LED screen correction coefficient adjusting method and system

Abstract

The present application relates to the technical field of display, and discloses a LED screen correction coefficient adjusting method and system, the method comprising: collecting original luminance and chrominance data of each LED unit; constructing a global topology connection relationship diagram; calculating initial correction coefficients; analyzing the residual error of the corrected adjacent unit boundary pixels; establishing a consistency constraint equation set with the minimum boundary luminance and chrominance difference as the target; jointly solving to obtain a global optimization correction matrix and write into the driving module. The system comprises data acquisition, topology construction, initial correction, residual error analysis, constraint construction, global optimization and coefficient writing modules, and integrates temperature drift compensation and visual verification mechanism. The present application realizes seamless fusion in the splicing area through cross-unit collaborative optimization, and improves luminance uniformity, chrominance consistency and long-term operation stability.

Description

Technical Field

[0001] This invention belongs to the field of display technology, specifically relating to a method and system for adjusting the calibration coefficient of an LED screen. Background Technology

[0002] With the widespread application of LED displays in large stages, command and control centers, and commercial advertising, high consistency and uniformity in display performance have become key indicators for evaluating system performance. To eliminate brightness and color unevenness caused by individual differences in LED chips, nonlinearity of driving circuits, and environmental factors, existing technologies generally adopt coefficient adjustment methods based on independent acquisition and correction of each screen.

[0003] This method achieves uniform processing within the screen by individually acquiring images, analyzing data, and writing correction parameters for each display unit. However, in large-size display systems composed of multiple screens, each unit module is often composed of components from different batches, and its electro-optical response characteristics will shift over time due to factors such as temperature drift, power fluctuations, and driver aging during long-term operation.

[0004] Calibration of multi-screen splicing systems is typically performed independently on a per-cabinet or per-module basis, lacking a cross-screen collaborative mechanism. When adjacent screens do not share boundary pixel information during calibration, even with good internal uniformity, noticeable color gradation jumps or brightness banding can easily appear at their splicing boundaries. These visual defects are particularly prominent in static images, severely impacting the overall viewing experience. More seriously, due to the dynamic and spatially non-uniform nature of temperature drift, traditional offline calibration cannot detect changes in the boundary region's state in real time, causing calibration coefficients to quickly become invalid. Maintenance personnel must rely on manual visual comparison and repeated adjustments to boundary region parameters, resulting in long debugging cycles, high labor costs, and difficulty in ensuring consistency and stability during long-term operation.

[0005] While some existing technologies attempt to introduce edge blending or local compensation algorithms, they remain limited to post-processing corrections within a single screen and fail to establish a cross-screen parameter linkage mechanism at the system level. Furthermore, the lack of a global optimization objective function based on boundary pixels prevents the correction process from automatically converging to a visually continuous optimal solution.

[0006] Therefore, in large-scale LED splicing screen applications, there is an urgent need for an intelligent correction method that can automatically identify the state of boundary areas, collaboratively optimize the correction coefficients of adjacent screens, and has environmental adaptability, so as to completely solve the boundary abruptness problem caused by independent correction and improve the system automation level and display quality consistency. Summary of the Invention

[0007] This invention provides a method and system for adjusting the correction coefficient of an LED screen. The problem stems from the fact that different LED units are affected by factors such as temperature drift, differences in driving current, and inconsistent aging rates during operation, resulting in brightness or color discontinuities at the splicing boundaries of adjacent screen units. Existing technologies rely on manual, repeated adjustments of the correction coefficients for each unit, which is not only inefficient but also fails to guarantee global visual consistency. This invention constructs a cross-unit collaborative correction model and introduces a boundary pixel state perception and dynamic compensation mechanism to achieve seamless correction of the splicing area.

[0008] This invention provides a method for adjusting the calibration coefficient of an LED screen, comprising:

[0009] The original brightness and original chromaticity data of all LED units in the multi-screen splicing system are acquired. The original brightness and original chromaticity data are collected by a high-precision area array optical measurement device under standard environmental conditions.

[0010] Based on the physical location information of each LED unit, a global topology connection graph is constructed, which defines the set of boundary pixels between any two adjacent LED units;

[0011] For each LED unit, perform initial correction coefficient calculation to generate an initial luminance correction matrix and an initial chromaticity correction matrix;

[0012] Extract the original luminance data and original chrominance data from the set of boundary pixels of all adjacent LED units, and combine them with the corresponding initial correction coefficients to calculate the corrected luminance residual and corrected chrominance residual in the boundary region.

[0013] Based on the corrected luminance residual and the corrected chrominance residual, a set of boundary consistency constraint equations is constructed. The boundary consistency constraint equations aim to minimize the corrected luminance difference and chrominance difference of adjacent unit boundary pixels.

[0014] The boundary consistency constraint equations are solved together with the initial correction coefficients of each LED unit to obtain the globally optimized luminance correction matrix and chromaticity correction matrix.

[0015] The globally optimized luminance correction matrix and chromaticity correction matrix are written into the driving control module of the corresponding LED unit to complete the dynamic update of the correction coefficients.

[0016] Preferably, the raw brightness and raw chromaticity data of all LED units in the multi-screen splicing system are obtained, including:

[0017] Image data of each LED unit under full white, full red, full green and full blue test patterns were collected using a high-precision area array optical measurement device under standard environmental conditions: ambient illuminance below 10 lux, ambient temperature of 25℃±0.5℃ and relative humidity of 50%±5%.

[0018] The image data is converted into raw luminance data in candela per square meter and raw chromaticity data in the CIE1976UCS chromaticity coordinate system.

[0019] The raw luminance and raw chrominance data are transmitted to the central processing unit via a Gigabit Ethernet interface using a custom binary stream format based on TCP / IP.

[0020] Preferably, based on the physical location information of each LED unit, a global topology connection graph is constructed, including:

[0021] Each LED unit is mapped to a two-dimensional Cartesian coordinate system to form a node;

[0022] An undirected edge is established between any two LED units that share a boundary in physical space. The data structure associated with the undirected edge contains a list of boundary pixel coordinates shared by the two units.

[0023] For horizontally adjacent LED units, their boundary pixel set consists of the rightmost 5 columns of pixels from the left unit and the leftmost 5 columns of pixels from the right unit; for vertically adjacent LED units, their boundary pixel set consists of the bottom 5 rows of pixels from the top unit and the top 5 rows of pixels from the bottom unit.

[0024] Preferably, an initial correction coefficient calculation is performed for each LED unit to generate an initial luminance correction matrix and an initial chromaticity correction matrix, including:

[0025] For each pixel, based on its original brightness data under the full white field test pattern and the preset standard brightness target value, the initial brightness gain coefficient is obtained by fitting using the least squares method, forming the initial brightness correction matrix.

[0026] For each pixel, based on its original XYZ tristimulus values ​​under the full red, full green and full blue test patterns, the initial chromaticity correction matrix is ​​obtained by solving the tristimulus value linear transformation matrix. The tristimulus value linear transformation matrix is ​​obtained by minimizing the sum of squared Euclidean distances between the corrected chromatic coordinates and the preset standard chromatic coordinates and by using the singular value decomposition method.

[0027] Preferably, the original luminance data and original chrominance data are extracted from the boundary pixel sets of all adjacent LED units, and combined with the corresponding initial correction coefficients, the corrected luminance residual and corrected chrominance residual of the boundary region are calculated, including:

[0028] Spatial registration is performed on the boundary pixel set of each pair of adjacent LED units, and a phase-correlation-based subpixel alignment algorithm is used to align the boundary images of the two units to the same coordinate system.

[0029] For the registered boundary pixels, the initial luminance correction matrix and initial chrominance correction matrix of the respective unit are applied to calculate the corrected luminance value and corrected chrominance coordinates.

[0030] Brightness residual after correction pass: Calculate and correct the chromaticity residual. Defined as:

[0031] ;

[0032] For boundary pixels, For the first In each LED unit, the boundary pixel Corrected brightness value For the first In each LED unit, the boundary pixel Corrected brightness value For the first In each LED unit, the boundary pixel Corrected Chromaticity coordinate components, For the first In each LED unit, the boundary pixel Corrected Chromaticity coordinate components, For the first In each LED unit, the boundary pixel Corrected Chromaticity coordinate components, For the first In each LED unit, the boundary pixel Corrected Chromaticity coordinate components.

[0033] Preferably, based on the corrected luminance residual and the corrected chromaticity residual, a set of boundary consistency constraint equations is constructed, including:

[0034] A weighting factor is assigned to the boundary pixel set of each pair of adjacent LED units, the weighting factor It is inversely proportional to the average brightness gradient of the boundary region;

[0035] objective function Defined as the weighted sum of squared residuals of all boundary pixels:

[0036] ;

[0037] To correct the brightness residual, This is the luminance-chromaticity balance factor;

[0038] The resulting optimization problem is a constrained quadratic programming problem:

[0039] ;

[0040] ;

[0041] ;

[0042] in The smoothness threshold is set to 0.01. This represents the optimization operation of finding the minimum value; It is an abbreviation for "subject to", meaning "subject to", and indicates that the following content is a constraint that must be met in the optimization process; For "all" The abbreviation for "" means that the constraint applies to the coordinates of every pixel within the LED unit; This is the brightness correction coefficient matrix; For discrete Laplace operators; It is a 3×3 transformation matrix; This is the brightness correction coefficient after global optimization.

[0043] Preferably, the boundary consistency constraint equations are solved together with the initial correction coefficients of each LED unit to obtain the globally optimized luminance correction matrix and chromaticity correction matrix, including:

[0044] The constrained quadratic programming problem is iteratively optimized using the alternating direction multiplier method.

[0045] In each iteration, the luminance correction subproblem and chrominance correction subproblem are solved alternately, and the Lagrange multipliers and penalty parameters are updated.

[0046] When the change in the objective function value in three consecutive iterations is less than When the iteration terminates, the globally optimized luminance correction matrix and chrominance correction matrix are output.

[0047] Preferably, the globally optimized luminance correction matrix and chromaticity correction matrix are written into the driving control module of the corresponding LED unit, including:

[0048] The correction coefficient data packet is transmitted to the drive control module of each LED unit through a high-speed serial interface. The data packet contains a checksum and a version number.

[0049] The drive control module stores the correction coefficients in a non-volatile memory, the non-volatile memory having a write life of no less than 100,000 times and a data retention time of no less than 10 years.

[0050] Once the writing is complete, the drive control module immediately switches to the new correction coefficient and returns an acknowledgment signal to the central processing unit.

[0051] Preferably, after the correction coefficients are written, the visual consistency verification process is automatically triggered, including:

[0052] The image of the stitched area is acquired again, and the standard deviation of brightness and the Euclidean distance of chromaticity at the boundary of adjacent units are calculated.

[0053] If the standard deviation of luminance is greater than 5 candela per square meter or the chromaticity Euclidean distance is greater than 0.01, a secondary fine-tuning correction is initiated, and local optimization is performed only on the boundary areas that exceed the standard.

[0054] The present invention also provides an LED screen calibration coefficient adjustment system, comprising:

[0055] The raw data acquisition module is used to acquire the raw brightness and raw chromaticity data of all LED units in the multi-screen splicing system;

[0056] The topology construction module is used to build a global topology connection graph based on the physical location information of each LED unit;

[0057] The initial calibration calculation module is used to calculate the initial calibration coefficients for each LED unit and generate the initial luminance calibration matrix and the initial chromaticity calibration matrix;

[0058] The boundary residual analysis module is used to extract the original luminance data and original chrominance data from the set of boundary pixels of all adjacent LED units, and calculate the corrected luminance residual and corrected chrominance residual of the boundary region by combining the corresponding initial correction coefficients.

[0059] The consistency constraint construction module is used to construct a set of boundary consistency constraint equations based on the corrected luminance residual and the corrected chrominance residual.

[0060] The global optimization solution module is used to jointly solve the boundary consistency constraint equations and the initial correction coefficients of each LED unit to obtain the globally optimized luminance correction matrix and chromaticity correction matrix.

[0061] The correction coefficient writing module is used to write the globally optimized luminance correction matrix and chromaticity correction matrix into the driving control module of the corresponding LED unit, respectively.

[0062] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0063] 1. This invention establishes a global topology connection diagram, transforming the calibration process of all LED units in a multi-screen splicing system from isolated calculation to collaborative optimization, fundamentally solving the problem of boundary color level abrupt changes caused by independent calibration between units.

[0064] 2. This invention introduces boundary pixel residual analysis and consistency constraint equations, so that the solution process of the correction coefficient explicitly considers the visual continuity requirements between adjacent units, and can achieve seamless fusion of the stitching area without relying on manual intervention.

[0065] 3. This invention further integrates a temperature drift compensation subsystem, which effectively suppresses the impact of operating temperature changes on correction accuracy by real-time temperature monitoring and pre-calibrated temperature-brightness mapping relationship, thereby improving the stability of the system during long-term operation.

[0066] 4. The alternating direction multiplier optimization strategy adopted in this invention balances computational efficiency and solution accuracy, and can realize real-time or near real-time correction coefficient updates on embedded platforms, significantly improving the deployment efficiency and maintenance convenience of large LED displays.

[0067] 5. This invention achieves simultaneous improvement in three dimensions of multi-screen splicing systems: brightness uniformity, color consistency, and long-term stability, providing reliable technical support for high-resolution, large-size LED display applications. Attached Figure Description

[0068] Figure 1 This is a schematic diagram of the overall technical architecture of the LED screen calibration coefficient adjustment method and system proposed in this invention;

[0069] Figure 2 This is a schematic diagram of the core principle framework of the cross-unit collaborative correction model in this invention;

[0070] Figure 3 This is a logical flowchart of the process of constructing the relationship between raw data acquisition and global topology connection in this invention;

[0071] Figure 4 This is a flowchart illustrating the logical flow of the initial correction coefficient calculation and boundary residual analysis in this invention.

[0072] Figure 5 This is a flowchart illustrating the logical process of constructing the boundary consistency constraint equation system and solving the global optimization problem in this invention.

[0073] Figure 6 This is a schematic diagram of the multi-level interaction relationship and data flow between the LED unit, the temperature drift compensation subsystem, and the drive control module in this invention. Detailed Implementation

[0074] Please refer to Figures 1 to 6 This invention provides a method and system for adjusting the correction coefficient of an LED screen, addressing the problem of abrupt color gradation changes in boundary areas caused by independent calculation of correction coefficients for each unit in a multi-screen splicing system. This problem stems from the influence of factors such as temperature drift, differences in driving current, and inconsistent aging rates among different LED units during operation, resulting in discontinuities in brightness or color at the splicing boundaries between adjacent screen units. Existing technologies rely on manual, repeated adjustments to the correction coefficients of each unit, which is not only inefficient but also fails to guarantee global visual consistency. This invention constructs a cross-unit collaborative correction model, introducing boundary pixel state perception and dynamic compensation mechanisms to achieve seamless correction of the splicing area.

[0075] The LED screen calibration coefficient adjustment method includes the following steps:

[0076] S1, acquire the original brightness data and original chromaticity data of all LED units in the multi-screen splicing system. The original brightness data and original chromaticity data are acquired by a high-precision area array optical measurement device under standard environmental conditions.

[0077] S2, Based on the physical location information of each LED unit, a global topology connection graph is constructed, wherein the global topology connection graph defines the set of boundary pixels between any two adjacent LED units;

[0078] S3, perform initial correction coefficient calculation for each LED unit to generate initial luminance correction matrix and initial chromaticity correction matrix;

[0079] S4. Extract the original luminance data and original chrominance data from the set of boundary pixels of all adjacent LED units, and calculate the corrected luminance residual and corrected chrominance residual in the boundary region by combining the corresponding initial correction coefficients.

[0080] S5. Based on the corrected luminance residual and the corrected chrominance residual, construct a set of boundary consistency constraint equations. The boundary consistency constraint equations aim to minimize the difference in luminance and chrominance between adjacent unit boundary pixels after correction.

[0081] S6. Solve the boundary consistency constraint equation set together with the initial correction coefficients of each LED unit to obtain the globally optimized luminance correction matrix and chromaticity correction matrix.

[0082] S7, the globally optimized luminance correction matrix and chromaticity correction matrix are written into the driving control module of the corresponding LED unit to complete the dynamic update of the correction coefficients.

[0083] In step S1, the raw brightness and chromaticity data of all LED units in the multi-screen splicing system are acquired. This process is completed using a high-precision area array optical measurement device, which has a spectral response range covering 380 nm to 780 nm, a spatial resolution of 0.1 mm per pixel, and a brightness measurement accuracy better than 0.5%. The measurement process is conducted under standard environmental conditions, including an ambient illuminance of less than 10 lux, an ambient temperature maintained at 25℃±0.5℃, and a relative humidity controlled at 50%±5%. Each LED unit is sequentially illuminated under four test patterns: full white, full red, full green, and full blue, with each pattern lasting for 5 seconds to ensure that the pixels reach a thermally stable state.

[0084] A high-precision area array optical measurement device synchronously acquires image data at a sampling frequency of 20 frames per second. Each frame contains a complete multi-screen stitched area, and the image data format is 16-bit grayscale RAW. Raw luminance data is quantized in candela per square meter, and raw chromaticity data is recorded in CIE1931 XYZ tristimulus values, which are then converted to u'v' coordinates in the CIE1976 UCS chromaticity coordinate system during subsequent processing. All acquired data is transmitted to the central processing unit via a gigabit Ethernet interface, encapsulated in a custom binary stream format based on TCP / IP. Transmission of a single frame of image data is completed within 200 milliseconds, ensuring data timeliness.

[0085] In step S2, a global topology connection diagram is constructed based on the physical location information of each LED unit. Each LED unit in the multi-screen splicing system has a unique physical coordinate identifier, which is determined by the mechanical positioning reference during installation and is affixed to the back of the unit via an RFID tag or QR code. After reading the physical coordinates of all LED units, the central processing unit maps them to a two-dimensional Cartesian coordinate system to form a regular or irregular grid layout.

[0086] For regular layouts, a two-dimensional mesh structure is used for modeling. Each LED unit is abstracted as a node, and an undirected edge is established between any two LED units that share a boundary in physical space. The data structure associated with this undirected edge contains a list of boundary pixel coordinates common to the two units. The boundary pixel coordinates are determined based on the physical size of the LED unit and the pixel pitch parameters. For example, if a single LED unit has a resolution of 192×144 pixels and a pixel pitch of 2.5 mm, then its physical size is 480 mm multiplied by 360 mm. When two units are adjacent in the horizontal direction, the rightmost 5 columns of pixels of the left unit and the leftmost 5 columns of pixels of the right unit together constitute the pixel set of the boundary; when two units are adjacent in the vertical direction, the bottom 5 rows of pixels of the upper unit and the top 5 rows of pixels of the lower unit together constitute the pixel set of the boundary.

[0087] For irregular layouts, such as curved or spherical splicing, adopt The triangulation algorithm determines adjacency relationships and uses a Euclidean distance threshold to determine whether a valid boundary is formed. The global topological connectivity graph is stored in memory as an adjacency list, with each node recording its list of neighboring nodes and corresponding boundary pixel set pointers, supporting fast traversal and querying.

[0088] In step S3, initial correction coefficients are calculated for each LED unit, generating an initial luminance correction matrix and an initial chromaticity correction matrix. The initial correction coefficient calculation is performed independently for all pixels within each LED unit. For luminance correction, the preset standard luminance target value is 100 candela per square meter, which can be adjusted according to the application scenario. For each pixel, its original luminance data under a full white field test pattern is used. Compared with standard brightness target value The initial brightness gain coefficient of the pixel is obtained by fitting using the least squares method. The calculation formula is: . These are the two-dimensional coordinates of a pixel within an LED unit. This process is performed pixel-by-pixel within the unit, forming an initial luminance correction matrix with the same resolution as the unit. For chromaticity correction, the preset standard chromaticity coordinate target value is in the CIE1976UCS coordinate system. , ( This represents the horizontal dimension component of the target color in the CIE1976UCS color space. (Representing the vertical dimension component of the target color in this color space) corresponds to the D65 standard light source. First, the original XYZ tristimulus values ​​of each pixel under three monochromatic fields (full red, full green, and full blue) are converted into RGB device-dependent values. Then, the initial chromaticity correction matrix of the pixel is obtained by solving the linear transformation matrix of the tristimulus values. Specifically, let... For raw RGB channel data ( , , (Data for red (R), green (G), and blue (B) channels, respectively). For the corrected target RGB value ( , , (where the values ​​of the red, green, and blue channels are obtained after the correction operation, respectively), then there exists a 3×3 transformation matrix. satisfy . The overdetermined equations are obtained by minimizing the sum of squared Euclidean distances between the corrected and target color coordinates of all pixels, and then solved using singular value decomposition. The initial color correction matrix is ​​stored in floating-point form, with each pixel corresponding to a 3×3 matrix, or, in resource-constrained scenarios, approximated using a lookup table.

[0089] In step S4, the original luminance and chrominance data are extracted from the set of boundary pixels of all adjacent LED units, and combined with the corresponding initial correction coefficients, the corrected luminance residual and corrected chrominance residual of the boundary region are calculated. For any pair of adjacent LED units... Obtain the set of boundary pixels from the global topology connectivity graph. .right Each boundary pixel in ,like Belongs to unit ,but , For the first In each LED unit, the boundary pixel Corrected brightness value For the first Boundary pixels in each LED unit The corresponding brightness correction factor, For the first In each LED unit, the boundary pixel The original luminance data; the corrected chromaticity coordinates ( For the first In each LED unit, the boundary pixel Corrected Chromaticity coordinate components, For the first In each LED unit, the boundary pixel Corrected The chromaticity coordinate components are derived from the initial chromaticity correction matrix. The original RGB value is obtained by applying the same method; similarly, if... Belongs to unit Then calculate , For the first In each LED unit, the boundary pixel After correction, the brightness values, while physically aligned at the boundary pixels, may exhibit sub-pixel offsets due to manufacturing tolerances, necessitating spatial registration. Registration employs a phase-correlation-based sub-pixel alignment algorithm to align the boundary images of the two units to the same coordinate system. The corrected brightness residual... pass: Calculate and correct the chromaticity residual. Defined as:

[0090] ;

[0091] For the first In each LED unit, the boundary pixel Corrected Chromaticity coordinate components, For the first In each LED unit, the boundary pixel Corrected Chromaticity coordinate components.

[0092] The residual values ​​of all boundary pixels are stored in a residual tensor with dimension . ,in For the total number of boundaries, and The height and width of the boundary region (in pixels).

[0093] In step S5, a set of boundary consistency constraint equations is constructed based on the corrected luminance residual and the corrected chrominance residual. This set of equations aims to minimize the corrected luminance and chrominance differences between adjacent unit boundary pixels. First, a weighting factor is assigned to the boundary pixel set of each pair of adjacent LED units. This weighting factor is inversely proportional to the average brightness gradient of the boundary region, and is calculated using the following formula: ,in The average brightness gradient of the boundary region. The scaling factor is 0.1. The average brightness gradient is calculated on the original brightness image using the Sobel operator, reflecting the texture complexity of the boundary region; high gradient regions are less sensitive to the human eye, so they are assigned lower weights. Objective function Defined as the weighted sum of squared residuals of all boundary pixels:

[0094] ;

[0095] in The luminance-chromaticity balance factor, with a value of 0.7, reflects the human eye's greater sensitivity to differences in luminance. Non-negativity and smoothness constraints are imposed on this objective function. The non-negativity constraint requires... , The brightness correction coefficients are globally optimized to prevent negative gain from causing display abnormalities; the smoothness constraint is achieved by adding a Laplace regularization term to the objective function, i.e. ,in The discrete Laplace operator is used to suppress high-frequency spatial noise in the correction coefficients. The resulting optimization problem is a constrained quadratic programming problem:

[0096] ;

[0097] ;

[0098] ;

[0099] in The smoothness threshold is set to 0.01. This represents the optimization operation of finding the minimum value. It is an abbreviation for "subject to", meaning "subject to", and indicates that the following content is a constraint that must be met in the optimization process. For "all" The abbreviation for "" means that the constraint applies to the coordinates of every pixel within the LED unit. This is the brightness correction coefficient matrix.

[0100] In step S6, the boundary consistency constraint equations and the initial correction coefficients of each LED unit are solved together to obtain the globally optimized luminance correction matrix and chrominance correction matrix. The solution process uses an alternating direction multiplier method for iterative optimization. This algorithm decomposes the original problem into a luminance correction subproblem and a chrominance correction subproblem, solves them alternately, and introduces Lagrange multipliers to coordinate the two. The specific iterative steps are as follows: Initialize the global correction coefficients to the initial correction coefficients; in the k-th iteration, fix the chrominance correction matrix, solve the constrained luminance correction subproblem, and obtain the updated luminance correction matrix; then fix the luminance correction matrix, solve the chrominance correction subproblem, and update the chrominance correction matrix; update the Lagrange multipliers and penalty parameters; calculate the current objective function value. ;like If this condition is met three times consecutively, the iteration terminates. For the first In each iteration, the objective function is defined. Interior-point methods are used to solve each subproblem to ensure convergence and numerical stability. The optimization process is executed on an embedded computing platform equipped with a quad-core ARM processor with a clock speed of 1.8 GHz, 4 gigabytes of RAM, a real-time Linux kernel operating system, and a task scheduling cycle of 10 milliseconds. A typical multi-screen system with 1 million pixels can complete global optimization within 15 seconds.

[0101] In step S7, the globally optimized luminance correction matrix and chromaticity correction matrix are written into the corresponding LED unit's drive control module. The drive control module has a built-in non-volatile memory for storing the correction coefficient matrix. This non-volatile memory has a write endurance of at least 100,000 cycles and a data retention time of at least 10 years. The writing process is completed via a high-speed serial interface, using differential signal transmission to resist interference. Each LED unit's drive control module receives a correction coefficient data packet from the central processing unit. The data packet includes a checksum and version number to ensure data integrity.

[0102] After the writing is complete, the drive control module immediately switches to the new correction coefficients and returns an acknowledgment signal to the central processing unit. After completing one global correction, the system automatically triggers the visual consistency verification process. This process involves re-acquiring images of the stitched area and calculating the standard deviation of luminance and the Euclidean distance of chromaticity at the boundaries of adjacent units. If either indicator exceeds a preset threshold (standard deviation of luminance greater than 5 candela per square meter, chromaticity Euclidean distance greater than 0.01), a secondary fine-tuning correction is initiated, performing local optimization only on the boundary areas exceeding the limits to reduce computational overhead.

[0103] The LED screen calibration coefficient adjustment system includes a raw data acquisition module, a topology construction module, an initial calibration calculation module, a boundary residual analysis module, a consistency constraint construction module, a global optimization solution module, and a calibration coefficient writing module. The raw data acquisition module is connected to a high-precision area array optical measurement device via a gigabit Ethernet interface, responsible for acquiring and transmitting raw luminance and chromaticity data. The topology construction module reads the physical location information of the LED units and constructs and maintains a global topology connection graph. The initial calibration calculation module performs unit-level calibration coefficient calculations. The boundary residual analysis module extracts boundary pixel data and calculates residuals. The consistency constraint construction module generates a weighted objective function and constraints. The global optimization solution module is deployed on an embedded computing platform and performs alternating direction multiplier optimization. The calibration coefficient writing module manages the transmission and storage of calibration coefficients to the drive control module.

[0104] The system also includes a temperature drift compensation subsystem. This subsystem comprises a temperature sensor array distributed on the back of each LED unit for real-time monitoring of the operating temperature of each unit. The temperature sensors are digital output thermistors with a temperature measurement accuracy of ±0.5℃ and a sampling frequency of once per second, communicating with the main control unit via an I2C bus. Before each correction coefficient update, the global optimization solution module performs temperature-brightness mapping compensation on the original brightness data based on the current temperature data. The temperature-brightness mapping relationship is obtained through a step-by-step temperature rise experiment within the range of 25℃ to 70℃ before shipment, with each 5℃ step recording the brightness decay curve of each pixel at different temperatures, and fitting it to a quadratic polynomial function. ,in This is the brightness data after temperature compensation. This refers to the real-time operating temperature of the LED unit. These are pixel-level fitting coefficients. The compensated brightness data is used for subsequent correction calculations, effectively suppressing the effects of temperature drift.

[0105] This embodiment achieves fully automatic, high-precision, and seamless calibration of multi-screen splicing LED displays through the above-described method and system, significantly improving visual consistency and long-term operational stability.

Claims

1. An LED screen correction coefficient adjustment method, characterized in that, The method comprises the following steps: Obtaining original luminance data and original chrominance data of all LED units in a multi-screen splicing system, wherein the original luminance data and the original chrominance data are collected by a high-precision area array optical measuring device under standard environmental conditions; Based on the physical position information of each LED unit, a global topology connection relationship graph is constructed, which defines the boundary pixel set between any two adjacent LED units; Performing initial correction coefficient calculation on each LED unit to generate an initial luminance correction matrix and an initial chrominance correction matrix; Extracting the original luminance data and the original chrominance data in the boundary pixel set of all adjacent LED units, and combining the corresponding initial correction coefficients to calculate the corrected luminance residual and the corrected chrominance residual of the boundary area, including: Performing spatial registration on the boundary pixel set of each pair of adjacent LED units, and aligning the boundary images of the two units to the same coordinate system by using a sub-pixel alignment algorithm based on phase correlation; Applying the initial luminance correction matrix and the initial chrominance correction matrix of the unit to which the registered boundary pixels belong, respectively, to calculate the corrected luminance value and the corrected chrominance coordinate; corrected luma residual by: calculating, a corrected chroma residual defined as: ； is a boundary pixel, is a first LED unit, the boundary pixel has a corrected luminance value, is a second LED unit, the boundary pixel has a corrected luminance value, is a third LED unit, the boundary pixel has a corrected chromaticity coordinate component, is a fourth LED unit, the boundary pixel has a corrected chromaticity coordinate component, is a fifth LED unit, the boundary pixel has a corrected chromaticity coordinate component, is a sixth LED unit, the boundary pixel has a corrected chromaticity coordinate component; Based on the corrected luminance residual and the corrected chrominance residual, a boundary consistency constraint equation set is constructed, including: assigning a weight factor to the set of border pixels of each pair of adjacent LED units, the weight factor being inversely proportional to the average luminance gradient of the border region; Objective function is defined as the weighted sum of squared residuals of all the boundary pixels: ； to correct the post-luminance residual, is a luminance-chrominance balance factor; The final optimization problem is a constrained quadratic programming problem: ； ； ； is a smoothness threshold with a value of 0.01; represents an optimization operation that seeks a minimum value; is an abbreviation for "subject to" and indicates that the following content is a constraint that the optimization process must satisfy; is an abbreviation for "for all " that indicates that the constraint applies to every pixel coordinate within the LED unit; is a luminance correction coefficient matrix; is a discrete Laplacian operator; is a 3x3 transformation matrix; is a luminance correction coefficient after global optimization; The boundary consistency constraint equation set aims to minimize the corrected luminance difference and the corrected chrominance difference of the adjacent unit boundary pixels; Solving the boundary consistency constraint equation set and the initial correction coefficients of each LED unit jointly to obtain a globally optimized luminance correction matrix and a globally optimized chrominance correction matrix; Writing the globally optimized luminance correction matrix and the globally optimized chrominance correction matrix into the corresponding LED unit drive control module to complete dynamic updating of the correction coefficients.

2. The LED screen correction coefficient adjustment method according to claim 1, wherein, Obtaining original luminance data and original chrominance data of all LED units in a multi-screen splicing system, including: Collecting image data of each LED unit under full white field, full red field, full green field and full blue field test patterns by a high-precision area array optical measuring device under standard environmental conditions, i.e. ambient illuminance below 10 lux, ambient temperature at 25℃±0.5℃, and relative humidity at 50%±5%; Converting the image data into original luminance data in units of candela per square meter and original chrominance data represented by CIE1976 UCS chrominance coordinate system; Transmitting the original luminance data and the original chrominance data to a central processing unit in a custom binary stream format based on TCP / IP through a gigabit Ethernet interface.

3. The LED screen correction coefficient adjustment method according to claim 2, characterized in that, Based on the physical position information of each LED unit, a global topology connection relationship graph is constructed, including: Mapping each LED unit to a two-dimensional Cartesian coordinate system to form a node; Establishing a non-directional edge between any two LED units sharing a boundary in physical space, wherein the data structure associated with the non-directional edge contains a list of boundary pixel coordinates shared by the two units; For horizontally adjacent LED units, the set of boundary pixels is composed of the rightmost 5 columns of pixels of the left unit and the leftmost 5 columns of pixels of the right unit; for vertically adjacent LED units, the set of boundary pixels is composed of the lowermost 5 rows of pixels of the upper unit and the uppermost 5 rows of pixels of the lower unit.

4. The LED screen correction coefficient adjustment method according to claim 3, characterized in that, Performing initial correction coefficient calculation on each LED unit to generate an initial luminance correction matrix and an initial chrominance correction matrix, comprising: For each pixel point, an initial luminance gain coefficient is obtained by least square fitting according to its original luminance data under the full white field test pattern and a preset standard luminance target value, forming an initial luminance correction matrix; For each pixel point, an initial chrominance correction matrix is obtained by solving a tristimulus value linear transformation matrix according to its original XYZ tristimulus values under the full red field, full green field and full blue field test patterns, wherein the tristimulus value linear transformation matrix is solved by minimizing the squared sum of Euclidean distances between corrected color coordinates and a preset standard color coordinate and using a singular value decomposition method.

5. The LED screen correction coefficient adjustment method according to claim 4, characterized in that, Solving the set of boundary consistency constraint equations and the initial correction coefficients of each LED unit to obtain globally optimized luminance correction matrices and chrominance correction matrices, comprising: Iterative optimization of the constrained quadratic programming problem using an alternating direction multiplier method; Alternately solving the luminance correction sub-problem and the chrominance correction sub-problem in each iteration, and updating the Lagrange multipliers and the penalty parameters; When the variation of the objective function value in three successive iterations is less than the iteration is terminated and the global optimized luminance correction matrix and the chroma correction matrix are output.

6. The LED screen correction coefficient adjustment method according to claim 5, characterized in that, Writing the globally optimized luminance correction matrices and chrominance correction matrices into the drive control modules of the corresponding LED units, comprising: Transmitting a correction coefficient data packet to the drive control modules of each LED unit through a high-speed serial interface, the data packet containing a checksum and a version number; The drive control module stores the correction coefficients in a non-volatile memory, the write endurance of the non-volatile memory is not less than 100,000 times, and the data retention time is not less than 10 years; After writing is completed, the drive control module immediately switches to the new correction coefficients and returns an acknowledgement signal to the central processing unit.

7. The LED screen correction factor adjustment method of claim 6, wherein, After completing the correction coefficient writing, automatically triggering a visual consistency verification process, comprising: Collecting the image of the spliced area again, calculating the luminance standard deviation and the chrominance Euclidean distance at the boundary of adjacent units; If the luminance standard deviation is greater than 5 candela per square meter or the chrominance Euclidean distance is greater than 0.01, a secondary fine tuning correction is started, which only optimizes the local area of the over-standard boundary region.

8. An LED screen correction factor adjustment system, comprising: The LED screen correction coefficient adjustment system is executed to implement the method of any one of claims 1 to 7, comprising: An original data acquisition module for obtaining original luminance data and original chrominance data of all LED units in a multi-screen splicing system; A topological relationship construction module for constructing a global topological connection graph based on the physical location information of each LED unit; An initial correction calculation module for performing initial correction coefficient calculation on each LED unit to generate an initial luminance correction matrix and an initial chrominance correction matrix; An initial correction calculation module for performing initial correction coefficient calculation on each LED unit to generate an initial luminance correction matrix and an initial chrominance correction matrix; a boundary residual analysis module configured to extract original luminance data and original chrominance data in a set of adjacent LED unit boundary pixels, and calculate corrected luminance residual and corrected chrominance residual in the boundary region by combining the initial correction coefficients; a consistency constraint construction module configured to construct a boundary consistency constraint equation set according to the corrected luminance residual and the corrected chrominance residual; a global optimization solution module configured to jointly solve the boundary consistency constraint equation set and the initial correction coefficients of each LED unit to obtain a globally optimized luminance correction matrix and a globally optimized chrominance correction matrix; a correction coefficient writing module configured to write the globally optimized luminance correction matrix and the globally optimized chrominance correction matrix into the drive control module of the corresponding LED unit, respectively.

Images