An LED module correction parameter generation method and system

Abstract

The present application relates to the technical field of LED, and particularly relates to a LED module correction parameter generation method and system, the method comprising: acquiring acquisition response data of a LED module to be corrected in dark state, single base color step adjustment and multi-base color combination state, and associating driving code, acquisition condition and module position; establishing a measurement conversion relationship of acquisition channel response to standard light color space based on a reference light emitting sample, to obtain measured standard light color data; performing dark state deduction, condition correction and abnormality rejection on the measured standard light color data, to generate measurement uncertainty; establishing a reverse light color characterization model introducing a multi-base color mutual influence term according to the driving code and the corrected data; determining an effective area and residual distribution after verification by a verification sample, to generate correction parameters containing conversion parameters and reliability information. The present application can solve the problems of inaccurate conversion between acquisition response and standard light color data and difficulty in reflecting multi-base color coupling influence in the existing LED module correction process.

Description

Technical Field

[0001] This invention relates to the field of LED technology, and in particular to a method and system for generating calibration parameters for LED modules. Background Technology

[0002] LED modules are widely used in direct-view displays, splicing displays, virtual shooting background screens, and high-precision visual display devices. Due to factors such as batch differences in light-emitting chips, packaging differences, drive chain differences, differences in optical acquisition conditions, and long-term use degradation, problems such as inconsistent brightness, color shift, and unstable grayscale response can easily occur between different light-emitting units within the same module and between different modules. In the prior art, US9276766B2 discloses a display calibration system and related methods, which utilizes the photosensitivity of LEDs to calibrate differences between LEDs. EP2367348B1 discloses a method for generating a color correction lookup table for display devices, which generates the color correction lookup table by comparing the grayscale stimulus values ​​of primary colors and target stimulus values. WO2012030719A2 discloses an adaptive color correction scheme that generates a color correction matrix for different backlight levels. The above solutions can improve the color or brightness consistency of display devices to a certain extent. However, in the scenario of on-site or batch calibration of LED modules, there are still problems such as the difficulty in unifying the channel response of the acquisition device to the standard color space, the difficulty in characterizing the mutual influence of channels when multiple primary colors emit light together, and the difficulty in quantifying the effective range and parameter reliability of the calibration model. As a result, the generated calibration parameters are not stable enough in different brightness ranges, different acquisition conditions and different module areas. Summary of the Invention

[0003] In view of the above technical problems, the present invention provides a method and system for generating calibration parameters for LED modules, so as to solve the problems of inaccurate conversion between the acquired response and standard light color data, difficulty in reflecting the influence of multi-primary color coupling, and unclear applicability and reliability of calibration models and parameters in the existing LED module calibration process.

[0004] Other features and advantages of this disclosure will become apparent from the following detailed description, or may be learned in part from practice of this disclosure.

[0005] According to one aspect of the present invention, a method for generating calibration parameters for an LED module is provided, the method comprising: Acquire the acquisition response data of the LED module to be calibrated in dark state, single primary color different tone states, and multi-primary color combination state with at least two primary colors lit at the same time, and associate the acquisition response data with the corresponding driving code, acquisition conditions and module position information; Based on the response of the reference luminescent sample in each acquisition channel of the acquisition device and the reference standard color data of the reference luminescent sample, a measurement conversion relationship from the acquisition channel response to the standard color space is established, and the acquisition response data is converted using the measurement conversion relationship to obtain the measured standard color data of the LED module to be calibrated. The measured standard light color data is subjected to dark state subtraction, acquisition condition correction and abnormal sample removal, and measurement uncertainty corresponding to the preset brightness range is generated based on the repeated acquisition results; A reverse colorimetric representation model is established based on the driving code and the modified measured standard colorimetric data. The reverse colorimetric representation model is used to represent the mapping relationship between the input target standard colorimetric data and the driving code, and the mutual influence term between multiple primary colors is introduced when establishing the reverse colorimetric representation model. The reverse light color characterization model is validated using a verification sample. Based on the visual color difference between the target standard light color data of the verification sample and the re-acquired measured standard light color data, the effective area and residual distribution of the reverse light color characterization model are determined. Based on the inverse light color characterization model, the effective region, the residual distribution, and the measurement uncertainty, correction parameters for the LED module to be corrected are generated. The correction parameters include conversion parameters from the target standard light color data to the driving code and confidence information corresponding to the conversion parameters.

[0006] Furthermore, the module position information is used to characterize the target correction area in the LED module to be corrected, and the target correction area includes a display partition composed of multiple light-emitting units; different target correction areas correspond to different reverse light color characterization models and correction parameters.

[0007] Furthermore, the acquired response data includes dark state response data, single primary color response data, and mixed primary color response data; the dark state response data is used to determine the black level compensation amount, the single primary color response data is used to determine the tone response of each primary color channel, and the mixed primary color response data is used to determine the mutual influence of channels when multiple primary colors emit light together.

[0008] Furthermore, when establishing the measurement conversion relationship, a weighted average is performed on the multiple acquisition responses of the reference luminescent sample. The reference response of the acquisition channel is determined based on the weighted average result, and the parameters of the measurement conversion relationship are formed based on the deviation between the reference response of the acquisition channel and the reference standard light color data. The parameters of the measurement conversion relationship are used to correct for differences in channel response, acquisition gain, and environmental background effects of the acquisition device.

[0009] Furthermore, the abnormal sample removal includes removing samples that reach the nonlinear response region of the acquisition device, removing samples that are below the effective signal-to-noise ratio requirement, and removing samples that deviate from the residual distribution of similar samples after conversion by the measurement conversion relationship.

[0010] Furthermore, when establishing the inverse light color representation model, one or more of the following are trained: a multi-order regression sub-model, a lookup interpolation sub-model, and a neural estimation sub-model. The visual color difference statistics of the multi-order regression sub-model, the lookup interpolation sub-model, or the neural estimation sub-model are obtained using the verification samples. The target sub-model used to generate the correction parameters is determined based on the visual color difference statistics and the model complexity.

[0011] Furthermore, the multi-order regression sub-model includes a driving encoding term, a higher-order term, and a primary color interaction term; before training the multi-order regression sub-model, black level compensation is performed on the measured standard light color data, and the model order is limited according to the visual color difference change of the verification sample during the training process.

[0012] Furthermore, the lookup interpolation sub-model uses the corrected measured standard light color data as index nodes and the corresponding driving code as node values; when the target standard light color data is input, the set of nodes surrounding the target standard light color data is determined, and the interpolation driving code is generated according to the relative position of each node in the set of nodes.

[0013] Furthermore, after generating the correction parameters, the LED module to be corrected is driven to display the target verification screen according to the correction parameters, and the measured standard light color data of the target verification screen is re-acquired; when the visual color difference between the measured standard light color data and the corresponding target standard light color data exceeds the allowable range corresponding to the confidence information, samples are collected in the corresponding effective area, and the reverse light color characterization model and the correction parameters are updated.

[0014] According to another aspect of the present invention, an LED module calibration parameter generation system is provided, comprising: The response acquisition module is used to acquire the acquisition response data of the LED module to be calibrated in the dark state, the single primary color at different tones, and the multi-primary color combination state in which at least two primary colors are lit at the same time, and associate the acquisition response data with the corresponding driving code, acquisition conditions and module position information; The light color conversion module is used to establish a measurement conversion relationship from the acquisition channel response to the standard light color space based on the response of the reference luminescent sample in each acquisition channel of the acquisition device and the reference standard light color data of the reference luminescent sample, and to convert the acquisition response data using the measurement conversion relationship to obtain the measured standard light color data of the LED module to be calibrated. The data correction module is used to perform dark state subtraction, acquisition condition correction and abnormal sample removal on the measured standard light color data, and generate the measurement uncertainty corresponding to the preset brightness range based on the repeated acquisition results; The reverse modeling module is used to establish a reverse light color representation model based on the driving code and the modified measured standard light color data. The reverse light color representation model is used to represent the mapping relationship between the input target standard light color data and the driving code, and introduces the mutual influence term between multiple primary colors when establishing the reverse light color representation model. The model validation module is used to validate the reverse light color representation model using validation samples, and to determine the effective area and residual distribution of the reverse light color representation model based on the visual color difference between the target standard light color data of the validation samples and the re-acquired measured standard light color data. The parameter generation module is used to generate correction parameters for the LED module to be corrected based on the reverse light color characterization model, the effective area, the residual distribution, and the measurement uncertainty. The correction parameters include conversion parameters from the target standard light color data to the driving code and confidence information corresponding to the conversion parameters.

[0015] The technical solution of the present invention has the following beneficial effects: This invention associates the acquired response data of the LED module to be calibrated with the driver code, acquisition conditions, and module location information, and converts the channel response of the acquisition device to a standard color space. This allows subsequent calibration to no longer rely solely on the original acquired grayscale or single brightness data, improving the consistency of data across different acquisition devices and conditions. By correcting the measured standard color data and establishing a reverse color characterization model from the target standard color data to the driver code, the invention can consider the mutual influence of primary color tone response and multiple primary colors emitting light together when generating calibration parameters, reducing the mutual interference between brightness consistency calibration and color consistency calibration. By determining the effective area of ​​the model, residual distribution, and measurement uncertainty through verification samples, and outputting the confidence information along with the calibration parameters, the invention provides a basis for judgment on the writing, verification, and subsequent updates of calibration parameters, thereby improving the accuracy, stability, and traceability of LED module calibration parameters. Attached Figure Description

[0016] Figure 1 This is a flowchart of a method for generating LED module calibration parameters in an embodiment of this specification; Figure 2 This is a structural block diagram of an LED module calibration parameter generation system as described in the embodiments of this specification. Detailed Implementation

[0017] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided to make this disclosure more comprehensive and complete, and to fully convey the concept of the example embodiments to those skilled in the art. The described features, structures, or characteristics can be combined in any suitable manner in one or more embodiments. In the following description, numerous specific details are provided to give a full understanding of embodiments of this disclosure. However, those skilled in the art will recognize that the technical solutions of this disclosure can be practiced with one or more of the specific details omitted, or other methods, components, systems, steps, etc., can be employed. In other instances, well-known technical solutions are not shown or described in detail to avoid obscuring various aspects of this disclosure.

[0018] Furthermore, the accompanying drawings are merely illustrative of this disclosure. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted. Some block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities may be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor systems and / or microcontroller systems.

[0019] This invention provides a method for generating calibration parameters for LED modules. (Refer to...) Figure 1 The diagram shown is a flowchart illustrating a method for generating LED module calibration parameters according to an embodiment of the present invention. Specifically, this method may include the following steps S101~S106: In step S101, the acquisition response data of the LED module to be calibrated is obtained in the dark state, the single primary color at different tones, and the multi-primary color combination state in which at least two primary colors are lit at the same time. The acquisition response data is then associated with the corresponding driving code, acquisition conditions, and module position information.

[0020] The module position information is used to characterize the target correction area in the LED module to be corrected. The target correction area includes a display partition composed of multiple light-emitting units. Different target correction areas correspond to different reverse light color characterization models and correction parameters. The acquired response data includes dark state response data, single primary color response data, and mixed primary color response data. The dark state response data is used to determine the black level compensation amount, the single primary color response data is used to determine the tone response of each primary color channel, and the mixed primary color response data is used to determine the mutual influence of channels when multiple primary colors emit light together.

[0021] In this embodiment, the LED module to be calibrated can be an RGB module, an RGBW module, or an LED module with more primary color channels. The driving code can be the digital input value of each primary color channel, and the acquisition device can be an industrial camera, an imaging colorimeter, an image sensor with filter channels, or other optical acquisition devices capable of obtaining channel-specific responses. Acquisition conditions can include one or more of the following: exposure time, acquisition gain, acquisition distance, acquisition angle, acquisition duration, environmental background, and acquisition device identifier.

[0022] Dark state refers to the state of the LED module under calibration when no valid illumination code is input. Dark state response data is used to characterize the background response of the acquisition device, the residual black level response of the module, and the environmental background response. Single primary color different tone states refer to only illuminating one primary color channel and changing the driving code of that primary color channel according to a preset tone interval, while keeping other primary color channels unlit or below the valid illumination threshold. The single primary color response data obtained in this way can reflect the tone response of each primary color channel as the driving code changes, and provides a sample basis for subsequent black level compensation, tone response determination, and model building.

[0023] A multi-primary-color combination state can refer to at least two primary color channels being lit simultaneously, forming multiple driving code combinations according to preset combination rules. Driving code combinations can include regular grid combinations, random combinations, key brightness range encrypted combinations, or color gamut boundary combinations. Mixed primary color response data is used to reflect the actual acquisition response when multiple primary colors emit light together, enabling the subsequent inverse light color characterization model to identify situations where the sum of single-primary-color responses cannot fully represent the mixed emission response, thereby reducing the impact of mutual influence between primary color channels on the generation of correction parameters.

[0024] Module location information can include the region number, region coordinates, row and column positions of the light-emitting units, display partition identifier, or module splicing position of the target correction area. The target correction area can be divided according to pixel array, LED array, display partition, or splicing edge area. For areas with significant brightness attenuation, color shift, or differences in optical acquisition angle, a smaller target correction area can be set; for areas with high response consistency, a larger target correction area can be set. By associating the acquired response data with the module location information, subsequent models and correction parameters can be generated by region, avoiding the loss of local correction accuracy caused by the mixing of response differences from different regions.

[0025] During the acquisition process, each set of acquired response data forms an acquisition sample record along with the corresponding drive code, acquisition conditions, and module location information. The acquisition sample record may include a dark state identifier, a single primary color identifier, a multi-primary color combination identifier, the drive code for each primary color channel, the channel response value of the acquisition device, the target correction area identifier, and the acquisition condition identifier.

[0026] In step S102, based on the response of the reference luminescent sample in each acquisition channel of the acquisition device and the reference standard color data of the reference luminescent sample, a measurement conversion relationship from the acquisition channel response to the standard color space is established, and the acquisition response data is converted using the measurement conversion relationship to obtain the measured standard color data of the LED module to be calibrated.

[0027] In establishing the measurement conversion relationship, a weighted average is performed on the multiple acquisition responses of the reference luminescent sample. The reference response of the acquisition channel is determined based on the weighted average result, and the parameters of the measurement conversion relationship are formed based on the deviation between the reference response of the acquisition channel and the reference standard light color data. The parameters of the measurement conversion relationship are used to correct for differences in channel response, acquisition gain, and environmental background effects of the acquisition equipment.

[0028] The reference emission sample can be a standard light source with known reference standard colorimetric data, a standard color block emission element, an LED sample calibrated by a standard photometric device, or other stable emission samples. The reference standard colorimetric data can include one or more of the following: standard luminance values, standard chromaticity coordinates, XYZ tristimulus values, CIELAB data, or standard photometric channel data. The acquisition channels of the acquisition device can be a red acquisition channel, a green acquisition channel, and a blue acquisition channel, or multiple spectral response channels corresponding to the filter structure of the acquisition device.

[0029] When acquiring reference luminescent samples multiple times, multiple sets of channel responses can be obtained under the same acquisition conditions, or multiple sets of channel responses can be obtained under different acquisition conditions while retaining the corresponding acquisition condition identifiers. For the c-th acquisition channel, the weighted average result of multiple acquisition responses can be expressed as: ; In the formula, This represents the reference response of the c-th acquisition channel. This represents the response value of the c-th acquisition channel in the q-th acquisition. This represents the weight corresponding to the q-th acquisition, where n represents the number of acquisitions. The weight can be determined based on exposure stability, signal-to-noise ratio, linearity of the acquisition equipment, or consistency of acquisition conditions. Weighted averaging can reduce the impact of fluctuations in single acquisitions on measurement conversion relationships.

[0030] In one implementation, the measurement conversion relationship can be represented using a matrix conversion relationship as follows: ; In the formula, This represents the standard color vector corresponding to the reference standard color data. This represents a reference response vector composed of reference responses from multiple acquisition channels. This represents the measurement conversion matrix. The measurement conversion matrix can be obtained by fitting the reference response vectors of multiple reference emission samples with the corresponding reference standard color data, and is used to unify the channel response of the acquisition device to the standard color space.

[0031] In another implementation, when the standard color space is represented by photometric or equivalent channel quantity, the channel photometric can be obtained first based on the acquisition channel response and the corresponding zero-point correction. The channel photometric can be expressed as: ; In the formula, This represents the channel photometric value of the c-th acquisition channel. This represents the effective response strength of the c-th acquisition channel. This represents the zero-point correction value corresponding to the c-th acquisition channel. The zero-point correction value can be determined based on the difference between the reference standard colorimetric data of the reference emission sample and the reference response of the acquisition channel. In this way, the acquisition channel response can be converted into a photometric value with a unified reference, and then further mapped to the standard color space.

[0032] When converting the acquired response data of the LED module to be calibrated, the acquired response data can be organized into a response vector to be converted according to the acquisition channel, and the measurement conversion relationship can be input to obtain the corresponding measured standard light color data. The measured standard light color data continues to be associated with the driver code, acquisition conditions, and module position information, so that subsequent steps can correct and model data under different lighting states, different target calibration areas, and different acquisition conditions within the same standard light color space.

[0033] In step S103, the measured standard light color data is subjected to dark state subtraction, acquisition condition correction and abnormal sample removal, and measurement uncertainty corresponding to the preset brightness range is generated based on the repeated acquisition results.

[0034] The abnormal sample removal includes removing samples that reach the nonlinear response region of the acquisition device, removing samples that are below the effective signal-to-noise ratio requirement, and removing samples that deviate from the residual distribution of similar samples after conversion by the measurement conversion relationship.

[0035] In this embodiment, dark-state subtraction is used to remove the influence of black spot residue, background response of the acquisition device, and environmental background on the measurement results from the measured standard color data. For the i-th acquired sample, the standard color data after dark-state subtraction can be expressed as: ; In the formula, This represents the standard light color data after dark state subtraction. This represents the measured standard light color data obtained through conversion. This represents the dark-state standard color data corresponding to the target correction region p to which the i-th acquired sample belongs. The dark-state standard color data can be obtained by converting the dark-state response data through measurement conversion relationships, or it can be obtained by statistically analyzing the results of multiple dark-state acquisitions under the same acquisition conditions.

[0036] Acquisition condition correction is used to reduce response differences caused by variations in exposure time, acquisition gain, acquisition distance, acquisition angle, and acquisition duration. For acquired data that is approximately linearly correlated with exposure time and acquisition gain, the standard colorimetric data after acquisition condition correction can be expressed as: ; In the formula, This represents the standard light and color data after correction of the acquisition conditions. This represents the acquisition gain corresponding to the i-th acquired sample. This represents the exposure time corresponding to the i-th sample. and These represent the reference acquisition gain and reference exposure time, respectively. For situations with significant variations in acquisition distance, acquisition angle, or environmental background, a correction factor can be applied based on pre-calibrated acquisition conditions. Make corrections.

[0037] Outlier removal can be performed at both the acquisition channel response level and the standard optical color data level. When the raw response of any acquisition channel reaches the upper limit of the linear response of the acquisition device, or falls below the effective signal-to-noise ratio requirement, the corresponding acquired sample will not participate in subsequent uncertainty statistics and model building. The effective signal-to-noise ratio can be expressed as: ; In the formula, This represents the effective signal-to-noise ratio of the i-th collected sample. This represents the average of multiple acquisitions of the response under the same target correction area and the same illumination state. This indicates the corresponding response fluctuation. When When the signal-to-noise ratio is below the preset threshold, the corresponding sample is discarded.

[0038] For the collected samples after conversion based on measurement conversion relationships, similar sample sets can be formed according to the same brightness range, the same target correction area, or the same point illumination state. Deviation samples are then removed based on the residual distribution of these similar sample sets. The residual deviation index can be expressed as: ; In the formula, This represents the residual deviation index of the i-th sample relative to the h-th class of samples. This represents the standard light color mean of the h-th class of samples. This represents the residual fluctuation of the h-th class of samples. When the residual exceeds the preset residual threshold, the corresponding sample is judged to deviate from the residual distribution of the same type of sample.

[0039] Measurement uncertainty can be generated separately according to preset brightness ranges. For the b-th preset brightness range, the measurement uncertainty formed by repeated acquisitions can be expressed as: ; In the formula, This represents the measurement uncertainty corresponding to the b-th preset brightness range. This represents the number of samples collected within the b-th preset brightness range that participated in the statistics. This represents the corrected standard color value within the b-th preset brightness range.

[0040] In step S104, a reverse color representation model is established based on the driving code and the modified measured standard color data. The reverse color representation model is used to represent the mapping relationship between the input target standard color data and the driving code, and the mutual influence term between multiple primary colors is introduced when establishing the reverse color representation model.

[0041] When establishing the inverse light color representation model, one or more of the following are trained: multi-order regression sub-model, lookup value interpolation sub-model, and neural estimation sub-model.

[0042] The multi-order regression sub-model includes a self-encoding term, a higher-order term, and a primary color interaction term. Before training the multi-order regression sub-model, black level compensation is performed on the measured standard light color data, and the model order is limited according to the visual color difference changes of the verification samples during the training process.

[0043] The lookup interpolation sub-model uses the corrected measured standard light color data as index nodes and the corresponding driving code as node values. When the target standard light color data is input, the set of nodes surrounding the target standard light color data is determined, and the interpolation driving code is generated according to the relative position of each node in the set of nodes.

[0044] In this embodiment, when establishing the inverse light color characterization model, the modified measured standard light color data can be used as the model input sample, and the driving code corresponding to the measured standard light color data can be used as the model output sample, thereby forming an inverse mapping relationship from the target standard light color data to the driving code. The target standard light color data may include one or more of the following: target XYZ tristimulus values, target CIELAB data, target luminance values, and target chromaticity coordinates. The driving code may include the digital input values ​​of each primary color channel, such as the red channel code, green channel code, and blue channel code, and may also include the digital input values ​​of the white channel or other extended primary color channels.

[0045] A multi-order regression sub-model is used to describe the nonlinear relationship between the target standard light color data and the driving code. For a three-primary-color LED module, the target standard light color data can be represented as... and based on Construct multi-order feature terms. A second-order feature term can be represented as: ; In the formula, Represents a second-order eigenvector. , , These represent the tristimulus components in the target standard colorimetric data. , }、 This represents the interaction terms between components. Interaction terms characterize situations where, when multiple primary color channels emit light together, the change in target color is not independently determined by a single primary color channel. The corresponding driving coding prediction relationship can be expressed as: ; In the formula, Indicates prediction-driven coding. This represents the multi-order regression parameter matrix. The multi-order regression parameter matrix can be obtained by fitting sample pairs formed using modified measured standard light and color data and corresponding driving codes. When constructing multi-order regression sub-models based on different orders, the model order can be limited according to the visual color difference changes of the validation samples to avoid overfitting of local samples due to excessively high order.

[0046] The lookup interpolation sub-model is used to generate continuous driving codes between discretely acquired samples. Corrected measured standard color data can be used as index nodes in the standard color space, with the corresponding driving codes serving as node values. When target standard color data is input, a set of nodes surrounding the target standard color data is determined in the standard color space, and interpolation weights are generated based on the position of the target standard color data relative to each node in the node set. The interpolation weights satisfy: ; In the formula, Indicates the target standard light color data. This represents the j-th index node in the node set. Let represent the interpolation weight corresponding to the j-th index node, and r represent the number of nodes in the node set. Generating interpolation-driven codes based on the same interpolation weights can be represented as: ; In the formula, This represents the driving code corresponding to the j-th index node. Using this method, the lookup interpolation sub-model can generate the driving codes corresponding to the target standard light and color data across a limited number of acquisition nodes without needing to build a complete table for all driving code combinations.

[0047] The neural estimation sub-model can employ a feedforward structure to learn the nonlinear mapping relationship between the target standard color data and the driving code. The number of input nodes in the neural estimation sub-model can correspond to the number of components in the target standard color data, and the number of output nodes can correspond to the number of primary color channels. The number of hidden layers and hidden nodes can be determined based on the sample size, visual color difference statistics, and model complexity. For target correction regions with strong multi-primary color coupling or those that are difficult to represent with a single regression model, the neural estimation sub-model can be used to improve the nonlinear mapping capability; for target correction regions with a small sample size or requiring lower computational cost, a multi-order regression sub-model or a lookup interpolation sub-model can be used.

[0048] The interaction terms between multiple primary colors can be determined from the mixed primary color response data. If there is a difference between the predicted results of a single primary color response and the measured results of the mixed primary colors, this difference is considered as a manifestation of channel interaction and is expressed through the interaction term in the multi-order regression sub-model, the mixing node in the lookup interpolation sub-model, or the nonlinear weight in the neural estimation sub-model. Therefore, after receiving the target standard light color data, the inverse light color characterization model can output a driving code that more closely approximates the actual light emission result.

[0049] In step S105, the reverse light color characterization model is verified using a verification sample. Based on the visual color difference between the target standard light color data of the verification sample and the measured standard light color data obtained by re-acquisition, the effective area and residual distribution of the reverse light color characterization model are determined.

[0050] When validating the inverse color representation model, the visual color difference statistics of the multi-order regression sub-model, the lookup interpolation sub-model, or the neural estimation sub-model are first obtained using the validation samples. Then, the target sub-model used to generate the correction parameters is determined based on the visual color difference statistics and the model complexity.

[0051] In this embodiment, the verification samples can be samples that did not participate in the establishment of the reverse colorimetric representation model, or they can be independent target points selected from the target standard color space. The verification samples cover the target brightness range, target chromaticity range, target correction region, and multi-primary color combination region, so that the verification results can reflect the actual applicability of the reverse colorimetric representation model under different target standard color data.

[0052] During verification, the target standard light color data of the verification sample is input into the inverse light color characterization model to obtain the verification driving code. Then, the LED module to be calibrated is driven to emit light according to the verification driving code, and the corresponding measured standard light color data is re-acquired by the acquisition device. In this way, the verification process can establish a correspondence between the target standard light color data, the verification driving code, and the re-acquired measured standard light color data, avoiding the judgment of model effectiveness based solely on the model's internal fitting error.

[0053] Visual color difference can be used to characterize the difference between the target standard light color data of a calibration sample and the re-acquired measured standard light color data. When the standard light color data is represented using CIELAB data, the visual color difference of the i-th calibration sample can be expressed as: ; In the formula, This represents the visual color difference of the i-th verification sample. This represents the target CIELAB vector corresponding to the i-th verification sample. This represents the measured CIELAB vector obtained by re-collecting and converting the i-th verification sample. The L2 norm is used to represent color difference. Visual color difference can also be represented by the CIEDE2000 color difference or the minimum perceptible difference index, as long as it can reflect the difference that the target standard light color data and the measured standard light color data are perceptible to the human eye.

[0054] For the same inverse light color representation model, a visual color difference statistic can be formed based on the visual color differences of multiple calibration samples. The visual color difference statistic can be expressed as: ; In the formula, The visual color difference means, where n represents the number of validation samples. Visual color difference statistics can also include the maximum visual color difference, visual color difference quantiles, and visual color difference standard deviation, used to determine the overall and local errors of the model within the target standard color space.

[0055] The effective region can be determined based on visual color difference statistics. If the mean and quantile values ​​of visual color difference for a validation sample within a target standard color range do not exceed a preset threshold, then that target standard color range is determined as the effective region of the inverse color representation model. If a validation sample exhibits persistently large visual color differences in low-brightness regions, color gamut boundary regions, or regions with strong coupling of multiple primary colors, then the corresponding range is not included in the effective region, or is considered as a candidate range for subsequent supplementary data collection and model updates.

[0056] The residual distribution can be determined based on the target standard color data of the calibration sample and the reacquired measured standard color data. The residual vector of the i-th calibration sample can be expressed as: ; In the formula, This represents the residual vector of the i-th validation sample. This represents the target standard color data of the i-th calibration sample. This represents the measured standard light color data obtained by re-collecting and converting the i-th calibration sample. The residual distribution can be statistically analyzed according to the luminance range, chromaticity range, target correction area, or primary color combination type, and is used to characterize the deviation direction and deviation magnitude of the inverse light color characterization model in different ranges.

[0057] When selecting a target sub-model, the multi-order regression sub-model, lookup interpolation sub-model, and neural estimation sub-model can be validated using the same batch of validation samples to obtain the visual color difference statistics for each sub-model. Model complexity can be represented by the number of model parameters, the number of lookup nodes, the size of the hidden layers, or the computational complexity. The evaluation metric used to select the target sub-model can be expressed as: ; In the formula, This represents the evaluation metric for the s-th sub-model. This represents the mean visual color difference corresponding to the s-th sub-model. Let represent the model complexity of the s-th sub-model. This represents the complexity weights. Sub-models with smaller evaluation values ​​can be used as target sub-models. In this way, the target sub-model can maintain low visual color difference while avoiding excessive model complexity that could lead to increased computation or local overfitting.

[0058] In step S106, correction parameters for the LED module to be corrected are generated based on the reverse light color characterization model, the effective area, the residual distribution, and the measurement uncertainty. The correction parameters include conversion parameters from the target standard light color data to the driving code and confidence information corresponding to the conversion parameters.

[0059] In this embodiment, the correction parameters can be generated separately according to the target correction region, or separately according to the luminance range, chromaticity range, or primary color combination type. The correction parameters may include one or more of the following: conversion parameters from target standard light color data to driving code, model parameters of the target sub-model, lookup interpolation node parameters, effective region identifier, residual level, measurement uncertainty level, and allowable visual color difference range. For different target correction regions, if the reverse light color characterization model is different, the corresponding correction parameters are stored separately, so that each target correction region can obtain an independent correction result according to its own luminous response and acquisition error.

[0060] For the p-th target correction region, the correction parameter can be expressed as: ; In the formula, This represents the correction parameters for the p-th target correction region. This represents the inverse light color representation model or target sub-model. Indicates the effective region of the model. Represents the residual distribution. Indicates measurement uncertainty. This indicates confidence level information. Confidence level information is used to identify the availability of the conversion parameters within the corresponding target standard color data, target correction area, and brightness range.

[0061] When inputting target standard light color data, it can first be determined whether the target standard light color data is within the valid region. If it is within the valid region, the corresponding inverse light color representation model is invoked to generate driving codes; if it is not within the valid region, the confidence level can be reduced, the output of the driving codes can be limited, or a prompt can be made to supplement the sample collection. The driving code can be represented as: ; In the formula, This indicates the driving code corresponding to the target standard light color data. This represents the inverse light color representation model corresponding to the p-th target correction region. This represents the target standard light color data. Using this conversion relationship, the target brightness, target chromaticity, or target tristimulus values ​​can be converted into executable drive code for the LED module to be calibrated.

[0062] Confidence information can be determined jointly based on the effective area, residual distribution, and measurement uncertainty. If the target standard color data is located within the effective area, and the residual distribution and measurement uncertainty within the corresponding range are small, then the confidence level is set to a higher level. If the target standard color data is close to the boundary of the effective area, or the measurement uncertainty within the corresponding brightness range is high, then the confidence level is set to a lower level. The permissible visual color difference range can be set corresponding to the confidence information, allowing different judgment criteria to be used based on different confidence levels during subsequent verification.

[0063] After generating the correction parameters, the LED module to be corrected is driven to display the target verification screen according to the correction parameters, and the measured standard light color data of the target verification screen is re-acquired; when the visual color difference between the measured standard light color data and the corresponding target standard light color data exceeds the allowable range corresponding to the confidence information, samples are collected in the corresponding effective area, and the reverse light color characterization model and the correction parameters are updated.

[0064] After generating the calibration parameters, the LED module to be calibrated can be driven to display the target calibration screen according to the calibration parameters. The target calibration screen can include a grayscale gradient screen, a single primary color screen, a multi-primary color combination screen, a color gamut boundary screen, or a target application scene screen. After re-acquiring the measured standard light color data of the target calibration screen, it is compared with the corresponding target standard light color data. If the visual color difference does not exceed the allowable range corresponding to the confidence information, the calibration parameters can be used as valid parameters for the current target calibration area. If the visual color difference exceeds the allowable range, additional samples are collected near the corresponding target standard light color data, within the corresponding target calibration area, or within the corresponding brightness range, and the inverse light color representation model and calibration parameters are updated using the additional collected samples.

[0065] Through closed-loop verification, the correction parameters not only include the conversion results from the target standard color data to the driving code, but also the usable range and reliability corresponding to the conversion results. For low-brightness regions, color gamut boundary regions, and regions with strong coupling of multiple primary colors, additional samples can be collected and the model updated after verification to reduce correction deviations caused by insufficient samples, large model residuals, or high measurement uncertainties in local areas.

[0066] Based on the same line of thought, such as Figure 2 As shown, an LED module calibration parameter generation system is provided, comprising: The response acquisition module 201 is used to acquire the acquisition response data of the LED module to be calibrated in the dark state, the single primary color at different tones, and the multi-primary color combination state in which at least two primary colors are lit at the same time, and associate the acquisition response data with the corresponding driving code, acquisition conditions and module position information; The light color conversion module 202 is used to establish a measurement conversion relationship from the acquisition channel response to the standard light color space based on the response of the reference luminescent sample in each acquisition channel of the acquisition device and the reference standard light color data of the reference luminescent sample, and to convert the acquisition response data using the measurement conversion relationship to obtain the measured standard light color data of the LED module to be calibrated. The data correction module 203 is used to perform dark state subtraction, acquisition condition correction and abnormal sample removal on the measured standard light color data, and generate the measurement uncertainty corresponding to the preset brightness range based on the repeated acquisition results; The reverse modeling module 204 is used to establish a reverse light color representation model based on the driving code and the modified measured standard light color data. The reverse light color representation model is used to represent the mapping relationship between the input target standard light color data and the driving code, and introduces the mutual influence term between multiple primary colors when establishing the reverse light color representation model. The model verification module 205 is used to verify the reverse light color characterization model using verification samples, and to determine the effective area and residual distribution of the reverse light color characterization model based on the visual color difference between the target standard light color data of the verification samples and the measured standard light color data obtained by re-acquisition. The parameter generation module 206 is used to generate correction parameters for the LED module to be corrected based on the reverse light color characterization model, the effective area, the residual distribution and the measurement uncertainty. The correction parameters include conversion parameters from the target standard light color data to the driving code and confidence information corresponding to the conversion parameters.

[0067] This system associates the acquired response data of the LED module to be calibrated with the driver code, acquisition conditions, and module location information, and converts the channel response of the acquisition device to the standard color space. This ensures that subsequent calibration no longer relies solely on the original acquired grayscale or single brightness data, improving data consistency across different acquisition devices and conditions. By correcting the measured standard color data and establishing a reverse color characterization model from the target standard color data to the driver code, the system can consider the mutual influence of primary color tone response and multiple primary colors emitting light together when generating calibration parameters, reducing the mutual interference between brightness consistency calibration and color consistency calibration. By determining the effective area of ​​the model, residual distribution, and measurement uncertainty through verification samples, and outputting the confidence information along with the calibration parameters, the system provides a basis for judgment in writing, verifying, and updating calibration parameters, thereby improving the accuracy, stability, and traceability of LED module calibration parameters.

[0068] It should be noted that although several modules or units for the device used to perform actions have been mentioned in the detailed description above, this division is not mandatory. In fact, according to exemplary embodiments of this disclosure, the features and functions of two or more modules or units described above can be embodied in one module or unit. Conversely, the features and functions of one module or unit described above can be further divided and embodied by multiple modules or units.

[0069] Other embodiments of this disclosure will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not disclosed herein. The specification and embodiments are to be considered exemplary only, and the true scope and spirit of this disclosure are indicated by the claims.

[0070] It should be understood that this disclosure is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this disclosure is limited only by the appended claims.

Claims

1. A method for generating calibration parameters for an LED module, characterized in that, The method includes: Acquire the acquisition response data of the LED module to be calibrated in dark state, single primary color different tone states, and multi-primary color combination state with at least two primary colors lit at the same time, and associate the acquisition response data with the corresponding driving code, acquisition conditions and module position information; Based on the response of the reference luminescent sample in each acquisition channel of the acquisition device and the reference standard color data of the reference luminescent sample, a measurement conversion relationship from the acquisition channel response to the standard color space is established, and the acquisition response data is converted using the measurement conversion relationship to obtain the measured standard color data of the LED module to be calibrated. The measured standard light color data is subjected to dark state subtraction, acquisition condition correction and abnormal sample removal, and measurement uncertainty corresponding to the preset brightness range is generated based on the repeated acquisition results; A reverse colorimetric representation model is established based on the driving code and the modified measured standard colorimetric data. The reverse colorimetric representation model is used to represent the mapping relationship between the input target standard colorimetric data and the driving code, and the mutual influence term between multiple primary colors is introduced when establishing the reverse colorimetric representation model. The reverse light color characterization model is validated using a verification sample. Based on the visual color difference between the target standard light color data of the verification sample and the re-acquired measured standard light color data, the effective area and residual distribution of the reverse light color characterization model are determined. Based on the inverse light color characterization model, the effective region, the residual distribution, and the measurement uncertainty, correction parameters for the LED module to be corrected are generated. The correction parameters include conversion parameters from the target standard light color data to the driving code and confidence information corresponding to the conversion parameters.

2. The method for generating LED module calibration parameters according to claim 1, characterized in that, The module position information is used to characterize the target correction area in the LED module to be corrected. The target correction area includes a display partition composed of multiple light-emitting units. Different target correction areas correspond to different reverse light color characterization models and correction parameters.

3. The method for generating LED module calibration parameters according to claim 1, characterized in that, The acquired response data includes dark state response data, single primary color response data, and mixed primary color response data; the dark state response data is used to determine the black level compensation amount, the single primary color response data is used to determine the tone response of each primary color channel, and the mixed primary color response data is used to determine the mutual influence of channels when multiple primary colors emit light together.

4. The method for generating LED module calibration parameters according to claim 1, characterized in that, When establishing the measurement conversion relationship, the multiple acquisition responses of the reference luminescent sample are weighted and averaged. The reference response of the acquisition channel is determined based on the weighted average result, and the parameters of the measurement conversion relationship are formed based on the deviation between the reference response of the acquisition channel and the reference standard light color data. The parameters of the measurement conversion relationship are used to correct for differences in channel response, acquisition gain, and environmental background of the acquisition equipment.

5. The method for generating LED module calibration parameters according to claim 1, characterized in that, The abnormal sample removal includes removing samples that reach the nonlinear response region of the acquisition device, removing samples that are below the effective signal-to-noise ratio requirement, and removing samples that deviate from the residual distribution of similar samples after conversion by the measurement conversion relationship.

6. The method for generating LED module calibration parameters according to claim 1, characterized in that, When establishing the inverse color representation model, one or more of the following are trained: a multi-order regression sub-model, a lookup interpolation sub-model, and a neural estimation sub-model. The visual color difference statistics of the multi-order regression sub-model, the lookup interpolation sub-model, or the neural estimation sub-model are obtained using the validation samples. The target sub-model for generating the correction parameters is determined based on the visual color difference statistics and the model complexity.

7. The method for generating LED module calibration parameters according to claim 6, characterized in that, The multi-order regression sub-model includes a self-encoding term, a higher-order term, and a primary color interaction term. Before training the multi-order regression sub-model, black level compensation is performed on the measured standard light color data, and the model order is limited according to the visual color difference changes of the verification samples during the training process.

8. The method for generating LED module calibration parameters according to claim 6, characterized in that, The lookup interpolation sub-model uses the corrected measured standard light color data as index nodes and the corresponding driving code as node values. When the target standard light color data is input, the set of nodes surrounding the target standard light color data is determined, and the interpolation driving code is generated according to the relative position of each node in the set of nodes.

9. The method for generating LED module calibration parameters according to claim 1, characterized in that, After generating the correction parameters, the LED module to be corrected is driven to display the target verification screen according to the correction parameters, and the measured standard light color data of the target verification screen is re-acquired; when the visual color difference between the measured standard light color data and the corresponding target standard light color data exceeds the allowable range corresponding to the confidence information, samples are collected in the corresponding effective area, and the reverse light color characterization model and the correction parameters are updated.

10. A system for generating calibration parameters for an LED module, characterized in that, The system includes: The response acquisition module is used to acquire the acquisition response data of the LED module to be calibrated in the dark state, the single primary color at different tones, and the multi-primary color combination state in which at least two primary colors are lit at the same time, and associate the acquisition response data with the corresponding driving code, acquisition conditions and module position information; The light color conversion module is used to establish a measurement conversion relationship from the acquisition channel response to the standard light color space based on the response of the reference luminescent sample in each acquisition channel of the acquisition device and the reference standard light color data of the reference luminescent sample, and to convert the acquisition response data using the measurement conversion relationship to obtain the measured standard light color data of the LED module to be calibrated. The data correction module is used to perform dark state subtraction, acquisition condition correction and abnormal sample removal on the measured standard light color data, and generate the measurement uncertainty corresponding to the preset brightness range based on the repeated acquisition results; The reverse modeling module is used to establish a reverse light color representation model based on the driving code and the modified measured standard light color data. The reverse light color representation model is used to represent the mapping relationship between the input target standard light color data and the driving code, and introduces the mutual influence term between multiple primary colors when establishing the reverse light color representation model. The model validation module is used to validate the reverse light color representation model using validation samples, and to determine the effective area and residual distribution of the reverse light color representation model based on the visual color difference between the target standard light color data of the validation samples and the re-acquired measured standard light color data. The parameter generation module is used to generate correction parameters for the LED module to be corrected based on the reverse light color characterization model, the effective area, the residual distribution, and the measurement uncertainty. The correction parameters include conversion parameters from the target standard light color data to the driving code and confidence information corresponding to the conversion parameters.

Images