Method and system for determining color coordinates of light emitters of matrix display

Abstract

The present application relates to a method and system for determining the color coordinates (XYZ) of the light emitted by each emitter (R, G, B) of a matrix display (M) using already available information about the color coordinates (XYZ) of similar light emitters, the matrix display has similar light emitters having color characteristics similar to those of the emitters of the matrix display but not those present in the matrix display, the method comprising: acquiring at least two intensity images of each emitter of the matrix by different filtering, at least one of the intensity images is acquired by filtering by at least one of the acquisition filters, and determining, for each transmitter of the matrix, an estimated color coordinate of the transmitter based on the information that has been available and the intensity images.

Description

Technical Field

[0001] The present invention relates to matrix displays, such as are found in large display panels used for displaying information outdoors. Background Art

[0002] Matrix displays typically include thousands of pixels made of red, blue, and green light emitters (e.g., LEDs) connected to one or more electronic drivers. Such displays can "trick" the eye into reproducing digital image content that will accurately correspond to the captured digital content. The accuracy of the colors of the reproduced content is key to the perceived visual quality of the display.

[0003] Fine control of RGB color mixing is critical to ensure that the human eye cannot notice the difference between the original scene and the reproduced digital content.Observer metamerism may increase the complexity of mixing RGB components.

[0004] Matrix displays include active displays and passive displays.

[0005] A passive matrix display corresponds to a display in which each pixel maintains its current state until addressed by an external drive circuit. In a passive display, this circuit drives the transitions between pixel states, whereas in an active matrix display, each pixel is individually addressed by a switching device that actively maintains that pixel state while addressing other pixels.

[0006] An example of a passive matrix display driver is given in US11361707.

[0007] LEDs are typically powered by a pulsed current of constant intensity (constant current pulse width modulation) or, as described in US11361707, by a combination of pulses of variable intensity.

[0008] Light emitters such as LEDs have spectral characteristics that vary with current, temperature, and time, and two light emitters with the same manufacturing number may exhibit slightly different spectral characteristics due to manufacturing tolerances.

[0009] Under certain conditions, such as when displaying pure colors, these spectral differences can produce localized variations in color that are easily noticeable to the human eye, thereby negatively affecting display quality.

[0010] In order to optimally calibrate an LED display, it is critical to accurately determine the spectrum of each individual LED to predict the visual effect on the observer's visual system due to the significant dispersion of LED characteristics across the display. Although such measurements are possible using scanning spectrometers, imaging spectrometers, or hyperspectral cameras, such instruments are typically expensive or do not provide adequate performance. The optical dynamic range required for a display with a dynamic range of, for example, 1,000,000:1 in the prior art is not easily achievable.

[0011] Therefore, it is necessary to calibrate the matrix display by measuring each light emitter of the matrix display individually to adjust the control signal sent to that light emitter in order to obtain a full reproduction of the desired colors.

[0012] As explained above, conventional display calibration techniques utilizing, for example, a spectroradiometer may prove to be prohibitively expensive and time consuming when applied to matrix displays having a very large number of pixels.

[0013] To solve this problem, US8246408 proposes a method for calibrating such a display. The method includes using multiple spectrometers mounted on a movable head to simultaneously measure the Normalized Power Spectral Density (NPSD) function of multiple light emitters of the display. Then, the panel of the display is imaged to measure the relative brightness between each LED of each color, and a standard brightness meter is used to determine the average absolute brightness of the entire display. Calibration data is calculated based on the measured chromaticity and brightness of each light emitter.

[0014] This approach still involves the use of multiple spectrometers to calibrate any new matrix display and requires a movable measurement head, so it remains relatively complex and time consuming.

[0015] US6222631 discloses a relatively complex two-dimensional spectral sensor having multiple bandpass filters and associated lenses to form multiple images of a sample on an imaging device, thereby allowing the spectral characteristics of the sample to be measured.

[0016] US5432609 and US20140300753A1 define solutions for performing 2D color measurements that can be described as conventional solutions, which are known in the prior art. In the described methods, RGB filters alone (or in groups in a typical Bayer arrangement) are in a first order estimated to have a spectral transmission sufficiently similar to the spectral response of the human visual system (a simple model of the human visual system), where each of the three types of cones responds to a different frequency range known as short, medium and long distance (LMS model).

[0017] Assuming the measurement apparatus is calibrated, the measured energy filtered through the RGB filters is considered to be close enough to be approximated to XYZ values ​​(absolute CIE 1931 color space).

[0018] Given a certain type of emitter, the conventional method of US5432609 cannot accurately measure slight dominant wavelength deviations between similar spectra. More specifically, since the transmission of the RGB filters in certain areas of interest is flat, the energy measured by the camera through those RGB filters will be minimally affected by slight shifts in the spectrum.

[0019] Taking into account the typical sensitivity of color cameras, the minimum dominant wavelength shift detectable by this method is greater than 0.6 nm near the green sRGB primary or even 1.2 nm near the red sRGB primary, or at least + / - 0.003 CIE2000 distance, which is similar to the advertised performance of conventional 2D colorimeters.

[0020] US20140300753A1 and US2021215536A1 describe similar methods for calibrating a standard color camera to use it as a 2D colorimeter. The correlation matrix between the RGB values ​​and the XYZ values ​​measured by the color camera is intended to improve the measurement accuracy by calibrating the color camera across a larger set of reference points (measured by both the spectroradiometer and the color camera). Summary of the invention

[0021] There remains a need for improved calibration methods and systems for calibrating matrix displays having many light emitters in an accurate, fast and efficient manner.

[0022] An exemplary embodiment of the invention relates to a method of determining the color coordinates of light emitted by each emitter of a matrix display, the method using already available information of similar light emitters having similar color characteristics to emitters of the matrix display but not being emitters present in the matrix display, the method comprising:

[0023] a) acquiring at least two intensity images of each emitter of the matrix by different filtering, at least one of the intensity images being acquired by filtering by at least one acquisition filter,

[0024] b) determining, for each emitter of the matrix, estimated color coordinates of said emitter based on said already available information and said intensity image.

[0025] The at least one collection filter preferably has an attenuation that varies by more than 10%, such as by more than 20%, or even by more than 25% or 30%, as the wavelength varies by 25 nm around a central wavelength of light emitted by the light emitter.

[0026] "Similar light emitters having similar color characteristics" include light emitters having the same manufacturing number or a manufacturing number that the manufacturer claims is equivalent.

[0027] The already available information may include the light energy of the similar light emitter filtered by the corresponding collection filter; in a variant, the light energy is calculated based on the knowledge of the transmission properties of the collection filter. The already available information may also be a spectral model of the emitter.

[0028] The invention does not require moving the measuring head along the pixels of the matrix display, since the image acquisition device can image many light emitters while being stationary.

[0029] The present invention exploits a priori knowledge of emitters of this type, and the already available information may comprise a reference database of such emitters.

[0030] For example, the reference database may contain information for predicting the color of an emitter of a display to be calibrated when varying the emitter current and modulation width.

[0031] Thus, the method may comprise using the reference database which stores, for a plurality of light emitters, color coordinates of the emitters and reference normalized light energy vectors corresponding to light of the emitters filtered by one or more corresponding collection filters.

[0032] In the present invention, the spectral response of the acquisition filter is used as an encoding function of the emitter spectrum and allows the scalar product of the actual spectrum of the emitter and the traditional color function of the selected color space to be accurately determined from a simple transmission measurement. The present invention is suitable for color calibration of narrow pixel pitch displays. In the measurement method of the present invention, different acquisition filters are selected not for their ability to reproduce the spectral sensitivity of the human eye cone, but for their ability to use spectral transmission as an encoding function.

[0033] Preferably, the collection filter has a sharp transmission slope near the wavelength of interest. Therefore, the brightness effectively measured by the CCD or CMOS camera after filtering by the collection filter is very affected by slight changes in the dominant wavelength of the emitter.

[0034] In an exemplary embodiment, the color accuracy of the present invention is better than + / - 0.0015 in CIE2000.

[0035] The determination at b) may include decoding the transfer function by calculating the moment of the emitter spectrum and calculating the color coordinates (e.g., XYZ values) of the emitter. The energy measured by a monochrome camera through the acquisition filter is not directly related to the XYZ values, contrary to prior art methods where even a simple linear calibration from a few reference points can approximate the XYZ values.

[0036] The present invention does not require expensive spectrometers or cameras, can accommodate a wide dynamic range, and is capable of having 10 -4 The relative accuracy is comparable to that of high-end spectroradiometers.

[0037] In an exemplary embodiment, the emitter is an LED and the color space coordinates (eg, in linear CIE1931XYZ) are estimated solely from a generic LED spectral model and measurements of the absorbance of light emitted by the emitter through the collection filter.

[0038] This is in contrast to the prior art cited above which relies on analysis through filters having a spectral response that mimics the human eye.

[0039] The method may include:

[0040] - calculating, based on said intensity image obtained at a), a corresponding normalized light energy vector for each emitter of the matrix,

[0041] - determining for each emitter of the matrix an estimated color coordinate of the emitter based on reference color coordinates and a normalized light energy vector using an interpolation technique, or using a neural network trained using the reference database to estimate the color coordinates of the emitter based on the normalized light energy vector.

[0042] Once the reference database has been generated, the method may only require the successive calibration of two different displays in order to acquire a plurality of images of these displays using the image acquisition device, for example by only making some changes to the filters upstream of the image acquisition device.

[0043] This change can be easily performed by rotating a movable bracket (eg a wheel) on which the filter is mounted.

[0044] The color coordinates stored in the reference database can be calculated after measuring the emission spectrum of the light emitter with a spectroradiometer. The normalized light energy of the light emitter in the reference database can be calculated based on the emission spectrum of the light emitter measured based on the response curve of the corresponding acquisition filter.

[0045] The method may be repeated for each different supply current of the transmitter.

[0046] "Different filtering" may include not filtering one of the intensity images, or using two corresponding filters, or using two different transmission slopes of the same filter.

[0047] For example, one filter may have a slope from opaque to transparent for one color (e.g., blue), and another slope from transparent to opaque that matches green. Thus, one filter may be used twice during calibration for two different colors.

[0048] The method may include acquiring at least one intensity image through at least one neutral density filter without an acquisition filter.

[0049] Thus, the method may comprise acquiring at a) at least one filtered intensity image of each emitter of the matrix filtered by at least one of the acquisition filters, and acquiring at least one intensity image without an acquisition filter or with different acquisition filters. In an exemplary embodiment, the method comprises acquiring at a) at least two intensity images of each emitter of the matrix filtered by at least two different acquisition filters, the at least two different acquisition filters having a positive slope and a negative slope at the dominant wavelength of each emitter of the same color (e.g. green, red or blue), respectively. The absorption curves of the at least two acquisition filters at the dominant wavelength of the emitter are preferably linear around the dominant wavelength in an interval of + / - 10 nm around the dominant wavelength, with a linearity better than 10%, preferably better than 5%, and more preferably better than 1%. In other words, if A approx (λ) is the dominant wavelength of the filter around the emitter λ 0 =(λ 1 +λ 2 ) / 2 interval [λ 1 ,λ 2 ] is the best linear approximation of the absorbance A(λ) within , which is equivalent to max [λ1,λ2] (|A approx (λ)-A(λ)| / A(λ)]<0.10, preferably max [λ1，λ2] (|A approx (λ)-A(λ)| / A(λ)]<0.05, more preferably max [λ1，λ2] (|A approx (λ)-A(λ)| / A(λ)]<0.01.

[0050] Preferably, the interval [λ 1 ,λ 2 ] has a width of 20nm.

[0051] The method may comprise acquiring at a) at least one filtered intensity image of each emitter of the matrix filtered by at least one transmission slope and acquiring at least one filtered intensity image of each emitter of the matrix filtered by at least one different transmission slope.

[0052] The at least two collection filters may have response curves with opposite slopes.

[0053] The at least one collection filter preferably has a slope of at least 0.05 nm at the central wavelength of the light emitted by the light emitter. -1 , preferably about 0.1 nm -1 Response curve.

[0054] The normalized light energy in the reference database may be calculated from the emission spectrum measured based on the response curve of the acquisition filter.

[0055] The color coordinates stored in the reference database can be calculated after measuring the emission spectrum of the light emitter with a spectroradiometer. The normalized light energy can be calculated given the transmission curve and emission spectrum of the filter.

[0056] The collection filter can be a transmission filter.

[0057] Preferably, the method comprises repeating steps a) to b) for each different supply current of the transmitter.

[0058] Calibration system

[0059] An exemplary embodiment of the invention also relates to a calibration system for determining the color coordinates of light emitted by each emitter of a matrix display, comprising:

[0060] - a memory storing available information about the color coordinates measured for light emitters other than those present in the matrix,

[0061] - image acquisition device,

[0062] - A selectable set of acquisition filters,

[0063] - a processor, the processor being used to control the image acquisition device and select the acquisition filter, for:

[0064] o acquiring by said image acquisition means at least one filtered intensity image of each emitter of said matrix filtered by at least one selected acquisition filter, said memory preferably storing the light energy of said other light emitters filtered by said acquisition filter,

[0065] o calculating, for each emitter of the matrix, estimated color coordinates of the emitter based on the available information stored in the memory and the acquired intensity image.

[0066] Optical Transmitter

[0067] The light emitter may be any type of LED (including organic LEDs).

[0068] LEDs can be primary colors such as red, blue, and green.

[0069] The LEDs can be of various types, such as top-emitting LEDs, flip-chip LEDs, or other types. The LEDs can be packaged in SMD packages or different types of electronic packages, or they can be mounted directly on the screen surface (chip on board).

[0070] The LEDs may include blue or violet LEDs for pumping the phosphor.

[0071] The light emitters may include white emitters and / or cyan and amber emitters.

[0072] Matrix Display

[0073] The matrix display may comprise at least 384 pixels, for example at least 16 columns and at least 8 rows.

[0074] A matrix display may comprise assembled unit elements (also called "tiles" or "panels"), each formed by a smaller matrix arrangement, the matrix display having, for example, 384 to 28,800 pixels, for example 16 to 160 columns and 8 to 180 rows.

[0075] Image acquisition device

[0076] The image acquisition device preferably comprises a two-dimensional sensitive sensor (such as a CCD or CMOS imaging sensor) whose spectral sensitivity covers the required range for calibration.

[0077] The sensor can be a monochrome or a color camera.

[0078] The camera shutter electronics may be connected to the control electronics of the matrix display to ensure stable synchronization between the camera shutter and the time refresh of the matrix.

[0079] The method according to the invention can be applied to both active-matrix and passive-matrix displays.

[0080] Scanning a passive matrix LED screen: The row driver turns on only one row of LEDs at a time. The column driver adjusts the brightness of the LEDs in the selected row. The row and column drivers then switch in sync to control the next row of LEDs.

[0081] The camera is preferably calibrated by performing dark field measurements and white field measurements to improve the accuracy of its measurements. Calibration of the camera may also be aimed at improving its linearity.

[0082] Preferably, more than one pixel of the sensor is assigned to one pixel of the matrix display.

[0083] The camera's resolution and focal length are chosen to ensure good discrimination of each light emitter to be measured.

[0084] To optimize the dynamic range of the sensor, it is usually best to spread the light from one emitter over several pixels of the camera, for example using slight defocus to cover about 4 x 4 pixels per light emitter.

[0085] Another possible implementation is to use a diffractive optical element in front of the camera to spread the Airy disk.

[0086] For example, to calibrate a matrix display of at least M x N pixels, a sensor resolution of at least 5M x 5N is preferred.

[0087] The resolution of the camera can exceed 2K pixels.

[0088] Acquisition filter

[0089] The collection filter is selected to characterize the light emitter around its dominant wavelength.

[0090] The filters act as encoding functions, thereby converting wavelengths into intensity measurements. By combining several such filters (preferably with opposite slopes), and provided that the approximate spectrum of each emitter is known as the available information, the color coordinates of the emitter in any color space suitable for human vision can be derived. The available information can be used as initial values ​​in an iterative algorithm to solve equations that provide actual color coordinates based on the measured intensity and the approximation of the filter's absorption curve.

[0091] The collection filter is preferably a transmission filter, such as a simple colored glass transmission filter. Absorptive glass filters are relatively inexpensive and can be chosen to exhibit a relatively sharp transmission variation across the line width of the emitter being measured.

[0092] The collection filter can also be a color filter or a specially designed interference filter.

[0093] All collection filters may be mounted on a movable carriage (eg, wheels) that is driven to change the filter placed in front of the collection device.

[0094] Preferably, there are at least two collection filters for each primary color. The first collection filter can be a filter with a positive transmission slope from low wavelength to high wavelength. The second collection filter can be a filter with a negative transmission slope from low wavelength to high wavelength. Using more filters with different slopes tends to improve the accuracy of the measurement.

[0095] In other words, for each primary color, the filter set preferably includes at least two filters having opposite slope transmission characteristics versus wavelength to provide good DC bias rejection in intensity measurements.

[0096] Preferably, the transmission of the collection filter varies greatly over the wavelength emitted by the emitter being measured, which means that for the filter to be useful for measurement, the variation in attenuation across the typical line width of the emitter should be large. For typical emitters used in video displays, the half-line width is about 25nm, so in order to specify the slope of the filter, it is preferred that the filter's attenuation vary by more than 10% when the wavelength changes by 25nm.

[0097] Each collection filter preferably exhibits a relatively steep response slope around the central emitter wavelength of the associated light emitter. Each collection filter may have a slope of at least 0.05 nm at the central wavelength of light emitted by the light emitter. -1 The response curve has a slope of about 0.1 nm. -1 , a less steep slope improves linearity. The center wavelength is chosen to match a standard color space, such as Rec2020, where, for example, the blue wavelength is about 470nm, the green wavelength is about 532nm, and the red wavelength is about 635nm.

[0098] The transmittance of the collection filter relative to the wavelength is preferably as linear a function as possible throughout the transition to improve the accuracy of the method. Preferably, for an interval around the dominant wavelength of the emitter of + / -10 nm, the transmittance relative to the wavelength deviates from the linear approximation by less than 10%, and more preferably by less than 5%.

[0099] To best suppress nonlinearities in the filter's response, it is preferred to use multiple filters with different transmittances. Complementary filter sets exhibiting positive and negative curvatures are preferred.

[0100] The method may include calculating the transmission response of the collection filter as a power series.

[0101] The method may include placing a collection filter k around the emitter's dominant wavelength λ 0 The transmittance F(λ) is expressed as the power of λ:

[0102]

[0103] Among them, the measurement intensity Tk It can be expressed as:

[0104]

[0105] Among them, I(λ) LED is the actual spectral irradiance of the emitter.

[0106] The method may include placing I LED The moment of is calculated as:

[0107]

[0108] The measured intensity T k Corresponding to the matrix with those moments M i The product of the vector M.

[0109]

[0110] The method may include calculating the coefficients of the matrix [F i ] and moment M, given the measured intensity T k Vector.

[0111] This available information can provide irradiance model I Model (λ), the model is used to determine a set of best-fit parameters that lead to the moments M estimated from the measurements. i .

[0112]

[0113] The main wavelength of the actual transmitter λ 0 It can be calculated iteratively to solve M 1 ≡0.

[0114] The method may include obtaining a i The value of determines the X, Y, Z coordinates of the emitter in the selected color space.

[0115] Intensity image

[0116] The intensity image is an image of the display's light emitters.

[0117] As described above, the image of a given light emitter of the display may extend over a group of detection pixels (eg, 4x4 pixels) of the image acquisition device. In this case, the sum of the intensity values ​​of the group of detection pixels may be taken as the intensity value of the light emitter.

[0118] The filtered intensity image of an emitter is the emitter image filtered by the corresponding acquisition filter.

[0119] The method may comprise acquiring, for each emitter of the matrix, at least two intensity images filtered by at least two respectively different acquisition filters, and calculating at least two corresponding normalized light energies, which form components of a vector for that light emitter, as explained in more detail below. The at least two acquisition filters preferably have response curves with opposite slopes.

[0120] For example, at least two, better at least three, even better at least four images of red light emitters of a matrix display can be collected by respective collection filters associated with red. Thus, the vector of each red light emitter can have a dimension ranging from 2 to 4, preferably at least 3, and more preferably 4.

[0121] For example, at least two, preferably at least three images of green light emitters of a matrix display can be captured by respective capture filters associated with the color green. Thus, the vector of each green light emitter can have a dimension ranging from 2 to 3, preferably 3.

[0122] For example, at least two, preferably at least three images of blue light emitters of a matrix display can be captured by respective capture filters associated with the color blue. Thus, the vector of each blue light emitter can have a dimension ranging from 2 to 3, preferably 3.

[0123] The method may include acquiring at least one intensity image without an acquisition filter. The intensity image may be used to normalize the light energy.

[0124] The slope of one filter can be used for a wavelength region of one color, and another suitable slope of the same filter can be used for a different wavelength region of another color. In this way, the total number of color filters installed on the device can be reduced without affecting the measurement performance.

[0125] In cases where the brightness range of the matrix display is wider than the dynamics of the imaging sensor, a neutral density filter can be added in front of the imaging sensor to prevent saturation of the imaging sensor.

[0126] Drive

[0127] The drive may be manufactured according to US11361707, which is incorporated herein by reference. BRIEF DESCRIPTION OF THE DRAWINGS

[0128] Figure 1 shows a calibration system according to an exemplary embodiment of the present invention,

[0129] Figure 2 yes Figure 1 A partial schematic diagram of detail II,

[0130] Figure 3 shows the response curve T (transmittance) of an example of a collection filter and the emission spectra of the R, G and B emitters,

[0131] Figure 4 is similar to Figure 3 Another view of the acquisition filter,

[0132] Figure 5 is a block diagram illustrating an example of reference database generation,

[0133] Figure 6 is a block diagram illustrating an example of determining estimated color coordinates,

[0134] Figure 7 An example of a calculated best fit applied to the transmission of a GS16 filter between 500nm and 550nm is given,

[0135] Figure 8 shows the spectral irradiance of green InGaN relative to the fitted LED model,

[0136] Fig. 9 The GS16, BS21, and BS12 filter transmissions and their respective second and third order power series approximations are shown, and

[0137] Fig.10 The selection of two absorptive collection filters for a green emitter is shown. DETAILED DESCRIPTION

[0138] Figure 1 An example of a calibration system 1 according to the invention is shown.

[0139] The system 1 is used to calibrate a matrix display M connected to a driver D.

[0140] The matrix display M includes an array of light emitters that respectively emit lights of various colors.

[0141] Each pixel of the matrix display M may include a triplet of R, G and B emitters emitting red, green and blue light respectively, such as Figure 2 shown.

[0142] The calibration system 1 is intended to determine the color coordinates of each light emitter of a matrix display for at least one power supply intensity of the light emitter.

[0143] The calibration system 1 comprises an image acquisition device 2, which may comprise an imaging sensor, such as a CCD or CMOS imaging sensor, such as a monochrome CCD camera.

[0144] A set of collection filters 3 carried by a movable carriage, such as wheels 4, is present in front of the objective of the camera 2 to filter the incident light.

[0145] Wheel 4 can carry:

[0146] - Collection filter for filtering red light,

[0147] - Collection filter for filtering green light,

[0148] - Collection filter for filtering blue light,

[0149] -Neutral density (ND) filters.

[0150] For example, the wheel carries four collection filters for filtering red light, which can be collection filters of item numbers BG38 (Schott), CM500 (Hoya), RG610 (Schott) and RG630 (Schottt), three collection filters for filtering green light, which can be collection filters of item numbers OG515 (Schott), OG530 (Schott) and B460 (Hoya), and three collection filters for filtering blue light, which can be collection filters of item numbers BG3 (Hoya), EB390 and GG475 (Schott).

[0151] As an example, in Figure 3 and Figure 4 The transmittance T of examples of collection filters B460 (Hoya) and OG515 (Schott) for green are given in FIG. Fig.10 Another example of collection filters BS12 and GS16 for a green emitter is provided.

[0152] In these figures, the emission spectra of the red R, green G and blue B emitters are also shown.

[0153] Each collection filter exhibits a transmittance that spans the central wavelength of the emission spectrum of its associated emitter (in Figure 3 and Figure 4 For each color, preferably at least one collection filter has a positive slope (from low wavelength to high wavelength), such as Figure 4 As shown, the other filter has a negative slope, as Figure 3 shown.

[0154] The acquisition system 1 comprises a computing device 5, such as a computer or a dedicated electronic circuit, which has access to an electronic memory, such as a hard drive or solid state drive or a server, in which a reference database 6 is stored.

[0155] The (unfiltered) light energy E can be defined for sample m with manufacturing number i by mi :

[0156]

[0157] Among them, E mi (λ) is the light energy of the emitter measured without the collection filter.

[0158] The ends of the integration interval may be other than 400 nm and 800 nm as long as they contain the emission spectrum of the light emitter.

[0159] The normalized light energy NLE of the light from emitter m with manufacturing number i filtered by collection filter k with transmittance Uk mik Given by:

[0160]

[0161] The reference database 6 stores, for each sample m of a plurality (p) of samples of light emitters having a manufacturing number i, the color coordinates X of the emitter in a given color space (e.g., CIE 1931XYZ) measured by a spectroradiometer. mi Y mi Z mi and the associated normalized light energy vector NLE mi (NLE mi1 , ..., NLE mik , ..., NLE min ), n is an integer that can vary depending on the color of the light emitter having manufacturing number i, NLE mik The transmittance is U k The normalized light energy of sample m with emitter number i is filtered by filter k.

[0162] For example, for green and blue, n = 3, and for red, n = 4. Using more filters makes the measurement more accurate.

[0163] To measure the color coordinate X mi Y mi Z mi , a spectroradiometer that measures the emission spectrum and calculates the color coordinates can be used, such as the Konika Minolta CS2000.

[0164] This corresponds to Figure 5 The block diagram of step 11.

[0165] To calculate NLE miThe component k of the vector can be used to calculate the light energy when the emission spectrum associated with this component and the transmittance Uk of filter k are known.

[0166] This corresponds to Figure 5 Step 12.

[0167] NLE mi The component k of the vector m can also be measured directly by acquiring the intensity of such an emitter of the emitter samples mi filtered by the filter k using a calibration device.

[0168] Reference databases can also be generated from theoretical transmission functions of filters and theoretical spectra of emitters.

[0169] For a given color, 100-1000 light emitter samples may be measured, so the number of NLE vectors may range from 100 to 1000.

[0170] However, the database can be enriched with theoretical filters derived from a limited database of measured emitter spectra but shifted to different central wavelengths. This ensures that all emitters to be calibrated are within the coverage area of ​​the reference database.

[0171] In an example, for a green emitter, one may have n=3 and three collection filters OG515 (k=1), OG530 (k=2), and B460 (k=3).

[0172] ComponentNLE i1 The normalized light energy, NLE, of the green light of emitter i filtered by filter OG515 (Schott) i2 The normalized light energy corresponding to the green light of the emitter filtered by filter OG530 (Schott), and the NLE i3 May correspond to the normalized light energy of the green light of the emitter filtered by filter B460 (Hoya).

[0173] Therefore, the reference database 6 may contain the following information (for n=3):

[0174]

[0175] The matrix display M is preferably made of light emitters having a manufacturing number similar to that in the database 6 .

[0176] The invention allows the estimated color coordinates of each light emitter q of a given color (eg green) of a matrix display M to be determined without measuring the emission spectrum with a spectroradiometer.

[0177] Instead, all light emitters of a given color of the matrix display are powered with a given intensity and a plurality of intensity images of the matrix display are taken by the camera 2 to enable calculation of the NLE vector for each light emitter q of the matrix display M that is switched on. The image may comprise all light emitters of a given color of the matrix display M; in this case the process may be performed simultaneously for all light emitters of the image.

[0178] Take an image without the acquisition filter to calculate the light energy of the light emitter q:

[0179]

[0180] This corresponds to Figure 6 Step 13.

[0181] The step of acquiring an intensity image using successive acquisition filters for the color corresponds to Figure 6 Step 14.

[0182] This enables Figure 6 At step 15, the NLE of the light emitter q is calculated q The components of the vector (NLE q1 、NLE q2 , …):

[0183]

[0184] Among them U k is the transmittance of filter k.

[0185] Once the NLE of the light emitters q of the matrix display is known q Vector, just Figure 6 Step 16 uses the reference database 6 to estimate the color coordinates X of the light emitter q , Y q , Z q .

[0186] For example, a triangulation is performed using a reference database. The triangulation may be a Delaunay triangulation. A binary barycentric interpolation may then be performed to estimate the color coordinates of the light emitter (e.g., XYZ or Lab color space). An example of binary barycentric interpolation is disclosed in "Barycentric Interpolation" Kai Hormann, Faculty of Informatics, Universita della Svizzera italiana, Hormann thesis (2014).

[0187] Other mathematical or computational methods may be used.

[0188] For example, given the color coordinate X q , Y q , Z q In the case of regression, a fully connected multilayer perceptron (MLP) with an appropriate activation function can be used to compute the NLE vector of the light emitter q.

[0189] MLP can be composed of several dense layers with nonlinear activation functions, such as Rectified Linear Unit (ReLU), LeakyReLU or Hyperbolic Tangent (Tanh). The number of neurons in each layer is determined based on the complexity of the data and the desired performance level.

[0190] Given a limited number of input and output dimensions, a shallow architecture with one or two hidden layers is sufficient for this task. However, deeper architectures can be used if desired.

[0191] Three neural networks, one for each color of light emitter, can be trained under supervision using a reference database 6 .

[0192] The input layer may have n neurons, for example, n=3 for green and blue, and n=4 for red.

[0193] The output layer may have 3 neurons using an appropriate linear activation function to perform the regression task.

[0194] Possible architectures could be:

[0195] Input: 3 neurons

[0196] Hidden layer 1: 8 neurons with ReLU activation

[0197] Hidden layer 2: 4 neurons with Tanh activation

[0198] Output layer: 3 neurons with linear activation

[0199] The architecture can be adjusted according to the complexity and amount of data stored in the reference database 6 .

[0200] These steps are repeated for all light emitters of the same color of the matrix display M, and the same method is performed for all emitters of other colors using corresponding collection filters.

[0201] The color coordinates of each emitter of the matrix display M can thus be estimated for the different currents used in the power supply scheme.

[0202] This enables the establishment of a calibration database used by the screen controller and driver to compensate for manufacturing tolerances by adjusting the individual average current of each light emitter for a given pixel, thereby producing uniform color when displaying pure colors.

[0203] For example, the screen controller and driver will adjust the average current of each LED individually to ensure that image content representing pure green has the same brightness on different panels of the display matrix.

[0204] In addition, the screen controller and driver can adjust the brightness of complementary colors (e.g. red and blue) in each RGB pixel (usually formed by 3 LEDs, one red, one green, and one blue) to ensure that the green to be displayed is the same green on different panels of the display matrix.

[0205] Determine the actual spectrum of the light emitter using the emitter's spectral irradiance model as already available information Example of irradiance

[0206] A. Moment vector

[0207] The light from an LED emitter transmitted through a given collection filter can be approximated by expressing the transmission response of that collection filter as a power series.

[0208] A simple power series is used here as a possible implementation, but other techniques are known to be possible, such as Legendre's polynomials.

[0209]

[0210] Among them, I(λ) LED is the actual spectral irradiance of the emitter, k is the index of the acquisition filter used for the measurement, and T k is the intensity of the emitter as measured through acquisition filter k.

[0211] The acquisition filter k is around λ 0 The transmission of is expressed as a power of λ:

[0212]

[0213] Therefore, the intensity T k It can be expressed as:

[0214]

[0215] Figure 7 An example of the calculated best fit applied to the transmission of a GS16 filter (supplied by Nantong Guoguang Optical Glass Co., Ltd., Nantong, Jiangsu, China) between 500 nm and 550 nm is given.

[0216] In practice, the smooth spectral variation of this filter allows the emission of 0 The exact power series expansion of .

[0217] Will I LED The moment of is defined as:

[0218]

[0219] The measured intensities can be understood as matrices with those moments M i The product of the vector M.

[0220]

[0221] Here, F 1 It is considered to measure the optical power emitted by an LED emitter in the special case where no collection filter is placed in front of the LED emitter and thus provides a standard value.

[0222]

[0223] Therefore, given the measured intensity T k The vector and the calculated [F k i ], Equation 2a can now be used to solve for M, as described in further detail in Section C below.

[0224] This method can constrain the moment vector M by determining the transmission intensity through the collection filter set.

[0225] B.LED Model

[0226] Using available information about the spectrum of the emitter under consideration, such as a spectral model or a database about the spectra of similar emitters, its moment M i Equal to the measured M.

[0227] To constrain the solution, a parametric model of the spectral irradiance I(λ) of the red, green, and blue LEDs can be used so that known nonlinear fitting techniques can be used to determine the M estimated from the measurements. i A set of best fitting parameters.

[0228]

[0229] For example, from a modified Shockley-Read-Hall principle, the following model can be used, which is pseudo-phenomenological.

[0230]

[0231] Ω center =Cled -Ω led ·In(S led ) (3c)

[0232] Among them, A led Essentially represents the amplitude, C led is the dominant wavelength, S led is the entropy of the carrier, Ω led Depends on line width.

[0233] Figure 8 The best nonlinear fit of this model to a typical green InGaN LED used in narrow pixel pitch LED displays is depicted.

[0234] Since the semi-phenomenological model chosen here cannot perfectly capture all physical effects, the difference between the actual LED emitter and its model can be quantified when the model is applied to determine the X, Y, and Z coordinates of the CIE 1931 color space.

[0235]

[0236] Then, the Hunt-Pointer-Estevez relationship is given by the CIE 1931 tristimulus color as perceived by the observer: Vector:

[0237]

[0238] If we calculate the expectation from Figure 8 The measured LED irradiance of the emitter and its model, we find that:

[0239]

[0240] and

[0241]

[0242] This is an acceptable consensus given the limitations of the model and the illustrative nature of this example.

[0243] In a practical measurement system, preferably a large database of similar actual emitters replaces the spectral irradiance model as available information. More elaborate physical models can also be used.

[0244] C. Solution

[0245] There are many different known methods to solve for M.

[0246] For simplicity, a simple singular value decomposition shall be used here to compute the least squares solution to equation 2a. In fact, this method can handle the case where there are more measurements than the number of parameters to be solved, thus further improving accuracy.

[0247] For example, Fig. 9 As shown, the measured vector (T k )for:

[0248]

[0249] The F calculated from the power series of each filter around the wavelength λ = 532.8 nm k i The matrix of coefficients is:

[0250]

[0251] The least mean square solution using the singular value decomposition method gives:

[0252]

[0253] From the moment M i The value of derives the X, Y, and Z coordinates of the emitter in the selected color space. Equation (1) converts the moment M i is combined to the spectrum I(λ) of the emitter. Using a nonlinear best fit method, an optimal emitter spectral model with respect to the metric selected for the optimization algorithm is obtained. A suitable model may be, for example, the Shockley-Read-Hall model as given above. This best fit spectrum is an accurate representation of the actual emitter spectral properties. Therefore, calculations X, Y and Z can then be done according to LMS as described above (see Equations 4a and 5a).

[0254] D. Accuracy

[0255] The estimated transmitted power through the filter can be evaluated for the exact acquisition filter transmission and its approximation as a power series.

[0256]

[0257] To calculate the expected power through the filters for the exact case, as well as a polynomial approximation, the scalar product of the emitter's spectrum with each acquisition filter's response to the second and third order (or its power series) can be calculated, e.g. for a GS16 filter:

[0258]

[0259] The dominant wavelength selected here is, for example, M1 ≡0, corresponding to the wavelength λ of the green LED emitter 0 =532.795nm.

[0260]

[0261] The spectral response of the acquisition filter modeled as a power series can be substituted for the actual response, and when the third order of the power series is used, Equation 1a can be numerically estimated to a few 1×10 -3 relative accuracy.

[0262] Furthermore, in practice, for stability of the results, it is advantageous to iterate the method, thereby calculating the average wavelength of the emitters (i.e., M 1 ≡0).

[0263] Therefore, after each iteration, the best average wavelength is determined and substituted as the seed for the next iteration.

[0264] This exemplary semi-analytical approach shows that the spectral irradiance of a quasi-monochromatic LED emitter can be inferred from a simple set of intensity measurements via a glass filter that serves as a discriminant function.

[0265] This method corrects the nonlinearity in the spectral response of the filter with respect to wavelength.

[0266] The exemplary method relies on inexpensive optical components and cameras and can be applied to color calibration of narrow pixel pitch LED displays. The method is also very fast considering that all pixels of the display are measured simultaneously.

[0267] The invention is not limited to the disclosed embodiments.

[0268] For example, the dichroic transmission filter can be replaced by a dichroic beam splitter cube.

[0269] The dimensions of the NLE vector for a given color may be different, for example four or more instead of three for green. To improve accuracy, it is preferred to collect larger NLE vectors for any given color so that systematic deviations can average out.

[0270] The collection filter may be mounted on a support other than wheels, such as a translatable support.

[0271] The acquisition filters may be fixed on the camera sensors, one for each filter for an embodiment with multiple camera sensors. One possible embodiment is to use a dichroic filter to separate the light to be characterized towards multiple camera sensors.

Claims

1. A method for determining the color coordinates of light emitted by each emitter of a matrix display (M), the method using already available information of similar light emitters having similar color characteristics to an emitter of the matrix display but not being an emitter present in the matrix display, the method comprising: a) acquiring at least two intensity images of each emitter of the matrix by different filtering with acquisition filters, at least one of the intensity images being acquired by filtering with at least one of the acquisition filters, the at least one acquisition filter having an attenuation that changes by more than 10% when the wavelength changes by 25 nm around a central wavelength (λ0) of the light emitted by the light emitter, b) determining, for each emitter of the matrix, estimated color coordinates of said emitter based on said already available information and said intensity image.

2. The method of claim 1, comprising: At a), at least one filtered intensity image of each emitter of the matrix is ​​acquired filtered by at least one of the acquisition filters, and at least one intensity image is acquired without an acquisition filter.

3. The method of claim 1, comprising: At a) at least one filtered intensity image of each emitter of the matrix filtered by at least one transmission slope is acquired and at least one filtered intensity image of each emitter of the matrix filtered by at least one different transmission slope is acquired.

4. The method of claim 1, wherein: The already available information is about color coordinates, and the method comprises: using a reference database (6) which stores, for a plurality of light emitters, the color coordinates (XYZ) of the emitters and a reference normalized light energy (NLE) vector corresponding to the light of the emitters filtered by one or more corresponding collection filters.

5. The method of claim 1, wherein: The already available information is about color coordinates, and the method comprises: c) calculating a corresponding normalized light energy vector for each emitter of said matrix based on said intensity image obtained at a), d) determining, for each emitter of the matrix, estimated color coordinates of the emitter based on reference color coordinates and a normalized light energy vector using an interpolation technique, or using a neural network trained using the reference database to estimate the color coordinates of the emitter based on the normalized light energy vector.

6. The method of claim 1, wherein: The at least one collection filter has a slope of at least 0.05 nm at the center wavelength of light emitted by the light emitter. -1 , preferably about 0.1 nm -1 Response curve.

7. The method of claim 4, wherein: The normalized light energy in the reference database is calculated from the emission spectrum measured based on the response curve of the acquisition filter.

8. The method of claim 4, wherein: The color coordinates stored in the reference database are calculated after measuring the emission spectrum of the light emitter with a spectroradiometer.

9. The method of claim 1, comprising: At least one intensity image is acquired through at least one neutral density filter without an acquisition filter.

10. The method of claim 1, wherein: The image acquisition device is a camera.

11. The method of claim 1, wherein: The light emitter is an LED.

12. The method of claim 1, wherein: The collection filter is a transmission filter. Preferably, for the interval around the main wavelength (λ0) of the emitter + / - 10 nm, the relationship between the transmittance of the filter and the wavelength deviates from the linear approximation by less than 10%, and better less than 5%.

13. The method of claim 1, comprising: At least two intensity images of each emitter of the matrix are acquired, respectively filtered by at least two different acquisition filters.

14. The method of claim 13, wherein: The at least two collection filters have response curves with opposite slopes.

15. The method of claim 1, comprising: Steps a) to b) are repeated for each different supply current of the transmitter.

16. A calibration system for determining the color coordinates of light emitted by each emitter of a matrix display (M), comprising: - a memory storing available information about the color coordinates measured for light emitters other than those present in the matrix, - image acquisition device, - A selectable set of acquisition filters, - a processor, the processor being used to control the image acquisition device and select the acquisition filter, for: acquiring, by means of said image acquisition means (2), at least one filtered intensity image of each emitter of said matrix filtered by at least one selected acquisition filter (3), said at least one collection filter having an attenuation that varies by more than 10% when the wavelength varies by 25 nm around a central wavelength (λ0) of light emitted by the light emitters of said matrix, An estimated color coordinate of the emitter is calculated for each emitter of the matrix based on the available information stored in the memory and the acquired intensity image.

Images