A temperature compensation method for LED display screen

Abstract

A method for temperature compensation of an LED display screen relates to the field of LED display control technology. To address the technical defects of the prior art, namely, the lack of a pixel-level temperature compensation strategy in the existing temperature compensation technology and the inability to effectively handle brightness spikes at the edge of the screen, the present invention provides a technical solution comprising: collecting the cold and hot grayscale matrices of the screen to form a cold grayscale matrix and a hot grayscale matrix; calculating the brightness attenuation ratio of the lamp beads pixel by pixel based on the cold grayscale matrix and the hot grayscale matrix to generate a ratio matrix; constructing and iteratively calculating a correction matrix so that the hot brightness matrix approximates the cold brightness matrix after processing; constructing a temperature compensation surface for the screen and performing normalization processing; calculating the brightness ratio of the screen edge and performing mirror filling on the edge out-of-bounds pixels; and dynamically adjusting the driving current in combination with the temperature compensation matrix to compensate for the impact of temperature changes on color performance.

Description

Technical Field

[0001] The invention relates to the technical field of LED display control, and in particular to a temperature compensation method for an LED display screen. Background Art

[0002] LED (light-emitting diode) displays, a key component of modern display technology, are widely used in advertising, sporting events, stage performances, and command and dispatch. As LED displays evolve toward ultra-high definition, high brightness, low power consumption, and long lifespans, the impact of temperature on the brightness, color, and uniformity of LED displays is becoming increasingly prominent. Over long-term operation, LED chips, due to their own heat and ambient temperature fluctuations, can experience brightness decay, color temperature drift, and uneven brightness between modules, directly impacting the visual experience and product lifespan.

[0003] 1. Current status of existing technology research

[0004] At present, there are mainly the following methods for temperature compensation technology of LED display screens:

[0005] (1) Compensation method based on temperature sensor

[0006] This method installs a temperature sensor on the LED module or driver circuit to collect real-time junction temperature data and adjust the LED drive current to compensate for brightness loss. For example, some high-end LED displays use NTC thermistors or PTC resistors to measure temperature and dynamically adjust the LED drive current through PWM (pulse width modulation). However, this method has the following problems:

[0007] Sensor response lag: The measurement accuracy of the temperature sensor is affected by the environment and it is difficult to reflect the temperature changes at the LED pixel level in real time.

[0008] High cost and complex installation: Each LED module needs to be equipped with a sensor, which increases manufacturing costs and system complexity, making it difficult to apply to large-scale LED displays.

[0009] Difficulty compensating for spatial temperature gradients: Even if the sensor can measure the temperature, it is difficult to accurately reflect the uneven local temperature of the LED and the sudden change in brightness at the module joints.

[0010] (2) Correction method based on global brightness compensation

[0011] This method adjusts the driving current of the entire LED screen or a large area and sets a uniform compensation coefficient based on the brightness measurement results. Common implementation methods include:

[0012] Using the full-screen brightness attenuation model, the variation curve of LED luminous efficiency at different temperatures was measured experimentally, and the compensation coefficient was calculated based on the curve.

[0013] The captured image of the LED screen is analyzed through image processing algorithms, the brightness change trend is calculated, and the PWM duty cycle is adjusted to compensate for the brightness attenuation.

[0014] However, this method has the following main problems:

[0015] It is impossible to compensate at the individual module or pixel level, which can easily lead to uneven brightness in local areas of the screen.

[0016] There is a lack of correction for color temperature drift, especially red LEDs are significantly affected by temperature, and red light attenuation is prone to occur, resulting in color distortion.

[0017] The dynamic compensation capability is limited, making it difficult to respond to temperature changes in real time, resulting in compensation lag.

[0018] (3) Temperature compensation method based on software correction

[0019] Some LED display control systems use software algorithms to optimize temperature compensation. For example:

[0020] Through artificial neural networks (ANN) or machine learning models, the LED brightness attenuation law is trained based on a large amount of experimental data, and a temperature-brightness compensation model is established.

[0021] A correction method based on LUT (lookup table) is used to pre-store brightness compensation coefficients at different temperatures and perform dynamic interpolation calculations during display.

[0022] Although these methods have improved the compensation accuracy to a certain extent, they still have problems such as high dependence on training data, large amount of calculation, and poor real-time performance. Especially in large-size LED displays, the calculation cost is high and it is difficult to respond to temperature changes in real time.

[0023] 2. Main technical problems of existing technologies

[0024] Although existing temperature compensation technology has alleviated the brightness attenuation and chromaticity shift problems of LED displays to a certain extent, the following core technical issues remain unresolved:

[0025] The lack of a pixel-level temperature compensation strategy results in uneven brightness in local areas of the LED screen, making it difficult to meet high-precision display requirements.

[0026] The sudden brightness changes at the edge of the screen cannot be effectively handled, resulting in unnatural brightness transitions at the splicing points, affecting the overall display effect.

[0027] The chromaticity shift problem has not been effectively solved, especially the wavelength of red LEDs shifts red at high temperatures, resulting in color distortion. Existing methods mainly focus on brightness compensation and ignore color stability.

[0028] The global temperature compensation method is difficult to adapt to the individual differences of different LED modules, resulting in inconsistent compensation effects in different areas of the same LED screen.

[0029] The existing compensation algorithm has high computational complexity and insufficient real-time performance, and cannot adjust the compensation coefficient in real time during the dynamic operation of the LED display. Summary of the Invention

[0030] To address the technical shortcomings of the existing temperature compensation technology, such as the lack of a pixel-level temperature compensation strategy and the inability to effectively handle sudden brightness changes at the edge of the screen, the present invention provides the following technical solutions:

[0031] A temperature compensation method for an LED display screen, comprising:

[0032] The step of collecting the cold state and hot state grayscale matrices of the LED screen to form the cold state grayscale matrix C and the hot state grayscale matrix H;

[0033] Based on the cold grayscale matrix C and the hot grayscale matrix H, the step of calculating the LED lamp bead brightness attenuation ratio pixel by pixel to generate a ratio matrix R;

[0034] The steps of constructing and iteratively calculating the correction matrix K so that the hot state brightness matrix H approaches the cold state brightness matrix C after being processed by K;

[0035] Based on the ratio matrix R, the temperature compensation surface of the LED screen is constructed and normalized;

[0036] Calculate the brightness ratio of the LED screen edge and perform mirror filling on the pixels that cross the edge boundary;

[0037] The steps include constructing an RGB chromaticity correction matrix for the CIE-XYZ color space and combining it with a temperature compensation matrix to dynamically adjust the LED drive current to compensate for the effect of temperature changes on LED color performance.

[0038] Based on the same inventive concept, the present invention also provides a temperature compensation device for an LED display screen, comprising:

[0039] A module that collects the cold and hot grayscale matrices of the LED screen to form a cold grayscale matrix C and a hot grayscale matrix H;

[0040] A module for calculating the LED lamp bead brightness attenuation ratio pixel by pixel based on the cold grayscale matrix C and the hot grayscale matrix H to generate a ratio matrix R;

[0041] A module that constructs and iteratively calculates the correction matrix K so that the hot state brightness matrix H approaches the cold state brightness matrix C after being processed by K;

[0042] Based on the ratio matrix R, a module is built to construct the temperature compensation surface of the LED screen and perform normalization processing;

[0043] A module that calculates the brightness ratio of the LED screen edge and performs mirror filling on the pixels that cross the edge boundary;

[0044] A module that constructs an RGB chromaticity correction matrix for the CIE-XYZ color space and combines it with a temperature compensation matrix to dynamically adjust the LED drive current to compensate for the impact of temperature changes on LED color performance.

[0045] Based on the same inventive concept, the present invention also provides a computer storage medium for storing a computer program. When the computer program is read by a computer, the computer executes the method described above.

[0046] Based on the same inventive concept, the present invention also provides a computer, comprising a processor and a storage medium. When the processor reads the computer program stored in the storage medium, the computer executes the method described above.

[0047] Based on the same inventive concept, the present invention also provides a computer program product, which is a computer program. When the computer program is executed, the method described above is implemented.

[0048] Compared with the prior art, the technical solution provided by the present invention is beneficial in that:

[0049] This solution employs a dual-state grayscale matrix acquisition and ratio analysis method. Using a high-resolution industrial camera, it captures the brightness matrix of the LED display in both cold and hot states and calculates a ratio matrix to quantify the temperature impact. Compared to traditional global brightness compensation, this method can achieve pixel-level brightness attenuation measurement, accurately characterizing the local temperature characteristics of the LED screen. This makes compensation more precise and avoids the localized unevenness caused by global compensation.

[0050] This solution uses gradient descent to optimize the correction matrix and, based on the Nesterov accelerated optimization algorithm, adjusts the temperature compensation coefficients to ensure that the LED screen's brightness matrix in the hot state closely matches the brightness distribution in the cold state. Compared to LUT (lookup table) correction methods, this method can adapt to dynamic changes under different temperature gradients, ensuring the LED screen's brightness uniformity after long-term operation, and improving the real-time and adaptability of compensation.

[0051] This solution uses Makima interpolation to smooth the compensation matrix, interpolating the temperature peaks of each LED module to generate a smooth temperature compensation surface. Compared to traditional cubic spline interpolation, this method avoids overshoot, better fits the temperature curve of the LED display, ensures a smooth distribution of compensation coefficients, and improves the brightness stability of the LED screen under different temperature environments.

[0052] This solution uses logarithmic ratio analysis and exponential interpolation to process edge pixels, achieving nonlinear smoothing of brightness variations at the edges of LED screen splices. Compared to traditional linear interpolation, this method more naturally eliminates sudden brightness changes at module joints, resulting in a more uniform overall brightness transition for the LED display. This avoids visual artifacts caused by edge brightness mismatches and improves the splicing effect.

[0053] This solution integrates chromaticity correction to construct an RGB adjustment coefficient matrix for the CIE-XYZ color space, and combines this with a temperature compensation matrix to adjust color temperature consistency. Compared to existing methods that only compensate for brightness, this method also considers the impact of temperature on LED color. Specifically, it addresses the wavelength redshift of red LEDs at high temperatures, compensating for this color drift. This allows LED displays to maintain color stability across different temperature conditions, improving display quality.

[0054] This solution utilizes FPGA hardware optimization and employs SVD (Singular Value Decomposition) noise reduction and low-rank matrix approximation in the compensation algorithm, significantly reducing computational complexity. Compared to temperature compensation methods based on deep learning or complex neural network models, this solution offers greater computational efficiency and can adjust compensation coefficients in real time during the dynamic operation of the LED screen. This avoids the computational lag associated with traditional algorithms, improving the system's real-time performance and engineering applicability.

[0055] It can be used for brightness uniformity correction, chromaticity compensation and splicing optimization of large LED displays, and is widely used in high-precision display applications such as advertising displays, stage performances, command and dispatch, and sports events. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 The figure is a flow chart of a method for temperature compensation of an LED display screen.

[0057] Figure 2 This is a schematic diagram of the temperature compensation coefficient for the entire screen.

[0058] Figure 3 Schematic diagram of the module temperature peak extraction method.

[0059] Figure 4 Schematic diagram of universal compensation coefficient surface interpolation.

[0060] Figure 5 Schematic diagram comparing the similarity of standard temperature compensation coefficients of horizontal single module and vertical single module.

[0061] Figure 6 A comparison of the cold screen brightness (left) and the thermal equilibrium temperature compensated brightness (right). DETAILED DESCRIPTION

[0062] In order to make the advantages and benefits of the technical solution provided by the present invention more clearly reflected, the technical solution provided by the present invention is now further described in detail with reference to the accompanying drawings, specifically:

[0063] Implementation method 1: This implementation method further explains the technical solution in detail through specific steps:

[0064] A method for temperature compensation of an LED box display screen is as follows:

[0065] Step 1: Start the temperature compensation process and initialize the system parameters, including screen resolution , module pixel size , and load the camera calibration parameters.

[0066] Step 2: Use a high-resolution industrial camera (resolution ≥ 1440 × 810) to collect data from the entire M × N cabinet screen. The matrix elements are calculated using the grayscale integral formula: Place the entire screen upright in a constant temperature darkroom (25±0.5℃), light it up immediately after cooling to room temperature, and use a telecentric lens to collect every other point to generate the cold screen grayscale matrix C; keep the white field lit until the screen temperature stabilizes (thermal balance time ≥2 hours) to generate the hot screen grayscale matrix , calculate the ratio matrix (element-by-element division) as the temperature compensation coefficient matrix for the entire M×N cabinet (before correction).

[0067] in, Indicates that the display is at its initial or lower temperature (cold screen). The grayscale of the pixels covered by the LEDs on the camera, represents the integral, Represents the grayscale function in the "cold screen" state, It means that when the screen continues to work under 600nit white field display and heats up to a certain thermal equilibrium state (hot screen), the pixel grayscale of the same LED Represents the integral value of the grayscale function under the "hot screen" state.

[0068] Step 3: Collect the grayscale matrix that reaches thermal equilibrium after loading the compensation coefficient Ideally, and They should be exactly the same, but due to multiple reasons there are errors between the two, so it is necessary to construct and optimize the correction matrix , so that the grayscale distribution Adjust to a brightness distribution as close to the cold state as possible . Correction matrix Gradient descent iterative algorithm: Until the convergence condition

[0069] ( ) is satisfied. The gradient descent iterative algorithm adopts Nesterov acceleration optimization, and the learning rate Dynamic Adjustment: Initial learning rate , maximum number of iterations . As a temperature compensation coefficient matrix for the entire M×N cabinet (after correction);

[0070] in, represents element-wise multiplication, is the Frobenius norm, To adjust the balanced brightness matrix, is the cold brightness matrix, represents a pair of adjacent submatrices, and are the correction factor matrices of adjacent sub-matrices, The subscripts i and j represent rows and columns respectively. and represents the regularization parameter, Represents the updated correction matrix To minimize the error .

[0071] Step 4: In the temperature compensation coefficient matrix dedicated to the entire screen, calibrate the temperature peak area of ​​each module (i.e., the mean of the 9×9 neighborhood near the peak of the ratio matrix), extract the compensation coefficients of all module peak points, and divide the array into X×Y (for example, 2×2) matrix peak areas based on the dense arrangement characteristics of the back driver IC. The coordinate set is:

[0072]

[0073] Where X represents the number of rows where the driver ICs are concentrated, and Y represents the number of columns where they are concentrated. In a typical scenario, the distance between adjacent peak areas is Δx = ⌊M / X⌋, and Δy = ⌊N / Y⌋.

[0074] Perform a×a (e.g., 9×9) neighborhood averaging on each peak region. Define the compensation coefficient matrix as R, and the neighborhood coordinate set of the module (m',n') in the (i,j)th peak region as Ω_{m',n'}^{(i,j)}. Then the compensation coefficient is calculated as:

[0075]

[0076] express (element-wise division).

[0077] The neighborhood window satisfies:

[0078]

[0079] Each module (m',n') generates X×Y temperature characteristic points in the X×Y peak area:

[0080]

[0081] Based on each module Generated Temperature characteristic points , for the same peak position index in all modules The global interpolation is performed at the point, the specific process is: traverse all modules , extract the index of the same peak area Temperature characteristic points under , forming the data set corresponding to the location:

[0082]

[0083] For example, if the total number of modules is ,but Include data points, each point corresponds to the original physical position of the screen The compensation coefficient value at . Each data point in The physical coordinates of its module Associate (e.g. pixel or driver IC coordinates) to construct a two-dimensional interpolation input:

[0084]

[0085] Generate a continuous compensation function covering the entire screen through the Makima interpolation algorithm ,satisfy:

[0086] right Peak Area Index Execute the above operations separately and finally get Independent interpolation functions:

[0087]

[0088] Each Characterizes the full screen in the peak area Temperature compensation distribution characteristics under .

[0089] Take All The mean of is calculated as:

[0090] Finally, the interpolation surface is normalized and used as the universal compensation coefficient for the erection of M×N boxes. The Makima interpolation uses extrapolation constraints in the boundary region and defines the slope weights:

[0091]

[0092] To avoid oscillation of the interpolation surface at the module joints, use Divide by , and divide the result according to the actual module and add each module to get the average value, then the standard temperature compensation coefficient of the vertical single module can be obtained , used for comparative verification.

[0093] Step 5: Place the two independent boxes horizontally in a vertical dark box (ambient light <1 lux), cool them to room temperature, and then light them up simultaneously. Use a top-mounted linear array CCD camera (sampling rate ≥1kHz) to scan point by point and record the initial cold state brightness distribution. After thermal equilibrium, repeat the acquisition and calculate the horizontal dedicated single module compensation coefficient matrix , and for the middle 4 modules (index range ) to generate the horizontal single module standard compensation matrix .

[0094] Step 6: Verify the compensation effect and optimize the verification. Set up the single module standard temperature compensation coefficient. Standard compensation matrix for horizontal and horizontal single modules Verification indicator: brightness root mean square error ,

[0095] Step 7: Edge pixel processing

[0096] Perform nonlinear smoothing on the edge area of ​​the screen (the outer two rows of pixels): calculate the brightness ratio of the left and right edges , generating the index adjustment coefficient ( ). Gradual adjustment of the left half row / right half row pixels:

[0097]

[0098] Represents the left half row of pixels after processing,

[0099] Indicates the left half of the row range: column index , whose pixels are: ;

[0100] Represents the adjustment coefficient. For the left half of the row, the adjustment coefficient is arrive Incrementally; for the right half of the row, start from arrive Decreasing.

[0101] Represents the right half row of pixels after processing,

[0102] Indicates the right half of the row range: column index , whose pixels are:

[0103] .

[0104] Indicates a rank change.

[0105] Step 8: Temperature compensation coefficient expansion and integration

[0106] The horizontal standard compensation matrix Expand to full screen size by module period: and erection general compensation coefficient Multiply element by element to generate the final compensation matrix:

[0107] Step 9: Compensation effect verification

[0108] Load the compensation coefficient to the display control system, including:

[0109] Constructing a complete temperature compensation matrix . This matrix is ​​given by A composite matrix consisting of sub-matrices, where each sub-matrix is The matrix of is the row and column label of the pixel. The subscript X means the XYZ component corresponding to the correction coefficient applied to the compensation coefficient. The component means that the X (red), Y (green), and Z (blue) components of the light source spectrum are collected. For example, the blue (B) component in the red (R) is collected:

[0110] (1) Matrix

[0111]

[0112] (2) Matrix

[0113]

[0114] (3) Matrix

[0115]

[0116] (3) Matrix

[0117]

[0118] Loading the correction matrix of the temperature compensation coefficient

[0119] definition is the chromaticity correction coefficient matrix calculated by collecting the CIE-XYZ color space components corresponding to each RGB color, where the subscript is the row and column label of the pixel, as well as its position. The subscript RGB indicates the displayed color, and the subscript XYZ indicates the corresponding components, corresponding to X (red), Y (green), and Z (blue), respectively. For example What is meant is that when displaying red, the correction coefficient of the X (red) component is What is expressed is the correction coefficient of the Y (green) component when displaying blue. is a The matrix of and Multiply the corresponding elements of point by point, the formula is:

[0120]

[0121] The specific form is as follows:

[0122]

[0123]

[0124] The construction of the cold screen grayscale matrix C and the hot screen grayscale matrix H includes: collecting the pixel grayscale values ​​of each LED lamp bead in the cold screen and hot screen states through a camera, and accumulating the grayscale values ​​of each LED lamp bead respectively. , and arranged in m×n matrix form, where the cold screen grayscale function and thermal screen grayscale function Grayscale accumulation is achieved through double integration or discrete summation.

[0125] The ratio matrix R is constructed using element-by-element division operation and is defined as R=C H, whose matrix elements , which is used to quantify the brightness attenuation rate of LEDs in hot and cold states, and intuitively reflect the temperature sensitivity distribution of the entire screen in matrix form.

[0126] The optimization objective function of the correction matrix K contains three constraints: the brightness error minimization term , submatrix consistency term and correction constraint , where L is the graph Laplacian matrix, and Grid search is used to fine-tune the balance between correction accuracy and smoothness.

[0127] The submatrix consistency constraint is achieved by the difference in the Frobenius norm of the correction factors K_i,j and K_k,l of adjacent modules. The graph Laplacian matrix L is used to encode the spatial adjacency relationship between modules to ensure that the corrected brightness distribution has no mutation at the splicing location.

[0128] Small singular values ​​are truncated to remove noise, and the first r principal components are retained to construct a low-rank matrix X, satisfying rank(X)≤r, thereby improving the computational efficiency and robustness of the optimization process.

[0129] The temperature peak extraction adopts a modular partitioning strategy, dividing the screen into M_x×M_y modules according to the module splicing properties, extracting the adjacent neighborhood grayscale mean as the feature point in the driver IC clustering area of ​​each module, and generating a continuous compensation surface of the entire screen pixels through Makima interpolation.

[0130] The Makima interpolation adopts an improved cubic spline algorithm, and establishes interpolation surfaces at different peak points of the module and then performs weighted fusion, wherein the weight is dynamically adjusted according to the symmetry of the module's thermal distribution.

[0131] The edge pixel processing adopts a method combining logarithmic domain ratio analysis and exponential interpolation: the natural logarithm r = ln(M / m) of the left and right / upper and lower edge brightness ratios is calculated, an exponential adjustment coefficient α_k = exp(kr / w) is generated, and nonlinear gradient adjustment is performed on half-row / half-column pixels to eliminate optical mutations at the splicing boundary.

[0132] When processing edges, mirror filling is used to process out-of-bounds pixels. G(p+i,q+j)=G(-pi,q+j) is defined. When p+i<0, the periodic boundary characteristics are maintained to avoid edge artifacts.

[0133] The chromaticity correction generates a final compensation matrix G by dividing the dedicated compensation matrix R_c by the universal compensation surface S element by element, where , K is the correction matrix iteratively optimized by the gradient descent method, which ensures that the wavelength red shift of the three primary color LEDs caused by temperature is differentially compensated.

[0134] The gradient descent iterative algorithm uses an adaptive learning rate η, and the update formula is , where the gradient term contains the brightness error gradient , adjacent sub-matrix difference gradient and the correction constraint gradient .

[0135] The single-module standard compensation matrix Ĝ is obtained by calculating the position-by-position mean of the remaining (M_x-2)×(M_y-2) middle modules after removing the outer circle modules, effectively eliminating the compensation deviation introduced by the abnormal heat dissipation of the screen frame.

[0136] The structural similarity index (SSIM) was used to evaluate the consistency of the vertical and horizontal acquisition modules, and SSIM>0.93 was set as the acceptance threshold. and theoretical matrix The brightness structure and contrast similarity of the two images are used to verify the effectiveness of compensation.

[0137] The temperature compensation coefficient is back-calculated by periodically expanding the single module matrix After reaching full screen size, it is multiplied point by point with the general compensation curve. The expansion formula is: , maintaining the periodic repetitive characteristics of the module compensation pattern.

[0138] Final compensation matrix By converting the color space and loading it into each RGB channel, the compensation matrix results in setting a higher compensation gain for the red LED to compensate for the higher temperature sensitivity of its AlGaInP material and achieve color temperature consistency correction.

[0139] Implementation Method 3: Combination Figure 1 This embodiment further describes the above technical solution in detail through specific examples, specifically:

[0140] The method is as follows: First, capture data for the entire M*N cabinet and calculate and apply correction coefficients to provide baseline data for subsequent brightness uniformity and temperature compensation. After applying the correction coefficients, the entire screen is placed upright and allowed to completely cool to room temperature. Immediately, every other pixel is captured from a distance for each component of each color. After the white field is illuminated for a long period of time to reach thermal equilibrium, every n*n pixel is captured from a distance. The ratio of cold-state capture to hot-state capture is calculated to obtain the temperature compensation coefficient specific to the M*N cabinet. The temperature peak point x of each module is calibrated, and the temperature compensation coefficients for all modules in the area with the same temperature peak position are selected. A smooth surface is interpolated based on the displayed pixels. Since there are x peak points, x smooth surfaces are obtained. Finally, the average of these x smooth surfaces is calculated as the universal compensation coefficient for the M*N cabinet. The M*N cabinet-specific temperature compensation coefficient is divided by the universal compensation coefficient for the M*N cabinet. The result is divided by the actual module and averaged to obtain the standard temperature compensation coefficient for a single module in an upright position for comparison and verification.

[0141] The prerequisites for horizontal acquisition are to collect data from two separate cabinets, calculate, and apply correction coefficients. Place the upper and lower screens horizontally in a vertical darkroom and allow them to cool completely to room temperature. Both cabinets are suddenly illuminated, and the top camera immediately acquires data point by point. After the white field is illuminated for a long period of time until thermal equilibrium is reached, the top camera acquires data point by point. The ratio of cold to hot acquisitions is the temperature compensation coefficient for the horizontal module. The average of the four central modules is used as the standard temperature compensation coefficient for the horizontal module.

[0142] Check whether the currently calculated horizontal single module compensation coefficient and the vertical single module standard temperature compensation coefficient meet the requirements of brightness uniformity and temperature compensation. If they do not meet the requirements, further adjustments or re-collection of data are required. And calculate the correction matrix K by optimizing the objective function so that the corrected thermal brightness matrix Approximate the cold brightness matrix C; process the special pixels at the edge of the screen separately to optimize screen splicing;

[0143] The standard temperature compensation coefficient of a single horizontal module is periodically extended to the compensation coefficient of a single horizontal module for the entire screen. The universal compensation coefficient for M*N cabinets is multiplied by the standard temperature compensation coefficient of single horizontal modules of different models to infer the temperature compensation coefficient for the entire vertical screen. The inferred temperature compensation coefficient is combined with the existing correction coefficient to verify whether the final compensation effect meets the design requirements.

[0144] Specifically include:

[0145] Step 1: Dual-state grayscale matrix acquisition and ratio analysis: At an initial temperature of 25°C, light the screen with a 600nit white field. Use an area array CCD camera (resolution ≥ 4K) with 650nm / 530nm / 450nm filters to collect RGB components three times. Calculate the grayscale integral value of each LED lamp bead covering the pixel and construct a cold matrix. ,satisfy: ; Continue working until the temperature rise is less than 0.5℃ / min in the thermal equilibrium state, and repeat the acquisition process to build the thermal state matrix ;

[0146] Step 2: Ratio matrix construction and correction matrix optimization calculation

[0147] Calculate the hot-cold ratio element by element , identify the attenuation rate of each pixel. Establish the objective function: in is the graph Laplace matrix, which constrains the consistency of the correction coefficients of adjacent modules; SVD decomposition is used to Matrix denoising, retaining the top 10% singular values; gradient descent iterative algorithm to solve the optimal correction matrix , termination condition .

[0148] Step 3 Peak extraction and Makima interpolation: Divide the screen into Module, extract the 9×9 pixel mean of the same peak position in each module as the feature point; construct a piecewise cubic polynomial along the x / y axis to satisfy: The slope calculation uses the adjacent four-point weighted method to avoid the overshoot phenomenon of traditional cubic spline; generate a smooth compensation surface with a resolution of 1440×810 .

[0149] Step 4: Chroma Correction Fusion

[0150] Calculate the temperature compensation matrix Then, construct the chromaticity correction matrix , each element corresponds to the RGB adjustment coefficient of the CIE-XYZ space and the temperature compensation matrix and Multiply element-wise to generate a composite correction matrix: The RGB drive current is adjusted in real time through PWM modulation to synchronously compensate for brightness and chromaticity.

[0151] Step 3: Optimization effect: Content spatial resolution is improved: sub-pixel temperature compensation is achieved through a 1440×810 compensation surface, and the local brightness difference is less than 1%; edge transition optimization: the exponential smoothing algorithm makes the SSIM of the stitching area ≥ 0.95, eliminating mutations visible to the naked eye; color consistency is improved: the ΔE color difference is reduced from 5.2 to 2.3, meeting professional display requirements.

[0152] Cold and hot grayscale acquisition: Obtain the grayscale value corresponding to each LED lamp bead on the camera sensor under cold and hot conditions of the LED display screen, and form a cold grayscale matrix and a hot grayscale matrix.

[0153] Temperature compensation coefficient calculation: Based on the cold and hot grayscale matrices, the temperature compensation coefficient matrix is ​​calculated. This step includes calculating the correction coefficients and applying Gaussian filtering and Makima interpolation to optimize the distribution and accuracy of the compensation coefficients.

[0154] Temperature compensation application: By loading the temperature compensation coefficient matrix, the LED display screen is temperature compensated to achieve uniformity of brightness and color.

[0155] Camera imaging and exposure

[0156] In a digital camera, the image sensor converts received light signals into electrical signals, which are then converted into digital grayscale values ​​through analog-to-digital conversion. These grayscale values ​​then enter the image signal processing pipeline and are ultimately stored as image files. During this process, the output of each photosensitive pixel unit has a nearly linear response relationship with the accumulated light flux during a given exposure time.

[0157] Relationship between exposure and brightness: Under ideal conditions, exposure is the product of illuminance per unit area and exposure time. Exposure is determined by the actual brightness of the object, the lens transmission coefficient, and the aperture factor. By properly controlling exposure time and aperture settings, the camera can operate within the linear region of the sensor, resulting in a nearly linear relationship between digital grayscale values ​​and actual brightness.

[0158] Grayscale calculation: In LED display inspection scenarios, each LED generates a concentrated brightness across several pixels on the camera sensor. To assess the overall brightness of a single LED in a digital image, it's typically necessary to superimpose or integrate the output values ​​of all pixels within the pixel area covered by the LED on the sensor. This method yields the combined grayscale values ​​of each LED under both cold and hot conditions, which can then be used to calculate the temperature compensation coefficient.

[0159] Calculation of temperature compensation coefficient for the entire screen

[0160] In practical applications, LED displays can experience temperature increases due to the operating environment or their own heat generation, which in turn affects their luminous efficiency. To effectively compensate for this temperature, it is necessary to construct grayscale matrices for both cold and hot states and calculate the temperature compensation coefficient based on these grayscale matrices.

[0161] Cold and hot acquisition: First, a full-screen image of the LED display is captured under cold conditions (e.g., room temperature) to form a cold grayscale matrix. Subsequently, a full-screen image is captured again under thermal equilibrium conditions (e.g., after a long period of white light), forming a hot grayscale matrix. These two grayscale matrices reflect the brightness distribution of the LED display under different temperature conditions.

[0162] Ratio matrix construction: By dividing the hot grayscale matrix element by element by the cold grayscale matrix, a ratio matrix is ​​constructed. This ratio matrix reflects the brightness change of the LED under hot conditions relative to the cold conditions.

[0163] Correction Coefficient Calculation: After constructing the ratio matrix, correction coefficients must be calculated to achieve accurate temperature compensation. This step involves optimizing the objective function, taking into account factors such as minimizing brightness errors, consistency between submatrices, and the proximity of the correction matrix. Using optimization algorithms such as gradient descent, the optimal correction matrix can be iteratively calculated, thereby optimizing the temperature compensation effect.

[0164] Gaussian filtering and low-rank approximation: To improve the robustness and efficiency of the algorithm, Gaussian filtering is performed on the contrast matrix to remove noise and extract the main brightness patterns. Singular value decomposition and low-rank approximation can further simplify the matrix structure, reduce the amount of calculation, and speed up the optimization process.

[0165] Single module temperature compensation coefficient

[0166] In order to ensure the consistency and uniformity of the temperature compensation coefficient across the entire display, the display needs to be modularized.

[0167] Module division: The entire display is divided into multiple modules by rows and columns. After removing the outermost modules, the number of remaining intermediate modules is reduced, making further analysis and processing easier. By averaging the temperature compensation coefficients of each intermediate module, a standard temperature compensation ratio matrix for each module is obtained. This matrix represents the average brightness change of the entire display screen under cold and hot conditions.

[0168] Calculation of the standard temperature compensation ratio matrix for a single module: The compensation ratio matrices of all intermediate modules are added point by point at the corresponding position, and then divided by the number of modules to obtain the standard temperature compensation ratio matrix for a single module. This matrix reflects the average brightness change of the display at each module position, ensuring the uniformity and accuracy of the compensation coefficient.

[0169] Gaussian filtering: Gaussian filtering is applied to the single-module standard temperature compensation ratio matrix to further smooth the distribution of the compensation coefficients and ensure the continuity and consistency of the compensation effect. Mirror filling or extrapolation is used to process the boundaries during the filtering process to avoid distortion of edge pixels.

[0170] Surface Similarity Evaluation: This evaluation evaluates the similarity between the filtered compensation matrix and the theoretical reference matrix by calculating the root mean square error (RMS) and structural similarity index. If the RMS error is below a preset threshold and the structural similarity index is above a preset standard, the temperature compensation matrix is ​​considered to meet the design requirements.

[0171] Edge pixel processing

[0172] In scenarios with large-scale images or spliced ​​screens, the inconsistency of edge pixels can lead to poor visual effects. To ensure a natural brightness transition at the module splicing area, effective smoothing and adjustment of edge pixels is required.

[0173] Defining and segmenting edge pixels: Dividing the image's edge region into left and right half-rows and top and bottom half-columns facilitates targeted adjustments based on brightness or color differences during subsequent processing. By segmenting the edge region, you can more precisely control the adjustment range of edge pixels.

[0174] Nonlinear edge smoothing: This method uses logarithmic transformation and exponential interpolation to perform nonlinear smoothing on edge pixels. First, edge pixel values ​​are mapped to the logarithmic domain to enhance the stability of brightness ratio calculations. Adjustment coefficients are then generated based on the logarithmic ratios, and exponential interpolation is used to achieve a smooth transition between brightness differences, avoiding the abrupt effects that can occur with linear adjustments.

[0175] Determine and apply the adjustment direction: The adjustment direction is determined based on the brightness relationship between the left and right edges and the top and bottom edges. For the lower brightness side, gradually increase the brightness; for the higher brightness side, gradually decrease the brightness. This ensures a smooth brightness transition at the module joints, achieving overall brightness and color uniformity.

[0176] Example 1

[0177] This embodiment uses an LED display screen with 12 rows and 6 columns as the experimental object. First, the entire screen is placed vertically and cooled to room temperature (25°C), then the white field is lit and each pixel is collected at a distance to form a cold grayscale matrix. Subsequently, the white field is kept lit for a long time until it reaches a thermal equilibrium state (about 50°C), and then pixel collection is performed at intervals to form a hot grayscale matrix. By calculating the ratio matrix, that is, dividing the hot grayscale matrix by the cold grayscale matrix point by point, a ratio matrix is ​​obtained. Then, Gaussian filtering and Makima interpolation methods are applied to optimize the contrast matrix and generate a temperature compensation coefficient matrix. After loading the temperature compensation coefficient matrix, the brightness and chromaticity of the LED display screen under different temperature conditions are significantly improved in uniformity, verifying the effectiveness of the present invention.

[0178] Example 2

[0179] This embodiment further optimizes the processing of edge pixels of the horizontal single-module standard temperature compensation coefficient. By performing logarithmic transformation and exponential interpolation on edge pixels, it ensures a smooth brightness transition at the edge of the entire screen and avoids sudden brightness changes caused by inconsistent compensation coefficients. The specific steps include:

[0180] Logarithmic ratio calculation: Maps the brightness values ​​of edge pixels to the logarithmic domain to enhance the stability of brightness ratio calculation.

[0181] Adjustment coefficient generation: Generate adjustment coefficients based on logarithmic ratios, and use exponential interpolation to achieve a natural transition of brightness differences.

[0182] Apply adjustment coefficient: Based on the adjustment direction, the adjustment coefficient is applied to the left and right edge pixels and the top and bottom edge pixels respectively to ensure a smooth brightness transition at the module joints.

[0183] Experimental results show that the visual effects of the LED display screen after edge pixel processing are more consistent under different temperature conditions, the display effect is more stable, and the overall visual quality of the display is further improved.

[0184] Example 3

[0185] In this embodiment, a temperature compensation method is applied and verified for a high-resolution, large-size LED display screen. The specific steps are as follows:

[0186] Modular division: The large display is divided into multiple modules, each containing multiple LED beads. After removing the outermost modules, the remaining middle modules are used to calculate the temperature compensation coefficient.

[0187] Temperature compensation coefficient calculation: Under cold and hot conditions, the grayscale values ​​of the middle module are collected and a ratio matrix is ​​constructed. Through Gaussian filtering, a standard temperature compensation ratio matrix for a single module is generated.

[0188] Generation of compensation coefficients for the entire screen: The compensation ratio matrix of a single module is expanded periodically according to rows and columns to generate a temperature compensation coefficient matrix for the entire display screen.

[0189] Chromaticity correction application: Load the temperature compensation coefficient matrix into the chromaticity correction matrix and implement full-screen temperature compensation by point-by-point multiplication.

[0190] After the above steps, the brightness and chromaticity uniformity of the high-resolution large-size LED display screen under different temperature conditions were significantly improved, verifying the effectiveness and scalability of the method of the present invention in practical applications.

[0191] Example 4

[0192] This embodiment applies a temperature compensation method to a large outdoor LED display. The outdoor ambient temperature varies greatly, and the display screen is large, resulting in uneven temperature distribution. This embodiment performs temperature compensation by following the steps below:

[0193] Multi-point acquisition: Set multiple acquisition points in different areas of the display screen to collect grayscale values ​​in cold and hot states respectively, forming multiple cold grayscale matrices and hot grayscale matrices.

[0194] Regional compensation coefficient calculation: Calculate the corresponding temperature compensation coefficient matrix based on the grayscale value changes in different regions, taking into account the temperature distribution differences in each region.

[0195] Local optimization: Perform local optimization on the temperature compensation coefficient matrix of each region to ensure the accuracy and consistency of the compensation effect in each region.

[0196] Full-screen application: Integrate the compensation coefficient matrices of each area and apply them to the entire display screen to improve the uniformity of overall brightness and color.

[0197] The experimental results show that after regional temperature compensation, the brightness and color uniformity of large outdoor LED displays at different ambient temperatures have been significantly improved, the display effect is stable, and it meets high visual requirements.

[0198] Example 5

[0199] This embodiment studies and applies temperature compensation for LED displays in high humidity environments. High humidity environments can cause performance degradation of LED display circuit components, which in turn affects brightness and chromaticity. This embodiment performs temperature compensation through the following steps:

[0200] Environmental simulation and grayscale acquisition: Simulate high humidity conditions in a laboratory environment, collect the grayscale values ​​of the LED display in cold and hot states, and form corresponding cold grayscale matrices and hot grayscale matrices.

[0201] Compensation coefficient calculation and optimization: Based on the collected grayscale matrix, the ratio matrix is ​​calculated, and the compensation coefficient matrix is ​​optimized through Gaussian filtering and Makima interpolation method.

[0202] Compensation coefficient application and effect evaluation: The optimized compensation coefficient matrix is ​​applied to the LED display screen, operated under high humidity conditions, and the improvement effect of brightness and color uniformity is evaluated.

[0203] The experimental results show that the brightness and chromaticity uniformity of the LED display screen in a high-humidity environment are significantly improved after temperature compensation treatment, and the display effect is stable, which further verifies the applicability and effectiveness of the present invention in complex environmental conditions.

[0204] The above further describes the technical solution provided by the present invention in detail through several specific embodiments in order to highlight the advantages and benefits of the technical solution provided by the present invention. However, the several specific embodiments described above are not intended to limit the present invention. Any reasonable modification and improvement of the present invention, combination of embodiments and equivalent replacement based on the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A temperature compensation method for an LED display screen, characterized in that: include: The step of collecting the cold state and hot state grayscale matrices of the LED screen to form the cold state grayscale matrix C and the hot state grayscale matrix H; The step of calculating the LED lamp bead brightness attenuation ratio pixel by pixel based on the cold state grayscale matrix C and the hot state grayscale matrix H to generate a ratio matrix R; The steps of constructing and iteratively calculating the correction matrix K so that the hot state brightness matrix H approaches the cold state brightness matrix C after being processed by K; Among them, the correction matrix Gradient descent iterative algorithm: Until the convergence condition ( ) meets the requirements; the gradient descent iterative algorithm adopts Nesterov acceleration optimization, and the learning rate Dynamic Adjustment: Initial learning rate , maximum number of iterations ; As the modified M×N cabinet-wide temperature compensation coefficient matrix; in, represents element-wise multiplication, is the Frobenius norm, To adjust the balanced brightness matrix, is the cold brightness matrix, represents a pair of adjacent submatrices, and are the correction factor matrices of adjacent sub-matrices, The subscripts i and j represent rows and columns respectively. and represents the regularization parameter, Represents the updated correction matrix To minimize the error ; Based on the ratio matrix R, the temperature compensation surface of the LED screen is constructed and normalized; In constructing the temperature compensation surface of the LED screen, the Makima interpolation algorithm is used to generate a continuous compensation function covering the entire screen. ,satisfy: right Peak Area Index Execute the above operations separately and finally get Independent interpolation functions: Each Characterizes the full screen in the peak area Temperature compensation distribution characteristics under ; Take All The mean of is calculated as: ; Calculate the brightness ratio of the LED screen edge and perform mirror filling on the pixels that cross the edge boundary; The steps include constructing an RGB chromaticity correction matrix for the CIE-XYZ color space and combining it with a temperature compensation matrix to dynamically adjust the LED drive current to compensate for the effect of temperature changes on LED color performance.

2. A temperature compensation method for an LED display screen according to claim 1, characterized in that: The brightness data of the LED display screen are obtained respectively when the ambient temperature is stable and when it is lit for a long time and reaches a thermal equilibrium state, forming a cold grayscale matrix C and a hot grayscale matrix H.

3. The temperature compensation method for an LED display screen according to claim 1, characterized in that: The Nesterov accelerated gradient descent optimization algorithm is used to construct and iteratively calculate the correction matrix K.

4. The temperature compensation method for an LED display screen according to claim 1, characterized in that: The temperature compensation surface of LED screen is constructed using Makima interpolation method.

5. A temperature compensation method for an LED display screen according to claim 4, characterized in that: The specific method for constructing the temperature compensation surface of the LED screen is as follows: the screen is divided into Modules, extract the 9×9 pixel mean of the same peak position of each module as the feature point; Construct a piecewise cubic polynomial along the x / y axis that satisfies: , a is a constant, and the slope is calculated using the adjacent four-point weighted method to generate a smooth compensation surface with a resolution of 1440×810.

6. A temperature compensation device for an LED display screen, characterized in that: include: A module that collects the cold and hot grayscale matrices of the LED screen to form a cold grayscale matrix C and a hot grayscale matrix H; A module for calculating the LED lamp bead brightness attenuation ratio pixel by pixel based on the cold grayscale matrix C and the hot grayscale matrix H to generate a ratio matrix R; A module that constructs and iteratively calculates the correction matrix K so that the hot state brightness matrix H approaches the cold state brightness matrix C after being processed by K; Among them, the correction matrix Gradient descent iterative algorithm: Until the convergence condition ( ) meets the requirements; the gradient descent iterative algorithm adopts Nesterov acceleration optimization, and the learning rate Dynamic Adjustment: Initial learning rate , maximum number of iterations ; As the modified M×N cabinet-wide temperature compensation coefficient matrix; in, represents element-wise multiplication, is the Frobenius norm, To adjust the balanced brightness matrix, is the cold brightness matrix, represents a pair of adjacent submatrices, and are the correction factor matrices of adjacent sub-matrices, The subscripts i and j represent rows and columns respectively. and represents the regularization parameter, Represents the updated correction matrix To minimize the error ; Based on the ratio matrix R, a module is built to construct the temperature compensation surface of the LED screen and perform normalization processing; In constructing the temperature compensation surface of the LED screen, the Makima interpolation algorithm is used to generate a continuous compensation function covering the entire screen. ,satisfy: right Peak Area Index Execute the above operations separately and finally get Independent interpolation functions: Each Characterizes the full screen in the peak area Temperature compensation distribution characteristics under ; Take All The mean of is calculated as: ; A module that calculates the brightness ratio of the LED screen edge and performs mirror filling on the pixels that cross the edge boundary; A module that constructs an RGB chromaticity correction matrix for the CIE-XYZ color space and combines it with a temperature compensation matrix to dynamically adjust the LED drive current to compensate for the impact of temperature changes on LED color performance.

7. A computer storage medium for storing a computer program, characterized in that When the computer program is read by a computer, the computer executes the method according to claim 1 .

8. A computer comprising a processor and a storage medium, characterized in that When the processor reads the computer program stored in the storage medium, the computer executes the method according to claim 1 .

9. A computer program product, being a computer program, characterized in that When the computer program is executed, the method according to claim 1 is implemented.

Images