A small-pitch LED splicing screen cross-module color and brightness consistency correction method and system

Abstract

The present application relates to LED display correction technical field, disclose a kind of small-pitch LED splicing screen cross module color and bright consistency correction method and system.The method obtains module basic correction parameter, splicing relationship, batch information and low gray correction code-light output response, collects reference gray scale and low gray point bright image, determines the boundary distance layer of two sides of splicing boundary, internal reference area and boundary optical residual;Corresponding batch and boundary distance layer are constructed to low gray execution response, generate cross-boundary continuous target in light output domain, determine the adjacent executable correction code of two sides of target, and according to cumulative brightness, chroma quantization residual, select correction code along splicing boundary tangential in the same boundary distance layer.The system includes imaging acquisition device, display control device and correction server, for generating and loading relative to the boundary increment correction table of basic correction parameter.The present application can inhibit low gray brightness band and color deviation band, and support local update after module replacement.

Description

Technical Field

[0001] This invention relates to the field of LED display calibration technology, and in particular to a method and system for calibrating color brightness consistency across modules in a small-pitch LED splicing screen. Background Technology

[0002] Small-pitch LED video walls are typically formed by splicing together multiple LED modules, cabinets, video controllers, transmitting cards, and receiving cards. Before leaving the factory, the pixel differences within the modules can be reduced through point-by-point brightness and color correction. However, after the entire screen is assembled, delivered to the site, and after subsequent module replacements, bright lines, dark lines, color difference lines, and low gray bands of a certain width may still appear between adjacent modules due to differences in LED device batches, driver IC low gray response, basic correction parameters, mask reflection characteristics, and assembly gaps.

[0003] Chinese patent CN104332133B discloses a method for adjusting the brightness compensation of the seams in a high-density LED display screen. Based on point-by-point uniformity correction, a second brightness uniformity correction is performed on the seam area, and the first and second correction matrices are combined to form the final correction matrix. This scheme can reduce the bright and dark lines at the seams, but it mainly generates the second correction parameters based on the overall brightness of the seam correction unit, and does not address the differences in the correction code-light output response between different batches of modules under low grayscale conditions.

[0004] Chinese patent CN116343652A discloses a method for compensating for splicing lines in an LED display screen. This method involves illuminating the screen, capturing an image, determining the pixels along the splicing lines, setting a target value, and iteratively adjusting the correction coefficients to compensate for the splicing lines. Other existing technologies use distance weighting to smoothly transition the correction coefficients between adjacent correction areas. While these technologies can reduce spatial abrupt changes in correction coefficients, continuous correction coefficients do not necessarily ensure continuous actual light output.

[0005] Under low grayscale display conditions, the combined photoelectric response of LED devices, constant current drive circuits, and PWM grayscale expansion exhibits significant nonlinearity, and the same correction code from different batches of modules may correspond to different light output increments. Furthermore, the point-by-point correction coefficients, low grayscale correction parameters, or full grayscale correction parameters that the receiver card can recognize have finite bit widths; theoretically continuous correction values ​​ultimately need to be quantized into discrete executable correction codes.

[0006] When directly rounding the continuous correction values ​​pixel by pixel, if the continuous correction values ​​of several adjacent pixel columns are in the same discrete code range, these pixel columns will collectively select the same correction code. When the continuous correction value crosses the quantization boundary, the next pixel column may simultaneously jump to another correction code. For different batches of modules, the actual brightness increment and overall chromaticity difference corresponding to this jump are not the same, which may transform the original single splicing difference into a low gray brightness band or color shift band with a certain width and parallel to the splicing boundary.

[0007] Therefore, given that adjacent modules already have their own basic correction parameters and the boundary transition targets are continuous, how to convert the continuous light output target into a discrete executable correction code based on the low gray response of different batches on both sides of the splicing boundary, and avoid the quantization residual generated by the conversion from accumulating along the direction perpendicular to the splicing boundary, is a specific technical problem that still needs to be solved. Summary of the Invention

[0008] To address the shortcomings of existing technologies, the present invention aims to provide a method and system for correcting color brightness consistency across modules in small-pitch LED splicing screens.

[0009] To achieve the above objectives, the present invention provides the following technical solution:

[0010] A method for correcting color brightness consistency across modules in a small-pitch LED splicing screen includes:

[0011] Acquire the basic calibration parameters of each module in the small-pitch LED splicing screen, the module splicing relationship, the module batch information, the reference grayscale image, the low gray lighting images under multiple low gray test grayscales, and the calibration code-light output response of each module batch;

[0012] Based on the module splicing relationship, reference grayscale image and low gray point bright image, the splicing boundary between adjacent modules is determined. Multiple boundary distance layers are divided on both sides of the splicing boundary and an internal reference area is set. The pixels in each boundary distance layer are determined as sequentially arranged correction processing units along the tangent of the splicing boundary. The correction processing unit is a single pixel or a pixel group composed of multiple pixels adjacent along the tangent. The boundary optical residual is determined based on the brightness difference and chromaticity difference between each boundary distance layer and the corresponding internal reference area.

[0013] Based on the correction code-optical output response and the optical output characteristics of each boundary distance layer, the low gray execution response corresponding to each correction processing unit is determined, and based on the internal reference optical output on both sides of the splicing boundary and the boundary optical residual, a continuous target optical output spanning the splicing boundary is generated.

[0014] Based on the low-gray execution response, adjacent executable correction codes on both sides of the continuous target light output are determined; for correction processing units arranged sequentially along the tangential direction within the same boundary distance layer, a target executable correction code is selected from the adjacent executable correction codes based on the cumulative luminance quantization residual and the cumulative chrominance quantization residual, and the cumulative luminance quantization residual and the cumulative chrominance quantization residual are updated based on the difference between the actual light output corresponding to the target executable correction code and the continuous target light output;

[0015] The cumulative luminance quantization residual and the cumulative chrominance quantization residual are propagated only along the tangential direction within the same boundary distance layer, and are not propagated to adjacent boundary distance layers.

[0016] Based on the target executable correction code of each correction processing unit, a boundary incremental correction table relative to the basic correction parameters is generated, and the boundary incremental correction table is loaded into the display control system of the small-pitch LED splicing screen.

[0017] In a preferred embodiment, determining the splicing boundary and boundary distance layer includes:

[0018] The reference grayscale image and the low grayscale bright image are subjected to dark field subtraction, flat field correction, exposure normalization and multi-frame fusion.

[0019] Based on the module splicing relationship, boundary search bands are set on both sides of the nominal splicing position of adjacent modules, and the position of the splicing boundary is corrected according to the brightness gradient of the reference grayscale image and the chromaticity gradient of the low gray point bright image.

[0020] The number of boundary distance layers and the width of the boundary compensation strip are determined based on the attenuation of the boundary optical residual along the direction perpendicular to the splicing boundary.

[0021] In a preferred embodiment, determining the low-gray execution response includes:

[0022] Read the calibration code-light output response samples stored according to module batch, color channel, and low gray test grayscale; when the corresponding response sample does not exist, load multiple trial calibration codes into the module of the corresponding batch, and measure the light output corresponding to each trial calibration code to obtain the calibration code-light output response sample.

[0023] The measurement confidence of each response sample is determined based on at least one of multi-frame measurement variance, effective pixel ratio of exposure, and repeated measurement stability, and the response samples are monotonicized according to the measurement confidence to obtain segmented monotonic batch responses.

[0024] Based on the ratio between the light output of each boundary distance layer and the light output inside the corresponding batch module, edge optical correction is performed on the batch response to obtain the low gray execution response of the corresponding boundary distance layer.

[0025] In a preferred embodiment, generating the continuous target light output includes:

[0026] Based on the low-gray execution response, executable correction code range and safety drive margin on both sides of the splicing boundary, the reachable optical output range on both sides of the splicing boundary is determined, and the intersection of the reachable optical output range on both sides is determined as the common reachable optical output range.

[0027] Using the internal reference light outputs on both sides of the splicing boundary as anchor points, an initial continuous target light output with a smooth transition is generated according to the position of each correction processing unit relative to the splicing boundary, and the initial continuous target light output is restricted within the common reachable light output range to obtain the continuous target light output;

[0028] When the common reachable light output range is empty, or when the preset correction conditions cannot be met within the common reachable light output range, the corresponding splicing boundary is marked as a mechanically uncorrectable boundary.

[0029] In a preferred embodiment, selecting the target executable correction code from the adjacent executable correction codes includes:

[0030] Based on the low-gray execution response, determine the upper neighbor executable correction code and the lower neighbor executable correction code located on both sides of the continuous target light output, respectively;

[0031] Combine the upper and lower adjacent executable correction codes of the R, G, and B color channels into multiple candidate correction code groups;

[0032] Based on the differences between the actual luminance and the target luminance, the differences between the actual chrominance and the target chrominance corresponding to each candidate correction code group, the cumulative luminance quantization residuals and chrominance quantization residuals of the preceding correction processing units, and the code value arrangement constraints of adjacent correction processing units, the selection cost of each candidate correction code group is determined, and the candidate correction code group with the minimum selection cost is determined as the target executable correction code of the current correction processing unit.

[0033] In a preferred embodiment, it also includes:

[0034] Multiple correction processing units arranged continuously along the tangential direction within the same boundary distance layer are set as tangential residual constraint windows;

[0035] When the absolute value of the sum of the brightness differences between the target light output and the actual light output of each correction processing unit within the tangential residual constraint window exceeds the preset brightness residual threshold, or the norm of the sum of the chromaticity difference vectors exceeds the preset chromaticity residual threshold, the selection combination of the upper neighbor executable correction code and the lower neighbor executable correction code within the tangential residual constraint window is adjusted.

[0036] Among the multiple boundary distance layers, adjacent boundary distance layers adopt opposite tangential traversal directions.

[0037] In a preferred embodiment, generating and loading the boundary increment correction table includes:

[0038] Using the splicing boundary identifier, low gray test gray level, boundary distance layer, tangential sequence number of the correction processing unit and color channel as indexes, store the boundary increment of each target executable correction code relative to the corresponding basic correction parameter;

[0039] When the display control system supports an independent calibration layer, the boundary incremental calibration table and the basic calibration parameters are loaded separately; when the display control system does not support an independent calibration layer, the boundary incremental calibration table and the basic calibration parameters are loaded together.

[0040] For input gray levels that have not been tested, interpolation is performed based on the boundary increments corresponding to adjacent low-gray test gray levels to determine the boundary increments corresponding to the input gray levels that have not been tested.

[0041] In a preferred embodiment, it also includes:

[0042] After loading the boundary incremental correction table, the small-pitch LED splicing screen is controlled to re-display the low gray test pattern and acquire the retest image. The retest brightness residual and retest chromaticity residual of each splicing boundary are determined based on the retest image.

[0043] For splicing boundaries that do not reach the preset acceptance threshold, update the continuous target light output or low gray execution response according to the corresponding retest brightness residual and retest chromaticity residual, redetermine the target executable correction code and update the boundary incremental correction table.

[0044] After replacing the module, the splicing boundary connected to the replaced module is determined according to the updated module splicing relationship, and only the boundary increment correction parameter corresponding to the splicing boundary connected to the replaced module is updated.

[0045] A cross-module color brightness consistency correction system for small-pitch LED splicing screens is used to correct small-pitch LED splicing screens. The system includes a correction server, which comprises a processor, a memory, an image data interface, and a display control interface.

[0046] The image data interface is used to receive the illuminated images of the small-pitch LED splicing screen acquired by the imaging acquisition device under a reference grayscale and multiple low grayscale test grayscales.

[0047] The display control interface is used to send test pattern display instructions and calibration parameter loading instructions to the display control device that is communicatively connected to the small-pitch LED splicing screen;

[0048] The memory is used to store programs; the calibration server is also used to acquire module basic calibration parameters, module splicing relationships, module batch information, calibration code-optical output response, and boundary incremental calibration table;

[0049] When the program is executed by the processor, it causes the correction server to perform the steps of the method described above.

[0050] In a preferred embodiment, it also includes an imaging acquisition device, a display control device, and a calibration parameter storage device;

[0051] The imaging acquisition device is connected to the image data interface and includes an industrial camera or an imaging colorimeter, used to acquire the illuminated image;

[0052] The display control device is connected to the display control interface and includes a video controller, a sending card, and a receiving card. It is used to control the small-pitch LED splicing screen to display test patterns and load the boundary incremental correction table.

[0053] The calibration parameter storage device is located within the calibration server, or is configured as a database server that is communicatively connected to the calibration server, for storing the module's basic calibration parameters, module splicing relationships, module batch information, calibration code-optical output response, boundary incremental calibration table, and retest results.

[0054] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0055] This invention acquires the basic calibration parameters, splicing relationships, batch information, and low gray correction code-light output response data of each module constituting a small-pitch LED splicing screen. Based on the whole-screen reference grayscale and low gray illumination image, it determines the boundary distance layer, internal reference area, and boundary optical residual on both sides of the splicing boundary. For different batches, color channels, low gray test grayscale, and boundary distance layers, it constructs low gray execution responses, generates continuous targets across modules in the light output domain, and uses the corresponding low gray execution responses to convert the continuous targets into adjacent executable calibration codes. Furthermore, based on the cumulative brightness quantization residual and the cumulative color quantization residual, it selects the target executable calibration code tangentially along the splicing boundary within the same boundary distance layer, thereby making the average actual light output within the tangential window approximate the light output of the continuous target, reducing the low gray brightness band and color shift band formed by discrete calibration code quantization.

[0056] Furthermore, this invention determines the commonly reachable optical output range based on the low-gray execution response, the range of executable correction codes, and the safety drive margin on both sides of the splicing boundary, and restricts continuous targets within the commonly reachable optical output range. When the commonly reachable optical output range is empty or the acceptance threshold cannot be met within the range, the corresponding splicing boundary is marked as a mechanically uncorrectable boundary, avoiding the loading of correction parameters exceeding the safety drive range onto the receiver card to compensate for excessively wide splicing seams, edge LED failures, mask obstruction, or cabinet misalignment.

[0057] Furthermore, this invention stores the boundary incremental correction table and the module's basic correction parameters in layers. When the display control system supports an independent correction layer, the basic correction parameters and the boundary incremental correction table are loaded separately; when the display control system only supports a single correction database, the boundary incremental correction table and the basic correction parameters are merged and loaded. Thus, it can retain the module's factory correction results, perform incremental compensation on splicing boundaries, and support version tracking and rollback of correction parameters.

[0058] Furthermore, when replacing modules on-site, this invention determines the splicing boundary that connects to the replaced module based on the updated screen topology. Only the determined splicing boundary is re-acquired, the boundary optical residual is calculated, the continuous target is generated, the correction code is determined, and the boundary incremental correction table is updated, while the correction parameters of other splicing boundaries and non-boundary areas are retained. This reduces the recalibration range after module replacement and improves on-site maintenance efficiency.

[0059] Furthermore, this invention uses a combination of static spatial correction codes to represent continuous targets located between two discrete response values, eliminating the need to switch correction codes between adjacent video frames and preventing time flickering introduced by the correction process of this invention. The method can be implemented using industrial cameras or imaging colorimeters, video controllers, transmitting cards, receiving cards, and correction servers, and is applicable to production correction, field delivery, and after-sales maintenance scenarios. Attached Figure Description

[0060] Figure 1 This is a flowchart illustrating the cross-module color brightness consistency correction method for small-pitch LED splicing screens according to the first embodiment of the present invention;

[0061] Figure 2 This is a schematic diagram illustrating the division of the splicing boundary, boundary distance layer, tangential pixel group, and internal reference area in the first embodiment of the present invention;

[0062] Figure 3 This is a schematic diagram of the process of converting continuous target light output into discrete executable correction code and performing quantization residual directional allocation in the first embodiment of the present invention;

[0063] Figure 4 This is a schematic diagram of the structure of the cross-module color brightness consistency correction system for small-pitch LED splicing screen according to the second embodiment of the present invention;

[0064] Figure 5 This is a schematic diagram of the process of locally updating the boundary incremental correction table after replacing the module in an embodiment of the present invention. Detailed Implementation

[0065] In the following description, several technical details are presented to enable those skilled in the art to better understand the present invention. However, those skilled in the art will understand that the technical solution of the present invention can be achieved even without using all of the following technical details, or by making corresponding changes and modifications to the following embodiments.

[0066] To make the objectives, technical solutions, and advantages of the present invention clearer, the embodiments of the present invention will be described in further detail below with reference to the accompanying drawings.

[0067] like Figure 1 As shown, the first embodiment of the present invention relates to a method for cross-module color brightness consistency correction of a small-pitch LED splicing screen, the method including steps S100 to S800. This embodiment takes a small-pitch LED display screen spliced ​​from multiple modules from different batches as the object. Each module has completed factory point-to-point brightness and color calibration, and the display control system stores the corresponding basic calibration parameters.

[0068] Step S100: Obtain the basic data of the small-pitch LED splicing screen.

[0069] Specifically, a module record is established for each module. The module record includes at least the module identifier m, the cabinet identifier h, the module's row and column position in the entire screen, the batch identifier b, the basic calibration parameters, the receiver card address, the output port mapping relationship, and the range of executable calibration codes.

[0070] Preferably, the module identifier is read via a QR code, barcode, RFID tag, or the module's non-volatile memory. The module identifier is associated with and stored in conjunction with the batch identifier, basic calibration parameters, and factory calibration version, so that the low gray response data of the corresponding batch can be obtained after the entire screen is assembled or the module is replaced.

[0071] Furthermore, a screen topology diagram is established based on the row and column positions of each module within the entire screen. Nodes in the screen topology diagram represent modules, and the edges between two adjacent nodes represent splicing boundaries. For each splicing boundary e, the modules on both sides of the boundary, the boundary direction, the nominal start coordinates, the nominal end coordinates, and the corresponding receiving card area are recorded. The boundary direction includes both horizontal and vertical directions.

[0072] In the following explanation, c represents the R, G, and B color channels, g represents the low gray test gray level, k represents the boundary distance layer divided along the direction perpendicular to the splicing boundary, i represents the sequence number of the correction processing unit arranged tangentially along the splicing boundary within the same boundary distance layer, b represents the module batch, and q represents the executable correction code that the display control system can recognize.

[0073] Furthermore, a single pixel or a group of pixels tangentially adjacent along the splicing boundary is defined as a correction processing unit. When the display control system supports pixel-level correction, each correction processing unit corresponds to one pixel; when the display control system supports pixel-group-level correction, each correction processing unit corresponds to the smallest pixel group that the display control system can independently configure. The following explanation uses a pixel group as a correction processing unit.

[0074] Specifically, the basic correction parameters can be point-by-point luminance coefficients, point-by-point luminance-chrominance matrices, low-grayscale correction coefficients, full-grayscale correction coefficients, or other correction parameters that the display control system can recognize. The subsequently generated boundary increment correction table only generates increments for pixels or pixel groups within the boundary compensation band; increments in non-boundary areas are set to zero increments or unity gain.

[0075] Preferably, the screen is preheated before image acquisition to bring the light output into a stable range during calibration. Specifically, the preheating time can be 20 to 40 minutes, and the calibration ambient temperature can be 23°C to 27°C. The above temperature control is used to ensure stable measurement conditions during this calibration process and does not involve compensation for thermal drift during normal operation of the screen.

[0076] Step S200: Control the small-pitch LED splicing screen to display the test pattern and collect the image of the entire screen lit up.

[0077] Specifically, multiple low-grayscale test levels are selected. Taking an 8-bit equivalent input grayscale as an example, 2, 4, 8, 16, and 32 can be selected as low-grayscale test levels, and 128 or 192 can be selected as reference grayscale. For display control systems with 10-bit, 12-bit, or higher input bit depths, the above grayscale levels are converted into input codes of the corresponding bit depth.

[0078] Furthermore, the control display screen sequentially displays the reference grayscale white point, the R monochrome pattern, G monochrome pattern, B monochrome pattern and white point pattern of each low grayscale test grayscale, and uses an industrial camera or imaging colorimeter to acquire the corresponding images.

[0079] Preferably, the industrial camera outputs linear RAW images of at least 12 bits. In one specific embodiment, the industrial camera outputs 14-bit linear RAW images, with 16 frames continuously acquired for each test pattern.

[0080] Specifically, for each test pattern, a dark field image D, a flat field image F, and a bright field image I are acquired, and the flat field corrected image is obtained according to the following formula:

[0081]

[0082] in, This represents the corrected pixel value at camera pixel coordinates (x, y); This indicates the illuminated image at camera pixel coordinates. Pixel value at; This represents the pixel value of the dark field image at camera pixel coordinates (x, y); This represents the pixel value of the flat field image at camera pixel coordinates (x, y); To prevent positive numbers with a denominator of zero; This represents the average value of the effective area of ​​the flat field.

[0083] The aforementioned dark field subtraction is used to reduce the impact of camera dark current and fixed background on low grayscale images, while flat field correction is used to reduce the impact of lens vignetting and inconsistent imaging sensor response on the overall screen brightness distribution.

[0084] Furthermore, for multiple frames of the same test pattern, the median, truncated average, or arithmetic mean after removing outliers is used. For the lowest grayscale, images are preferably acquired using two to four exposure times, and after removing saturated pixels, they are normalized and synthesized according to the exposure time.

[0085] The above-mentioned processing of multiple frames of the same test pattern by applying median, truncated mean, or arithmetic mean after removing outliers is collectively referred to as multi-frame fusion processing.

[0086] Specifically, the effective pixel ratio and multi-frame variance of each exposure setting are used as the confidence basis for subsequent low gray response fitting. The smaller the multi-frame variance and the higher the effective pixel ratio, the greater the confidence weight of the corresponding response sample.

[0087] Furthermore, a nominal pixel grid is established based on the known number of rows and columns of the module and the screen topology. The geometric transformation is estimated using the LED light-emitting center, periodic texture, or box positioning mark in the reference grayscale image, and the camera coordinates are registered to the full-screen pixel coordinates.

[0088] Optionally, for screens with smaller pixel pitches that cannot stably separate individual physical pixels, two to four adjacent physical pixels along the splicing boundary are divided into a pixel group, and each pixel group is used as a correction processing unit. The luminance and chromaticity statistics of each physical pixel in the pixel group are used as the light output data of the correction processing unit.

[0089] The advantage of this setup is that subsequent corrections are based on the light output statistics of the boundary distance layer and tangential pixel groups, rather than relying on a single measurement of a single camera pixel, which can improve the signal-to-noise ratio and repeatability of low-gray image measurements.

[0090] Step S300: Determine the splicing boundary based on the screen topology and the illuminated image of the entire screen, and calculate the boundary optical residual of the splicing boundary.

[0091] Specifically, a set of splicing boundaries is generated based on the screen topology. For each splicing boundary e, a boundary search band is set on both sides of its nominal coordinates. The luminance gradient of the reference grayscale image and the chrominance gradient of the low grayscale image are calculated. The local extremum lines of the weighted amplitudes of the luminance gradient and chrominance gradient are determined as the corrected boundary center lines.

[0092] Furthermore, such as Figure 2 As shown, using the corrected boundary centerline as the dividing line, multiple boundary distance layers are defined on both sides of the stitching boundary along a direction perpendicular to the normal of the stitching boundary. The boundary distance layers on the A side are numbered sequentially from closest to furthest from the stitching boundary, and the B side is numbered in the same way. Internal reference areas for both Module A and Module B are set at positions located outside the boundary compensation band and far from the stitching boundary. These internal reference areas do not contain other stitching boundaries, bad pixels, or saturated pixels, and belong to the same module and batch as the corresponding boundary distance layers.

[0093] Specifically, after color calibration of the camera's RGB response, tristimulus values ​​X, Y, and Z are obtained, where Y is the luminance value. The CIE 1976 UCS color coordinates are determined according to the following formula:

[0094]

[0095] Where X, Y, and Z represent the tristimulus values ​​corresponding to the correction processing unit, and This indicates the color coordinates of the correction processing unit in the CIE 1976 UCS color space.

[0096] For one side of module A, the splicing boundary e, the gray level g, and the distance layer k, the relative brightness residual is determined by the following formula:

[0097]

[0098] in, This represents the median or segmental median along the tangent of the splice boundary. This indicates the brightness of the boundary layer. This indicates the brightness of the reference area inside module A.

[0099] Furthermore, the chromaticity residual vector is determined according to the following formula:

[0100]

[0101] The boundary optical residuals are calculated in the same way on the B side.

[0102] Using the chromaticity residual vector instead of the chromaticity scalar alone can preserve the direction of color bias, thus allowing the determination of the influence direction of different candidate correction codes on the chromatic coordinates during subsequent R, G, and B joint code selection.

[0103] Specifically, such as Figure 2 As shown, the pixel groups within the same boundary distance layer are numbered sequentially along the tangent parallel to the stitching boundary. .in, This represents the total number of correction processing units contained within the current boundary distance layer corresponding to the stitching boundary e. The boundary distance layer is used to limit the propagation range of the quantization residual in the boundary normal direction, and the tangential correction processing unit sequence is used to determine the allocation order of the quantization residual. Preferably, when the absolute value of the relative brightness residual between two consecutive distance layers is not greater than the brightness residual threshold... Furthermore, the L2 norm of the chroma residual vector is not greater than the chroma residual threshold. At that time, the corresponding outer distance layer is determined as the end point of the boundary compensation zone.

[0104] Specifically, the boundary compensation band can include 4 to 12 layers on each side of the splicing boundary, preferably 6 to 8 layers on each side. This setting allows the compensation range to be determined based on the actual influence width of different splicing boundaries, avoiding setting the same fixed compensation width for all splicing boundaries.

[0105] Step S400: Construct a low-gray execution response based on the module batch and measured light output.

[0106] Specifically, for each module batch b, color channel c, and low grayscale test gray level g, one or more representative modules are selected. J trial correction codes are loaded near the basic correction parameters. Measure the corresponding light output The correction code—optical output response sample—is obtained.

[0107] Preferably, when the module's factory database has already saved the response sample, the corresponding response sample is read directly according to the batch identifier; when the module's factory database has not saved the response sample, the non-boundary reference area of ​​the corresponding batch of modules is selected in the entire screen, and the trial correction code is loaded and the light output is collected in sequence.

[0108] Furthermore, the trial correction codes are limited to the correction code range allowed by the receiving card and the LED safety driving range. The trial correction codes are preferably distributed on both sides of the basic correction codes to reduce the impact of the testing process on the normal correction state of the screen.

[0109] Specifically, to reduce the impact of measurement noise and the inversion of low-gray local responses, a weighted order-preserving fitting is performed on the original response samples:

[0110]

[0111] Where J represents the number of trial correction codes; This represents the measured optical output corresponding to the j-th trial correction code; This represents the estimated optical output of the j-th response sample point during the weighted order-preserving fitting process; This represents the fitted light output of the j-th response sample point after weighted order-preserving fitting; This represents the confidence weight of the j-th response sample, determined based on at least one of multi-frame measurement variance, effective pixel ratio for exposure, and repeated measurement stability. Under constraints... The fitted light output that minimizes the weighted squared error is obtained by finding the fitted light output and connecting adjacent fitted points to obtain a piecewise monotonic batch response. .

[0112] Furthermore, considering that the edge optical efficiency at the seam may differ from that inside the module due to variations in mask edges, LED spacing, and seam reflection, the edge optical correction coefficient is determined based on the ratio between the current light output of the boundary distance layer and the corresponding batch's internal response. The boundary execution response is obtained:

[0113]

[0114] Optionally, three to five trial correction codes within a safe range can be directly applied to the boundary distance layer and the light output measured to establish the boundary execution response of that boundary distance layer. The direct measurement method is suitable for screens with high requirements for edge optical effect correction accuracy; the proportional correction method is suitable for screens that need to shorten the overall screen correction time.

[0115] The advantage of this setting is that subsequent processing only requires finding adjacent response values ​​on both sides of the continuous target light output. It does not require the original low gray response to satisfy global linearity, nor does it require the original response to be strictly reversible across the entire correction code range, which can improve the stability of low gray correction code inversion.

[0116] Step S500: Determine the common reachable light output range based on the low gray execution response on both sides of the splicing boundary, and generate cross-module continuous targets.

[0117] Specifically, for each color channel, low grayscale test grayscale, and boundary distance layer on both sides of the splicing boundary e, the reachable light output range is determined based on the set of allowed correction codes. Taking one side of module A as an example:

[0118]

[0119] in, This represents the set of executable correction codes after applying safety-driven margin constraints.

[0120] Furthermore, the reachable optical output range on one side of module B is determined in the same manner, and the intersection of the reachable optical output ranges on both sides is calculated. The intersection is used as the common reachable light output range of the corresponding splicing boundary, boundary distance layer, color channel and low gray test gray level.

[0121] Specifically, let t be the normalized position from the outer edge of the compensation strip on module A to the outer edge of the compensation strip on module B, and The transition weights are determined according to the following formula:

[0122]

[0123] The first derivative of the aforementioned transition weights is zero at both ends, so that the compensation band smoothly connects with the internal regions of module A and module B, respectively.

[0124] Furthermore, using the internal reference light output vectors of module A and module B... and Using the anchor point, the initial continuous target is obtained according to the following formula:

[0125]

[0126] Specifically, the color channel components of the initial continuous target are projected onto the corresponding common reachable light output range to obtain the continuous target. .

[0127] When the commonly reachable light output range is empty, the correction code is not increased further; instead, the corresponding stitching boundary is marked as a mechanically uncorrectable boundary. When the commonly reachable light output range is not empty, but the projected target still fails to meet the acceptance threshold for the boundary optical residual after prediction, the corresponding stitching boundary is also marked as a mechanically uncorrectable boundary.

[0128] The preset correction conditions include: the predicted luminance residual determined based on the continuous target light output after projection and the corresponding low grayscale execution response does not exceed a preset luminance acceptance threshold, the predicted chromaticity residual does not exceed a preset chromaticity acceptance threshold, and the executable correction code required to achieve the continuous target light output meets the safety drive margin requirements. If any of the above conditions are not met, it is determined that the preset correction conditions cannot be met within the commonly reachable light output range.

[0129] Specifically, uncorrectable mechanical boundaries may be caused by excessively wide seams, malfunctioning edge LEDs, mask obstruction, misalignment of the housing, or insufficient safety drive margin. The system outputs the corresponding boundary coordinates, predicts the remaining error, and suggests handling methods such as assembly adjustment, module replacement, or mask inspection.

[0130] Optionally, for the intersection region of the horizontal and vertical stitching boundaries, horizontal and vertical boundary targets are generated separately, and then weighted and synthesized based on the residual confidence of the two boundaries; or the intersection region is divided into independent two-dimensional boundary units, and continuous targets are generated using two-dimensional smoothing weights. The correction results for the intersection region still need to satisfy the common reachable light output range constraint.

[0131] The advantage of this setup is that it can distinguish between electrically compensable color and brightness differences and uncompensable differences caused by mechanical structures, thus avoiding loading correction parameters beyond the safe range onto the receiver card.

[0132] In step S600, based on the batch low-gray execution response, the continuous target light output is converted into adjacent executable correction codes, and the quantization residuals generated by the conversion are directionally allocated.

[0133] like Figure 3 As shown, step S600 includes: querying the corresponding low-gray execution response based on the module batch, color channel, low-gray test grayscale, and boundary distance layer of the pixel group; determining the upper and lower neighboring executable correction codes on both sides of the continuous target light output; combining them to form R, G, and B candidate correction code groups; selecting the target executable correction code based on the cumulative luminance quantization residual, cumulative color quantization residual, and continuous code value arrangement constraints; updating the cumulative quantization residual; and performing residual checks on the tangential constraint window.

[0134] Specifically, for pixel group i, color channel c, and low gray test gray level g, the corresponding boundary execution response is selected according to the module batch and boundary distance layer to which the correction processing unit belongs, and the selected boundary execution response is denoted as... The continuous correction code is represented as:

[0135]

[0136] in, This represents the continuous target light output of the i-th correction processing unit under color channel c and low gray test gray level g; This represents the continuous correction code obtained through inversion; Indicates in Response to low gray within the segmented monotonic interval The execution of a local inversion does not imply a global inversion of the entire executable correction code range.

[0137] Since the calibration codes that the display control system can actually execute are a discrete set, the next nearest executable calibration code is determined from the set of executable calibration codes. And adjacent executable correction codes To satisfy:

[0138]

[0139] Specifically, the executable correction code whose optical output is not greater than the continuous target optical output and has the smallest distance from the continuous target optical output is determined as the lower neighbor executable correction code, and the executable correction code whose optical output is not less than the continuous target optical output and has the smallest distance from the continuous target optical output is determined as the upper neighbor executable correction code. When the continuous target optical output is exactly equal to the optical output corresponding to an executable correction code, the upper neighbor executable correction code and the lower neighbor executable correction code are the same correction code.

[0140] Furthermore, such as Figure 3 As shown, for pixel groups arranged tangentially along the splicing boundary within the same boundary distance layer, a target executable correction code is selected sequentially from the upper and lower adjacent executable correction codes. The luminance quantization residual and chrominance quantization residual generated after code selection for the current pixel group are then transferred to the next tangential pixel group within the same boundary distance layer. The quantization residual does not cross over to adjacent boundary distance layers.

[0141] Specifically, for pixel groups in the same boundary distance layer, according to the tangential sequence number Process them sequentially. When each color channel has two adjacent executable correction codes, the R, G, and B color channels can form a maximum of eight candidate correction code groups.

[0142] Based on the batch low-gray execution response and camera color calibration relationship, predict the brightness corresponding to each candidate correction code group q. and chromaticity vector .

[0143] Before processing the i-th pixel group, save the cumulative luminance quantization residual of the previous pixel group. and cumulative colorimetric quantization residuals For each candidate correction code group, the selection cost is calculated according to the following formula:

[0144]

[0145] in, and These represent the target brightness and target chromaticity of the i-th pixel group, respectively. This indicates a constraint on the arrangement of consecutive code values. , and Indicates the weight.

[0146] Specifically, the consecutive code value arrangement constraint is used to limit the consecutive arrangement length of the same code value or to limit the code value abrupt change of adjacent pixel groups, and is not used to replace the luminance quantization residual and chrominance quantization residual.

[0147] Furthermore, the candidate correction code group with the lowest cost is selected. The target executable correction code for the current pixel group is used, and the cumulative quantization residual is updated according to the following formula:

[0148]

[0149]

[0150] Preferably, W correction processing units arranged continuously along the tangential direction of the splicing boundary within the same boundary distance layer are set as tangential residual constraint windows, and the tangential residual constraint windows satisfy:

[0151]

[0152]

[0153] Where s represents the tangential sequence number of the starting correction processing unit in the tangential residual constraint window; W represents the number of correction processing units contained in the tangential residual constraint window; This represents the target brightness of the i-th correction processing unit; This indicates that the i-th correction processing unit uses a candidate correction code group. The actual brightness at that time; This represents the target chromaticity vector of the i-th correction processing unit; This indicates that the i-th correction processing unit uses a candidate correction code group. The actual chromaticity vector at that time; This indicates the preset brightness residual threshold. This indicates the preset chromaticity residual threshold. The L2 norm of a vector.

[0154] When the result obtained by sequentially selecting the correction code cannot simultaneously satisfy the above-mentioned luminance residual constraint and chrominance residual constraint, the selection combination of the upper neighbor executable correction code and the lower neighbor executable correction code of one or more correction processing units is adjusted within the current tangential residual constraint window, and the correction code combination that minimizes the cumulative luminance residual and cumulative chrominance residual of the window is determined as the target correction code combination of the tangential residual constraint window.

[0155] Specifically, the window length W can be from 8 to 32 pixel groups, preferably 16 pixel groups. The window length can be adjusted according to the pixel pitch, viewing distance, and camera spatial resolution.

[0156] Furthermore, odd-numbered boundary distance layers traverse from the start point to the end point of the splicing boundary, while even-numbered boundary distance layers traverse from the end point to the start point of the splicing boundary. This configuration reduces the probability of different boundary distance layers experiencing simultaneous code value changes at the same tangential position, preventing the formation of regular textures perpendicular to the splicing boundary.

[0157] Optionally, pixel groups with the same tangential position on both sides of the stitching boundary are formed into cross-boundary pairing groups, and the difference in actual light output on both sides is added to the selection cost of candidate correction code groups, so that the target on both sides of the stitching boundary can be jointly determined for correction codes.

[0158] For receiver cards that only support pixel group-level correction, the smallest pixel group that the receiver card can independently configure is used as the processing unit mentioned above.

[0159] The advantage of this setting is that the quantization residual is only distributed tangentially along the splicing boundary within the same boundary distance layer, and is not propagated to adjacent boundary distance layers. This avoids synchronization code value jumps in multiple distance layers at the same normal position, thereby avoiding the formation of low gray brightness bands or color offset bands parallel to the splicing boundary.

[0160] To facilitate understanding of the quantization residual orientation allocation process of this invention, a specific numerical example is provided below. This numerical example is for illustrative purposes only and does not constitute a limitation on the parameter range of this invention.

[0161] Within a certain low gray test gray level, a certain color channel, and the same boundary distance layer, the normalized target light output of eight consecutive pixel groups is 0.926; the batch low gray execution response shows that the actual light outputs corresponding to two adjacent executable correction codes are 0.920 and 0.930, respectively.

[0162] If rounding is applied independently to each pixel group, all eight pixel groups will select the corresponding executable correction code of 0.930, resulting in an average window light output of 0.930 and a same-direction deviation of +0.004 relative to continuous targets.

[0163] After adopting the cumulative residual selection method of this embodiment, five upper neighbor codes corresponding to 0.930 and three lower neighbor codes corresponding to 0.920 are selected from the eight pixel groups, and the average light output of the window is:

[0164]

[0165] Therefore, the difference between the average light output of the window and the continuous target is 0.00025. The five upper neighbor codes are spaced apart along the splicing boundary direction according to the cumulative quantization residual and the consecutive code value arrangement constraints.

[0166] The advantage of this setup is that each pixel group still uses the discrete correction code that the display control system can execute, while the average light output of the tangential window can represent a continuous target located between two discrete response values. Furthermore, when jointly selecting the R, G, and B color channels, it avoids in-direction chromaticity errors caused by independent rounding of each color channel.

[0167] Step S700: Generate a boundary incremental correction table based on the target executable correction code, and load the correction parameters into the display control system.

[0168] Specifically, the boundary increment relative to the basic correction parameters is stored using the stitching boundary identifier e, the low gray test gray level g, the boundary distance layer k, the tangential pixel group number i, and the color channel c as indices. .

[0169] For a display control system that uses a multiplicative correction coefficient, the final correction parameters are generated according to the following formula:

[0170]

[0171] in, These represent the basic correction parameters corresponding to pixel group p, color channel c, and grayscale g in module m. This indicates hardware quantization performed according to the receiver card's parameter bit width and data format.

[0172] Optionally, for display control systems that employ additive bias, color matrix, or other correction parameter forms, the aforementioned multiplication operations can be replaced by the native merging operations of the display control system.

[0173] Furthermore, when the receiving card supports multiple correction layers among the luminance and chromaticity correction layer, seam correction layer, low gray correction layer, or full gray correction layer, the basic correction parameters and boundary incremental correction table are loaded respectively.

[0174] When the receiving card only supports a single calibration database, the calibration server merges the boundary incremental calibration table with the basic calibration parameters before sending the parameters, and then loads them into a native data format that the receiving card can recognize.

[0175] Preferably, regardless of whether an independent loading method or an offline merging method is used, independent versions of the basic correction parameters and the boundary incremental correction table are retained in the correction parameter storage device to facilitate version tracking and parameter rollback.

[0176] Furthermore, for adjacent input gray levels that have not been measured, piecewise linear interpolation, monotonic spline interpolation, or switching of the correction layer according to the method specified by the display control system are performed between the boundary increment parameters of adjacent low gray test gray levels. The interpolation results still need to be quantized by executable correction codes and checked for window residuals.

[0177] Step S800: After loading the boundary correction parameters, a retest is performed, and the relevant splicing boundaries are locally updated after the module is replaced.

[0178] Specifically, the display screen is re-controlled to show each low gray test pattern and images are acquired, and the boundary optical residuals are recalculated according to step S300.

[0179] For example, the acceptance criteria can be set as follows: the relative brightness difference between the boundary distance layer and the internal reference areas on both sides is no greater than 2%, and the CIE 1976 UCS color difference is within acceptable limits. It is not greater than 0.003, and no same-direction code value jump exceeding the preset length occurs within any omnidirectional constraint window.

[0180] Furthermore, for splicing boundaries that do not reach the preset acceptance threshold but still have non-empty common reachable light output ranges, the continuous target light output is corrected based on the remeasured luminance residual and the remeasured chromaticity residual, or the edge optical correction coefficient is corrected and the low gray execution response of the corresponding boundary distance layer is updated based on the corrected edge optical correction coefficient; subsequently, based on the corrected continuous target light output or the updated low gray execution response, the adjacent executable correction codes are re-determined, the quantization residual orientation allocation is performed, and the boundary incremental correction table is updated.

[0181] If the number of iterations reaches the preset upper limit but the acceptance threshold is still not met, output the manual review information and save the coordinates of the corresponding splicing boundary, the remaining brightness residual, the remaining chromaticity residual, and the reason why the machine cannot correct it.

[0182] Specifically, such as Figure 5 As shown, when module M is replaced on-site, the module identifier, batch identifier, and basic calibration parameters of the new module are read, and the screen topology version is updated. Based on the updated screen topology, the upper, lower, left, and right splicing boundaries connected to module M are found to form a set of boundaries to be locally updated.

[0183] Furthermore, the control display screen only displays the low gray test pattern of the coverage module M and its adjacent modules, and acquires the corresponding local images; the boundary optical residual calculation, common reachable light output interval calculation, continuous target generation, response inversion and quantization residual orientation allocation are re-executed for each splicing boundary in the boundary set to be updated locally.

[0184] Replace the regenerated boundary incremental correction parameters with the corresponding boundary records in the boundary incremental correction table, retain the spliced ​​boundary records that are not connected to module M and the non-boundary basic correction parameters; after completing the local parameter loading, perform local retesting on the boundary set to be updated locally.

[0185] By combining continuous targets in the optical output domain, batch low-gray execution response, and quantization residuals distributed tangentially along the splicing boundary, a statically executable calibration code with limited precision can express continuous optical output targets across modules. Furthermore, through a common reachable optical output range, retest closed loop, and local update mechanism, the safety of the calibration process, production availability, and field maintenance efficiency are improved.

[0186] The second embodiment of the present invention relates to a cross-module color brightness consistency correction system for small-pitch LED splicing screens.

[0187] In one server-centric implementation, the calibration system includes a calibration server 300, an imaging acquisition device 100, and a display control device 200, which are external devices connected to the calibration server 300 via corresponding interfaces. The calibration server 300 acquires a lit image of the small-pitch LED splicing screen 700 to be calibrated through the imaging acquisition device 100, and controls the display control device 200 to display test patterns and load calibration parameters on the small-pitch LED splicing screen 700 to be calibrated.

[0188] like Figure 4 As shown, in a complete device implementation, the calibration system includes an imaging acquisition device 100, a display control device 200, a calibration server 300, a calibration parameter storage device 400, a module identification reading device 500, and a communication device 600, and is communicatively connected to the small-pitch LED splicing screen 700 to be calibrated.

[0189] Specifically, the imaging acquisition device 100 includes an industrial camera or imaging colorimeter, a lens, a camera mount, an exposure controller, and a trigger interface. The field of view of the imaging acquisition device 100 covers the entire screen, or covers a correction zone that can be registered with adjacent fields of view.

[0190] The imaging acquisition device 100 is used to acquire R monochrome images, G monochrome images, B monochrome images and white field images at reference gray levels and multiple low gray test gray levels, and outputs linear image data, exposure time and acquisition timestamp to the calibration server 300.

[0191] Specifically, the display control device 200 includes a video controller, a transmitting card, and multiple receiving cards. The video controller is communicatively connected to the calibration server 300 and is used to receive test pattern display commands, brightness setting commands, and calibration parameter loading commands.

[0192] Furthermore, the transmitting card communicates with multiple receiving cards to send test image data and correction parameters to the corresponding cabinet or module; the receiving card is used to save and recall basic correction parameters, low gray correction parameters, full gray correction parameters, or stitching correction parameters.

[0193] Specifically, the calibration server 300 can be an industrial computer set up at the production calibration station, a mobile workstation set up at the delivery site, or a server connected to a display screen via a network.

[0194] Preferably, the calibration server 300 includes a processor, a graphics processor, memory, non-volatile memory, a network interface, and a device control interface. The graphics processor is optional and is used to accelerate image preprocessing, color conversion, and parallel computation of candidate calibration code groups.

[0195] Furthermore, the network interface includes an image data interface, and the device control interface includes a display control interface. The image data interface is used to receive the illuminated image, exposure time, and acquisition timestamp output by the imaging acquisition device 100; the display control interface is used to send test pattern display instructions, brightness setting instructions, and calibration parameter loading instructions to the display control device 200.

[0196] Furthermore, the calibration server 300 is configured with a basic data management module 310, an image preprocessing and registration module 320, a boundary recognition and optical residual calculation module 330, a batch low gray response management module 340, a continuous target and correctability determination module 350, an executable calibration code determination module 360, a quantized residual orientation allocation module 370, a calibration table generation and loading module 380, and a retesting and local update module 390.

[0197] Specifically, the basic data management module 310 is used to read module identifiers, cabinet identifiers, module batch information, module basic calibration parameters, receiver card mapping relationships, and calibration parameter formats supported by the display control system; establish screen topology based on the row and column positions of each module in the whole screen, and establish boundary identifiers for each splicing boundary.

[0198] Specifically, the image preprocessing and registration module 320 is connected to the data output terminal of the imaging acquisition device 100, and is used to receive dark field images, flat field images, reference grayscale images and low gray point bright images, perform dark field subtraction, flat field correction, exposure normalization, multi-frame robust averaging, color conversion and screen pixel coordinate registration, and output light output data organized according to module coordinates.

[0199] Specifically, the boundary recognition and optical residual calculation module 330 is connected to the basic data management module 310 and the image preprocessing and registration module 320. It is used to determine the splicing boundary search area according to the screen topology, correct the splicing boundary position according to the brightness gradient and chromaticity gradient, divide the boundary distance layer, tangential pixel group and internal reference area, calculate the brightness residual and chromaticity residual, and output the brightness residual and chromaticity residual as the boundary optical residual.

[0200] Specifically, the batch low gray response management module 340 is connected to the calibration parameter storage device 400 and is used to read the calibration code-light output response sample according to the module batch, color channel and low gray test gray level.

[0201] Furthermore, when the corresponding batch of response samples is not available, the batch low gray response management module 340 controls the display control device 200 to load the trial correction code and controls the imaging acquisition device 100 to measure the corresponding light output; it performs monotonic processing on the response samples and forms a low gray execution response based on the edge optical efficiency of the boundary distance layer.

[0202] Specifically, the continuous target and correctability determination module 350 is connected to the boundary recognition and optical residual calculation module 330 and the batch low gray response management module 340 to form a cross-boundary continuous target based on the internal reference light output on both sides of the splicing boundary, and to determine the common reachable light output range based on the range of the correction code that the receiving card can execute, the low gray execution response and the safety drive margin.

[0203] Furthermore, the continuous target and correctability determination module 350 limits the amplitude of targets that exceed the common reachable light output range, and outputs a mechanically uncorrectable mark for splicing boundaries that still cannot meet the acceptance threshold within the common reachable light output range.

[0204] Specifically, the executable correction code determination module 360 ​​is connected to the continuous target and correctability determination module 350 and the batch low gray response management module 340. It is used to query the corresponding low gray execution response according to the batch to which the pixel group belongs, the color channel, the low gray test gray level and the boundary distance layer, determine the upper neighbor executable correction code and the lower neighbor executable correction code on both sides of the continuous target light output, and form R, G and B candidate correction code groups.

[0205] Specifically, the quantization residual orientation allocation module 370 is connected to the executable correction code determination module 360, and is used to process pixel groups according to the splicing boundary tangential direction within the same boundary distance layer, and select the target executable correction code by combining the cumulative luminance quantization residual, the cumulative chrominance quantization residual and the continuous code value arrangement constraint.

[0206] Furthermore, the quantization residual orientation allocation module 370 is also used to perform residual checks on the tangential constraint window, and adjust the upper neighbor code and lower neighbor code combination within the window when the window does not meet the residual constraint conditions.

[0207] Specifically, the calibration table generation and loading module 380 is connected to the quantization residual orientation allocation module 370 and the display control device 200, and is used to generate a boundary increment calibration table according to the splicing boundary identifier, low gray test gray level, boundary distance layer, tangential pixel group number and color channel.

[0208] Furthermore, when the receiving card supports an independent calibration layer, the calibration table generation and loading module 380 loads the boundary incremental calibration table as an independent calibration layer; when the receiving card does not support an independent calibration layer, the boundary incremental calibration table is merged with the basic calibration parameters and loaded after being converted into a data format that the receiving card can recognize.

[0209] Specifically, the retest and local update module 390 is connected to the imaging acquisition device 100, the boundary recognition and optical residual calculation module 330, and the calibration table generation and loading module 380. It is used to control the display screen to re-display the low gray test pattern after the calibration parameters are loaded, and to determine whether the corresponding splicing boundary has reached the acceptance threshold based on the retest results.

[0210] Furthermore, after the module is replaced, the retest and local update module 390 determines the splicing boundary that connects to the replaced module from the screen topology, and only triggers image acquisition, correction calculation and parameter update of the splicing boundary.

[0211] Specifically, the calibration parameter storage device 400 can be located in the calibration server 300, or it can be a database server that is communicatively connected to the calibration server 300.

[0212] Preferably, the calibration parameter storage device 400 includes a module basic calibration parameter library, a batch low gray response library, a screen topology library, a boundary incremental calibration table library, a retest result library, and a release version library.

[0213] Among them, the module basic calibration parameter library stores the basic calibration parameters and calibration versions generated by the module during the factory point-to-point calibration stage; the batch low gray response library stores the calibration code-light output response corresponding to each batch, each color channel and each low gray test gray level; and the screen topology library stores the correspondence between the module, the cabinet, the receiver card and the splicing boundary.

[0214] Furthermore, the boundary incremental correction table library stores the incremental parameters and generation conditions of each splicing boundary; the retest result library stores the boundary residuals before and after correction, acceptance status, and mechanically uncorrectable marks; and the version distribution library stores the correction database version, loading time, and operation records for each time it is loaded to the display control device 200, in order to support version traceability and parameter rollback.

[0215] Specifically, the module identification reading device 500 includes a QR code reader, a barcode reader, an RFID reader, or a communication interface connected to the module's non-volatile memory, used to send the module identification and batch identification to the basic data management module 310.

[0216] Optionally, when the module identification and batch information are already provided by the production management system, the independent module identification reading device 500 can be omitted, and the calibration server 300 can directly read the corresponding information from the production management system.

[0217] Specifically, the communication device 600 includes at least one of an Ethernet switch, a USB interface, an image acquisition interface, and a serial communication interface. The imaging acquisition device 100, the display control device 200, the calibration parameter storage device 400, and the module identification reading device 500 exchange data with the calibration server 300 through the communication device 600.

[0218] The small-pitch LED splicing screen 700 to be calibrated includes multiple cabinets and multiple LED modules. Each cabinet is equipped with at least one receiver card, which is connected to the driving circuit of the corresponding module to drive the LED pixels to emit light according to video data, basic calibration parameters, and boundary incremental calibration parameters.

[0219] In one specific embodiment, the system is deployed at the production calibration station. The imaging acquisition device 100 is fixedly installed in a darkroom or a light-shielding calibration area, and the calibration server 300 is connected to the production management system to automatically read the module batch and the factory basic calibration parameters. After calibration, the boundary incremental calibration table, retest results, and screen topology version are written into the calibration parameter storage device 400.

[0220] In another specific embodiment, the system is deployed at the display delivery or maintenance site. The calibration server 300 is a mobile workstation, and the imaging acquisition device 100 is a portable industrial camera or imaging colorimeter. The mobile workstation reads the receiver card mapping relationship and current calibration parameters from the display control system, and loads the final calibration parameters into the receiver card after completing boundary calibration.

[0221] Furthermore, the mobile workstation synchronizes the basic calibration parameters, boundary incremental calibration table, retest results, and release version to the remote database server to enable cross-project screen file management and after-sales maintenance.

[0222] Optionally, the calibration server 300 and the database server are configured separately. The calibration server 300 performs image processing and calibration calculations, while the database server performs batch response sharing, calibration version management, and screen file management.

[0223] In another alternative implementation, the calibration server function and the database server function are centrally located on an industrial computer to reduce the deployment complexity of the production calibration station or the on-site calibration system.

[0224] Furthermore, the aforementioned functional modules can be implemented by computer programs running on the processor, or by a processor, graphics processor, field-programmable gate array, application-specific integrated circuit, or a combination thereof.

[0225] The memory is used to store computer programs, screen topology, image data, batch low gray response data, and correction parameters. When the processor executes the computer program, the control system completes the image acquisition, boundary recognition, response construction, continuous target generation, adjacent correction code determination, quantization residual orientation allocation, correction table loading, retesting, and local update processing in the first embodiment.

[0226] This implementation method forms a closed loop for full-screen illumination image acquisition and calibration parameter loading through an imaging acquisition device, a display control device, and a calibration server. Furthermore, through a batch low-gray response management module, a continuous target and correctability determination module, an executable calibration code determination module, and a quantization residual orientation allocation module, the continuous light output targets of different batches of modules can be converted into static calibration codes that are actually executable by the receiving card. Furthermore, through a calibration parameter storage device and a retest and local update module, hierarchical storage and local maintenance of basic parameters, boundary incremental parameters, retest results, and issued versions are achieved. Therefore, the system can be implemented based on existing LED display production calibration equipment and display control systems.

[0227] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any ordinary changes and substitutions made by those skilled in the art within the scope of the technical solution of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for correcting color brightness consistency across modules in a small-pitch LED splicing screen, characterized in that, include: Acquire the basic calibration parameters of each module in the small-pitch LED splicing screen, the module splicing relationship, the module batch information, the reference grayscale image, the low gray lighting images under multiple low gray test grayscales, and the calibration code-light output response of each module batch; Based on the module splicing relationship, reference grayscale image and low gray point bright image, the splicing boundary between adjacent modules is determined. Multiple boundary distance layers are divided on both sides of the splicing boundary and an internal reference area is set. The pixels in each boundary distance layer are determined as sequentially arranged correction processing units along the tangent of the splicing boundary. The correction processing unit is a single pixel or a pixel group composed of multiple pixels adjacent along the tangent. The boundary optical residual is determined based on the brightness difference and chromaticity difference between each boundary distance layer and the corresponding internal reference area. Based on the correction code-optical output response and the optical output characteristics of each boundary distance layer, the low gray execution response corresponding to each correction processing unit is determined, and based on the internal reference optical output on both sides of the splicing boundary and the boundary optical residual, a continuous target optical output spanning the splicing boundary is generated. Based on the low-gray execution response, adjacent executable correction codes on both sides of the continuous target light output are determined; for correction processing units arranged sequentially along the tangential direction within the same boundary distance layer, a target executable correction code is selected from the adjacent executable correction codes based on the cumulative luminance quantization residual and the cumulative chrominance quantization residual, and the cumulative luminance quantization residual and the cumulative chrominance quantization residual are updated based on the difference between the actual light output corresponding to the target executable correction code and the continuous target light output; The cumulative luminance quantization residual and the cumulative chrominance quantization residual are propagated only along the tangential direction within the same boundary distance layer, and are not propagated to adjacent boundary distance layers. Based on the target executable correction code of each correction processing unit, a boundary incremental correction table relative to the basic correction parameters is generated, and the boundary incremental correction table is loaded into the display control system of the small-pitch LED splicing screen.

2. The method for correcting cross-module color brightness consistency in small-pitch LED splicing screens according to claim 1, characterized in that, Determining the splicing boundary and boundary distance layer includes: The reference grayscale image and the low grayscale bright image are subjected to dark field subtraction, flat field correction, exposure normalization and multi-frame fusion. Based on the module splicing relationship, boundary search bands are set on both sides of the nominal splicing position of adjacent modules, and the position of the splicing boundary is corrected according to the brightness gradient of the reference grayscale image and the chromaticity gradient of the low gray point bright image. The number of boundary distance layers and the width of the boundary compensation strip are determined based on the attenuation of the boundary optical residual along the direction perpendicular to the splicing boundary.

3. The method for correcting cross-module color brightness consistency in small-pitch LED splicing screens according to claim 1, characterized in that, Determining the low-gray execution response includes: Read the calibration code-light output response samples stored according to module batch, color channel, and low gray test grayscale; when the corresponding response sample does not exist, load multiple trial calibration codes into the module of the corresponding batch, and measure the light output corresponding to each trial calibration code to obtain the calibration code-light output response sample. The measurement confidence of each response sample is determined based on at least one of multi-frame measurement variance, effective pixel ratio of exposure, and repeated measurement stability, and the response samples are monotonicized according to the measurement confidence to obtain segmented monotonic batch responses. Based on the ratio between the light output of each boundary distance layer and the light output inside the corresponding batch module, edge optical correction is performed on the batch response to obtain the low gray execution response of the corresponding boundary distance layer.

4. The method for correcting cross-module color brightness consistency in small-pitch LED splicing screens according to claim 1, characterized in that, Generating the continuous target light output includes: Based on the low-gray execution response, executable correction code range and safety drive margin on both sides of the splicing boundary, the reachable optical output range on both sides of the splicing boundary is determined, and the intersection of the reachable optical output range on both sides is determined as the common reachable optical output range. Using the internal reference light outputs on both sides of the splicing boundary as anchor points, an initial continuous target light output with a smooth transition is generated according to the position of each correction processing unit relative to the splicing boundary, and the initial continuous target light output is restricted within the common reachable light output range to obtain the continuous target light output; When the common reachable light output range is empty, or when the preset correction conditions cannot be met within the common reachable light output range, the corresponding splicing boundary is marked as a mechanically uncorrectable boundary.

5. The method for correcting cross-module color brightness consistency in small-pitch LED splicing screens according to claim 1, characterized in that, Selecting a target executable correction code from the adjacent executable correction codes includes: Based on the low-gray execution response, determine the upper neighbor executable correction code and the lower neighbor executable correction code located on both sides of the continuous target light output, respectively; Combine the upper and lower adjacent executable correction codes of the R, G, and B color channels into multiple candidate correction code groups; Based on the differences between the actual luminance and the target luminance, the differences between the actual chrominance and the target chrominance corresponding to each candidate correction code group, the cumulative luminance quantization residuals and chrominance quantization residuals of the preceding correction processing units, and the code value arrangement constraints of adjacent correction processing units, the selection cost of each candidate correction code group is determined, and the candidate correction code group with the minimum selection cost is determined as the target executable correction code of the current correction processing unit.

6. The method for correcting cross-module color brightness consistency in small-pitch LED splicing screens according to claim 5, characterized in that, Also includes: Multiple correction processing units arranged continuously along the tangential direction within the same boundary distance layer are set as tangential residual constraint windows; When the absolute value of the sum of the brightness differences between the target light output and the actual light output of each correction processing unit within the tangential residual constraint window exceeds the preset brightness residual threshold, or the norm of the sum of the chromaticity difference vectors exceeds the preset chromaticity residual threshold, the selection combination of the upper neighbor executable correction code and the lower neighbor executable correction code within the tangential residual constraint window is adjusted. Among the multiple boundary distance layers, adjacent boundary distance layers adopt opposite tangential traversal directions.

7. The method for correcting cross-module color brightness consistency in small-pitch LED splicing screens according to claim 1, characterized in that, Generating and loading the boundary incremental correction table includes: Using the splicing boundary identifier, low gray test gray level, boundary distance layer, tangential sequence number of the correction processing unit and color channel as indexes, store the boundary increment of each target executable correction code relative to the corresponding basic correction parameter; When the display control system supports an independent calibration layer, the boundary incremental calibration table and the basic calibration parameters are loaded separately; when the display control system does not support an independent calibration layer, the boundary incremental calibration table and the basic calibration parameters are loaded together. For input gray levels that have not been tested, interpolation is performed based on the boundary increments corresponding to adjacent low-gray test gray levels to determine the boundary increments corresponding to the input gray levels that have not been tested.

8. The method for correcting cross-module color brightness consistency in small-pitch LED splicing screens according to claim 1, characterized in that, Also includes: After loading the boundary incremental correction table, the small-pitch LED splicing screen is controlled to re-display the low gray test pattern and acquire the retest image. The retest brightness residual and retest chromaticity residual of each splicing boundary are determined based on the retest image. For splicing boundaries that do not reach the preset acceptance threshold, update the continuous target light output or low gray execution response according to the corresponding retest brightness residual and retest chromaticity residual, redetermine the target executable correction code and update the boundary incremental correction table. After replacing the module, the splicing boundary connected to the replaced module is determined according to the updated module splicing relationship, and only the boundary increment correction parameter corresponding to the splicing boundary connected to the replaced module is updated.

9. A cross-module color brightness consistency correction system for small-pitch LED splicing screens, used for correcting small-pitch LED splicing screens, characterized in that... It includes a calibration server, which includes a processor, a memory, an image data interface, and a display control interface; The image data interface is used to receive the illuminated images of the small-pitch LED splicing screen acquired by the imaging acquisition device under a reference grayscale and multiple low grayscale test grayscales. The display control interface is used to send test pattern display instructions and calibration parameter loading instructions to the display control device that is communicatively connected to the small-pitch LED splicing screen; The memory is used to store programs; the calibration server is also used to acquire module basic calibration parameters, module splicing relationships, module batch information, calibration code-optical output response, and boundary incremental calibration table; When the program is executed by the processor, it causes the correction server to perform the steps of the method according to any one of claims 1 to 8.

10. The cross-module color brightness consistency correction system for small-pitch LED splicing screen according to claim 9, characterized in that, It also includes an imaging acquisition device, a display control device, and a calibration parameter storage device; The imaging acquisition device is connected to the image data interface and includes an industrial camera or an imaging colorimeter, used to acquire the illuminated image; The display control device is connected to the display control interface and includes a video controller, a sending card, and a receiving card. It is used to control the small-pitch LED splicing screen to display test patterns and load the boundary incremental correction table. The calibration parameter storage device is located within the calibration server, or is configured as a database server that is communicatively connected to the calibration server, for storing the module's basic calibration parameters, module splicing relationships, module batch information, calibration code-optical output response, boundary incremental calibration table, and retest results.

Images