RGB pixels 18. Light source 10 and light modulator 16 are variable components. The light source can be changed (modulated) by applying a different voltage to the LED 12, and the light modulator can be changed to the LC pixel 18 by applying a different input (digital image) signal. There are fixed components attached to the LCD display system, such as the usually fixed light guide and light diffuser, which can be considered part of the light source. Polarizers, color filters, and even optional sensors can be fixed and viewed as part of the light modulator. Figure 1 is further applicable to flat panel displays that do not have a backlight panel, such as an OLED. The organic compound in the light modulator panel itself emits a different amount of light in response to the input signal. In the following discussion, the exemplary embodiment can be processed by setting the backlight panels to be uniform. Figure 2 shows two common types of LED backlight mechanisms. One is a direct LED backlight 20, where the LED is facing forward and is illuminated directly in the viewing direction to illuminate the LCD panel. The LED panel can also be an edge-lit backlight 30 where the LED is perpendicular to the viewing direction. Additional optics, such as reflective prisms, are needed to redirect the emitted light toward the LCD panel. This makes the edge-lit backlight more advantageous for reducing system thickness. In the exemplary embodiment of Figure 2, the LEDs are positioned upwardly at the bottom of the panel. However, in practice, the LEDs can be located on either side, at the bottom and at the top or on either side. In both typical cases, the LEDs are shown in three (RGB) groups to produce white light, although the method shown can also be used for white LEDs. Uniformity is defined as the brightness and color difference across the display in response to the platform input signal set. The platform signal or level is 1, where a constant number of RGB 分配 is assigned to all pixels (x, y): (R(x, y), G(x, y), B(x, y)) = (R 〇, GO, BO), for all (x, y) (1) Here, R(x, y) refers to the R (red) component 値, G(x, y) and the pixel position U, y) B(x, y) refers to the G (green) component 値 and B (blue) component 値 of the pixel position (x, y), respectively. The pixel position (x, y) is an integer 値, meaning column X and row y, and the range of resolution WX 显示器 of the display 'here W is the level resolution and Η is the vertical resolution: 8 201243808 〇<= X <= (W-1) and 0 <= y <= (Hl) (2) However, in the mathematical formula disclosed herein, (x, y) is allowed to include any real number. The pixel color 値(R0, GO, B0) is an integer in [0, 2bn-l], where bn is the bit depth of the display, such as 8'10. In internal calculations, the color 値 can be allowed to be any real number and can be adjusted to the allowed integer range when input as a signal into the display. The uniformity of brightness and color can be defined in the form of a measurable physical quantity. Define a variety of different measurable quantities to describe brightness and color. The most widely used is the International Illumination Commission (CIE) XYZ Tristimulus, from which other related quantities can be calculated. The Y component is a luminance 値, denoted L in the art, which is commonly referred to as brightness or intensity and has a unit kan / m 2 (cd/m 2 ). Minimizing the difference in luminance is the most critical for achieving a uniform display. The X and Z components are the additional amount of color required for positively defined colors, and their unit of measure is also cd/m2. The color used for display, with decomposed luminance, is easier to understand in terms of chroma 値 than XYZ. The chrominance component (xc, yc) is the derivative quantity given by the following formula:

X

' ~ X + Y+Z z〇=i-Xc-yc Because in color science, (x, y) is usually used for chroma 値, the c subscript is included in the symbol to distinguish between chrominance and spatial pixels. coordinate. Z-chrominance zc is not an argument and is therefore not used. The color chromaticity 値 is used as a coordinate on the CIE chromaticity diagram, and the solid color plane of the color is defined, and the luminance or luminance components are decoupled. Thus XYZ and xcycY can be viewed as two orthogonal systems that fully define the brightness and solid color of any displayed color. Therefore, XYZ is usually a measured amount, and xcycY is usually used to describe the amount of color. The typical chromaticity 値(xc,yc) of the RGB primary colors in the display is adjacent to R (0.640, 0.330), adjacent to G (0.300, 0.600), and adjacent to B (0.150.0.060). Pure white color (or gray level) can also be called color temperature or - point. Most color measuring instruments report XYZ値 and chromaticity 値. Therefore, in this spatial uniformity or non-uniformity, for the component measured at a specific level (platform signal)]\4={also 1^(^;} can be defined as: (4) % uniformity = 100* ( 1 - (Mmax - Mmin)/Mmax) 201243808 % Non-uniformity = 100 - % Uniformity Here, Mmax and Mmin are at a specific level, across all pixels in the display (ie all (x, y)) The maximum 値 and minimum 测量 of the measured components. In particular, the luminance (brightness) uniformity is given as follows: % luminance uniformity = 1 〇〇 * (1 - (Ymax - Ymin) / Ymax) % luminance non-uniformity = 100 * (Ymax - Ymin)/Ymax @ In practice, in contrast to each pixel, uniformity can be calculated by measuring Μ in a subset of pixels. The terms uniformity and non-uniformity used herein are understood to mean that they are only Two different perspectives. For any platform signal, an ideal uniform display will have the same measurement 横跨 across all pixels. This will be 100% uniform across all levels, which is not possible in practice. The goal of calibration techniques is to improve uniformity to within acceptable limits. For example, In consumer displays, 'for pure white (for 8 bits, level R0 = G0 = B0 = 255), > = 70% gray uniformity is generally considered sufficient. Uniformity in other components is not considered. The method provides an efficient way to provide a tool that achieves higher uniformity across multiple levels, which is necessary for professional displays and can be used to improve the standards of consumer displays. Furthermore, uniformity of luminance and chromaticity Sex is processed because the color temperature or white point is a subset of the color 値. Once the chromaticity processing is completed, the color temperature and white point correction are automatically completed. Once the non-uniformity is quantified and obtained, they can be electronically corrected. A hardware system for applying color and geometric correction is described in the pending patent application 11/649,765. Figure 3 shows a color correction system 100 and its components are incorporated herein by reference. Sources of non-uniformity in LCD displays can be classified In two groups: a) non-uniformity derived from the backlight panel (light source); b) non-uniformity derived from the LC panel (light modulator). This means that the non-uniformity correction can be made by adjusting the variable control of the backlight panel or the LC panel, or in the most common example by adjusting the variable control of the backlight panel and the LC panel. In the typical case of an LCD backlight, the variable control of the backlight panel is the voltage applied to the LED. Other components, such as the diffuser, are fixed and immutable. We represent these voltages as vectors 201243808 where 'Vi is the voltage of the ith LED and L is the total number of LEDs. If a common voltage is applied to all LEDs, then all components will have the same enthalpy. This mourning and mathematics will be applied to CCFL and LD backlight panels with i tubes. The variable control of the LC panel is a single pixel digital RGB, which is the input digital signal itself. The following symbols will be used alternately to indicate the pixel 値: (R(x, y), G(x, y), B(x, y)), or - (7) C (x, y) = (G(x, y), G(x, y), G(x, y)), {1, 2, 3} = {R, G, B} If the platform level is used' then the pixel position correlation will decrease: (R, G, B), M " C = (C., C2, C3) (8) Figure 4 shows an embodiment of the invention in which the steps for the non-uniformity correction method 4A are indicated and explained herein. In a first step 4, the uniformity of the display is reset in one or more backlights (voltages) in response to one or more platform signal groups, the uniformity of the display being by any one or more The physical quantity Mi e {χ, γ, ζ, χΐ} can be measured. The measured data in 42 is expressed as: (9), 刖, due to non-uniformity, the measurement 値 depends on the pixel position (x, y), the backlight of the backlight, and the face number f. In the setting of each light setting, all the light source points are usually set to public 値. The thief only has a bright piano, and the dirty 雠 Μ = γ will be dirty. Hanging evenly shows 'μ, all the chicks are constant and independent of (x, y). The purpose of the non-uniformity correction is to vary the general variable number depending on the spatial constant of the input signal and the backlight voltage: (1〇) From the data measured in step 42, in step 43 reconstruct the display back to any RGB signal 201243808 should. The response is represented in step 44 as a function that will be used to calculate the correction. Each object-quantity has its own response, allowing the measurement data to be written as follows: 岣=0,%(?/) (11) In the recalibration phase 45, determine the signal ί and voltage 所需 required for uniform measurement Correction. It looks for a constant solution: (12) where 'f is a new signal or a new pixel 値, which is due to non-uniformity variations in (X, y) and also depends on the input pixel 値. The user is the new voltage setting of the LED, which can also depend on the pixel 値. However, as will be shown, the pixels can be made independent. The solution essentially computes the reciprocal in multiple variables: {(?'/}=mine (dead) (13) Therefore, the non-uniformity correction can be written as: 0/'} is calculated as a grid The subset of coordinates in the parameter space of the point is completed. For example, (x, y) can be a subset of 17x17 pixel positions in the 1920x1080 resolution display. There are various reasons for this: 1) Measurement data can be used only at discrete points . 2) To speed up processing, only point subsets are measured. 3) The non-uniformity difference is smooth and can be accurately measured from the subset of points. 4) This measurement point is generally considered to correspond to the LED geometry, so a smaller subset is used because the number of LEDs is much smaller than the pixel resolution. IX 5) Accurate calculation of the inverse solution (in the form of variables) is not possible and must be done digitally, thus significantly reducing computation time by using a subset of points. Thus, at step 46, the calculation provides corrected grid data. In the correction reconstruction phase 47, the corrected grid data obtained in step 46 is converted into a functional form such that it can be applied to all pixel positions and color chirps. Similar to equation (11), a correction function can be obtained at step 48: 12 201243808 - &=Pc(x,y,^) V; = Fv (1,^,8) (15) The function Fv is at the ith LEDs, given their initial voltage and input pixel 値 to provide a new voltage level. This correction function F. It refers to an abbreviation such as a pixel correction map or a pixel map, and the correction function Fv refers to an abbreviation such as a backlight correction map or a backlight map. The format of the correction function can be determined by the hardware that applies the correction to the display. A general hardware effective format is described in U.S. Patent 7,324,706, which uses a polynomial surface functional form to represent a set of data points. In a final step 49, the calibration map can be applied to the input signal and backlight control using a hardware circuit. The hardware evaluates the map and sends a new pixel to the display controller to send a new voltage to the backlight controller. The pending patent application 11/649,765 describes a hardware system 100 (Fig. 3) for applying color correction at the pixel level. A similar system can be implemented in a low cost FPGA due to the efficient function form. For backlight adjustment, the same hardware can be used to evaluate the backlight and provide a new voltage to the LED. The steps of the non-uniformity correction method 40 provide a highly efficient and accurate method of correcting for brightness and color non-uniformity in an LCD display. Fig. 5 shows a schematic view of an embodiment of the present invention, such as a typical system employing the method. The system includes a capture device 52, such as a colorimeter that captures and provides measurable features of the reference image 51, which is typically a platform level, created by the input image generator 58 and displayed on the LCD panel 50. . A processing unit (which may be an embedded processor or software running independently on a computer) 54 analyzes the measurement 针对 for the to-be-determined 値 and generates corrected grid data including a new pixel map of correctable non-uniformity And LED voltage settings. The hardware processor 56, as described in one of the patent applications 11/649,765, implements and applies the correction to the input image and submits it to the display 50. The function of each component of the system is described in great detail here. The measurement of the physical quantity 41 is typically performed using a colorimeter or a spectroradiometer. These devices 52 come in two forms: a luminance meter type and a 2D imaging camera type. A spectroradiometer is typically a point luminance meter that measures physical quantities at a given point or pixel. The average 値 in a defined neighborhood of a particular pixel or measurable specific pixel can be measured. Spectroradiometers are very accurate devices and are commonly used in calibration colorimeters. The colorimeter can be used as a point luminance meter or a 2D imaging camera in which measurements can be performed at or near a single point. In the latter form, a very large number of pixels are simultaneously measured in the 2D (x, y) position space. The user can specify the pixel location of the measurement. Usually the regular grid of points 'represented as N, row by R column, is extracted from the 2D camera image and used for calculation. Similar to a point luminance meter, the camera performs an average over a small defined neighborhood of measurement pixels. In a preferred embodiment, a 2D imaging camera type colorimeter is used for the measurement because by definition, the non-uniformity is a measurement at the 2D (x, y) position. In addition, the pixel neighborhood averaging of the colorimeter allows itself to characterize non-uniformity well, since visual perception does not detect a single pixel (except for "broken" pixels), but rather averages on neighboring pixels. The point luminance meter can be used as an accurate measure of the 2D grid of points. However, this requires physically moving the point luminance meter across the display, and this will waste time unless a small number of points are measured. Multiple point luminance meters can be used, but this would be expensive. In any case, the method is independent of any particular measuring instrument and only needs to measure the grid of data points sampled in the X and y directions. The 2D and point colorimeter can measure all relevant physical quantities' such as XYZ tristimulus. In the typical discussion below, the term camera refers to a 2D colorimeter. The measurement process further requires selecting a set of platform reference signals 51 (also referred to as signal levels or levels) to be input to the display. Commercially available tools, such as test pattern generators, can be used to provide or provide the desired level set. The number of levels to be measured and their corresponding pixels 取决于 depends on several factors, including: 1) The level at which the non-uniformity is to be characterized. In the most common example, pure white RGB = (255, 255, 255) is calculated and corrected for non-uniformity, where a smaller amount of level needs to be measured. 2) Type of correction, which requires only brightness correction, only color correction or both Brightness and color correction are required. 3) Accuracy of calibration. If the required uniformity is very high, more electrons need to be captured. 0,, 4) Display non-uniformity characteristics. Depending on how uneven the non-uniformity of the display is, if you only correct the brightness in pure white, you may need a lot of levels. 5) The speed of the entire calibration process. In a manufacturing environment, speed is the most important factor depending on the production yield. Measuring many levels may be impractical. These factors not only determine the reference level, but also the method 201243808 method and optimization used in the steps following the measurement phase. By using the color superposition principle, the reference level can be reduced to a large extent in number. In terms of XYZ値, this principle stipulates that the color produced by the combination of two independent color sources in an additional color system (such as LCD) has a tristimulus that is the sum of the three sources of tristimulus. In the equation new color = color 1 + color 2 color 1: (HZ,), color 2: (K, z2) 2 2 2 (16) (x, y, Z) = (x1 + x2^ + y2} z, +z2) The image displayed on the LCD consists of three independent color components r, g and B. This superposition principle means that the tristimulus of any color (R, G, B) can be calculated by increasing the tristimulus R of the R, G, and B components: ^(RyG,B) = (I?) Z(R,〇 , B) = + ZG + Ζβ Therefore, the non-uniformity of the display can be fully characterized by measuring the levels of pure red, pure green and pure blue, pure means easy [J in addition to other components (pixels). Any combination of colors, including gray levels, can be obtained by the appropriate sum of individual components (R, G, Β). Let Nc, C = {R, G, B}, represent the pure level of component c. In the most common example of correcting the brightness and color across all pixel colors, the next level to measure: Pure Red Level: (and, G, 5) = (Must, 0, 0), ί = 1. Pure green level: (foot G, 5) = (0, , 〇), / = , 1 ... 乂 (18) Pure blue level: (foot G, 5) = (0,0,), Moxibustion = 1.. is & indicating the amount of position (such as 8). In practice, when the two components are stripped to produce a solid color (by setting 15 201243808 pixels to ο), the rejected liquid crystals and their filters are continuously leaked from the backlight. This destroys equation (18) by a small but non-negligible number. This light leakage refers to, for example, black level compensation. The amount of light leaked is measured when all of the pixels are set to 0 (R = G = B = 0). In order to correct the black level compensation, as a test of the superposition principle, the pure gray level is also measured, where (R = G = B) 〇 In fact, if only the luminance correction is completed, it is usually only sufficient to measure the pure gray level. Therefore, the amount of pure gray scale can be increased to the reference level for measurement (19) 〇 < W, < 2b " -1 % indicates the common pixel 分配 assigned to each color component for pure gray level, In the most common cases, the level groups to be measured are summarized as follows: Measurement signal level: (20) End)}, {(%,%,%)}}

i = = \".NG\k = \·······1 = \..,NW In fact, a smaller subset can be measured based on the factors discussed above. In addition to different levels of data, measurement data is also required at different backlight voltage settings. In principle, the voltage drop of each LED varies individually and the tristimulus point spread function (PSF) is measured. Thus, PSF refers to the expansion of light in a (X, y) space by a single LED. In practice, it is very difficult and time consuming to change individual LEDs and determine the three stimuli in the voltage and (X, y) space. Alternatively, for all LEDs, the voltage setting can be changed to a common 値 and the three stimuli can be measured to determine the voltage dependence. This essentially ignores the (X,y) correlation contained in the PSF. The different voltage settings are indicated by the following: Measurement voltage settings: (21)

Vsi,i = l...Nv In the ith setting:= For a given voltage, apply the same settings to all LEDs. The common voltage Vsi is referred to as a common or initial backlight setting. This needs to be seen as a backlight control that determines the common state of the full voltage 201243808 (common state). Such control is typically provided in the display OSD. In (21), the voltage is set equal to the control, but the control is usually a normalized amount, which is directly proportional to the LED voltage. The LED voltage can be varied to provide a higher or lower intensity (luminance) in the non-uniform region, as will be described below. They can also be changed to adjust the color uniformity of the RGB LEDs. However, in practice, this is unreliable because the chromaticity correction is more subtle and requires more precise control. Therefore, color correction can be better handled by the pixmap. In practice, RGB LEDs can be viewed as units in the same position as white LEDs in terms of voltage variations. We use this to simplify the equation below, although this method is easy to extend to have separate RGB voltage control. In order to generate a backlight correction map, it is necessary to know the PSF function, which stipulates that light from a single LED is spread out in the (X, y) space. A single LED will illuminate many pixels and change its voltage drop through the backlight correction map, which will affect many pixels. The PSF function can be measured in the backlight panel by turning on the intensity variation of a single LED to its maximum chirp and capture (X, y). The PSF can also be modeled by a suitable mathematical model such as Gaussian. Or, without the PSF function, an iterative method can be used to calculate the backlight correction map. These two situations will be discussed. In one embodiment, the PSF is given (provided, estimated, or directly measured). In another embodiment, the PSF is unknown. In a typical example of an RGB LED backlight, the light from the three LEDs combine to form a white light with a broad spectral distribution that is incident on the LC. panel. If a single LED is not individually tuned to change the chromaticity of the white light, a single PSF, similar to the white LED PSF, can be used to describe the combined effect of the 3 LEDs. This is similar to treating them as one from the point of view of voltage regulation. For the particular system illustrated in Figure 5, the measurement is performed by the following steps: • The input image generator 58 inputs each level in (20) to the display 50' at a given backlight (voltage) setting and Camera 52 is caused to capture the display output. As shown in Figure 5, camera 52 is placed in front of display 50 and positioned to capture the entire display as much as possible onto its sensor. The camera user views the display as accurately as the camera provides such non-angular uniformity measurements that the user can detect non-uniformity measurements that are perceived by the user. This means that when the image is perceived by the user's visual system, the present invention corrects the non-uniformity of the entire display system at the final point, without regard to sources of non-uniformity (LEDs, diffusers, LC panels, etc.). Correcting the final image at the output of the display is often critical in obtaining high quality display halo #. This process is repeated for all backlight settings of (21). For images captured by the camera, this physical quantity can essentially be extracted for all pixels. Although the resolution of the 17 201243808 camera limits the number of pixels that can be accurately measured, the number is much larger than the number of data points that are typically acquired. All colorimeters are equipped with software for extracting data for any user-specified grid point group. Data can be used for all pixels, but a much smaller subset is typically used for calculations because uniformity is a smoothly varying function that changes over a large number of pixels and is not based on each pixel component. Obtain the pixel position of the measurement data, take the rule grid of Ny row by N* column, which can be expressed as: In the measured α column, the pixels of 6 rows: (Ά), a = 1...%, Z? = 1 ... (22) In a typical setup, Radiant Imaging Inc's 2D colorimeter PM-1423F was used for illustrative purposes. The measurement data 42 is partially displayed in a different chart. All three stimuli were measured in units of cd/m2. Figure ό shows the default backlight setting to 0, for the level (192, 192, 192) (that is, the medium-high gray level), the measured stimuli across the display Υ (also known as luminance L) ·). The display is 1920x1080 resolution, the χ coordinate range is [0,1920], and the y coordinate range is [0,1080]. The origin of the display is the upper left, the X coordinate level is increased to the right, and the y coordinate is increased vertically to the bottom. The voltage setting is a "normalized" (i.e., linear proportional and shifted element) -16 to +16 unitless range, with -16 corresponding to near 〇. Figure 7 shows a 3D chart of the same data. Figure 8 shows a 2D contour plot of the chrominance 値(xc, yc) across the screen at the same level. In order to make the chart clearer, the contour is removed. For all levels, a 2D/3D chart can be obtained. Figure 10 shows a graph of the luminance (Y) R of the R, G and B levels of a subset (grid point) of the 7x7 pixel position of the backlight set to 0, 8 W (grayscale). (A larger number of grid points are actually used in the calculation, but only a smaller subset is shown in the chart for clarity). As shown in Fig. 9, the grid point position is equally spaced from 中央 and y from the center and superimposed on the image of the display. In this typical data, this level is taken (for 8-bit display): ··· · Measure JJ1 level: {32,64,96,128, 160,192, 224,255} (23) For a given level, in different pixels Measurements can be seen as being vertically displaced from each other (ie, 'the pixel's Y is different'). If there is no non-uniformity, all pixels will be consistent for a given level. Figures 11 and 2 show the same evidence for the three stimuli 値X and Z. It should be noted that the highest contributions of 'X' Y and Z are derived from the R, G and B components, respectively, as predicted from the tristimulus response score 201243808. In addition, the data average follows a power law functional form. As can be seen from the different charts, in brightness and color, the display has non-uniformities across all levels, characterized by changes in XYZ or medium. At medium-gray levels (192, 1 pair 192), uniformity in luminance and color, and corresponding non-uniformities, are shown in Table 1. Table 1 - Uniformity statistics before calibration. Halo% uniformity % Non-uniformity Average tristimulus 57X 57.56 42.44 37.53 cd/m2 Tristimulus 値Y (luminance) 58.73 41.27 40.58 cd/m2 Tristimulus 値Z 52.07 47.93 52.12 cd/m2 ** Chromatic Heart 93.85 6.15 0.2882 ** Chromaticity W 91.03 8.97 0.3119 The non-uniformity in the triple stimuli is very large, and the luminance across the display is only 58.73% uniform. The non-uniform number of chromaticity coordinates is misleading (hence, ** mark) and suggests that the color is uniform. Since the chromaticity 値 is ~0.3 order, even when the non-uniformity is significant, the % uniform enthalpy calculated from the equation (4) is large. For chromaticity, a change of 0.02 steps is visible. Therefore, considering the three stimuli is more accurate, the non-uniformity in X and Z here will result in a perceptible significant non-uniformity in the color. In a similar manner, non-uniform chirps can be calculated at all measurement levels. A particular measurement display has a very large amount of non-uniformity in luminance and color, which is represented by an LCD display. Here, the data from the example experiments are used to describe various embodiments of the present invention in detail with reference to FIG. Changing the backlight voltage setting, at the same 7x7 pixel position, for the gray levels 192 and 255, the sample results can be seen from the graphs of Figures 13 and 14, respectively. The voltage control can be set to [0Λ16]. As discussed in (21), the same settings are applied to all LEDs. For a given set of pixel locations, the non-uniformity is treated as a three-way vertical shift. Unlike the difference across the pixel ( (Figure 10-12), the difference in voltage space appears to be very linear. This is a common behavior of LCD backlight units and can be used to simplify response calculations. Figure 15 shows the difference in luminance Y as a function of pixel 値 (for pure gray level) and backlight settings - these points are connected by the mesh 201243808 mesh. The power form in the pixel space and the linearity from the voltage space are clearly visible. After the physical measurement is completed, the next step is to reconstruct the display response 43 and determine the response function 44. This essentially means converting the discrete measurement grid data 42 to a functional form (11) so that corrections can be calculated for all levels, all LED voltage settings, and all pixel locations. First, different response functions 44 are defined, each physical quantity XYZ, having an independent response function 44 (i = X, Y, Z): (24) The response function 44 can be decoupled to two components. As shown in Fig. 1 'from the architecture on the LCD display, the backlight panel 10 and the LC panel 16 continue to function. The backlight panel 10 is a light source that produces a front incident light onto the LC panel 16 'and then' corrects the light with pixels ,, the final response being the product of the light from the backlight and the pixel correction. This means that the response can be written as the product of the backlight response and the pixel response mine, where the former depends on the voltage and the latter depends on the level:

Ft (X, = Ft (x, y, hx Ftp (X, y, 6 (25) In addition, the product form of (25) means that the response can be studied at a fixed voltage] and then functionally connected in the voltage parameters Fixed voltage response. This connection can be made during the calibration phase. Thus, the response required for each data set at a given voltage is as follows: Calculated for each setting, less, 6, 〆 (26) lit By reducing the fact that each pixel operates alone in the LCD display and is unaffected by neighboring pixels, the reduction can be further reduced. The correction at a given pixel depends only on the response at that pixel, which can be calculated separately for all pixels. Response and Correction. The (x,y) correlation of the response can be reduced with an understanding of the following steps calculated at each measurement pixel location: Factory ((5), calculate each measurement pixel position (xfl, h) And each set (27) Next, the superposition principle can be used to further simplify the form of response. The XYZ response of any RGB color according to equation (17) is the sum of the components of R, G and B. Let β be the measured quantity, _e {U, phantom response function 44, which is applied pure level; '£{/^, 5} The results are then (27) 20 201 243 808 means:

Ft (3⁄4 = FtR (R) + (G) + FtB (β), ie {X, Y, Z} - Red component response function green component response function (28) if - blue component response function simplifies response function 44 The determination is to look for a 9-function treatment that follows the data described in Figures 2-12 of Figure 10-12. These responses are set in a fixed backlight and can be called pixel responses because they rely on pixel color levels. In the luminance correction requiring gray level (W=R=G=B), (28) is reduced to a single function:

Fy(^) = Frw(W), ^=(W, W, W) (29) The function (29) will follow the gray level data in the first graph of Fig. 10. Gray-only luminance correction is a common requirement for non-uniformity correction in LCD displays, especially when it comes to speed and cost. In the unified method of the present invention, only luminance correction is processed in the same scheme. Similar to the gray level Y response, the functions of X and Z are defined π), however these are rarely used in practice. As mentioned above, voltage correlation is taken into account after pixel correlation. This requires backlight (voltage) correlation, that is, the function represents data at a particular level in Figures 13-14. The backlight response can be expressed as: X = FJ (Π - backlight set X response y = < (Π - backlight set Y response (30) Z = 砑 (1〇 - backlight set z response although the pixel response is in Fixed voltage, but the backlight response is indeed at a fixed color level. For the most common corrections, use only a pure white backlight response (Figure 14) or a small gray level backlight response. Similar to the pixel response, the backlight response is Determined at each pixel position (~, %). The simplification of the response functions 44 to (28) depends on the validity of the superposition equation (Π). Given the superposition error, the black level produced by the light leakage is destroying the equation. One factor. The possible source of discrepancy is the difference between the display RGB primaries or the camera filter and the ideal state. Therefore, it is important to adjust these actual differences before calculating the response. In order to adjust the black level or other cold At each 倜 level, the sum of the XYZ measurement r of r, G, and B is compared with the XYZ measurement 同一 of the same gray level (r=G=B=W), and the difference due to non-ideal behavior is called 21 201243808 Do △ superposition (m s,...), is: AXs=Xw-(Xr+Xg+Xb) = ^W ~{Yr+^G+Yb) (31) = Z妒-(Zfi + ZG + Zfl) After averaging multiple measurement pixels The difference between the three stimuli 可见 can be seen in Figure 16. In general, the deviation from the ideal characteristic is very small except for the high level Y and Z. When calculating the pure R, G and B responses, the other 2 The pixel component is culled (set to 〇), however the light still leaks out of the rejected liquid crystal. After adding these three responses, the light leakage (also called black level) is compared to the example of calculating the corresponding gray level. It is incorrectly increased by a factor of 。. For luminance γ, the pure gradation is actually larger than the sum of 和, which is understandable because the total channel may be 〇' the intensity of the recording may be slightly higher. In order to correct the superposition mismatch, the measurement data is adjusted by this △ to ensure the matching of the R ' G, B and W data. The correction can be written as: corrected) = 忑 (corrected) = Z, + _rz (AZs (32) Ύ'/χι =Σ^< = R,g,b

The i i I r-factor determines the extension of the superposition correction in the R, G and B components and can be programmable. Setting them all to 〇 means that the overlay error is not corrected. For example, the following extensions can be used:

rXR = ^rXG ~rXB VYR ~ rYG ^ rra = (33) rZR = rZG ~ ^^rXB = ^ Λ'- The extension is based on the fact that X and z are more weighted towards R and B, respectively. And Y is more symmetrical about G. The data with the applied black level compensation correction for the triple stimulus 値Z is shown in FIG. It is possible that this correction results in XYZ negative 値 - these can be approximately equal to 〇 or alternative extensions can be used to avoid negative 値. This will be used to complete the correction (33). Various responses have been defined and the function 44 can be calculated using a method of data modeling. The two main methods associated with the present invention 22 201243808 are: data fitting or interpolation; and using known mathematical models to represent the data. However, it should be understood that any data modeling method can be used. In the first method, the data points are fitted and interpolated by a response function 44. Fitting is preferred because it is less susceptible to the effects of measurement errors. If the data is known to be very accurate, 'interpolation can be used. In the current typical discussion, the “least squares fit” method is used to model the data. The business suite software can be used to perform a least squares fit. The commonly used basis for fitting is a polynomial basis. In the exemplary illustration of the invention, a stereo (cubic) polynomial fit is performed on the data 'but the number of times in the equation is general (denoted as d). The deformation of the fitting method is to use a series of fitting functions; that is, the responses are represented by different fitting parts. Piecewise linear functions are just examples. Since the correction at a given level is local (which can be considered to be less variable), the function that best represents the response can be corrected at different levels. There is no need to use global functions. For a level (such as 255), the first polynomial works best, but for another level (such as 192), different polynomials can be used. The second method employs a specific model based on the known display. This method is especially useful if only a small number of data points are available or if the data points are unreliable. For LCD displays, the intensity is considered to be in accordance with the power law. This makes it possible to estimate the response function 44 from as little as one measurement level in the red, green, blue or white. If the luminance at the whitest WTM» (255) is Y-', then the power law function can be used to estimate the response function (29):

(w Y F^W)^ (34) \frmax / Power 値, called gamma r, is about 2.2. The actual system offset equation (34), so if accurate data is available, it is preferred to use a fit. However, equation (34) or similar mathematical models are usually the best solution due to speed and limited measurement limitations. In (34), inhomogeneity is inherent in the difference of Υπ- of all pixels. The model can be boosted _ black level compensated to (34), which gives a model that matches W=&5 and W=0. ^ These two methods can also be combined. One level can be corrected, a better result can be obtained using the fit, and at another level, the power law is optimal. Assuming that enough data is available, a three-fit fit of the first method can be implemented to determine the response function • 44. For the 7x7 pixel position of the exemplary embodiment, these have been shown above in Figure 18 for the measurement data for luminance (three thorns 23 201243808 値 γ). The grayscale response function is located on the first chart, the red response function is on the next chart, in a clockwise order, and so on. The response function 44 can be written as: F(10) π(Β)=ΣίΑ, (35) These functions can be calculated at each measurement pixel position (χα, _η). Similarly, the backlight response function is generated by fitting the data in Figures 13-14. The backlight response of the three stimuli 电平 at level 255 is shown in FIG. Since the correlation is linear, a linear polynomial can be used. The response function can be written as:

Fvx{V) = axvxV^axvo F;(V) = aYvlV + ary〇 (36)

Fzy(V) = azy]V + 4〇 〇 is a linear fit coefficient. These responses can also be calculated at all measurement pixel locations. The global response shown in Figure 18 is all a strictly monotonic function of the pixel ( (increasing as the pixel 値 increases) with a mathematical positive reciprocal. However, a global monotonic function may not be possible because the underlying data is not monotonic. This does happen for LCD displays, especially when low or high, the data may not be monotonic. Difficulties in capturing data will also result in non-monotonic data. The Z tristimulus data in Fig. 17 shows such non-monotonicity. Depending on the degree of non-monotonicity associated with the function 値, computational correction can be made difficult and result in erroneous results. If non-monotonicity becomes a problem, there will be a possible solution. One is to define the response function locally in the RGB space, as opposed to globally defining the response function, where it is monotonic, and then resolves the neighborhood correction at the corrected level. This is acceptable if the correction is within the monotonic neighborhood. A second possible solution is to shift and scale the global response function to make it monotonic. This is acceptable if the shift/scaling does not cause the response to significantly deviate from the real flaw. For example, for R to take the Z-stimulus, the global maximum and global minimum 回应 of the response are expressed as Zmx and Zmin, respectively, at Rzmax and Rzmin. The shifting element and the scaled response function are given by:" 24 (37)201243808

Pzr(R) = F R

+ and 2min

J β (9)=z—(and max ) = Zmax In this example, R-=255 and the maximum 8 bits are 値. This assumes that there is no local maximum/minimum 在 between global max/min 値, which is usually the case. In 〇 and ^-function (3" can be normalized to z〇 and B by: FZ\R) = Γ ζ, -Ζο ] f IT ( p and Z max and Z min ^^max ~~^ Min j V \ D , eight max /

+ Z0

J

FzR(〇) = Z0 (38) Sometimes this is used in calculations to bring data at R=〇 to 〇. In the following, it is conceivable to perform any non-monotonic adjustments and use the same symbols, if desired. Response functions (35) and (36) fully characterize the display for all level and backlight settings. In the next step, the recalibration phase 45, these responses will be used to construct the correction grid 46. The correction grid 46 provides a new set of RGB turns at each measurement point, which will cause the physical quantity or response to be constant in (x, y), i.e., uniform across the display. The measurement 値 will only depend on the level ^ and the backlight setting 〆 (see equation (10)). This uniform 値 lookup is calculated from the measured data. The most common choice is to take the average 値, the minimum 値 or the maximum 値. Let M* denote any measured quantity {X, Y, Z, x#} at the pixel position (~λ). Then you can write the required uniform amount: 1 Ν, Ν, —- ϋ NxxNyj^ti Average 値: Tolerance = (39) Minimum 値: M = min({Me), a = l".A^,6 = l."A^) Maximum 値:Λί = max({M0*}, a = = 1.. JVy) These quantities are calculated for each level & (foot and voltage Shanghai, however, in order to make the symbol simple, the correlation is not clearly shown. In the illustrated example, the average 値 and minimum 将 will be used. The barnotation details, F, 4, and fork} will represent the uniform enthalpy found by the correction. • For a fixed voltage, the recalibration step 45 is explained below. Assume a fixed level (or 25 2 〇 1243808 for gray level luminance) Correction, (?=(妒见趵), find a new pixel based on position 値5 = ' (or W), which will produce a position-independent uniform response. It should be noted that the level being corrected need not be measured ( 18) level, since all RGB値 responses have been calculated in (35). In mathematical terms, the following systems for solving nonlinear equations are required: I) Total level luminance + color correction R' = R + Ar, G' = G + Ac, B' = B + Ab (40) π) Only the gray level of the luminance correction: W-^W' γ=Σί〇^'η (41) (42 (43) (x, y) spatial correlation is inherent, where the coefficient depends on (x, y), that is, for each pixel position (xa, h) 'solve the above system. New 値 (R', G', B') is considered not to be significantly different from (R, G, B). The symbol (W~) represents a change in pixel ,, in which case the equation system can be represented and solved by the symbol. LCD, reporting luminance and chrominance (XcycY) instead of XYZ is a standard operation. These two descriptions are equivalent and can be converted between the two by equation (3). Similarly, the expression in the form is: ,Σ1〇^(〇ηΣΣ^Σ1〇<„(〇λ

Pc (44) ϋ=ΣΙΣΙΛλ^υ Here, m=l, 2, 3 indices correspond to R, G and B, C; W, c; = CT and C; = 5', and use (3) Similar to the definition, the magic to (39). In the case of using the average, 26 (45)201243808 one has · ί:Σ- ΐ ΐ Ι w xab

Xc Nx Parent Ny 〇=' b=\ Xab + Yab Ten Zab •' — Yab,, — (45) The right-hand side of the expression can be seen as the response function of the chrominance ,, however, they do not follow the component RGB color The principle of superposition (increase) of degrees. Therefore, the method described here (where the response function is defined in the form of XYZ and later § ten) is a more efficient and manageable method. Equations (41) and (44) define a system of three nonlinear equations that need to be solved. Once the calculations are made for all positions, levels and voltages, this will give the correction grid 46 写 The substitution formula ((41) and (43)) can be written directly in the form of ΧΥΖ 値 値. First define the inverse function of the response function: // // // c = R, G, B (46)

There is an inverse function of the response function (35), which can be calculated as a function of the RGB level by inverting (35) calculations or by fitting the RGB level as a function of the measurement XYZ. The same applies to the symbol (〇R, G, B): A: at level 〇 you) the measurement of component c: A for uniformity at the level C component C required pair | 4: at level c = Frc ( c part of the measurement of c r値G: for uniformity at the level c component c required r 値 (47) zc: at the level c = Fzc (C 汾 quantity C measurement Z 値 Z 沁 for uniformity in electricity The system of the required C値 equation (41) for the flat C component C becomes: 27 201243808 X = X'R+X'G+XfB Ϋ = Υ^+Υ'+Υ;

Z = Z'R+Z'G+Z'B 〇=/;σ'Λ)-//(Ζ'Λ) 〇= fGxdfGy(r,G) (48) 〇H)H) 〇=fBxm-fs (y'B) This is a system of nine unknown equations of unknown (XW0). The last 6 ancestors impose this limit (the actual independent variable is still RGB) so the counter-responses must be equal. Solve (48) and give the new RGB値 as: R'=fHxm=frAy\)=fRz(z\) G'=f〇x{x〇) = f^rG) = f^Z'G) B '=fBxm = f^rB) = fBz(rB) Similar to (48), an equation in the form of a chromaticity coordinate can also be obtained. Since this formula hides RGB correlation, it is preferred to use it in conjunction with (41) and (43). However, it should be noted that the two are equivalent. Equation (41) describes a system of three nonlinear equations in three variables and can be solved using a known nonlinear digital optimization program. However, for instant computing, this is not easy to implement. By using a suitable linear analogy, (41) can be converted into a linear system that can be solved quickly. Back to (41) you need to solve for each level being corrected. It is expected that the new enthalpy is in the neighborhood of the level being corrected. In particular, (Δ«, ‘Δ«) will be small, and the response function will be locally analogized to a linear function in Δ. Thus, for the correction level (R, G, B), the response is expanded as follows: = Χ 〇α^(-^ + ΔΛ)η 1 + «-^]=Ση=〇α».Λ +^R^n =〇a^R n 咖,)=+~)η * ΣΐΛβ”(ι+»苦)=+作·)=Σ:〇々5+Δβ)””Σ计Σ: Then equation (41) The 3χ3 linear system that becomes the equation: 28 201243808

A χΔ = b (51) where:

(52) fx-x] Σί〇<^ Σ:-^-'η Σί〇<Β^η) Ϋ-Υ ? ^XYZ = Zy^ Σί〇αΥ〇^ . Σΐ〇αΙηΒη-'η 7- Z \Δ Δ) X-〇<Rn-ln Σί〇α〇^η-'η Σί〇<Β"\ This is solved by the inverse matrix ,, giving the new RGB値 as follows: It can also be calculated based on the equation ( 44) The linear analogy of the chromaticity, where (44) becomes: 4^χΔ = 6.

XyY (54) here

Axe=xc

X

X + Y + Z δλ = yc- γ Χ + Υ + Ζ ΑΥ = Ϋ - Υ / 2 7 + Γ + Ζ, / „ = wide V w = foot

i=XJ, Z 4= Σΐ. (广广1 «, 4 = ΣΐΧ" (4) "_1«

'-Axcx/2, -Aycx/2 fS=S+L· g

(53) 29 (55) 201243808

\ldac+X)IR-IXRI (IAxc + X)Ig-IxgJ (/Axc Axyy= (IAxc+Y)Ir-IyrI (I/Sxc+Y)IG-IYaI (IAxc+Y)IB - IyBI

The total amount in { J 4 and & is known, and the solution is obtained again by the inverse matrix:

(56) (R) S'=S+L· g 不管 Whether based on (53) or based on (56). # are available. The chrominance 计算 calculated from the actual, fork) and from (foot F, Z) will usually not be the same, and the resulting RGB 値 will be slightly different. The derived equation provides full correction of brightness and color. If for the gray level, only the brightness is to be corrected, which is often sufficient for many consumer displays, then the recalibration step 44 will be simplified to solve only (43). This is a single nonlinear equation in a single variable and can be solved using a standard polynomial square root algorithm. Or, you can fit W as a function of Y, which is a similar reaction to (46) /; and directly read W, 通过 by evaluating /; (『,). Need to pay attention to unity and find solutions outside the domain. If the response is as in (34), further simplification can be obtained, and then an explicit formula for recalibration step 44 can be obtained: For scenarios where speed is critical and data is limited, equation (57) provides correction for grayscale A fast simulation of flat luminance. The solution of (56) can be calculated for each level at each pixel location (xa, h). For each correction level, this provides a set of grid points in the (X, y) space, '卩& thus the correction grid 46, which is expressed as: ((R&k, G heart, (π) The subscript refers to the corrected RGB 値 at the position, input color (level) f. For pure level luminance correction, similarly has the correction grid: 201243808 (59) {Κα„} (59) W indicates the gray level (W, W, W). For each correction level, each component (RGB) in the (x, y) space will be illustrated as a 2D surface. In practice, most (> 9亮度%) brightness and color correction focus on correcting gray level. The instructions in Figure 6-8 are for gray level, ie (192, 1 double 192). For the same LED display, Figure 20 and 21 shows a corrected R, G, and B grid for the 31x31 uniform spatial point ' at gray levels 192 and 195. The R surface has a '〇' mark with a 'G surface without a mark and a B surface with a mark '·'.値 will be used as a uniformity calculation. The surface shape of level 192 is compared to the XYZ surface shown in Figure 7, and we see that the RGB surface is essentially "reversed" with the XYZ surface. In the example, all calculations are performed in a fixed backlight setting applied to all light source components (eg, a single LED), also known as a common or initial backlight setting. So 'the correction is pure pixel correction, no digital signal is corrected If this is sufficient to achieve the desired uniformity, then there is no need to adjust the backlight voltage. For level 192 (Figure 20), this is sufficient because all new pixels are in the range of 8-bit [0···255] However, for level 255' as shown in Figure 21, pure pixel correction requires many pixels away from the center, which have a number 値 outside the range of 8 bits (> 255). This corresponds to the display not so Bright area. Because 'this is not possible for 8-bit displays (these will be cut to 255), it is impossible to achieve uniformity correction by adjusting only the pixels of high gray level. One solution is at (39 In the case of uniformity, the minimum 値 is taken. If χ#Υ is used, then only the minimum 辉 of the luminance Υ is needed. This will cause the pixel 値 to be lowered to match those pixels with lower brightness. If the brightness loss is not The solution is acceptable, however, if the average brightness is maintained at a high gray level, the backlight must be adjusted. Similar problems can occur at low gray levels, especially at 0 levels, where the correction grid can be Sending some pixels smaller than 0 is not possible for LCD displays. This will correspond to a brighter area on the display. In this case, the possible solution is at the expense of increasing the black level. The maximum 値 is taken in (39). Increasing the black level is not preferred because it reduces the contrast. Another possibility is to ignore the black level in the correction, but this is not the ideal solution. However, due to the single LED control, Black level and correction can be obtained by dimming the LED. Coordinating the LEDs from 255 and 0 will produce backlight correction, where the LED has its own voltage correction - 31 201243808 "Voltage Correction Grid". If a single LED is not adjustable and only global backlight adjustment is possible, the voltage correction can be based on the maximum change required to achieve uniformity at level 255 or ( (considering all pixels outside the range). First consider the general situation of local LED control. In the example of backlight adjustment, backlight correction as described in detail herein is required. This correction is made at a specific voltage level V, which is also referred to as an input or global backlight setting, or a brief backlight setting, or some normalized unit. The common voltage V is applied to all pixels. In the backlight setting V, pixel correction is determined for gray levels 255 and 0 (58 or 59 as analog 値). If no pixels are out of range, the backlight correction is consistent, ie the voltage is not corrected from V. It should be noted that this will depend on what uniformity metric is used from (39). From the calculation 値 (58), identify those points where the RGB components are outside the range. These points are marked as (in the case of 8-bit 値): Level (?=(0,0,0): (4,%) e {(x0, less A), a = 1"JVX,6 = 1... meaning} such one or more of the following points <〇, G'- <0,%A- <〇p _ level ί = (255,255,255): (;cf, less f) e {〇ca,h),a = 1..JVJ = 1···'} One or more of the following points: >255,%? >255,%ft- >255 The required level of uniformity is also used in equation (60), for levels 〇 and 255, respectively, labeled F°, F255, and for other quantities, as well. The positions of levels 0 and 255 are usually different ( Next, for the pixel identified in (60), the backlight response function (36) (shown in Figure 19) is used to determine the desired voltage. The required voltage can be given by solving the following equation:

At(^,^): Y°=FYy(V)

At(xf^f): F255 =F/(F) , · (61) An increase in voltage will result in an increase in luminance and a small change in chrominance, so we only consider Y値 when determining the new voltage level. The solution, expressed as, Qiao 5}, is 32 201243808 y〇 _

V〇— j ~ aVQab Vab ~ ~~T QV\ab f255 _ Y (62) ^S=—T ^ av\ob Note that the response function depends on the pixel position, and we have added additional markers to the function coefficients to Indicate this. When solving 4 and ^ respectively, the backlight response of level 〇 and 255 is also needed. If the positions of levels 0 and 255 are the same (this is rare), average 値 or other combinations may be used depending on the particularity of the correction (eg correction at 0 is more important, or at 255 minimizes loss of brightness, etc.) ). The full set of voltage adjustments are: Correction voltage: {&} at 吣W): at (Of): original F F-^+F_f ab ab (63) For some (ά, ΐ) 'if (xa?, K) = (xf ) Use one of the following: b 2 For most corrections, the two sets of positions do not need to be the same. The position of the 匕 has been incorporated into the levels 0 and 255, and is written (\-, slice), and removed 〇 and 255 subscripts.値(63) provides the voltage settings required for a particular pixel location. Usually, this does not need to correspond to the LED position. The pixel resolution is much larger than the LED resolution, that is, the number of LEDs. A single LED, combined with the scattering effect, illuminates many pixels. The LED voltage is Vi, i=l...L (see (6)) - such that the position of these LEDs is (xu, yu), i = l · · 1. The LED is assigned a voltage 通过 by averaging the total 最 which is closest to the LED position. Depending on the type of averaging, different "smoothing" of the voltage correction can be obtained. A little such a poor law was discussed. For each adjusted voltage position (1⁄2, slice), the closest LED, use a simple averaging method and assign it to it. The LED can obtain multiple assignments from different locations, denoted as 乂. From position, slice) W = i. · The voltage assigned to LED i is labeled ^Heart}. The final voltage used is the average 这些' of these giving the following corrections: 33 (64) 201243808

Vt ->K i = ^-L 1 Ny, ^Vi >1

The LED that Ke is not affected by will maintain its original voltage 値v. If the pixel position (\,Λ) sample reaches the resolution level of the LED and is positioned similar to the LED position, then the number of allocations from each (1⁄2, slice) will be essentially one. In this case, the above average becomes very important when selecting the closest & for each LED, that is, there is no sum. In fact, in the calculation of backlight correction, the sampled pixel locations may be small corresponding with certain subsets of LED positions. This speeds up the calculation. The deformation on (64) is based on the distance to the LED to assign weights ~'' V; = - Ϋα, Κ (65) NVi % v,} % weight depends on | (χ»(Ά-)| For each 匕 - Instead of picking up the nearest LED, it is possible to take the nearest n LEDs with the appropriate weight - this is essentially a variant of (65). The voltage can be calculated from the angle of the LED ❶ for each LED, within a certain distance, take the appropriate Weighted sum of hearts:

The palsy «(4)A ά<, ύ α (4) weight depends on d = ||(Wi,.)-(wA-)| (66) These methods are all different types of averaging. A slightly different approach is to interpolate or fit the smoothing function of 値 on (X, y). This gives a 2D voltage surface (which can be evaluated at (gas, %)) to determine the voltage at the 1st 1^0. If the function is expressed as simplistic, it has: F = heart (none, less): fit or interpolate, less defective) V>=FwixLnyu) (67) Fit is more preferable than interpolation because it Includes smoothing. 34 201243808 Calculate this correction voltage 値(10) for a specific initial backlight setting v. Each backlight setting will have a different set of correction voltages calculated using the above steps. Typically, this calculation is done in the set of voltages in (21) and the interpolation is interpolated between levels. It is also possible to use the same relative correction for all backlight settings. The backlight setting 匕, Z = 1··. 校正 correction voltage group can use the earlier vector symbol to indicate before correction: 甙}—after correction: (3⁄4 'V;: vsi, f;= V; 2 Μ Μ Λ η. Λ...Ny (68) Here, % means that for the jth LED at (\ is), the backlight is set to r = L, and the correction voltage obtained from one of (64) - (67) is obtained. This correction is equal to the change in GK at the jth LED for the initial voltage h 4. If there are RGB LEDs, the three will apply the same correction, ensuring that the backlight correction only adjusts the brightness and does not introduce colorartifacts. In theory, backlight correction can also vary with input pixel levels. However, the out of range condition (60) is primarily determined by the maximum and minimum gray levels. Due to monotonic responses, The color level will be in a range that is provided by the highest and lowest levels in the range. Other levels do not need to be considered separately to test out-of-range conditions. Thus, although levels 0 and 255 are used to determine their own Correction, but the voltage correction is independent of the color level. For uniformity correction, in a fixed backlight setting, the voltage can be adjusted once and the voltage does not change when the input pixel changes (unless the global backlight changes). Next, the pixel correction (58) can manage the desired content on its own. Correlation changes. The benefit of backlight correction independent of color levels is that uniformity correction will not interfere with other LCD display features such as local dimming, high dynamic range imaging. These features are all content (color levels) Relevant, and compete with non-uniformity correction (if the non-uniformity correction is color dependent). The backlight discussion in the exemplary embodiment focuses on the example where the LEDs can be individually adjusted, that is, in the example of a direct backlight configuration. As mentioned earlier, this method can also be used for individually controllable CCFL tubes or side-fired LEDs. The main change is that adjusting the LED or tube will affect a larger number of images, and this needs to be considered on average. If local adjustment is complex 'may be in, based on a side-lit backlight or tube backlight, global correction can be used Yan Guang, the possible choices are:

, average 値G Ρ = ·maximum 値kr (69) ‘min 値~ This selection is again specified by the calibrated feature. It should be understood that a large number of variations are possible, and depending on the calibration requirements and the above factors (calculation speed, number of levels being calibrated, etc.), different combinations will provide the best results for different standards. In particular, backlight correction can be skipped entirely if a minimum uniform amount of very fast luminance correction is required. Due to the limited resolution of LHD (CCFL is more limited), backlight correction (68) provides a coarse correction for brightness uniformity. Only backlight correction is insufficient to achieve a high level of uniformity, and in particular, it does not provide color uniformity correction. Roughly also means that the change in LED is more pronounced than the change in pixel 値. This means that once the LED is adjusted in accordance with (68), the pixel correction needs to be recalculated using (35). In order to make this process fast and efficient, the first pixel correction before the backlight correction calculation can be used as a rough estimate of the measurement levels 0 and 255 for a small number of points using the approximate 値(34). The backlight correction is calculated and the LEDs are adjusted, and then (35) will be used to make a more detailed calculation at more levels. If necessary, repeat the approximate pixel correction and backlight correction before proceeding to the detailed calculation to ensure that all are within range. An important benefit of the present invention is that it unifies all of the critical components that need to be corrected, i.e., backlight sources and pixels, in the same framework. These two components affect uniformity in an interdependent manner and must therefore be processed simultaneously for optimal results. The method described herein for generating backlight correction does not rely on known PSF LEDs and/or any scatterers. . It uses one or more simple iterations to determine backlight correction, and then determines the exact match pixel correction. This is very beneficial in practice because the PSF and diffuser effects are very difficult to determine accurately. This method is also very practical for the manufacturing process, where the details of the various optical/electronic components (LEDs, diffusers, claddings) typically provided by different suppliers are not required. This correction can also be applied in certain areas where backlight control is available - most displays allow for global backlight setting control. In one embodiment, a mathematical approach can be used in the less common case where PSF is available. Although in practice, the more "experience" approach described above is more expensive, it is expressed here as a formula. According to equation (25), the response is divided into a backlight component and a pixel component: (7〇) For backlight correction, only the luminance response is considered. The voltage dependence appears only in the backlight response ^ (^ less /). The response component <b, 3^) is independent of the voltage - this is not the same as the full response, 〆), and the full response is based on the measurement data of the fixed backlight color setting. In particular, 50c, ;; /) as the voltage changes, but a constant 'this component 〇,;;, (?) will be called the basic pixel response, the basic pixel response can be expressed as in (35), the following Give its representation:

Fy(^)^lX^L(x,yXcmy (? = (<^〇(()) This will still determine the coefficient of the defined response ^^ force. The function written by the coefficient more clearly represents the space Correlation. Now go back to 疔(Xj/), so that the normalized PSF for the ith LED includes the effect of the diffuser or any other component. This is, the PSF incident on the pixel panel 'passes through the diffuser Etc. At this time, only the ith LED is lit. Now, the assumption is known. The contribution of the ith LED with voltage to the backlight response is: Κ^^^Ρχχ, γ) (72) This assumes the voltage The correlation is linear, as the LCD, the display expects, and is shown in Figure 19. The backlight response is the sum of all LEDs: FY = ^{ViPi{x,y) (73) Next, the full response becomes : 37 201243808 ^(x^//) = £:;^.(x^)x£=lX;=〇a:(^)(Cj« (74) This means at any pixel position (x,y) and Any LED voltage 値, the complete luminance response of any RGB input. As mentioned above, this is rarely a known amount of reasoning. According to the component response (35), the full response of the fixed voltage 确定 can be determined from the above measurement data. The superscript # indicates that this is Specific responses calculated voltage must be equal to the measurement (74) and providing the following constraint equation:

Fy (x, y, ί, Ρ) = Fy (χ, y, ί) Σ# you, βχΣΐ=ιΣ:=Χ»(4)(cm)”=El=iELxfa,>o(cj” (75) coefficient Rewritten as 'to the two out of them depending on the position (χ, y) and the applied voltage Shanghai. There is a separate equation (75) at each position, and the equation (75) needs to be solved independently. In addition to the coefficients, All of the quantities in (75) are known. For all Jay, it is necessary to maintain this equation, which is only possible if the coefficients of each (CJ) term are equal. This gives siLkW The solution is as follows:

Kn(x,y)-^nix- (76) After the determination, the full response (74) is a known function and can be used to solve the backlight correction. In principle, (74) provides a solution for pixel and backlight correction, which includes X and Ζ three stimulus responses) is the solution to the following equation: (77) ^=T:m^y>Tm^lA^y\ cmY 1=Σ,ZL,Zl〇«i (^^xcj" Its constraints are:

0<Cm <24" -1,W7 = 1...3 V^[V,-S,V^Sli = \...NL 38 (78) 201243808 - The first constraint indicates that the pixel is located in the range [〇, 255], and the second constraint indicates that the solution of the voltage 値 is also within the same effective range. Equation (77) is a nonlinear system of the equation and exists as an independent equation at each position (\, _η). Thus, we have a 3 X % X nonlinear equation system in 3 <<><<\> variables, which come from different Cw値 for each position. This is a complex system that will require solutions, especially under time constraints. Alternatively, the effective two-step method described above can also be applied here. First, the solution of (56) is calculated for the gray level 〇 and 255 on the basis of the luminance constraint (or (57) is approximate 値). These can be expressed as G and. At these locations, G and Cf fall outside the valid range, with ciipped and c: c:<o^c°m=o

Cf >255 II C: =255 (79) If these defects are in range, then no clipping is required. Instructed to consider these limits as G and G55. Then, by solving the following equation, these adjustments are used to solve the corrected LED voltage π値: Strict 5=S, ye(x,)« (so) As previously known, each initial backlight can be set independently. The π値 group; looking back at this, when calculating C and C5, the common voltage is applied to all the LEDs. Using the same symbol in (68), write the correction voltage for the initial setting 4 (where 'j is the set index, 丨 is the LED index): Next, the corrected backlight voltage can be calculated from: Σί=1 Δ7<^&lt ; (Z!m=l Σμ=0 amn (Χο»}>b )(C° )n jf' -^NL ^ ^ ^ ^ a = l...Νχ, SJ i-1 is different from (77) This is a linear system of equations that is easier to solve. Each position (, 4) exists - for such an equation, so the heart is replaced, giving the mail variable heart (should be 39 201243808 note that each jth setting is separate Equation linear system in processing. The standard method previously mentioned, simultaneous processing of backlight correction and pixel correction is very clear in equation (82). Only one of the two equations in (82) can be used. 'For example, if the correction at 255 is more stringent, then only the second equation can be used. For LCDs, other optimizations can be used to make the solution (82) simpler. The number of points, and the number of equations can be reduced, so that The number of variables will be greater than the number of equations. This ensures that there is usually a solution. This simplification is always possible because the LED is looking rough. The brightness improvement varies over a large area. If the number of variables and equations are the same, then a system with a rectangular matrix can make this position correspond to the LED, which simplifies the matrix by reducing the correlation of adjacent LEDs. The role of the LED can be limited to the nearest LED 〇 which allows (82) to enter: 'block-diagonal' format (takes the effect from only 3 LEDs as an example):

«Η «12 «13 0 0 0 Λ, Μ 卜] «21 α22 αη 0 0 0 Λ Δ;2 b2 «3. «32 «33 0 0 0 Λ △>3 h 0 0 "43 α44 «45 0 Λ = 0 0 «53 «54 α55 0 Λ Δ;5 0 0 fl63 α64 «65 0 Λ K Μ Μ Μ Μ Μ Μ 〇 J 1 mJ , mJ (83) This system can be solved using well-known mathematical methods. The coarser the correction required, the more simplifications that can be made. Only the most recent LEDs can be used to greatly simplify (82). As shown above, when the PSF is known, the backlight correction calculation can be simplified to solve the problem in (82). In practice, PSF calculations are very complex and often difficult to implement, in which case the iterative method provides a faster, achievable alternative to determine backlight correction. At the end of the recalibration phase 45, pixel correction And backlight correction is known. The pixel correction 6 is used as a set of correction pixels 値 for the level on the entire grid of points ka) (3⁄4 common backlight setting 4. The backlight correction is as dependent on the common backlight setting ^ Correcting the LED voltage group is provided. The data is described as follows: 201243808 Pixel Correction: S = (R, G, B) = (3⁄4 Light correction: 1X1 K.1 (84)

The η index j indicates the initial common backlight setting. The next stage of reconstruction converts the grid data into a functional form 48. For pixel correction, this provides a new pixel for all levels, all pixel locations, and backlight settings. For this backlight correction, this will provide a new LED voltage 値 at any given common backlight. The constructor form essentially refers to the conversion of data from discrete points in different spaces to a continuous function using some fitting or interpolation methods. This is similar to the response from a set of point builds (35). The form of the function also depends on the hardware of the next and last application phases. A general form with very efficient hardware implementation has been introduced in 7,324,706. The general formula is introduced and included here. The range of all independent variables is divided into regions, and individual functions are fitted or interpolated into data (84) in each region. Similar to the response function, the fit is preferred and a polynomial basis is used. Pixel correction is first considered. Starting at the pixel position, the pixel space (X, y) is divided into 2D patches, and the polynomial is fitted to the grid points on each tile. The continuity of the function ensures that it spans small pieces. The number of patches and the fit are adjustable so that the polynomial represents the grid points very accurately. If the number of grid points on each tile is equal to the number of polynomial coefficients, the fit becomes an interpolation function. A variety of software programs are available for fitting and interpolation (eg MATLAB Spline Toolbox) ° (The result of the x, y-fitting is the following functional form: r R^=Tt〇T^Lxmyn: Q(4)=ΣίοΣίΑ〆/ ( 85) Here, dx and dy are the number of polynomials in X and y. It should be noted that the fit of (x, y) eliminates the Μ 'index and turns the discrete correlation into a continuous correlation. Level and backlight settings The discrete correlation 41 201243808 is still unchanged. The next fit is done in pixel pupil space (RGB) using a correction grid (58) calculated for different levels. The different levels are uniform. The RGB space is also divided. Make a small block' and fit the polynomial on each small block. Here, the small block is actually 3D cube because the RGB space is 3D. The result of the fitting is a functional form: print, from = (86) public times ( Common degree) d is used as an RGB fit. A similar form can be found for G and Β. By considering only the gray level, for the most common example of uniformity correction, (86) can be simplified to: R'jix^R) = T,^:JLtaUxmynR, (87) We use (88) to simplify the symbol. This only discrete correlation remains on the backlight setting. By dividing the backlight control into 1D tiles and fitting on each of these tiles is eliminated here. Chasing · R\x,y,R,V) = Σ;Ι〇Σ"〇Σ!:〇Σί〇^^>ηΛ^* G'(x,y,G,V) = ΣΖ〇Σί〇 Σί〇Σΐ〇α^χ^η^ (88) B\x,y,B,V) = ΣΐΧ,Σΐ^:-^χ^ηΒ, ν" V is used again in backlight control in (90), Backlight control may be in some normalization units. Equation (88) is written in the form of a general formula, but by using a linear polynomial (d = 1) or other optimization, it can be more simplified. Also, if the same correction is used for all backlights, the V correlation in (88) is eliminated. The number backlight correction function opens in the same way. Use the data calculated in (68) to fit the electricity; Μ data to the backlight control function, i is the LED hang ''...Nl (89) W) = Y^〇ajvJ^ i The function needs to satisfy: ν; (ν,) = νβ, j = \...Nv, 42 (90) 201243808 This backlight correction should be viewed as giving a corrected LED voltage in response to backlight settings or controls, which can be used to initiate a common voltage or some common Voltage dependent normalized amount. Equations (88) and (89) give the final pixel correction map and the backlight correction map for the uniformity correction process. They are rewritten as follows: ^ JC(XM=Σ;Ι〇Σ:=〇Σΐ:〇Σ::〇^^>η^^ κ=^(η=Σ·

, dL \vj *mntk *mntk *mntk 2B *mntk (91) i = LNt (91) The first expression needs to be understood as the component-wise (on the right side) of the evaluation (for example, alone) In R, G and B). This provides a very compact form of correction for the coefficient set, which can be conveniently stored and can be evaluated in hardware. The final stage 49 involves applying the correction map to the LCD display using a suitable hardware platform. This diagram can be easily applied using an FPGA due to the highly compact representation. The FPGA design basically consists of a multiplier and an adder that evaluates the above functions for inputting RGB signals and backlight control. Patent application 11/649,765 describes a hardware architecture implemented in an FPGA or ASIC for applying pixel correction in a fixed backlight setup. The architecture can be used for evaluation of backlight images and addition of pixel maps on backlight control. Function Correlation. For displays used in the specific embodiment (Figs. 6-8), the system 100 based FPGA is used to apply the correction. The tristimulus 値 XYZ and the same level are corrected using the method described herein. The chromaticity diagram at (192 '192 '192) is shown in Figures 21-22. The improvement in uniformity with the corresponding graphs in Figures 6 and 8 is clearly visible. Table 2 shows Statistical data after correction. Table 2 - Non-uniformity statistics before calibration Physical quantity % Uniformity % Non-uniformity Average 値 % Uniformity increased % Non-uniformity reduction Tristimulus 90 X 90.88 9.12 37.77 cd/m2 57.87 78.50 Stimulation 値Y (Hui.degree.) 91.31 8.69 41.00 cd/m2 55.48 78.95 43 201243808 Triple stimuli 87Z 87.35 12.65 52.49 cd/m2 67.75 73.61 98.45 1.55 0.2878 4.90 74.78 **Color y. 96.48 3.52 0.3124 5.99 60.80 Last two columns The percentage change in uniformity and non-uniformity is given. After the correction, the uniformity is significantly improved, and the luminance increases from 58.73% uniformity to 91.31% uniformity. This corresponds to a uniformity increase of 58%, that is, > 1.5X. Equally, the non-uniformity seems to drop by 79%. Similar improvements can be seen from the X and Z tristimulus, in particular, the Z uniformity is increased by 1.67 times, which is the key to chroma uniformity. The chromaticity coordinates also show a large reduction in non-uniformity, and more importantly, the △ simulation in (&, y.) is (% non-uniformity X average 値) is now less than 0.01, making the perceived chromaticity Uniformity across To further verify this, the distance is calculated in the perceived CIE L*u*v* space, which is the perceptibility measure of the color difference. The distance of <1 is considered imperceptible (the rain color will look It is the same), here, the 値 near 22 is considered to be perceptible, although it should be noted that this is a big one. In practice, color differences outside this range may or may not be perceived. In order to obtain an estimate, the following "perceptible surface" is generated. For each pixel (assuming (X., y.)), calculate the Δ^ν値 between L*u*v*値 of the <v and all other pixels, for a total of WxH (display resolution), all Average and assign to pixels (9), y. ). This process is repeated for all pixels and totals the distance calculated for all possible pixel pair combinations. This provides a surface in the (X,y) space that indicates the average perceived color distance between the pixel and the other pixels. The surface before and after the correction is shown in FIG. A plane with 値 2 is shown for reference. Prior to correction, a plurality of pixels have a chirp near 4, where all pixels are S1 after correction. Figure 24 clearly shows an improvement in color uniformity after the correction is employed. The average metric in (39) has been used for this correction and it is expected that the average 値 will not change. This can be verified by comparing the third column in Tables 1 and 2, which further provides support for the linear analogy in (50). The compact nature of the correction of equation (91), that is, only the storage coefficient & 〆 ,, means that any external variable that can affect uniformity can be corrected by storing the specific system, array associated with these variables. The example shown in 25 can be calculated for different ambient light levels or different temperatures 62. When ambient light changes, a suitable set of coefficients 64 can be loaded into processor 66 and applied to display 68. Ambient Light Coefficient The calculations for the group are accurate as described above, the only difference being the measurement of the amount of the 2012 43808. Another common external variable is the ambient temperature of the light source components, especially the LH) response is known to vary with temperature. Different corrections can be calculated and It is applied to different ambient temperatures. These corrections are calculated as above, but are now made at a specific monitored temperature. The present invention provides an accurate and efficient method of greatly improving brightness and color uniformity in a backlit LCD display because the correction is on the display The output is completed, that is, what the viewer sees, which corrects the non-uniformity of all sources. By using the three-stimulus, the observation can be accurately considered. Perceived uniformity. This specific method provides correction maps for pixels and light sources in a unified framework. Various embodiments provide different optimizations that can be used to make the method simple based on specific criteria (eg, computational speed, correction type, etc.) While the above description provides examples of the embodiments, it should be understood that certain features and domain functions of the above-described embodiments may be modified without departing from the spirit and the principles of operation of the embodiments described above. The description of the preferred embodiments of the present invention is intended to be illustrative and illustrative, and it is understood by those skilled in the art that the invention can be made without departing from the scope of the invention as defined by the appended claims. Modifications and Variations [Simplified Description of the Drawings] In order to better understand the embodiments and domain-related implementations described herein, and to more clearly show how they are effective, reference may be made to the accompanying drawings. In the drawings, at least one exemplary embodiment and/or related embodiments are illustrated, in which: Figure 1 shows LCD panel backlights for direct LED illumination (1-A) and CCFL illumination (1-B) sources. Although LEDs are shown as RGB, white LEDs are also available. For OLEDs, the backlight and display panel are the same; Figure 2. Two types of LED backlights, direct and side-lit; Figure 3 shows an exemplary prior art color and brightness correction system; Figure 4 illustrates the steps for color and brightness non-uniformity correction in the present invention; Figure 5 is a side elevational view of the color and brightness non-uniformity correction system of the present invention; Figure 3 is a diagram of three stimuli (stimulus valu6) of the gray level 192 measured prior to calibration in an exemplary experiment. 2D contour map; 45 201243808 Figure 7 is a 3D contour plot of three stimuli 値 XYZ as a function of pixel position of gray level 192 measured prior to correction in an exemplary experiment; Figure 8 is in an exemplary experiment 2D contour map of gray level 192 chromaticity 値(xc, yc) measured before correction; Figure 9 shows a 7x7 grid point map on the display selected for the exemplary experiment; Figure 10 is for Measurement of the grid points of Figure 9 at pure gray, red, green and blue levels (Blue level) Figure 3 is a graph of measured tristimulus 値X for the grid points of Figure 9 for pure gray, red, green, and blue levels; Figure 12 is for pure gray, red , green and blue level, graph of the measured tristimulus 値Z of the grid points of Figure 9; Figure 13 is the measured tristimulus 値 XYZ of Figure 9 as a function of the backlight voltage setting of the gray level 192 chart. This voltage 値 is within the normalized range; Figure 14 is a graph of the measured tri-stimulus 値 XYZ of Figure 9 as a function of the gray level 255 (pure white) backlight voltage setting. The voltage 値 is in the normalized range; Figure 15 is a 3D graph of tristimulus 値 γ (brightness) as a function of backlight voltage setting and gray level signal; Figure 16 is the measured gray level and RGB from the corresponding measurement The difference between the level and the calculated tristimulus △ (Δ); Fig. 17 is a graph of the tristimulus 値Z with the superimposed correction included; Fig. 18 is the pixel response function for the calculation of the tristimulus 値Y (brightness) Figure 19 is a graph of the backlight response function for the calculation of XYZ triads for gray level 255 (pure white). The voltage 値 is within the normalization range; Figure 20 is a 3D graph of corrected RGB 用于 for gray level 192 as a pixel position number; each R, G, and B is labeled as a separate plane 'its upper right The party shows a description;

Figure 21 is a 3D acid for correcting RGB turns of gray level 255 as a function of pixel position; each R, G, and B is a standard; Figure 22 is a third of the measured gray level 192 after correction. 2D contour map of the stimuli 图; Fig. 23 is a 2D line diagram of the chromaticity 値(xc, yc) of the measured gray level 192 after the correction; 46 201243808 - Fig. 24 is the position before and after the correction, across the pixel position Calculated, given a perceived difference in ΔΕ*ιιν値. Also shown is a constant 値 "2" surface with a higher than perceived and below imperceptible ,, with a description of 3 surfaces at the top right; and Figure 25 shows a storage correction for ambient light and temperature Coefficient selection. It should be understood that the elements shown in the figures are not required to For example, the dimensions of some units may be exaggerated to make them clearer than other units. Further, it is considered to be appropriate that in the figures, reference numerals may be repeated to refer to corresponding or similar elements. [Main component symbol description] LCD panel 50 Reference image 51 Capture device 52 Processing unit 54 Hardware processor 56 Input image generator 58 Light source 10 or 11 LED 12 Light modulator 16 Direct LED backlight 20 Side-lit backlight 30 Physical quantity 41 47

Claims (1)

201243808 VII. Patent Application Range: 1. A method for improving color and brightness uniformity of a liquid crystal display backlight by a plurality of light source components, the method comprising: providing at least one common backlight voltage for all light source components; Each of a backlight voltage setting displays a plurality of reference input images having a predetermined platform RGB pixel 横跨 across the display for at least one signal level; selected on the display a set of grid points to measure display uniformity in response to the plurality of reference input images, the uniformity being characterized by at least one tristimulus 辉, luminance, and chrominance components; generating a display response function from the measured uniformity data, The response function is a response sum of the R, G, and B components of each pixel of the display; a pixel corrected grid data map is generated, the pixel corrected grid data map being generated at each of the at least one signal level Constant uniformity of all pixels; converting the pixel corrected grid data map into a set of coefficients · Forms, and the pixel correction function is applied to the input signal and a display for the position and color of all the pixels Zhi. 2. The method of claim 1, wherein the response function is further dependent on a backlight voltage setting, and wherein the corrected grid data comprises a backlight correction map applied to the backlight component. 3. The method of claim 2, wherein the pixmap is regenerated after the application of the backlight correction, and the backlight and pixel correction are iterated until the desired uniformity is obtained. 4. The method of claim 2, wherein the backlight response function is expressed as a linear function. 5. The method of claim 2, wherein the backlight correction is obtained using a tristimulus point spread function (PSF) of each light source component. 6. The method of claim 5, wherein the PSF is proposed as a specification for a light source component. 7. The method of claim 5, wherein the psf is approximated using a mathematical 48 201243808 model. 8. The method of claim 5, wherein the psf is directly measured. 9. The method of claim 1, wherein the reference input image is a plurality of pure gray levels for luminance correction only. 10. The method of claim 1, wherein each of the light source components is a light emitting diode (LED). 11. The method of claim 1, wherein the light source component is a cold cathode fluorescent lamp (CCFL) tube. 12. The method of claim 1, wherein the light source component is a clockless diode (LD). 13. The method of claim 1, wherein the reference input image is a plurality of levels of pure red, pure green, and pure blue levels. The method of claim 13, wherein the reference input image further comprises a plurality of pure gray levels, and the plurality of pure gray levels are performed to compensate for the LCD black level of the bile component. Offset. 15. The method of claim 1, wherein the display response function is obtained using a multi-turn fit to the measurement grid point. 16. 雠• Turn the 15th face, where the polynomial is three-sided. 17 'root 1 tears the lion, where the leaky leg measurement grid point data obtains the display market response function. 18. The function according to the scope of the patent application 丨j; the order-du-side-term of the method wherein the function is estimated by the function of the model. A method of 1 leg, wherein the number of back faces is fitted to the === partial polynomial such that the local fit is a monotonic function. A method as described in item 15, wherein the response function is shifted and scaled to monotonize the function. 21. The method of claim 7, wherein the response function is constant 49 201243808 Uniformity 値 is set to an average measurement 値. 22. The method of claim 1, wherein for a high signal level, the constant uniformity of the response function is set to a minimum measurement when the corrected pixel 値 is above the allowed bit range. . 23. The method of claim 1, wherein for a low signal level, the constant uniformity of the response function is set to a maximum measurement when the corrected pixel 値 is below the allowed bit range. . 24. The method of claim 1, wherein for a plurality of pixel correction maps, the plurality of coefficient sets are calculated and stored corresponding to settings of a plurality of ambient temperatures. 25. The method of claim 1, wherein for a plurality of pixel correction maps, the plurality of coefficient sets are calculated and stored corresponding to settings of a plurality of ambient lights. 26. An electronic system for improving color and brightness uniformity of a backlit liquid crystal display, wherein the system comprises: a display panel comprised of a pixel array, each of said pixels being characterized by a controllable digital RGB ;; a backlight panel composed of components' each of the light source components is characterized by an adjustable voltage control; an image generator unit for displaying a plurality of reference input images on the display; a grid point group for selection on the display Measured display uniformity in response to an image capture and measurement unit of the plurality of reference input images, the uniformity being characterized by at least one of tristimulus 辉, luminance, and chrominance components; a first processing component for The measured uniformity data generates a display response function and calculates a pixel corrected grid data map that produces a constant uniformity of all pixels at each of said at least one signal level; a second processing component 'for converting the pixel corrected grid data map into a functional form represented by a set of coefficients; and a pixel correction function And a display input signal is used for all pixel positions and colors Zhi. 27. The system of claim 26, wherein the response function is further dependent on the backlight voltage setting, and the corrected grid data comprises backlight correction applied to the backlight voltage control fJ 201243808 - . 28. The system of claim 26, wherein each of said light source components is a light emitting diode (LED). The system of claim 27, wherein the LED component is disposed in a direct backlight architecture. The system of claim 27, wherein the LED component is disposed in a side-emitting backlight architecture. The system of claim 26, wherein the light source component is a cold cathode fluorescent lamp (CCFL) tube. 32. The system of claim 26, wherein the light source component is a laser diode (LD), the system of claim 26, wherein the first processing component and The second processing component is integrated into one processor. The system of claim 26, wherein the first processing component is a software tool that runs on a computer system. The system of claim 26, wherein the second processing member is integrated into the display. The system of claim 26, wherein the capture device is a two-dimensional (2D) camera. 37. The system of claim 26, wherein the capture device is a illuminometer. 38. An electronic system for improving color and brightness uniformity of an organic LED display (OLED), the system comprising: a display panel comprised of a pixel array, each of the pixels being an OLED component and controlled by digital RGB Characterizing; an image generator unit for displaying a plurality of reference input images on a display; measuring grid uniformity for selecting a set of grid points on the display to respond to image capture of the plurality of reference input images撵 and measurement unit 'the uniformity is characterized by at least one tristimulus 辉, luminance and chrominance component; 51 201243808 First processing means for generating a display response function from the measured uniformity data and calculating pixel corrected grid data The pixel correction grid data map produces a constant uniformity of all pixels at each of the at least one signal level; and a second processing component for converting the pixel corrected grid data map into a set of coefficients The form of the function represented; and the application of the pixel correction function to the input signal and display for all pixel positions and colors 39 39, root The system of claim 38, wherein the first processing member and the second processing member are integrated in a single display. 40. The system of claim 38, wherein the second processing member is integrated in the display. The system of claim 38, wherein the capture device is a two-dimensional (2D) camera. 42. The system of claim 38, wherein the capture device is a illuminometer. 43. A method for improving color and brightness uniformity of an organic LED display (OLED), wherein the method comprises: displaying a plurality of reference input images on a display, the reference input image having at least one signal Level, predetermined platform RGB pixels 値; selected grid point groups on the display measure display uniformity in response to the plurality of reference input images, the uniformity being at least one of three stimuli, luminance, and color Character component representation; generating a display response function from the measured uniformity data, the display response function being a response sum of the R, G, and B components of each pixel of the display; calculating a pixel corrected grid data map, the pixel Correcting a grid data map to produce a constant uniformity of all pixels at each of said at least one signal level; converting said pixel corrected grid data map into a functional form of a set of coefficient representations; and applying a pixel correction function to Input signal and display 'for all pixel locations and colors 44, according to the method described in claim 43 of the patent application, wherein Reference plurality of the input image is merely a pure gray level luminance correction. The method of claim 43, wherein the reference input image is a plurality of levels of pure red, pure green, and pure blue levels. 46. The method of claim 45, wherein the reference input image further comprises a plurality of pure gray levels 'performing the plurality of pure gray levels to compensate for LCD black levels of RGB components Offset. 47. The method of claim 43, wherein the display response function is obtained using a polynomial fit to the measurement grid point. 48. The method of claim 47, wherein the polynomial is a cubic function. 49. The method of claim 43, wherein the display response function is obtained by interpolating the measured grid point data. 50. The method of claim 43, wherein the display response function is estimated by modeling the function according to a power law. The method of claim 43, wherein the response function is a plurality of local polynomials fitted to the neighborhood of the measurement grid point such that the local fit is a monotonic function. 52. The method of claim 47, wherein the response function is shifted and scaled to monotonize the function. 53. The method of claim 43, wherein the constant uniformity 値 of the response function is set to an average measurement 値. 54. The method of claim 43, wherein for a high signal level, the constant uniformity 所述 of the response function is set to a minimum measurement when the corrected pixel 値 is above the allowed bit range. . 55. The method of claim 43, wherein the low uniformity 値 of the response function is set to a maximum measurement for a low signal level 'when the corrected pixel 値 is below the allowed bit range 値. 56. The method of claim 43, wherein for a plurality of pixel correction maps, the plurality of coefficient sets are calculated and stored corresponding to settings of a plurality of ambient temperatures. 57. The method of claim 43, wherein for a plurality of pixel correction maps, the plurality of coefficient sets are calculated and stored corresponding to settings of a plurality of ambient lights. 53