Image projector multi-light-source color balance evaluation method and system

Abstract

The invention relates to the technical field of projectors, in particular to a multi-light-source color balance evaluation method and system for an image projector, and the method comprises the steps: obtaining the spectral feature data of a multi-light-source system in the image projector, and carrying out the fusion analysis of the spectral feature data, so as to obtain a light source feature portrait; matching a color balance model corresponding to the light source feature portrait, so as to call a color adjustment template from a preset template library through the color balance model; simulating a color equalization process corresponding to the multi-light source system according to the color adjustment template; the color adjustment template is optimized according to the color equalization process, and a multi-light-source color equalization evaluation result is generated according to the optimized color adjustment template, so that automatic and accurate evaluation of color equalization of the multi-light-source system is realized, the color consistency and stability of a projection picture are remarkably improved, and the quality of the multi-light-source system is improved. The method is suitable for image projection equipment with different optical architectures.

Description

Technical Field

[0001] This invention relates to the field of projector technology, and in particular to a method and system for evaluating the color balance of multiple light sources in an image projector. Background Technology

[0002] As a core device for achieving high-quality visual presentation, image projectors utilize multi-light source systems that integrate light sources with different spectral characteristics (such as LEDs and lasers) to achieve high brightness and wide color gamut output. However, the spectral differences among multiple light sources (such as wavelength distribution, light intensity stability, and light mixing uniformity) can easily lead to problems such as color shift and uneven brightness in the projected image, severely impacting the visual experience. Traditional color balance calibration methods rely on manual experience to independently adjust the parameters of a single light source, lacking a systematic fusion analysis of the spectral characteristics of multiple light sources. This makes it difficult to accurately match the color requirements under complex optical architectures, and also results in low calibration efficiency and poor adaptability.

[0003] The above content is only used to help understand the technical solution of the present invention and does not represent an admission that the above content is prior art. Summary of the Invention

[0004] The main objective of this invention is to provide a method and system for evaluating the color balance of multiple light sources in an image projector. This aims to solve the technical problems of traditional color balance calibration methods that rely on manual experience to independently adjust the parameters of a single light source, lack systematic fusion analysis of the spectral characteristics of multiple light sources, make it difficult to accurately match the color requirements under complex optical architectures, and have low calibration efficiency and poor adaptability.

[0005] To achieve the above objectives, the present invention provides a method for evaluating the color balance of a multi-light source in an image projector, the method comprising: Acquire spectral feature data of a multi-light source system in an image projector, and perform fusion analysis on the spectral feature data to obtain a light source feature profile; Match the color balance model corresponding to the light source feature image, so as to call the color adjustment template from the preset template library through the color balance model; Simulate the color balance process corresponding to the multi-light source system based on the color calibration template; The color calibration template is optimized according to the color balance process, and a multi-light source color balance evaluation result is generated based on the optimized color calibration template.

[0006] Optionally, acquiring the spectral feature data of the multi-light source system in the image projector and performing fusion analysis on the spectral feature data to obtain a light source feature profile includes: Obtain the spectral distribution parameters of the multi-light source system, wherein the spectral distribution parameters include wavelength coverage and light intensity stability parameters; A spectral decomposition algorithm is used to extract principal component features and correlation indices from the spectral distribution parameters, and a spectral feature model and a mixing map are constructed based on the principal component features and correlation indices to obtain the fusion degree parameters. The spectral fusion vector is obtained based on the fusion degree parameter, and the spectral fusion vector is normalized to obtain the light source feature profile. The spectral fusion vector includes uniformity, color deviation, and consistency.

[0007] Optionally, the step of calling a color calibration template from a preset template library through the color balance model includes: Obtain the template spectral attributes corresponding to each template in the template library, and calculate the degree of fit between each template spectral attribute and the light source feature profile by using the spectral matching degree preset in the color balance model; When the fit exceeds a preset threshold, the preset template adjustment rules are invoked to adapt and optimize the current template, and a color adjustment template corresponding to the optimized template is generated.

[0008] Optionally, the step of simulating the color equalization process corresponding to the multi-light source system based on the color calibration template includes: A virtual spectral topology model is constructed based on the light source configuration structure of the color calibration template, and the optical path dependency of the virtual spectral topology model is analyzed by the spectral propagation diagram to obtain the optical path analysis model; A spectral embedding algorithm is used to establish an optical path relationship equation, and the spectral relationship data and brightness mapping data of the optical path analytical model are calculated by the optical path relationship equation and the optical path inference engine. The interspectral relationship data and the brightness mapping data are extracted using a convolutional neural network to obtain the constraint relationship between the interspectral relationship data and the brightness mapping data. The real-time spectral data of the multi-light source system is processed according to the constraint relationship to obtain the spectral evolution data and brightness adjustment data of the multi-light source system during the color equalization process. By fusing the spectral evolution data and the brightness adjustment data using a Gaussian mixture model, the actual optical path link and the actual brightness distribution are obtained. The rationality of the color calibration template is verified based on the actual optical path link, the actual brightness distribution, the interspectral relationship data, and the brightness mapping data.

[0009] Optionally, optimizing the color calibration template according to the color balance process and generating a multi-light source color balance evaluation result based on the optimized color calibration template includes: When the color calibration template is verified to be reasonable, an evaluator based on color difference entropy is used to calculate the spectral evaluation parameters and brightness evaluation parameters corresponding to the color calibration template, and a sequence of templates to be calibrated is generated based on the comparison results of the spectral evaluation parameters and the brightness evaluation parameters with preset thresholds respectively. A gradient boosting regressor is trained based on the template sequence to be adjusted. The template adjustment amount is predicted by the gradient boosting regressor. The light source parameters of the color calibration template are adjusted according to the template adjustment amount to obtain the light source relationship coordinates of the updated color calibration template. The light source relationship coordinates of the updated color calibration template are reconstructed using the minimum spanning tree algorithm, and the adjustment planning path is obtained by dynamically adjusting the plan under the constraints of maximum spectral depth and brightness convergence rate according to the minimum spanning tree algorithm. The centrality value of each light source node is calculated for the adjusted planning path. Compensation nodes are inserted in the centrality mutation region by a clustering algorithm to obtain the optimized color calibration template. The multi-light source color balance evaluation result is generated based on the optimized color calibration template.

[0010] Optionally, before acquiring the spectral feature data of the multi-light source system in the image projector, the method further includes: The environmental parameters of the projector's optical architecture are obtained to construct a light path governance dynamics model corresponding to the optical architecture and to perform an impact analysis on the light path governance dynamics model, thereby obtaining the light path dependence cloud map and brightness distribution cloud map corresponding to the multi-source system under each preset initial calibration template. Each initial calibration template is extracted from the nodes in the optical path dependency cloud map and the brightness gradient in the brightness distribution cloud map. The evolution rate and brightness change rate of each initial calibration template at each node are calculated to obtain the optical path dependency change curves of the multi-source system in the equalization process under different initial calibration templates. Based on the optical path dependence change curve, select multiple first calibration templates from a number of initial calibration templates; The first parameter in each of the first tuning templates is evolved according to the preset template evolution algorithm to obtain several evolved template combinations and save them to the preset template library.

[0011] Optionally, the environmental parameters include system compatibility parameters and optical path transmission strategies. The process of obtaining the environmental parameters of the projector's optical architecture, constructing an optical path management dynamics model corresponding to the optical architecture, and performing an impact analysis on the optical path management dynamics model to obtain the optical path dependency cloud map and brightness distribution cloud map corresponding to the multi-source system under each preset initial calibration template includes: The system compatibility parameters and optical path transmission strategy of the projector optical architecture are obtained to calculate the transmission impedance matrix corresponding to the optical architecture. The system compatibility parameters include the light source modulation depth and compatibility density, and the transmission impedance matrix is ​​determined by the optical path transmission strategy, the light source modulation depth and the compatibility density. The influence values ​​of spectral nodes are obtained through the transmission impedance matrix and mapped to optical path nodes to establish an optical diffusion equation. The optical path nodes are partitioned using a directed graph and the node density in the core optical path region is increased to twice the basic density. The node centrality value is calculated based on the light diffusion equation to obtain the light influence field calculation equation, i.e., the light path governance dynamic model; wherein, the node centrality value is determined by the incident light vector and the outgoing light vector; Based on the light influence field calculation equation, a set of nonlinear spectral evolution equations is established to obtain the optical path dependence cloud map and brightness distribution cloud map corresponding to each preset initial calibration template of the multi-source system; wherein, the set of nonlinear spectral evolution equations is solved by gradient descent until the loss function is less than a preset value.

[0012] Optionally, calculating the evolution rate and brightness change rate of each initial calibration template at each node to obtain the optical path dependence change curve of the multi-source system during the equalization process under different initial calibration templates includes: Obtain the template topology corresponding to each initial calibration template, and obtain the adjacency spectrum set of the nodes on the template topology; wherein, different initial calibration templates correspond to different topology shapes, topology depths and topology connectivity; The brightness distribution value of the node is calculated using an optical path ranking algorithm; wherein, the brightness distribution value is calculated through the mapping relationship between node importance and brightness. A spectral evolution dynamic equation is established based on the brightness distribution value, and the brightness change rate of the nodes is obtained by solving the Runge-Kutta method; wherein the brightness change rate is obtained by calculating the differential of brightness with respect to time. A state-space equation is constructed based on the brightness change rate of the node and the dependency on the optical path dependency cloud map, and the dependency time function is obtained by Kalman filtering. The optical path dependency change curve is obtained based on the dependency time function.

[0013] Optionally, the step of selecting multiple first calibration templates from a plurality of initial calibration templates based on the optical path dependence change curve includes: The optical path dependency change curve is used to determine the current optical path dependency of each initial calibration template. Constraints are constructed based on the preset maximum dependency of the multi-source system and the current optical path dependency of each initial calibration template. The dependency constraint optimization parameters are then calculated using the Lagrange multiplier method. Based on the aforementioned dependency constraint optimization parameters, a multi-objective optimization function is established, comprising a dependency objective function and a brightness objective function. The mapping relationship between the template vector and the constraint response vector corresponding to each initial calibration template is obtained through a support vector machine with a polynomial kernel function. Based on the mapping relationship, a constrained optimization equation system is constructed using the Karouch-Kun-Tucker conditions, and the optimal tuning template and dual variable corresponding to each initial tuning template are obtained by solving the interior point method. If the norm difference between the optimal tuning template and the initial template corresponding to the initial tuning template is less than a preset convergence threshold, the optimal tuning template is stored as the first tuning template.

[0014] Furthermore, to achieve the above objectives, the present invention also provides a multi-source color balance evaluation system for an image projector, the multi-source color balance evaluation system for an image projector comprising: The spectral profiling module is used to acquire spectral feature data of a multi-light source system in an image projector, and to perform fusion analysis on the spectral feature data to obtain a light source feature profile. The template calling module is used to match the color balance model corresponding to the light source feature portrait, so as to call the color adjustment template from the preset template library through the color balance model; The color balance simulation module is used to simulate the color balance process corresponding to the multi-light source system based on the color calibration template. The template optimization module is used to optimize the color calibration template according to the color balance process, and generate multi-light source color balance evaluation results based on the optimized color calibration template.

[0015] This invention provides a method for evaluating color balance in a multi-source image projector. The method acquires and fuses spectral feature data from a multi-source system to construct a light source feature profile including parameters such as uniformity, color cast, and consistency. This provides a systematic and quantitative description of the spectral characteristics of multiple light sources, avoiding the limitations of traditional methods that rely solely on adjusting individual parameters of a single light source. This lays a data foundation for accurate color balance evaluation. Based on the light source feature profile, a color balance model is matched, and calibration templates from a pre-defined template library are invoked. Semantic matching and template optimization rules enable automated adaptation of calibration strategies, allowing for rapid response to spectral differences in different light source combinations. This significantly improves the adaptability of calibration strategies to complex optical architectures and reduces reliance on manual experience. By virtually simulating the color balance process of a multi-source system and combining techniques such as optical path dependency analysis and spectral evolution data fusion, the rationality of the calibration template is verified in advance, and key indicators such as color cast and brightness distribution are quantitatively evaluated, avoiding the trial-and-error costs of actual hardware debugging. Simultaneously, template parameters are dynamically optimized based on simulation results, forming a simulation-evaluation-optimization closed loop. This achieves adaptive iteration of the color balance strategy, significantly improving calibration accuracy and stability. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the image projector structure of the hardware operating environment involved in the embodiments of the present invention; Figure 2 This is a flowchart illustrating an embodiment of the multi-light source color balance evaluation method for image projectors according to the present invention. Figure 3 This is a structural block diagram of an embodiment of the multi-light source color balance evaluation system for image projectors of the present invention.

[0017] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0018] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0019] Reference Figure 1 , Figure 1 This is a schematic diagram of the structure of an image projector in the hardware operating environment involved in the embodiments of the present invention.

[0020] like Figure 1 As shown, the image projector may include: a processor 1001, such as a central processing unit (CPU), a communication bus 1002, a user interface 1003, a network interface 1004, and a memory 1005. The communication bus 1002 is used to enable communication between these components. The user interface 1003 may include a display screen, and optionally, it may also include a standard wired interface or a wireless interface. In this invention, the wired interface of the user interface 1003 may be a USB interface. The network interface 1004 may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface). The memory 1005 may be a high-speed random access memory (RAM) or a non-volatile memory (NVM), such as a disk storage device. The memory 1005 may also optionally be a storage device independent of the aforementioned processor 1001.

[0021] Those skilled in the art will understand that Figure 1 The structure shown does not constitute a limitation on the image projector and may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0022] like Figure 1As shown, the memory 1005, which serves as a computer storage medium, may include an operating system, a network communication module, a user interface module, and a multi-source color balance evaluation program for an image projector.

[0023] exist Figure 1 In the image projector shown, the network interface 1004 is mainly used to connect to the backend server and communicate data with the backend server; the user interface 1003 is mainly used to connect to peripherals; the image projector calls the image projector multi-light source color balance evaluation program stored in the memory 1005 through the processor 1001, and executes the image projector multi-light source color balance evaluation method provided in the embodiment of the present invention.

[0024] Based on the above hardware structure, an embodiment of the multi-light source color balance evaluation method for image projectors of the present invention is proposed.

[0025] Reference Figure 2 , Figure 2 This is a flowchart illustrating an embodiment of the multi-light source color balance evaluation method for an image projector according to the present invention.

[0026] In one embodiment, the multi-light source color balance evaluation method for the image projector includes the following steps: Step S100: Obtain spectral feature data of the multi-light source system in the image projector, and perform fusion analysis on the spectral feature data to obtain a light source feature profile.

[0027] Spectral characteristic data can be a set of parameters such as the radiant intensity distribution of a light source at different wavelengths, peak wavelength, full width at half maximum (FWHM), chromaticity coordinates (e.g., x, y values ​​in the CIE 1931 XYZ chromaticity diagram), and correlated color temperature (CCT). These parameters can be directly measured using a spectral analyzer or spectrophotometer, or obtained by convolving the radiant spectrum of the light source with the spectral sensitivity function of the human eye. For example, this data could include the peak intensity value of an LED light source at a wavelength of 450 nm, or the FWHM parameters of a laser light source. Fusion analysis can be a technique for comprehensively processing multi-source spectral data using statistical methods or machine learning algorithms, including but not limited to principal component analysis, cluster analysis, random forests, or neural networks. For example, this analysis could use a weighted average method to calculate the combined chromaticity coordinates of a multi-source combination, or analyze the correlation of light intensity fluctuations between light sources using the covariance matrix.

[0028] A light source profile can be a digital description of the comprehensive characteristics of a multi-light source system, including quantitative indicators such as uniformity, color cast, and consistency. For example, uniformity can be calculated by the brightness differences in different areas of the projected image, and color cast can be obtained by summing the Euclidean distances between each light source and the target chromaticity coordinates. Technical operations can eliminate measurement noise through data preprocessing (such as denoising and normalization), followed by the following sub-steps: aligning the multi-light source spectral data by wavelength and calculating the weighted composite radiant intensity of each band; identifying the dominant spectral components of different light sources using clustering algorithms; and separating the inherent characteristics of the light source from environmental interference factors using feature extraction techniques such as wavelet transform. For instance, in one specific embodiment, by analyzing the difference in the peak blue light wavelength between an LED light source and a laser light source, the potential color cast risk during light mixing can be identified, thus providing data support for subsequent calibration.

[0029] Step S200: Match the color balance model corresponding to the light source feature image, so as to call the color adjustment template from the preset template library through the color balance model.

[0030] Color equalization models can be mathematical models based on the mapping relationship between light source feature profiles and calibration targets, including Support Vector Machine (SVM) classification models or Bayesian networks. For example, this model can classify the feature profile input into predefined scenarios such as "high brightness priority" and "wide color gamut mode." The preset template library can be a database storing calibration strategies for different light source combinations. Each template contains light source driving parameters (such as current and duty cycle), optical component control parameters (such as the micromirror angle of a DLP chip), and expected effect indicators. For example, a template for the "uneven mixing of three-color LEDs" problem might contain rules for adjusting the driving current ratio of each LED. Semantic matching can be a method of associating feature profiles with the template library through natural language processing or rule engines, including but not limited to calculating similarity using the TFIDF algorithm or triggering specific templates using the rule library. Technical operations can align the input dimensions through parameter standardization (such as converting uniformity indicators to 0-1 range values), followed by similarity calculations (such as cosine similarity) or rule triggering. For example, if the model detects that the dispersion of chromaticity coordinates of multiple light sources exceeds a threshold, the "multi-wavelength calibration" template can be automatically invoked; if the light intensity of a certain light source fluctuates significantly, the "dynamic compensation" template can be invoked. In addition, parameter constraints (such as limiting the adjustment range of LED current to no more than 10%) can be embedded in the template optimization rules.

[0031] Step S300: Simulate the color balance process corresponding to the multi-light source system based on the color calibration template.

[0032] Virtual simulation is a technique that simulates the imaging effect of light emitted from a light source after passing through optical components, based on optical models (such as Zemax and TracePro software). It needs to consider optical path dependence (such as the difference in polarization state of light at different angles) and spectral evolution (such as the change of filter absorptivity over time). Optical path dependence analysis can be a technique to decompose the independent influence of each component in an optical system on the optical path, including separating the effects of spherical aberration and chromatic aberration of lenses on imaging, or quantifying the offset of the beam convergence position by the micromirror deflection angle.

[0033] Spectral evolution data fusion can be a method of combining time-varying spectral data of a light source with a static optical model, such as generating simulation results that include a time dimension using the Monte Carlo method. Technical operation involves inputting calibration template parameters (such as LED current values) into the simulation model, followed by numerical calculations to simulate the light path transmission process. For example, if the template requires adjusting the blue LED drive current to 1.2A, its spectral radiant power needs to be recalculated, and the RGB composition of each pixel needs to be calculated in conjunction with the color wheel rotation speed. During the simulation, key indicators can be verified through iterative calculations: collecting brightness values ​​from the simulated image to calculate the standard deviation of brightness across the entire image, extracting chromaticity coordinates to calculate the ΔE value, or embedding a light source aging model (such as an exponential decay function) to predict long-term color stability.

[0034] Step S400: Optimize the color calibration template according to the color balance process, and generate multi-light source color balance evaluation results based on the optimized color calibration template.

[0035] Dynamic optimization of template parameters can be a technique that uses optimization algorithms (such as genetic algorithms or particle swarm optimization) to adjust template parameters to minimize the error between simulation results and target indicators. For example, color deviation can be used as the objective function to adjust the proportion of the light source driving current using gradient descent. The simulation evaluation optimization closed loop can be an iterative process of re-inputting the optimization parameters into the simulation model until a preset threshold is met. For example, the iteration might terminate when the brightness uniformity is optimized from 8% to 5%. The evaluation results can be output files containing quantitative indicators (such as the final ΔE value and brightness uniformity percentage) and visualization reports (such as a brightness distribution heatmap). For example, an optimized calibration parameter configuration file can be generated for hardware execution. Technical operations can be performed by defining optimization objectives (such as minimizing the weighted sum of color deviation and maximizing brightness), initializing the parameter search space (such as the LED current range of 0.8A-1.5A), and using a genetic algorithm to cross-reference different parameter combinations, selecting individuals with high fitness for iteration. The algorithm terminates when the improvement margin of several consecutive iterations falls below a threshold, and the final output includes numerical indicators and comparison charts (such as chromaticity coordinate distribution before and after optimization).

[0036] This embodiment provides a multi-source color balance evaluation method for image projectors. By acquiring spectral characteristic data of the multi-source system and constructing a light source characteristic profile to quantify system characteristics, it matches a color balance model and calls an appropriate calibration template. Combined with virtual simulation to verify the calibration effect and form a closed-loop optimization, this method systematically solves the technical shortcomings of traditional manual calibration, which relies on experience and lacks data support. This method improves color uniformity and accuracy through a multi-dimensional collaborative optimization strategy, reduces manual intervention and lowers debugging costs. Simultaneously, simulation exposes potential problems in advance, enhancing the long-term robustness of the multi-source system. It is suitable for scenarios with stringent visual quality requirements, such as high-brightness laser projection.

[0037] In one embodiment, spectral feature data of a multi-light source system in an image projector is acquired, and the spectral feature data is fused and analyzed to obtain a light source feature profile, including: Obtain the spectral distribution parameters of the multi-source system, including wavelength coverage and light intensity stability parameters; Wavelength coverage refers to the wavelength range of effective radiation in the output spectrum of a light source. For example, the wavelength coverage of an LED is 450±10nm (blue light), while a laser source may be concentrated at 532nm (green light). This parameter is used to determine the spectral complementarity of different light sources, such as whether the wavelength coverage of red and blue light sources can cover the entire visible light spectrum. Wavelength coverage can be obtained by recording the spectral curve of the light source under static conditions using a spectrometer (such as OceanOptics USB4000) or a photodiode array measurement system, and extracting the wavelength peak value and full width at half maximum (FWHM) of each light source.

[0038] Luminous intensity stability parameters quantify the degree of fluctuation in the output luminous intensity of a light source over time or temperature, and are typically expressed using statistics such as standard deviation, relative standard deviation (RSD), or Allan variance. For example, the luminous intensity stability of a laser source may be better than that of an LED, with an RSD below 0.5%, while an LED may have an RSD of up to 2% due to fluctuations in the driving current. Luminous intensity stability parameters are obtained by continuously collecting luminous intensity data of the light source under fixed conditions (e.g., sampling once per second for 10 minutes) and calculating the ratio of the standard deviation to the mean (RSD) of the time series. For example, if the standard deviation of the luminous intensity fluctuation of a light source over 30 minutes is 0.8%, then its stability parameter is 0.8%.

[0039] A spectral decomposition algorithm is used to extract principal component features and correlation indices from the spectral distribution parameters, and a spectral feature model and a mixing map are constructed based on the principal component features and correlation indices to obtain the fusion degree parameters. Principal component features are key feature vectors extracted from high-dimensional spectral data using dimensionality reduction algorithms (such as Principal Component Analysis, PCA). For example, 300-dimensional spectral data can be compressed into the first three principal components, retaining 95% of the variance information. These principal components characterize the differences in the light source spectra; for example, principal component 1 corresponds to the intensity differences in the blue-green bands, and principal component 2 corresponds to the fluctuations in the red region. Principal component features are obtained by inputting the spectral data matrix of multiple light sources into the PCA algorithm, calculating the covariance matrix, and extracting the first N principal components. Correlation indices quantify the statistical correlation between spectral parameters of multiple light sources, including Pearson correlation coefficients, mutual information, or partial correlation coefficients. For example, if the light intensity fluctuations of two light sources are positively correlated (r=0.8), their correlation index can be marked as high coupling. Correlation indices are obtained by calculating the Pearson correlation coefficient matrix of the light intensity stability parameters between light sources, identifying the similarity of light source fluctuation patterns. Mixing maps are charts generated from numerical simulations or experimental data, showing the mixing results under different light source combination ratios, such as chromaticity coordinates or brightness values. For example, when blue and green light sources are mixed in a 7:3 ratio, the color mapping can predict that their color coordinates fall within a specific region of the CIE 1931 diagram. The color mixing map is calculated through simulation for each combination by parameterizing the light source output ratio (e.g., 0%~100%), and the results are visualized as a chromaticity coordinate distribution map. The fusion degree parameter is obtained by combining principal component similarity and correlation coefficients using a weighted formula. This parameter is calculated by constructing a spectral feature model and the color mixing map, combined with principal component features and correlation indices.

[0040] The spectral fusion vector is obtained based on the fusion degree parameter, and the spectral fusion vector is normalized to obtain the light source feature profile. The spectral fusion vector includes uniformity, color deviation, and consistency.

[0041] The spectral fusion vector consists of three quantitative indicators: uniformity, color deviation, and consistency. Uniformity is calculated using the standard deviation of brightness in different regions of the projected image. Color deviation is calculated as the average Euclidean distance between the chromaticity coordinates of each light source and the target chromaticity coordinates. Consistency is quantified by the area ratio of spectral overlap between light sources. Normalization converts indicators of different dimensions into 0-1 intervals, for example, through linear scaling or Min-Max normalization. The light source feature profile is formed by normalizing the spectral fusion vector, ultimately outputting a multi-dimensional quantitative result.

[0042] This embodiment acquires the spectral distribution parameters of a multi-source system to characterize its fundamental properties. It employs a spectral decomposition algorithm to extract principal component features and correlation indices, constructing a mixing map to quantify the complementarity and fluctuation coupling relationships between light sources. Combined with the fusion degree parameter, a normalized light source feature profile is generated to achieve unified evaluation of multiple indices. This achieves the following technical effects: Principal component analysis (PCA) dimensionality reduction reduces the consumption of computational resources by redundant data while preserving key spectral differences, such as distinguishing wavelength shifts between different LED chips; quantification of correlation indices enables the system to identify fluctuation coupling relationships between light sources, allowing for targeted compensation strategies when the light intensity fluctuations of two light sources are synchronized; the introduction of the mixing map provides an intuitive ability to predict mixing effects, avoiding color gamut loss due to insufficient spectral complementarity; the normalized light source feature profile resolves the issue of inconsistent dimensions among multiple indices, enabling subsequent color balance models to more efficiently match calibration templates, such as prioritizing specific calibration templates to address high color shift risks; and the comprehensive evaluation of the fusion degree parameter allows the algorithm to dynamically determine the overall compatibility of the light source system, triggering an alarm if the fusion degree is below a threshold when mixing laser and LED light. Compared to traditional trial-and-error calibration methods that rely on human experience, this solution significantly reduces the debugging cycle and hardware wear risk through data-driven feature extraction and modeling. In particular, when dealing with dynamic changes such as light source aging and temperature drift, adaptive optimization can be achieved by recalculating feature profiles.

[0043] In one embodiment, a color calibration template is retrieved from a preset template library using a color balance model, including: Obtain the template spectral attributes corresponding to each template in the template library, and calculate the fit between the spectral attributes of each template and the light source feature profile by using the spectral matching degree preset in the color balance model; When the fit exceeds a preset threshold, the preset template adjustment rules are invoked to adapt and optimize the current template, and a color adjustment template corresponding to the optimized template is generated.

[0044] The template spectral attributes can be a set of light source parameters corresponding to each calibration template in a preset template library, including parameters such as target spectral distribution, expected chromaticity coordinates, and light intensity ratio. For example, template spectral attributes can be the peak wavelength range in the target spectral distribution (e.g., 450±5nm), the target x,y values ​​in the CIE1931 chromaticity coordinates, or the RGB channel power allocation ratio (e.g., 1:1.2). These parameters can be obtained by performing spectral analysis on the template using a preset light source testing device and storing them in the template library. Spectral matching degree can be an indicator that quantifies the similarity between the light source feature profile and the template spectral attributes, for example, calculated using spectral cosine similarity, root mean square error, or color difference ΔE value. In a specific embodiment, spectral cosine similarity can be calculated by dividing the dot product of two spectral vectors by the product of their moduli, while the color difference ΔE value is quantified based on the formula in the CIE1976Lab color space. The matching degree threshold can be dynamically set; for example, a cosine similarity ≥0.85 can be defined as a high matching degree standard. The template calibration rules can be a preset parameter adjustment rule library, containing correction logic for template parameters. For example, if a peak wavelength deviation from the target value is detected, a wavelength compensation rule is triggered to achieve a spectral redshift by reducing the LED drive current or adjusting the duty cycle; or, if the uniformity after multi-source mixing is insufficient, a micromirror angle fine-tuning rule is triggered to optimize the optical path distribution. These rules can be based on PID control algorithms or other dynamic control strategies to adjust parameters.

[0045] The technical operation can be achieved through the following steps: First, extract the spectral attribute data of all templates in the template library, and then calculate the spectral matching degree between each template and the current light source feature profile. For example, if the light source feature profile shows a blue light peak of 455nm, while a certain template requires a peak of 450nm, the matching degree score is calculated using the wavelength difference and weighting coefficient. When the matching degree exceeds a threshold (e.g., cosine similarity ≥ 0.9), the system will perform parameter mapping, matching the quantitative indicators (e.g., color cast, uniformity) in the light source feature profile with the conditional parameters in the rule library to determine the range of parameters that need to be adjusted; then, the corresponding rules are triggered, such as adjusting the micromirror deflection angle of the DLP chip or the LED driving current; finally, constraint verification is performed to ensure that the adjusted parameters meet hardware limitations (e.g., maximum LED current) and optical synchronization requirements (e.g., color wheel rotation speed matching the light source frequency). This process ultimately generates an optimized template that integrates the original template framework and the corrected parameters, for example, adjusting the red LED current from 1A to 0.95A and recording the basis for the adjustment.

[0046] This embodiment obtains the spectral attribute parameters of each template in the template library, calculates its spectral matching degree with the current light source feature profile, and calls the rule library to perform parameter adaptation and optimization when the matching degree meets the preset conditions, finally generating a customized color calibration template. This achieves the following technical effects: quantitative matching of spectral attributes shifts template selection from experience-driven to precise numerical comparison, significantly reducing calibration deviations caused by differences in light source characteristics; rule-driven dynamic optimization forms a closed-loop path through clear parameter adjustment logic, improving automation and adjustment interpretability; the threshold screening mechanism reduces invalid calculations of low-relevance templates, improving system operating efficiency; at the same time, this method enhances the responsiveness of color balance strategies to dynamic changes in multi-light source systems (such as temperature drift and component aging), reducing the need for manual intervention.

[0047] In one embodiment, the color equalization process corresponding to a multi-light source system is simulated based on a color calibration template, including: A virtual spectral topology model is constructed based on the light source configuration structure of the color calibration template, and the optical path dependency of the virtual spectral topology model is analyzed by the spectral propagation diagram to obtain the optical path analysis model. The virtual spectral topology model can be an abstract representation of the spectral characteristics and optical component layout of a light source system. It can define the physical location and emission angle of the light source using a three-dimensional coordinate system and establish topological relationships by combining spectral data. Examples include the spatial distribution of the light source and the geometric parameters of the optical components. The spectral propagation diagram can be a mathematical expression of the spectral signal transmission path. It can record parameters such as light energy attenuation and deflection angle in matrix form. Examples include path transfer probability or interference effect descriptions. The optical path analytical model can be a mathematical model describing the independent influence of each component in the optical system on the optical path. Examples include transmission efficiency or spatial distribution functions for each wavelength. By tracking the complete path loss of light of a specific wavelength in the optical system through the spectral propagation diagram—for example, the energy change of 450nm wavelength light from a blue LED after passing through a color wheel filter—quantitative analysis of the independent influence of optical components can be achieved, thereby obtaining the optical path analytical model.

[0048] The optical path relationship equation is established by using a spectral embedding algorithm, and the interspectral relationship data and brightness mapping data of the optical path analytical model are obtained by calculating the optical path relationship equation and the optical path inference engine. Among these methods, spectral embedding algorithms can be mathematical methods that map multidimensional spectral data to a low-dimensional feature space. They can linearize nonlinear relationships through principal component analysis or kernel functions, and, for example, extract spectral semantic features or key wavelength distribution patterns. Optical path relationship equations can be mathematical expressions describing the interaction between light source characteristics and optical components. Optical path inference engines can be rule-based or machine learning-based inference systems that can verify light path deviations through symbolic computation, such as focal offsets or wavelength-related path differences. By reducing the dimensionality of spectral data through spectral embedding algorithms and combining them with numerical solutions to optical path relationship equations—for example, quantizing the focal distance between red and blue light—inter-spectral relationship data and brightness mapping data can be generated, such as brightness attenuation curves at the edge of an image.

[0049] By using a convolutional neural network to extract features from the interspectral relationship data and the brightness mapping data, the constraint relationship between the interspectral relationship data and the brightness mapping data can be obtained. The convolutional neural network (CNN) can be a deep learning model for processing spatially hierarchical data. It can analyze spectral band correlations or brightness distribution gradients using two-dimensional convolutional kernels, exemplified by local feature extraction and fully connected layer parameter output. The constraint relationship can be a mathematical association between inter-spectral relationships and brightness distribution, exemplified by the quantized ratio of wavelength difference to brightness non-uniformity. By extracting focal offset matrix features from inter-spectral relationship data and spatial gradient patterns from brightness mapping data through convolutional layers—for example, learning that brightness non-uniformity increases by 20% when the wavelength difference between green and red light exceeds 50 nm—quantized parameters for the constraint relationship can be established.

[0050] Based on the constraint relationship, the real-time spectral data of the multi-light source system is processed to obtain the spectral evolution data and brightness adjustment data of the multi-light source system during the color equalization process; The real-time spectral data can be the result of dynamic light intensity fluctuation monitoring, which can be acquired in real time through a photodiode array. Examples include LED peak wavelength shifts or intensity changes. Spectral evolution data can be predicted data of the light source spectrum changing over time, examples include wavelength drift curves caused by temperature increases. Brightness adjustment data can be optimization schemes for light source driving parameters, examples include duty cycle adjustments or current compensation values. By combining constraints with real-time spectral data, such as forcing a blue to red light intensity ratio ≥1.2, spectral evolution prediction and brightness distribution adjustment schemes can be generated, such as dynamically compensating for edge brightness decreases caused by green light intensity attenuation.

[0051] By fusing spectral evolution data and brightness adjustment data using a Gaussian mixture model, the actual optical path link and actual brightness distribution are obtained; The Gaussian mixture model can be a probabilistic model that fits complex data distributions by weighted combination of multiple Gaussian distributions, exemplified by fitting the brightness and noise distributions of the red, green, and blue channels. The actual optical path can be a probability density description of light reaching the target region, exemplified by the probability of optical path offset due to optical component aging. The actual brightness distribution can be a description of the standard deviation of pixel brightness, exemplified by spatial uniformity indices. By combining wavelength shift predictions from spectral evolution data with micromirror angle changes from brightness adjustment data using the Gaussian mixture model, such as quantifying the probability of light intensity changes caused by temperature fluctuations, a probability density function for the actual optical path and brightness distribution can be generated.

[0052] The rationality of the color calibration template was verified based on the actual optical path link, actual brightness distribution, interspectral relationship data and brightness mapping data; By comparing the actual optical path link with the preset target, such as verifying whether the focal distance between red and blue light exceeds the tolerance range, the impact of optical path offset can be assessed. By calculating the standard deviation of the actual brightness distribution, such as requiring a uniformity of ≤3% across the entire screen, the brightness uniformity can be verified. By checking the key wavelength color difference in the spectral relationship data, such as whether the ΔE value is ≤2, color consistency can be confirmed. If the verification fails, the system will send the error data back to the optimization step, such as adjusting the arrangement of light sources or the combination of optical component parameters.

[0053] This embodiment achieves accurate optical path modeling by constructing a virtual spectral topology model and spectral propagation map. It combines spectral embedding algorithm and convolutional neural network to automatically discover hidden constraint relationships, uses Gaussian mixture model to fuse dynamic noise data to generate probability distribution, and finally verifies through multi-dimensional verification to ensure that the calibration template meets the requirements of optical path, brightness and color. This can achieve the technical effect of significantly improving prediction accuracy and computational efficiency. In particular, it can reduce trial and error costs by more than 90% in complex multi-light source interaction scenarios, while enhancing robustness to dynamic factors such as temperature fluctuations, making the simulation results closer to the real scene.

[0054] In one embodiment, the color calibration template is optimized according to the color equalization process, and a multi-light source color equalization evaluation result is generated based on the optimized color calibration template, including the following steps: When the color calibration template is verified to be reasonable, an evaluator based on color difference entropy is used to calculate the spectral evaluation parameters and brightness evaluation parameters corresponding to the color calibration template, and a sequence of templates to be calibrated is generated based on the comparison results of the spectral evaluation parameters and brightness evaluation parameters with preset thresholds.

[0055] Among them, color entropy can be an information theory metric that measures the uncertainty of color differences. It can be obtained by calculating the distribution difference between the target color gamut and the actual color gamut in the CIE color space. For example, it includes the distribution entropy value of ΔE value of each pixel quantified by the Shannon entropy formula. Spectral evaluation parameters can be quantitative indicators that characterize the spectral characteristics of the light source. They can be obtained by extracting parameters such as the dominant wavelength deviation, spectral bandwidth consistency, or chromaticity coordinate dispersion of the light source. For example, it includes cases where the dominant wavelength deviation exceeds ±2nm. Brightness evaluation parameters can be indicators that evaluate the quality of brightness distribution. They can be obtained by analyzing parameters such as the standard deviation of brightness across the entire screen, the brightness ratio between regions, or the dynamic range compression rate. For example, it includes scenes where the brightness standard deviation exceeds 5%. The template sequence to be adjusted can be a set of templates sorted by optimization priority, containing the light source parameters that need to be adjusted and their current deviation from the target value. For example, it includes the deviation values ​​of drive current or duty cycle parameters. By calling the color entropy evaluator to compare the simulated chromaticity coordinate distribution with the target color gamut (such as the Rec.2020 standard), calculating the spectral and brightness evaluation parameters, and comparing them with preset thresholds, the templates that need optimization can be selected. In one specific embodiment, if a template has a blue light dominant wavelength deviation of 3nm and a brightness non-uniformity of 6%, it will be preferentially included in the sequence, thus forming a sequence of templates to be calibrated. This operation identifies key parameter deviations through a data-driven approach, providing a quantitative basis for subsequent optimization.

[0056] A gradient boosting regressor is trained based on the template sequence to be calibrated. The template calibration amount is predicted by the gradient boosting regressor. The light source parameters of the color calibration template are adjusted according to the template calibration amount to obtain the light source relationship coordinates of the updated color calibration template.

[0057] The gradient boosting regressor can be an ensemble learning model that minimizes the loss function by stacking weak regressors layer by layer, such as the XGBoost model, which iteratively adjusts the leaf node weights of the tree model using gradient descent. The light source relationship coordinates can be multi-dimensional spatial coordinates describing the correlation between light source parameters, such as parameter space points constructed with LED current and laser duty cycle as axes. By inputting the parameter deviation data from the template sequence to be tuned as the training set into the gradient boosting regressor, the model can predict the parameter adjustment amount, for example, predicting that the LED current needs to be increased from 1.2A to 1.35A. The light source relationship coordinates are generated through coordinate transformation of the parameter combination, such as mapping the current and duty cycle parameters to a high-dimensional space, facilitating subsequent path planning. This operation utilizes a nonlinear model to predict the parameter adjustment path, which improves optimization accuracy and convergence speed compared to traditional empirical methods.

[0058] The light source relationship coordinates of the updated color calibration template are reconstructed using the minimum spanning tree algorithm. Based on the minimum spanning tree algorithm, dynamic adjustment planning is performed under the constraints of maximum spectral depth and brightness convergence rate to obtain the adjustment planning path.

[0059] The minimum spanning tree algorithm can be a weighted graph algorithm used to optimize parameter adjustment paths, exemplified by Prim's algorithm or Kruskal's algorithm. The maximum spectral depth can be the maximum allowable change in spectral characteristics during light source parameter adjustment, exemplified by constraints such as a color temperature drift range ≤200K. The brightness convergence rate can be a parameter measuring the speed at which the brightness uniformity index approaches the target value, exemplified by constraints such as a brightness standard deviation reduction of ≥15% per iteration. By treating the light source relationship coordinates as nodes in a graph, and the cost of parameter adjustment (such as energy consumption or hardware response time) as edge weights, combined with the maximum spectral depth and brightness convergence rate constraints, the adjustment path with the lowest cost and meeting engineering constraints can be selected. In a specific embodiment, if adjusting the red LED current requires simultaneous compensation of the green laser duty cycle, the algorithm will select the edge between the two parameters as the path, ensuring that the spectral depth change does not exceed a threshold. This operation avoids local optima through global path optimization while satisfying hardware constraints.

[0060] To adjust the planned path, the centrality value of each light source node is calculated. A clustering algorithm is used to insert compensation nodes in the centrality mutation region to obtain an optimized color calibration template. Based on the optimized color calibration template, a multi-light source color balance evaluation result is generated.

[0061] The centrality value can be an indicator of the importance of nodes in the graph, such as degree centrality or betweenness centrality. Compensation nodes can be virtual parameter configurations inserted at key inflection points in the adjustment path, such as intermediate parameter combinations used to smooth parameter abrupt changes or eliminate local extrema. By calculating the centrality values ​​of each light source parameter node in the adjustment planning path, key nodes with betweenness centrality > 0.8 are identified, and clusters are formed in centrality abrupt change regions (such as where the parameter adjustment direction suddenly changes), with compensation nodes inserted at the cluster boundaries. In one specific embodiment, if the red LED current adjustment path causes a sudden drop in brightness, a compensation node is inserted to reduce the current increase and simultaneously increase the green laser duty cycle to maintain balance. This operation enhances the control capability of key parameters, achieving a smooth transition in regions with significant multi-light source coupling effects.

[0062] This embodiment achieves technical effects such as improving template optimization efficiency by over 40%, reducing hardware damage risk, and providing dynamic path reliability prediction through parameter deviation screening based on color difference entropy, nonlinear prediction using gradient boosting regressors, constrained path planning using the minimum spanning tree algorithm, and compensation node insertion driven by centrality analysis. This solution, through multi-stage collaborative optimization, shifts parameter adjustment from experience-driven to data-driven, while also considering engineering constraints in the spectral and brightness dimensions. Ultimately, it generates a comprehensive result including static indicators and dynamic path evaluation, providing comprehensive guidance for hardware deployment.

[0063] In one embodiment, before acquiring the spectral feature data of the multi-light source system in the image projector, the method further includes: The environmental parameters of the projector's optical architecture are obtained to construct a light path governance dynamics model corresponding to the optical architecture and to perform an impact analysis on the light path governance dynamics model. The light path dependency cloud map and brightness distribution cloud map corresponding to each preset initial calibration template of the multi-source system are obtained. The environmental parameters of the optical architecture can be a collection of the physical characteristics of the projector's internal optical components, environmental conditions, and external interference factors. These can be obtained through sensor data collection or extraction from design documents. Examples include lens focal length, DMD chip pixel pitch, temperature, humidity, vibration frequency, and screen material absorptivity. The optical path governance dynamics model can be a dynamic transmission simulation system constructed by combining ray tracing algorithms and physical dynamic equations. Examples include finite element analysis to simulate lens deformation or Monte Carlo methods to simulate light scattering. The optical path dependency cloud map can be a visualization of the sensitivity of key nodes in the form of a three-dimensional heatmap. Examples include the sensitivity distribution of nodes such as lens focal point and micromirror deflection angle. The brightness distribution cloud map can be a two-dimensional matrix generated through numerical integration, containing the correspondence between spatial coordinates and brightness values, and indicating the direction of the brightness gradient.

[0064] The technical operation can be achieved as follows: First, optical architecture parameters, such as the micromirror deflection angle range of the DLP chip or the temperature profile of the heat dissipation system, are obtained through sensors or design documents. Then, the light path equation is coupled with thermodynamic equations, for example, simulating the effect of temperature changes on the lens focal length. Next, sensitivity analysis is used to determine the degree of influence of key parameters. Finally, a full optical path simulation is run on each initial calibration template, recording the path changes and brightness distribution data of key nodes, thereby generating the corresponding contour map. This operation can achieve the technical effect of establishing a model that highly matches the actual hardware and quantifying the impact of environmental parameters on image quality.

[0065] The nodes in the optical path dependency cloud map and the brightness gradient in the brightness distribution cloud map of each initial calibration template are extracted respectively. The evolution rate and brightness change rate of each initial calibration template at each node are calculated to obtain the optical path dependency change curves of the multi-source system in the equalization process under different initial calibration templates. In this context, a node can be a location in the optical path with a specific function, such as a lens surface or a pixel array region of a DMD chip, whose parameter changes directly affect the direction of light transmission. The brightness gradient can be the brightness difference between adjacent pixels in a brightness distribution cloud map, typically obtained through numerical differentiation. The evolution rate can be the derivative of node parameters with time or adjustment parameters, such as the rate of change in lens curvature due to temperature increases. The brightness change rate can be the derivative of the brightness value with adjustments to the adjustment parameters, such as the instantaneous change in brightness caused by changes in LED current.

[0066] The technical operation can be achieved as follows: First, key nodes, such as the center point of the micromirror array, are extracted from the optical path dependency cloud map; then, the evolution trend of node path offset over time is calculated, such as the exponentially increasing derivative; next, the brightness gradient field is calculated through numerical differentiation, such as the absolute value of the brightness difference between adjacent pixels; finally, the evolution rate is correlated with the brightness change rate to generate a curve, for example, the impact of LED current adjustment on the center brightness is calculated through parameter differentiation. This operation can achieve the technical effect of quantifying the dynamic characteristics of different initial calibration templates during the equalization process.

[0067] Based on the optical path dependency change curve, select multiple first calibration templates from several initial calibration templates; The technical operation can be implemented as follows: First, a stability threshold is set; for example, templates with a node evolution rate exceeding 0.1 mm / ℃ will be excluded. Then, the brightness change rate is evaluated to determine if it meets efficiency requirements; for example, the center brightness change rate must be ≥3% / A. Finally, Pareto front analysis is used to perform multi-objective optimization between the evolution rate and the brightness change rate, retaining the template corresponding to the Pareto optimal solution. This operation can achieve the technical effect of filtering unstable or inefficient templates while retaining potential candidate solutions.

[0068] The first parameter in each first tuning template is evolved according to the preset template evolution algorithm to obtain several evolved template combinations and save them to the preset template library.

[0069] The template evolution algorithm can be a swarm intelligence algorithm such as a genetic algorithm or differential evolution, where the template parameters are treated as gene sequences for crossover and mutation operations. The fitness function can be a set of metrics that quantify template performance, such as equilibrium completion time, ΔE value, or brightness standard deviation.

[0070] The technical operation can be implemented as follows: First, initialize the parameter search space, for example, setting the LED current range to 0.8A to 1.5A; then, generate initial parameter combinations through population initialization; next, evaluate the fitness value of each combination, for example, simulating its brightness uniformity; then, iteratively optimize the parameters through selection, crossover, and mutation operations, for example, combining the red LED current parameters of template A with the micromirror angle parameters of template B, and perturbing some parameters by ±5%; finally, when the maximum number of generations is reached or the fitness converges, the optimized template combination is stored in the template library. This operation can achieve the technical effect of generating highly adaptive and performance-optimized template combinations.

[0071] This embodiment constructs a dynamic model and generates a cloud map by acquiring environmental parameters, extracts nodes and gradients to calculate dynamic parameters, selects stable and efficient initial templates, and finally optimizes parameter combinations through evolutionary algorithms to establish a template library. This achieves the technical effects of shortening the hardware design to color balance deployment cycle, improving the applicability of the template library, and reducing trial-and-error costs. Through in-depth analysis and closed-loop design in the preprocessing stage, this method makes the template design closer to actual hardware characteristics, considers the impact of dynamic environmental changes in advance, and generates optimized templates with anti-interference capabilities through multi-objective optimization and swarm intelligence algorithms, thereby quickly providing reliable solutions in scenarios such as complex optical architectures or temperature fluctuations.

[0072] In one embodiment, environmental parameters include system compatibility parameters and optical path transmission strategies. The environmental parameters of the projector's optical architecture are obtained, a light path governance dynamics model corresponding to the optical architecture is constructed, and an impact analysis is performed on the light path governance dynamics model. This yields the light path dependency cloud map and brightness distribution cloud map corresponding to the multi-source system under each preset initial calibration template, including: The system compatibility parameters and optical path transmission strategy of the projector's optical architecture are obtained in order to calculate the transmission impedance matrix corresponding to the optical architecture. The system compatibility parameters include the light source modulation depth and compatibility density, and the transmission impedance matrix is ​​determined by the optical path transmission strategy, the light source modulation depth and the compatibility density. The influence values ​​of spectral nodes are obtained by using the transmission impedance matrix and mapped to the optical path nodes to establish the optical diffusion equation. The optical path nodes are divided using a directed graph and the node density in the core optical path region is increased to twice the basic density. The node centrality value is calculated based on the light diffusion equation to obtain the light influence field calculation equation, i.e., the light path governance dynamic model; where the node centrality value is determined by the incident light vector and the outgoing light vector. A set of nonlinear spectral evolution equations is established based on the light influence field calculation equation to obtain the optical path dependence cloud map and brightness distribution cloud map corresponding to each preset initial calibration template of the multi-source system; among them, the nonlinear spectral evolution equations are solved by gradient descent until the loss function is less than the preset value.

[0073] System compatibility parameters can be a set of parameters that quantify the matching degree between the light source and the optical system, including the light source modulation depth (such as the adjustable ratio of LED current) and compatibility density (such as the proportion of spectral overlap area of ​​multiple light sources), which can be obtained through sensor acquisition or extraction from design documents. For example, the light source modulation depth can be the dynamic adjustment range of the maximum output intensity of the LED, and the compatibility density can be the proportion of spectral overlap area of ​​different light sources in a specific wavelength band. The optical path transmission strategy can be the planning rules for light paths in the optical system, including the priority of reflection and refraction paths or the synchronization rules between the color wheel and the light source drive, which can be determined through system design documents or experimental measurements. The transmission impedance matrix can be a numerical matrix describing the obstruction effect of optical path nodes, whose element values ​​are jointly determined by the light source modulation depth, compatibility density, and optical path transmission strategy, for example, by discretizing the optical path nodes and calculating the dynamic response capability and mixing efficiency loss of each node. The technical operation can be implemented as follows: First, the optical path is divided into discrete nodes, each corresponding to an optical element or spatial location; second, the dynamic response capability of the nodes is calculated based on the modulation depth of the light source, with lower modulation depths resulting in higher impedance values; then, the multi-source mixing efficiency loss at the nodes is evaluated in conjunction with compatibility density, with smaller spectral overlap regions leading to higher impedance values; finally, the matrix weights are adjusted according to the optical path transmission strategy, for example, reducing the impedance value of nodes along preferred paths. This process quantifies the comprehensive obstruction effect of each node in the optical path through matrix operations; for example, a node may have an impedance value of 0.8 due to poor light source compatibility.

[0074] The optical diffusion equation can be a mathematical expression describing the propagation of light intensity between optical path nodes, for example, in the form of a partial differential equation, where the diffusion coefficient is determined by the inverse of the transmission impedance matrix. Optical path nodes can be topologically partitioned using a directed graph structure. Local simulation accuracy can be improved by doubling the node density in the core region, for example, increasing the number of nodes per millimeter in the core region from 5 to 10. The spectral node influence value can be a numerical value quantifying the degree of influence of a node on light transmission, achieved by mapping transmission impedance matrix elements to corresponding nodes. Technical operations can be implemented as follows: constructing the optical path nodes as a directed graph, with edge weights determined by transmission impedance matrix elements; implementing a node density doubling strategy in the core optical path region; calculating node centrality values. Finally, by integrating the diffusion equation and node centrality values, a light influence field calculation equation is formed, for example, by combining the impedance matrix and the diffusion equation to construct a dynamic model. The node centrality value can be a graph-based index used to quantify the importance of a node in the optical network, determined by the ratio of the incident light vector to the outgoing light vector. The optical influence field calculation equation can be a dynamic model integrating the transmission impedance matrix, diffusion equation, and node centrality values, used to predict the dynamic response of the optical path under different calibration templates. The technical operation can be achieved by combining the optical diffusion equation with node centrality calculation to form a system of partial differential equations containing spatial distribution and node weights; solving the equation system using numerical simulation methods, such as finite element analysis or finite difference method; and finally, the generated optical influence field calculation equation can characterize the light intensity distribution and transmission efficiency of each node in the optical path.

[0075] Nonlinear spectral evolution equations can be a set of coupled equations that include changes in light intensity and wavelength distribution over time, such as nonlinear terms describing LED spectral drift or temperature-dependent refractive index changes. Gradient descent can be a numerical method that iteratively optimizes and adjusts the equation parameters to bring the loss function (such as mean square error) between the predicted and target spectra to below a preset threshold. This can be achieved by: combining the light influence field model with the light source spectral data to establish a multivariate equation set including light intensity, wavelength, and time dimensions; introducing nonlinear terms such as light source aging models or temperature-sensitive parameters; defining the loss function and initializing the parameters; and calculating the gradient and adjusting the parameter step size, for example, using an adaptive learning rate strategy, until the loss function is less than 0.01.

[0076] This embodiment obtains system compatibility parameters and optical path transmission strategies, calculates the transmission impedance matrix, divides the optical path nodes into a directed graph structure, establishes an optical diffusion equation, constructs an optical influence field calculation equation by combining node centrality values, and finally generates a cloud map by solving a set of nonlinear spectral evolution equations and gradient descent method. The following technical effects can be achieved: quantifying the dynamic matching relationship between the light source and the optical architecture, identifying key control nodes of the optical path, accurately simulating the influence of complex factors such as light source aging and temperature fluctuations on the optical path, and improving the local simulation resolution through high-density node division, thereby providing a data-driven color equalization strategy for multi-light source systems, and significantly improving the imaging stability and color gamut expansion efficiency in dynamic environments.

[0077] In one embodiment, the evolution rate and brightness change rate of each initial calibration template at each node are calculated to obtain the optical path dependence change curves of the multi-source system during the equalization process under different initial calibration templates, including: Obtain the template topology corresponding to each initial calibration template, and obtain the adjacency spectrum set of the nodes on the template topology; wherein, different initial calibration templates correspond to different topology shapes, topology depths and topology connectivity; The template topology can be the connection relationships and hierarchical structure between light sources, optical components, and control parameters in the calibration template. It can be obtained by parsing the parameter configuration file or system architecture design document of the initial calibration template. For example, the template topology can include star, mesh, or ring topologies, with topological depth reflecting differences in the number of levels and topological connectivity reflecting differences in the connection density between nodes. The adjacent spectrum set can be the set of spectral characteristics of other nodes directly connected to a given node. It can be obtained by querying a spectral database or by measuring spectral data. For example, if a node is a red LED light source, its adjacent spectrum set may include the transmission spectrum curve of a red filter or the reflectance parameter of the red light channel of a DMD.

[0078] The brightness distribution value of the nodes is calculated using an optical path ranking algorithm; the brightness distribution value is obtained by calculating the mapping relationship between node importance and brightness. Optical path ranking algorithms can be graph-based methods for evaluating node importance, such as variations of the PageRank algorithm, used to quantify the weight of nodes in optical path transmission. The brightness distribution value can be a quantified result of a node's contribution to the final brightness, and its mapping relationship can be established through linear or nonlinear functions. For example, this process can be implemented through the following steps: constructing an optical path topology graph containing nodes such as light sources, lenses, and micromirrors; initializing node importance scores and iteratively calculating them, where the weight depends on the light energy transmission efficiency of the optical path (such as lens focal length matching); and finally mapping the importance scores to specific brightness values ​​through function transformation. For example, if a micromirror node obtains a high importance score due to influencing multiple light paths, its corresponding brightness distribution value may be significantly higher than that of a regular lens node.

[0079] The spectral evolution dynamics equation was established based on the brightness distribution values, and the brightness change rate of the nodes was obtained by solving the Runge-Kutta method; the brightness change rate was obtained by calculating the differential of brightness with respect to time. The spectral evolution dynamics equation can be a differential equation describing the change of spectral parameters with time or calibration parameters, including state variables such as light source intensity, temperature, and micromirror angle. The brightness change rate can be the derivative of brightness with time, and its calculation is achieved through numerical solution methods. For example, this process may include: defining state variables; establishing a system of equations, such as an exponential equation describing the decay of LED light intensity; and iteratively calculating discrete time points using the fourth-order Runge-Kutta method, for example, calculating the brightness change rate every 0.1 seconds. For instance, this method can quantify the rate at which LED brightness decreases by 0.3% per minute at high temperatures.

[0080] The state-space equation is constructed based on the brightness change rate of the node and the dependency on the optical path dependency cloud map, and the dependency time function is obtained by Kalman filtering. The optical path dependency change curve is obtained based on the dependency time function.

[0081] The state-space equation can be a vector equation representing the system state, containing a system matrix and an input matrix describing the interactions between states. The dependencies on the optical path dependency contour map can be a set of parameters characterizing parameters such as light deflection sensitivity. For example, this process can be achieved through the following steps: combining the brightness change rate with the dependencies into a state vector; defining the system matrix and input matrix to describe physical relationships such as the effect of temperature changes on the thermal expansion of the micromirror; applying Kalman filtering to combine simulation predictions with sensor measurement data, for example, correcting the deviation between the predicted and actual observed brightness change rate; and finally generating an optical path dependency curve through the time function relationship of the state variables, for example, showing that the optical path offset increases exponentially with time as the temperature increases.

[0082] This embodiment analyzes the topology of the initial calibration template and extracts the adjacent spectrum set, quantifies the importance weight of nodes in optical path transmission, and predicts the brightness change rate by combining dynamic equations and numerical algorithms. Then, by fusing dynamic parameters and measured data through state-space modeling and filtering techniques, it can significantly improve the accuracy and adaptability of the optical path-dependent change curve. This method can identify the propagation path of parameter adjustments, such as the indirect impact of red LED current on the brightness of the green channel, prioritize the optimization of nodes that significantly affect the global brightness, and suppress model errors through numerical stability and real-time data fusion, thus providing a reliable basis for dynamic response optimization and long-term stability assessment of multi-source systems.

[0083] In one embodiment, multiple first calibration templates are selected from several initial calibration templates based on the optical path dependency change curve. This includes: determining the current optical path dependency of each initial calibration template through the optical path dependency change curve; constructing constraint conditions based on the preset maximum dependency of the multi-source system and the current optical path dependency of each initial calibration template; calculating the dependency constraint optimization parameters using the Lagrange multiplier method; establishing a multi-objective optimization function based on the dependency constraint optimization parameters, including a dependency objective function and a brightness objective function; obtaining the mapping relationship between the template vector and the constraint response vector corresponding to each initial calibration template using a support vector machine with a polynomial kernel function; constructing a constraint optimization equation system using the Karouch-Kuntuak condition based on the mapping relationship; and solving it using the interior-point method to obtain the optimal calibration template and dual variables corresponding to each initial calibration template; if the norm difference between the optimal calibration template and the initial template corresponding to the initial calibration template is less than a preset convergence threshold, then storing the optimal calibration template as the first calibration template.

[0084] The current optical path dependency can be a quantitative indicator characterizing the sensitivity of an optical system to specific environmental parameters, such as the rate of focal length change caused by temperature or component deformation, obtained through analysis of the optical path dependency change curve. The preset maximum dependency can be a threshold defined by hardware design or user requirements, such as an upper limit for temperature sensitivity, constructed by comparing the current dependency with this threshold to establish inequality constraints. The Lagrange multiplier method can be a mathematical method to transform constraints into penalty terms in an optimization problem, for example, modifying the objective function to a weighted sum of the original objective function and constraint terms, where the weight coefficients are determined by Lagrange multipliers. The dependency objective function can be a mathematical expression used to quantify template robustness, such as a function that minimizes the current optical path dependency. The luminance objective function can be a mathematical expression optimizing luminance uniformity or chromaticity accuracy, such as a function that minimizes the luminance standard deviation. A support vector machine with a polynomial kernel function can be a nonlinear mapping tool, for example, using a quadratic polynomial kernel function to correlate a high-dimensional parameter space with the constraint response to capture the complex relationship between luminance and dependency. The Karouch-Kuntucker condition can be a necessary condition for a constrained optimization problem, such as requiring that the linear combination of the objective function gradient and the constraint gradient be zero. Interior-point methods can be iterative optimization algorithms, such as combining Newton's method and the center path algorithm to gradually approach the optimal solution while ensuring the constraints are satisfied. Norm difference can be a metric for measuring the difference in parameters before and after optimization, such as Euclidean distance or Mahalanobis distance, which is calculated by comparing the parameter vectors of the optimal calibration template and the initial template. A preset convergence threshold can be a condition for determining the termination of the optimization process, such as stopping iteration when the norm difference is below 0.05.

[0085] The technical operation is implemented as follows: First, based on the optical path dependency change curve analysis, the current optical path dependency of each initial template is obtained. Constraints are constructed by combining the preset maximum dependency, and the constraint terms are integrated into the objective function using the Lagrange multiplier method to form dependency constraint optimization parameters. Second, a nonlinear mapping relationship is constructed using a support vector machine with a polynomial kernel function to associate the template parameters with the constraint response (such as dependency and brightness change rate), thereby establishing a multi-objective optimization function containing both the dependency objective function and the brightness objective function. Next, a system of constraint optimization equations is constructed based on the Karouch-Kuntuak conditions, and the optimal solution is solved iteratively using the interior-point method. This method avoids constraint boundary violations through the center path algorithm and efficiently calculates the search direction using Newton's method. Finally, the convergence of the optimization results is verified by norm difference calculation. When the difference is lower than a preset threshold, the optimized template is stored as the first calibration template. This process ensures that the template meets physical constraints through constraints, balances robustness and performance indicators in multi-objective optimization, reduces computational resource consumption through efficient solution algorithms, and avoids invalid iterations through a convergence judgment mechanism.

[0086] This embodiment obtains the sensitivity index of each initial template based on the optical path dependence change curve analysis, constructs a mathematical optimization problem based on preset constraints, uses support vector machines to handle nonlinear relationships, efficiently solves the constraint optimization equations using the interior-point method, and determines convergence based on the norm difference value. This achieves the technical effect of improving the accuracy and robustness of template selection. Specifically, the constraint-driven optimization process ensures that the selected templates maximize brightness performance while meeting hardware physical limitations. Nonlinear mapping technology effectively handles the complex trade-offs between multiple objectives, iterative solution algorithms reduce computational complexity, and the convergence judgment mechanism avoids over-optimization. The final template set can maintain optical path stability and brightness distribution uniformity in dynamic environments, thereby reducing actual debugging requirements and improving the long-term reliability of multi-source systems.

[0087] Furthermore, this embodiment of the invention also proposes a storage medium storing a multi-source color balance evaluation program for an image projector. When the multi-source color balance evaluation program for an image projector is executed by a processor, it implements the steps of the multi-source color balance evaluation method for an image projector as described above.

[0088] In addition, refer to Figure 3 This invention also proposes a multi-light source color balance evaluation system for image projectors, the multi-light source color balance evaluation system for image projectors comprising: The spectral imaging module 10 is used to acquire spectral feature data of the multi-light source system in the image projector and perform fusion analysis on the spectral feature data to obtain a light source feature portrait. Template calling module 20 is used to match the color balance model corresponding to the light source feature image, so as to call the color adjustment template from the preset template library through the color balance model; The color balance simulation module 30 is used to simulate the color balance process corresponding to the multi-light source system according to the color calibration template. The template optimization module 40 is used to optimize the color calibration template according to the color balance process, and generate a multi-light source color balance evaluation result based on the optimized color calibration template.

[0089] Other embodiments or specific implementations of the multi-light source color balance evaluation system for image projectors described in this invention can be referred to the above-described method embodiments, and will not be repeated here.

[0090] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0091] The sequence numbers of the above embodiments of the present invention are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments. In the module claims listing several systems, several of these systems may be specifically embodied by the same hardware item. The use of the terms first, second, and third, etc., does not indicate any order and can be interpreted as names.

[0092] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as a read-only memory image (ROM) / random access memory (RAM), magnetic disk, optical disk), and includes several instructions to cause a terminal user device (which may be a mobile phone, computer, server, air conditioner, or network user device, etc.) to execute the methods described in the various embodiments of the present invention.

[0093] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. A method for evaluating color balance in a multi-light source image projector, characterized in that, The method includes: Acquire spectral feature data of a multi-light source system in an image projector, and perform fusion analysis on the spectral feature data to obtain a light source feature profile; Match the color balance model corresponding to the light source feature image, so as to call the color adjustment template from the preset template library through the color balance model; Simulate the color balance process corresponding to the multi-light source system based on the color calibration template; The color calibration template is optimized according to the color balance process, and a multi-light source color balance evaluation result is generated based on the optimized color calibration template.

2. The multi-light source color balance evaluation method for image projectors as described in claim 1, characterized in that, The process of acquiring spectral feature data of a multi-light source system in an image projector and performing fusion analysis on the spectral feature data to obtain a light source feature profile includes: Obtain the spectral distribution parameters of the multi-light source system, wherein the spectral distribution parameters include wavelength coverage and light intensity stability parameters; A spectral decomposition algorithm is used to extract principal component features and correlation indices from the spectral distribution parameters, and a spectral feature model and a mixing map are constructed based on the principal component features and correlation indices to obtain the fusion degree parameters; The spectral fusion vector is obtained based on the fusion degree parameter, and the spectral fusion vector is normalized to obtain the light source feature profile. The spectral fusion vector includes uniformity, color deviation, and consistency.

3. The multi-light source color balance evaluation method for image projectors as described in claim 1, characterized in that, The step of calling a color calibration template from a preset template library using the color balance model includes: Obtain the template spectral attributes corresponding to each template in the template library, and calculate the degree of fit between each template spectral attribute and the light source feature profile by using the spectral matching degree preset in the color balance model; When the fit exceeds a preset threshold, the preset template adjustment rules are invoked to adapt and optimize the current template, and a color adjustment template corresponding to the optimized template is generated.

4. The multi-light source color balance evaluation method for image projectors as described in claim 1, characterized in that, The process of simulating the color balance of the multi-light source system based on the color calibration template includes: A virtual spectral topology model is constructed based on the light source configuration structure of the color calibration template, and the optical path dependency of the virtual spectral topology model is analyzed by the spectral propagation diagram to obtain the optical path analysis model; A spectral embedding algorithm is used to establish an optical path relationship equation, and the spectral relationship data and brightness mapping data of the optical path analytical model are calculated by the optical path relationship equation and the optical path inference engine. The interspectral relationship data and the brightness mapping data are extracted using a convolutional neural network to obtain the constraint relationship between the interspectral relationship data and the brightness mapping data. The real-time spectral data of the multi-light source system is processed according to the constraint relationship to obtain the spectral evolution data and brightness adjustment data of the multi-light source system during the color equalization process. By fusing the spectral evolution data and the brightness adjustment data using a Gaussian mixture model, the actual optical path link and the actual brightness distribution are obtained. The rationality of the color calibration template is verified based on the actual optical path link, the actual brightness distribution, the interspectral relationship data, and the brightness mapping data.

5. The multi-light source color balance evaluation method for image projectors as described in claim 4, characterized in that, The step of optimizing the color calibration template according to the color balance process and generating a multi-light source color balance evaluation result based on the optimized color calibration template includes: When the color calibration template is verified to be reasonable, an evaluator based on color difference entropy is used to calculate the spectral evaluation parameters and brightness evaluation parameters corresponding to the color calibration template, and a sequence of templates to be calibrated is generated based on the comparison results of the spectral evaluation parameters and the brightness evaluation parameters with preset thresholds respectively. A gradient boosting regressor is trained based on the template sequence to be adjusted. The template adjustment amount is predicted by the gradient boosting regressor. The light source parameters of the color calibration template are adjusted according to the template adjustment amount to obtain the light source relationship coordinates of the updated color calibration template. The light source relationship coordinates of the updated color calibration template are reconstructed using the minimum spanning tree algorithm, and the adjustment planning path is obtained by dynamically adjusting the plan under the constraints of maximum spectral depth and brightness convergence rate according to the minimum spanning tree algorithm. The centrality value of each light source node is calculated for the adjusted planning path. Compensation nodes are inserted in the centrality mutation region by a clustering algorithm to obtain the optimized color calibration template. The multi-light source color balance evaluation result is generated based on the optimized color calibration template.

6. The multi-light source color balance evaluation method for image projectors as described in claim 1, characterized in that, Before acquiring the spectral feature data of the multi-light source system in the image projector, the method further includes: The environmental parameters of the projector's optical architecture are obtained to construct a light path governance dynamics model corresponding to the optical architecture and to perform an impact analysis on the light path governance dynamics model, thereby obtaining the light path dependence cloud map and brightness distribution cloud map corresponding to the multi-source system under each preset initial calibration template. Each initial calibration template is extracted from the nodes in the optical path dependency cloud map and the brightness gradient in the brightness distribution cloud map. The evolution rate and brightness change rate of each initial calibration template at each node are calculated to obtain the optical path dependency change curves of the multi-source system in the equalization process under different initial calibration templates. Based on the optical path dependence change curve, select multiple first calibration templates from a number of initial calibration templates; The first parameter in each of the first tuning templates is evolved according to the preset template evolution algorithm to obtain several evolved template combinations and save them to the preset template library.

7. The multi-light source color balance evaluation method for image projectors as described in claim 6, characterized in that, The environmental parameters include system compatibility parameters and optical path transmission strategies. The process involves acquiring the environmental parameters of the projector's optical architecture, constructing an optical path governance dynamics model corresponding to the optical architecture, and performing an impact analysis on the optical path governance dynamics model to obtain the optical path dependency cloud map and brightness distribution cloud map corresponding to the multi-source system under each preset initial calibration template. This includes: The system compatibility parameters and optical path transmission strategy of the projector optical architecture are obtained to calculate the transmission impedance matrix corresponding to the optical architecture. The system compatibility parameters include the light source modulation depth and compatibility density, and the transmission impedance matrix is ​​determined by the optical path transmission strategy, the light source modulation depth and the compatibility density. The influence values ​​of spectral nodes are obtained through the transmission impedance matrix and mapped to optical path nodes to establish an optical diffusion equation. The optical path nodes are partitioned using a directed graph and the node density in the core optical path region is increased to twice the basic density. The node centrality value is calculated based on the light diffusion equation to obtain the light influence field calculation equation, i.e., the light path governance dynamic model; wherein, the node centrality value is determined by the incident light vector and the outgoing light vector; Based on the light influence field calculation equation, a set of nonlinear spectral evolution equations is established to obtain the optical path dependence cloud map and brightness distribution cloud map corresponding to each preset initial calibration template of the multi-source system; wherein, the set of nonlinear spectral evolution equations is solved by gradient descent until the loss function is less than a preset value.

8. The multi-light source color balance evaluation method for image projectors as described in claim 6, characterized in that, The calculation of the evolution rate and brightness change rate of each initial calibration template at each node, to obtain the optical path dependence change curve of the multi-source system during the equalization process under different initial calibration templates, includes: Obtain the template topology corresponding to each initial calibration template, and obtain the adjacency spectrum set of the nodes on the template topology; wherein, different initial calibration templates correspond to different topology shapes, topology depths and topology connectivity; The brightness distribution value of the node is calculated using an optical path ranking algorithm; wherein, the brightness distribution value is calculated through the mapping relationship between node importance and brightness. A spectral evolution dynamic equation is established based on the brightness distribution value, and the brightness change rate of the nodes is obtained by solving the Runge-Kutta method; wherein the brightness change rate is obtained by calculating the differential of brightness with respect to time. A state-space equation is constructed based on the brightness change rate of the node and the dependency on the optical path dependency cloud map, and the dependency time function is obtained by Kalman filtering. The optical path dependency change curve is obtained based on the dependency time function.

9. The multi-light source color balance evaluation method for image projectors as described in claim 6, characterized in that, The step of selecting multiple first calibration templates from a plurality of initial calibration templates based on the optical path dependence change curve includes: The optical path dependency change curve is used to determine the current optical path dependency of each initial calibration template. Constraints are constructed based on the preset maximum dependency of the multi-source system and the current optical path dependency of each initial calibration template. The dependency constraint optimization parameters are then calculated using the Lagrange multiplier method. Based on the aforementioned dependency constraint optimization parameters, a multi-objective optimization function is established, comprising a dependency objective function and a brightness objective function. The mapping relationship between the template vector and the constraint response vector corresponding to each initial calibration template is obtained through a support vector machine with a polynomial kernel function. Based on the mapping relationship, a constrained optimization equation system is constructed using the Karouch-Kun-Tucker conditions, and the optimal tuning template and dual variable corresponding to each initial tuning template are obtained by solving the interior point method. If the norm difference between the optimal tuning template and the initial template corresponding to the initial tuning template is less than a preset convergence threshold, the optimal tuning template is stored as the first tuning template.

10. A multi-light source color balance evaluation system for an image projector, characterized in that, The multi-light source color balance evaluation system for the image projector includes: The spectral profiling module is used to acquire spectral feature data of a multi-light source system in an image projector, and to perform fusion analysis on the spectral feature data to obtain a light source feature profile. The template calling module is used to match the color balance model corresponding to the light source feature portrait, so as to call the color adjustment template from the preset template library through the color balance model; The color balance simulation module is used to simulate the color balance process corresponding to the multi-light source system based on the color calibration template. The template optimization module is used to optimize the color calibration template according to the color balance process, and generate multi-light source color balance evaluation results based on the optimized color calibration template.

Images