Self-adaptive calibration method for lightening parameters of liquid crystal screen

Abstract

The invention provides a liquid crystal display lightening parameter self-adaptive calibration method, and relates to the technical field of image display. The method comprises the following steps: firstly, establishing a characteristic matrix and a parameter library of liquid crystal molecules, and dynamically correcting according to a real-time working state; and based on the characteristic matrix and the parameter library, executing pixel-level forward prediction on the input image, and generating a color error distribution diagram by solving a coupling solution of a liquid crystal molecule dynamic response differential equation and a backlight attenuation compensation equation. And taking the error distribution diagram as input, obtaining a gradient optimization direction of the calibration parameter mapping diagram through back propagation calculation, and iteratively approaching the optimal pixel value compensation amount in each frame period. And injecting a test pulse into the characteristic pixel point, capturing a voltage response curve through a sampling circuit, and converting the voltage response curve into a real-time correction value mapping graph. And finally, carrying out weighted fusion on the real-time correction value mapping graph and the pixel value compensation amount to generate a calibration parameter mapping graph. The self-adaptive calibration of the display parameters is improved, and the display consistency and the color accuracy are effectively improved.

Description

Technical Field

[0001] This invention relates to the field of image display technology, specifically to an adaptive calibration method for liquid crystal screen illumination parameters. Background Technology

[0002] With the widespread adoption of high-resolution display technology in consumer electronics, professional design, and medical imaging, users' expectations for the final image display quality are constantly increasing. Ideally, the digital image data input to the display system should be accurately and without distortion reproduced as light signals in the physical world. However, this conversion process inherently involves time-varying nonlinear distortion, resulting in a significant deviation between the final displayed image and the original input data.

[0003] Currently, mainstream image display quality optimization solutions fall into two main technical paths and their inherent limitations: The first type is static calibration technology based on display terminals. This method measures the display hardware at the factory stage and generates a fixed color profile, which is then built into the device. This type of method cannot compensate for the dynamic characteristic drift of the screen caused by liquid crystal material aging, backlight brightness decay, and changes in ambient temperature during use.

[0004] The second category is software-based image processing algorithms. Although these methods can perform certain image enhancements, such as contrast stretching and color saturation adjustment, they are general post-processing operations that are independent of the content. Their algorithm models do not embed the physical characteristics of specific display hardware, so they cannot fundamentally "predict" and "counteract" the distortions introduced by the physical characteristics of the display device itself.

[0005] Therefore, existing technologies lack a method that can deeply integrate the physical model of display hardware to perform feedforward and accurate pre-correction of input image data. This results in unpredictable image quality loss at the end of the entire processing chain, which severely restricts the realization of image applications. Summary of the Invention

[0006] The purpose of this invention is to provide an adaptive calibration method for LCD screen illumination parameters, thereby solving the above-mentioned technical problems. This invention can achieve its purpose through the following technical solutions: S1. Establish a characteristic matrix of the electric field response of liquid crystal molecules containing pixel-level transmittance deviation and a parameter library of LED backlight attenuation, wherein the parameter library is constructed by applying multiple sets of driving voltages and measuring the actual transmittance during the factory calibration stage, and the parameter library is dynamically corrected according to the real-time collected backlight unit operating temperature and cumulative running time. S2. Based on the characteristic matrix and parameter library, perform pixel-level forward prediction calculation on the input image data. By solving the coupled solution of the differential equation of the dynamic response of liquid crystal molecules and the backlight attenuation compensation equation, generate a brightness and color error distribution map containing spatiotemporal characteristics. S3. Using the color error distribution map as input, the gradient optimization direction of the calibration parameter mapping map is obtained through backpropagation calculation. Multiple rounds of iterative calculation are performed within each frame display cycle to gradually approach the optimal pixel value compensation amount. S4. Based on the distribution characteristics of the pixel value compensation amount, a test pulse is injected into the selected feature pixel, wherein the amplitude and pulse width of the test pulse are determined by the statistical characteristics of the pixel value compensation amount; the voltage response curve of the feature pixel is captured by the built-in sampling circuit, and the voltage response curve is compared with the pre-stored standard response curve point by point, and the voltage difference between the two at key time nodes is extracted as the response difference amount; based on the voltage-transmittance correspondence in the liquid crystal molecule electric field response characteristic curve, the response difference amount is converted into a transmittance compensation amount, and the transmittance compensation amount is then used to obtain a real-time correction amount mapping map through inversion calculation; S5. The real-time correction amount mapping map and the pixel value compensation amount are weighted and fused in the digital image processing unit, wherein the weighting coefficient is dynamically adjusted according to the statistical characteristics of the backlight unit aging coefficient and the error distribution map to generate a calibrated calibration parameter mapping map, thereby completing the adaptive calibration of the image display.

[0007] As a further technical solution, the construction process of the characteristic matrix and parameter library involves sequentially applying multiple sets of driving voltages of different amplitudes to the LCD screen during the factory calibration stage, covering the full voltage range from the lowest to the highest grayscale level. The actual transmittance value of each pixel under the corresponding driving voltage is collected in real time, and an optical sensor is used to measure the transmittance to ensure the accuracy of data acquisition. The LCD driving voltage and transmittance value are paired to form an original measurement sequence. Using the original measurement sequence as input, the transmittance change gradient within adjacent voltage intervals is calculated using cubic spline interpolation, generating a characteristic matrix of the liquid crystal molecule electric field response. The characteristic matrix stores the voltage and transmittance of each pixel, with voltage as the row index and transmittance as the column index. Mapping relationship; During the construction of the parameter library, the backlight unit is controlled to run within a set temperature range by a constant temperature chamber, and the brightness decay is recorded under different cumulative running times. The backlight brightness is measured using a photometer. Time and temperature are used as dual parameters to establish a multi-dimensional lookup dual parameter table. Each entry in the table corresponds to the brightness decay coefficient under a specific combination of temperature and running time. During operation, the current temperature and running time of the backlight unit are monitored in real time. The corresponding decay amount is extracted from the parameter library by interpolation. Then, the decay amount is input into the brightness correction of the characteristic matrix. The transmittance gradient related to the backlight brightness is dynamically adjusted so that the characteristic matrix maintains the correction capability consistent with the actual optical response at different aging stages.

[0008] As a further technical solution, the forward prediction calculation process based on the characteristic matrix and parameter library specifically involves: in the digital image processing unit, the input image data is converted into the target brightness value of the corresponding pixel. The conversion process is based on the standard color space mapping relationship to ensure that the brightness value is consistent with the actual display requirements; based on the voltage and transmittance correspondence of each pixel in the characteristic matrix, for example, the theoretical transmittance distribution is calculated by interpolation or table lookup, which reflects the optical output of the pixel under ideal conditions; then, according to the attenuation parameters in the parameter library corresponding to the current backlight unit temperature and cumulative operating time, the brightness of the theoretical transmittance distribution is corrected, and the correction formula is used to estimate the actual transmittance. The value is equal to the theoretical transmittance multiplied by the attenuation coefficient, yielding an estimated actual transmittance. Based on this, the transmittance change trend of each pixel is calculated over time. Using a first-order difference equation with the difference between the transmittance at the previous moment and the current driving voltage as the input variable, the adjustment amount of the liquid crystal molecule response rate is obtained. The adjustment amount is calculated by a proportional-integral controller, and the current transmittance estimate is corrected accordingly, generating a pixel brightness distribution that changes over time. Finally, the corrected transmittance estimate is compared pixel by pixel with the target brightness value. The brightness difference and chromaticity difference data are calculated using the color difference formula, forming a brightness and color error distribution map. This color error distribution map is stored in matrix form for easy subsequent processing.

[0009] As a further technical solution, the generation of the color error distribution map is achieved by coupling the solution of the dynamic response differential equation of liquid crystal molecules and the backlight attenuation compensation equation. Specifically, the voltage-transmittance relationship of liquid crystal molecules in the characteristic matrix is ​​used as the initial boundary condition. The change in driving voltage of each pixel between adjacent frames is used as the input variable to construct a dynamic response differential equation of liquid crystal molecule orientation angle changing with time. This dynamic response differential equation is based on the elastic continuum theory of liquid crystals and is in the form of a first-order nonlinear differential equation. It is solved by time-step integration using numerical methods such as the Runge-Kutta method to obtain the instantaneous transmittance of each pixel in the current frame. At the same time, the attenuation parameters at the corresponding temperature and running time in the parameter library are used as input to establish a backlight brightness attenuation compensation equation. This backlight brightness attenuation compensation equation is an empirical model. Based on the exponential decay law, and combined with polynomial fitting of temperature and time, the instantaneous transmittance is obtained by integrating the differential equation of the dynamic response of liquid crystal molecules. This transmittance result is then substituted into the backlight attenuation compensation equation to calculate the actual brightness distribution affected by backlight changes. The change in pixel brightness in consecutive frames is compared with the target brightness difference of the input image to obtain the temporal brightness error, which is calculated using the root mean square method. Furthermore, based on the relative difference in transmittance of the red, green, and blue channels, the chromaticity error is calculated using the CIE chromaticity formula, ultimately generating an error distribution map containing spatiotemporal characteristics. This coupled solution method overcomes the limitations of static calibration. By simulating the dynamic behavior of liquid crystal molecules and the interaction of backlight attenuation, it achieves comprehensive capture of display errors, improving the accuracy and robustness of calibration.

[0010] As a further technical solution, the gradient optimization of the calibration parameter mapping map is achieved through a backpropagation algorithm. Specifically, within each frame display cycle, the color error distribution map is subjected to pixel-by-pixel error propagation and voltage response correlation calculation. The error propagation is based on the chain rule, and the color error of each pixel is weighted according to the slope of its corresponding voltage and transmittance curve in the characteristic matrix. The slope is calculated by numerical differentiation, so that the gradient direction reflects the actual response sensitivity of the liquid crystal molecules in the voltage range. In each iteration, the row or column driving signal of the pixel with the larger absolute gradient value is adjusted first, and key pixels are identified by a sorting algorithm to maximize the error reduction rate. After each round of gradient update, a dynamic correction coefficient is calculated based on the difference between the two transmittance estimates. This dynamic correction coefficient is obtained by the exponential smoothing method and is applied to the gradient weighting of the next round, thereby forming a time-related stepwise approximation process. The iteration process continues until the error is lower than the preset threshold or the maximum number of iterations is reached, ensuring that the pixel value compensation amount quickly converges to the optimal value.

[0011] As a further technical solution, the dynamic adjustment logic in the weighted fusion is executed in consecutive display frames. Specifically, it calculates whether the direction of change of the pixel value compensation amount corresponding to each pixel is consistent with the error correction direction of the previous frame. The consistency of direction is judged by vector dot product. When the two directions are consistent and the change rate slows down, it is considered that the pixel has approached the stable response range. At this time, the weight allocation is smoothly shifted to the real-time correction amount mapping map side. The shifting process is implemented by linear interpolation. Conversely, when the direction of the pixel value compensation amount is reversed or the rate of change increases, it indicates that the orientation of the liquid crystal molecules has not yet stabilized. The proportion of the real-time correction amount mapping map in the fusion is automatically reduced. The reduction ratio is calculated by the Sigmoid function, while retaining the dominant influence of the voltage compensation in the previous round. Through the logic of dynamic reverse adjustment with the change trend between frames, the weighting result is kept locally consistent in space and converges asymptotically frame by frame in time. This dynamic adjustment mechanism is based on the feedback principle in control theory. By monitoring the pixel state in real time, the weights are adaptively adjusted, avoiding over-adjustment or response lag during calibration, and ensuring the stability and consistency of the calibration process.

[0012] As a further technical solution, the test pulse injection and real-time correction mapping calculation process is as follows: After optimizing the calibration parameter mapping, based on the distribution gradient of pixel value compensation in the pixel array, feature pixels with prominent compensation value changes are preferentially selected using an edge detection algorithm. These points are usually located in high-contrast regions or severely aged regions. A test pulse is injected, and the pulse amplitude and pulse width are determined according to the dispersion of the voltage distribution in the region. The dispersion is obtained by calculating the variance, so that the acquired response curve can reflect the real changes of liquid crystal molecules under local extreme driving conditions. The built-in sampling circuit captures feature images at a high sampling rate. The voltage response curve of the sample point is stored after analog-to-digital conversion. The voltage response curve is compared with the pre-stored standard response curve at each time node. The standard response curve is pre-measured and stored during the factory calibration stage. The comparison process uses a dynamic time warping algorithm to extract the voltage difference between the two at the response stage, forming a response difference quantity. Based on the correspondence between voltage and transmittance in the characteristic matrix, the response difference quantity is inverted into a transmittance compensation quantity interval by interval. The inversion is achieved by lookup table and linear interpolation. On this basis, combined with the response delay characteristics of the time series, a delay compensation filter is used to correct the timing of the compensation quantity, and a real-time correction quantity mapping map is obtained.

[0013] As a further technical solution, the weighted fusion process in S5 is implemented in the digital image processing unit. Specifically, it involves weighted fusion of the real-time correction mapping map and the pixel value compensation amount, wherein the weighting coefficient is dynamically adjusted based on the statistical characteristics of the backlight unit aging coefficient and the error distribution map. The dynamic adjustment process first calculates the basic weight value based on the aging coefficient, which is obtained by fitting the backlight's cumulative running time and historical temperature data. Then, it extracts the mean of the brightness error and the standard deviation of the chromaticity error from the error distribution map as statistical features. The basic weight value and statistical features are combined through nonlinear combination to generate the adjusted weighting coefficient. The nonlinear combination introduces a product term of the aging coefficient and the error statistical features to amplify their interactive influence. The weight ratio of the real-time correction mapping map and the pixel value compensation amount is dynamically scaled according to the sign and magnitude of the product term. The scaling ratio is achieved through a piecewise function. After weighted fusion, a calibrated calibration parameter mapping map is generated to complete adaptive calibration. Thus, when the backlight aging is severe and the error distribution is dispersed, the real-time correction mapping map is used preferentially, while the pixel value compensation amount retains its dominant role when the error is concentrated.

[0014] This invention provides an adaptive calibration method for LCD screen illumination parameters, which has the following beneficial effects: This invention digitizes the physical characteristics of display hardware by establishing a liquid crystal molecule characteristic matrix and a backlight attenuation parameter library. When processing input image data, it can perform forward prediction based on these physical models, accurately calculate the amount of data compensation required for each pixel due to hardware nonlinearity and aging, avoid distortion, and greatly improve the fidelity of image display.

[0015] This invention collects the working status of the backlight unit in real time and injects test pulses for feedback. This allows the image data correction strategy to adapt to changes in the display device over time and in the environment, ensuring that the high-quality and stable output of the image is maintained throughout the entire lifespan of the device.

[0016] The error distribution map generated by this invention includes spatiotemporal characteristics, which can effectively pre-compensate dynamic distortions such as motion blur and chromatic lag in moving images, thereby significantly improving the clarity and response speed of moving images. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of the process of the present invention. Detailed Implementation

[0018] The technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] During the factory calibration phase, a characteristic matrix of the liquid crystal molecule electric field response and a parameter library of LED backlight attenuation are first established. In practice, a high-precision voltage source applies multiple sets of driving voltages to the LCD screen, ranging from 0 volts to 5 volts in 0.1-volt increments, covering all 256 gray levels. Simultaneously, an integrating sphere and a spectroradiometer are used to measure the actual transmittance value of each pixel under the corresponding driving voltage, with measurement accuracy controlled within 0.1%. After pairing the driving voltage with the transmittance value to form the original measurement sequence, a cubic spline interpolation algorithm is used to calculate the transmittance gradient within adjacent voltage intervals, generating a characteristic matrix. This characteristic matrix is ​​then used to... Voltage values ​​are row indices, and transmittance is column indices, storing the voltage-transmittance mapping relationship for each pixel. The parameter library is constructed by placing the backlight unit in a temperature-controlled chamber, with temperatures ranging from -10 degrees Celsius to 70 degrees Celsius in 10-degree intervals, and cumulative operating time ranging from 0 hours to 10,000 hours in 100-hour intervals. The brightness decay is recorded for each temperature and time combination, forming a table with temperature and time as dual parameters. For example, at a temperature of 25 degrees Celsius and an operating time of 2,000 hours, the measured brightness decay coefficient is 0.92, indicating that the backlight brightness has decayed to 92% of its initial value.

[0020] During operation, this method collects the operating temperature of the backlight unit in real time using a temperature sensor and obtains the cumulative running time using a timer, with the data update frequency being once every 10 milliseconds. Based on the current temperature and running time, a bilinear interpolation algorithm is used to extract the corresponding attenuation amount from the parameter library, and this attenuation amount is input into the brightness correction module of the characteristic matrix to dynamically adjust the transmittance gradient related to the backlight brightness. For example, when the backlight unit temperature is detected to be 45 degrees Celsius and the cumulative running time is 5000 hours, the corresponding attenuation coefficient of 0.85 is extracted from the parameter library and applied to the brightness correction calculation of the characteristic matrix, thereby adjusting the theoretical transmittance value to reflect the actual backlight state.

[0021] Based on a feature matrix and a parameter library, this method performs pixel-level forward prediction calculations on the input image data. First, the input image data is converted into target brightness values, following the sRGB color space standard to ensure the brightness values ​​match display requirements. Next, based on the voltage-transmittance correspondence of each pixel in the feature matrix, the theoretical transmittance distribution is calculated using a lookup table or linear interpolation. Then, based on the attenuation parameters extracted from the parameter library, the theoretical transmittance distribution is corrected for brightness to obtain an estimated actual transmittance value. For example, if a pixel's theoretical transmittance is 60% and the current attenuation coefficient is 0.85, then... The corrected actual transmittance estimate is 51%. Further time-based recursive calculations are performed on the transmittance change trend of each pixel, using the difference between the transmittance at the previous moment and the current driving voltage as the input variable. The adjustment amount for the liquid crystal molecule response rate is obtained through a first-order difference equation. Specifically, proportional-integral control is used for the calculation, with a proportional coefficient of 0.6 and an integral coefficient of 0.1. These coefficients are optimized based on experimental data. The current transmittance estimate is corrected according to the adjustment amount, generating a pixel brightness distribution that changes over time. Finally, the corrected transmittance estimate is compared pixel-by-pixel with the target brightness value, and the brightness difference and chromaticity difference are calculated using the CIE 1976Lab color difference formula, forming a brightness and color error distribution map. This error distribution map is stored in matrix form for easy subsequent processing.

[0022] The generation of brightness and color error distribution maps is achieved by solving the coupled differential equations of the dynamic response of liquid crystal molecules and the backlight attenuation compensation equation. The voltage-transmittance relationship of liquid crystal molecules in the characteristic matrix is ​​used as the initial boundary condition, and the change in driving voltage of each pixel between adjacent frames is used as the input variable to construct the dynamic response differential equation of the liquid crystal molecule orientation angle over time. This dynamic response differential equation is based on the elastic continuum theory of liquid crystals and is in the form of a first-order nonlinear differential equation. It is solved by time-step integration using the fourth-order Runge-Kutta method, with a time step size of 1 millisecond. Simultaneously, the backlight brightness attenuation compensation equation is established using attenuation parameters corresponding to the temperature and running time in the parameter library as input. This backlight brightness attenuation compensation equation is an exponential attenuation equation. The model, fitted based on experimental data, is as follows: for example, the attenuation equation can be expressed as the brightness attenuation equal to the initial brightness multiplied by e, minus k, multiplied by the time power, where k is the attenuation constant, which is temperature-dependent. By integrating the differential equation of the dynamic response of liquid crystal molecules, the instantaneous transmittance of each pixel in the current frame is obtained. Then, this transmittance result is substituted into the backlight attenuation compensation equation to calculate the actual brightness distribution affected by backlight changes. The change in pixel brightness in consecutive frames is compared with the target brightness difference of the input image to obtain the temporal brightness error, which is calculated using the root mean square method. Based on the relative difference in transmittance of the red, green, and blue channels, the CIE chromaticity formula is used to calculate the chromaticity error, ultimately generating an error distribution map containing spatiotemporal characteristics.

[0023] Using the color error distribution map as input, the gradient optimization direction of the calibration parameter mapping map is obtained through backpropagation calculation. Multiple rounds of iterative calculations are performed within each frame display cycle to gradually approach the optimal pixel value compensation amount. Specifically, the color error of each pixel is weighted according to the slope of its corresponding voltage and transmittance curve in the characteristic matrix. The slope is calculated using the central difference method; for example, at a voltage of 2.5 volts, the slope might be 0.3 per volt. In each iteration, the driving signals of the row or column containing pixels with larger absolute gradient values ​​are adjusted first to maximize the error reduction rate. After each round of gradient update, a dynamic correction coefficient is calculated based on the difference between two consecutive transmittance estimates. This dynamic correction coefficient is obtained using the exponential smoothing method, with a smoothing factor set to 0.7. The dynamic correction coefficient is applied to the gradient weighting of the next round, forming a time-dependent gradual approximation process. The iterative process continues until the error is below a preset threshold of 2% or the maximum number of iterations (50) is reached. These thresholds are set based on display quality requirements.

[0024] Based on the distribution characteristics of pixel value compensation, test pulses are injected into selected feature pixels. According to the distribution gradient of pixel value compensation in the pixel array, for example, feature pixels with prominent compensation value changes are preferentially selected using the Sobel edge detection algorithm. These feature pixels are typically located in high-contrast areas or severely aged areas. The amplitude and pulse width of the test pulse are determined by the statistical characteristics of the voltage distribution in that area. Specifically, the pulse amplitude is set to 1.5 times the standard deviation of the voltage in that area, and the pulse width is set to an adjustable value between 10 microseconds and 100 microseconds. For example, when the standard deviation of the voltage is 0... At 0.2 volts, the pulse amplitude is set to 0.3 volts. The voltage response curve of the feature pixel is captured by the built-in sampling circuit at a sampling rate of 1 MHz. The voltage response curve is compared with the pre-stored standard response curve point by point, and the voltage difference between the two at key time nodes is extracted as the response difference. Based on the correspondence between voltage and transmittance in the characteristic matrix, the response difference is converted into transmittance compensation. The transmittance compensation is then used to obtain a real-time correction mapping through inversion calculation. For example, the inversion calculation adopts the Newton-Raphson iteration method, and the number of iterations is limited to within 10 to ensure real-time performance.

[0025] The real-time correction mapping map and pixel value compensation are weighted and fused in the digital image processing unit. The weighting coefficient is dynamically adjusted based on the statistical characteristics of the backlight unit aging coefficient and the error distribution map. The aging coefficient is obtained by fitting the backlight's cumulative running time and historical temperature data. The specific calculation formula is: aging coefficient = 1 minus running time divided by 100,000 hours and then multiplied by the temperature influence factor. The temperature influence factor is calculated based on the ratio of the current temperature to the reference temperature of 25 degrees Celsius. For example, when the temperature is 50 degrees Celsius, the temperature influence factor may be 1.2. The dynamic adjustment process first calculates the basic weight value based on the aging coefficient, and then extracts the mean of brightness error and chromaticity error from the error distribution map. The standard deviation of the error is used as a statistical feature. The basic weight values ​​are combined with the statistical feature through nonlinear combination to generate an adjusted weighted coefficient. The nonlinear combination introduces a product term of the aging coefficient and the error statistical feature to amplify the interaction between the two. The weight ratio of the real-time correction amount map and the pixel value compensation amount is dynamically scaled according to the sign and magnitude of the product term. The scaling ratio is achieved through a piecewise linear function. For example, when the aging coefficient is high and the error distribution is dispersed, the weight of the real-time correction amount map can be increased to 0.8, while when the error is concentrated, the weight of the pixel value compensation amount remains dominant, for example, set to 0.7. After weighted fusion, a calibrated calibration parameter map is generated to complete the adaptive calibration of the image display.

[0026] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for adaptive calibration of LCD screen illumination parameters, characterized in that, Includes the following steps: S1. Establish a characteristic matrix of the electric field response of liquid crystal molecules containing pixel-level transmittance deviation and a parameter library of LED backlight attenuation, wherein the parameter library is constructed by applying multiple sets of driving voltages and measuring the actual transmittance during the factory calibration stage, and the parameter library is dynamically corrected according to the real-time collected backlight unit operating temperature and cumulative running time. S2. Based on the characteristic matrix and parameter library, perform pixel-level forward prediction calculation on the input image data. By solving the coupled solution of the differential equation of the dynamic response of liquid crystal molecules and the backlight attenuation compensation equation, generate a brightness and color error distribution map containing spatiotemporal characteristics. S3. Using the color error distribution map as input, the gradient optimization direction of the calibration parameter mapping map is obtained through backpropagation calculation. Multiple rounds of iterative calculation are performed within each frame display cycle to gradually approach the optimal pixel value compensation amount. S4. Based on the distribution characteristics of pixel value compensation, inject test pulses into selected feature pixels, wherein the amplitude and pulse width of the test pulses are determined by the statistical characteristics of the pixel value compensation. The voltage response curve of the feature pixel is captured by the built-in sampling circuit. The voltage response curve is compared with the pre-stored standard response curve point by point, and the voltage difference between the two at key time nodes is extracted as the response difference. Based on the voltage and transmittance correspondence in the electric field response characteristic curve of the liquid crystal molecule, the response difference is converted into transmittance compensation. The transmittance compensation is then calculated by inversion to obtain a real-time correction mapping. S5. The real-time correction amount mapping map and the pixel value compensation amount are weighted and fused in the digital image processing unit, wherein the weighting coefficient is dynamically adjusted according to the statistical characteristics of the backlight unit aging coefficient and the error distribution map to generate a calibrated calibration parameter mapping map, thereby completing the adaptive calibration of the image display.

2. The adaptive calibration method for LCD screen illumination parameters according to claim 1, characterized in that: The process of establishing the characteristic matrix and parameter library is as follows: During the factory calibration stage, multiple sets of driving voltages with different amplitudes are applied to the LCD screen in sequence. The actual transmittance value of each pixel under the corresponding driving voltage is collected in real time. The LCD driving voltage and transmittance value are paired to form an original measurement sequence. Using the original measurement sequence as input, the transmittance change gradient in adjacent voltage intervals is determined by interpolation calculation, generating a characteristic matrix of the electric field response of the liquid crystal molecules. During the construction of the parameter library, the brightness decay of the backlight unit in the set temperature range and cumulative running time range is recorded, and a dual-parameter table is established with time and temperature as indexes. During operation, the corresponding decay amount is extracted from the parameter library according to the current temperature and running time of the backlight unit. The decay amount is then input into the brightness correction of the characteristic matrix to adjust the transmittance gradient related to the backlight brightness, so that the characteristic matrix maintains the correction capability consistent with the actual optical response at different aging stages.

3. The adaptive calibration method for LCD screen illumination parameters according to claim 2, characterized in that: After establishing the feature matrix and parameter library, the input image data is converted into the target brightness value of the corresponding pixel, and the theoretical transmittance distribution is calculated based on the relationship between voltage and transmittance of each pixel in the feature matrix. Then, based on the attenuation parameters in the parameter library corresponding to the current backlight unit temperature and cumulative running time, the theoretical transmittance distribution is corrected for brightness to obtain the estimated value of actual transmittance. Based on this, the time-based recursive calculation is performed on the transmittance change trend of each pixel. The difference between the transmittance at the previous moment and the current driving voltage is used as the input variable to calculate the adjustment amount of the liquid crystal molecule response rate, and the current transmittance estimate is corrected accordingly to generate the pixel brightness distribution that changes over time. Finally, the corrected transmittance estimate is compared with the target brightness value pixel by pixel to extract the brightness difference and chromaticity difference data, forming a brightness and color error distribution map.

4. The adaptive calibration method for LCD screen illumination parameters according to claim 3, characterized in that: The color error distribution map is generated by solving the differential equation of the dynamic response of liquid crystal molecules and the backlight attenuation compensation equation. Specifically, the voltage-transmittance relationship of liquid crystal molecules in the characteristic matrix is ​​used as the initial boundary condition. The change in driving voltage of each pixel between adjacent frames is used as the input variable to construct the differential equation of the dynamic response of liquid crystal molecules as a function of time. At the same time, the attenuation parameters at the corresponding temperature and running time in the parameter library are used as input to establish the backlight brightness attenuation compensation equation. By integrating the differential equation of the dynamic response of liquid crystal molecules step by step in time, the instantaneous transmittance of each pixel in the current frame is obtained. The transmittance result is then substituted into the backlight attenuation compensation equation to calculate the actual brightness distribution after the backlight change. The change in pixel brightness in consecutive frames is compared with the target brightness difference of the input image to obtain the temporal brightness error. And the chromaticity error is calculated based on the relative difference of the transmittance of the red, green and blue channels.

5. The adaptive calibration method for LCD screen illumination parameters according to claim 1, characterized in that: In the gradient optimization of the calibration parameter mapping map in S3, it is achieved by performing pixel-by-pixel error propagation and voltage response correlation calculation on the color error distribution map within each frame display cycle. Specifically, when calculating the backpropagation gradient, the color error of each pixel is weighted according to the slope of its corresponding voltage and transmittance curve in the characteristic matrix, so that the gradient direction reflects the actual response sensitivity of the liquid crystal molecules in the voltage range. In each iteration, the row or column driving signal of the pixel with the larger absolute gradient value is adjusted first to maximize the error reduction rate. After each round of gradient update, the dynamic correction coefficient is calculated based on the difference between the two transmittance estimates, and the dynamic correction coefficient is applied to the gradient weighting of the next round, thereby forming a time-dependent stepwise approximation process.

6. The adaptive calibration method for LCD screen illumination parameters according to claim 5, characterized in that: In a continuous display of frames, the direction of change of the pixel value compensation amount corresponding to each pixel is calculated to be consistent with the error correction direction of the previous frame. When the two directions are consistent and the change rate slows down, it is considered that the pixel has approached the stable response range. At this time, the weight allocation is smoothly shifted to the real-time correction amount mapping map. Conversely, when the direction of the pixel value compensation amount is reversed or the rate of change increases, it indicates that the orientation of the liquid crystal molecules has not yet stabilized. The proportion of the real-time correction amount mapping map in the fusion is automatically reduced, while the dominant influence of the voltage compensation in the previous round is retained. Through the logic of dynamically adjusting in reverse according to the inter-frame change trend, the weighting result is kept locally consistent in space and gradually converged frame by frame in time.

7. The adaptive calibration method for LCD screen illumination parameters according to claim 1, characterized in that: In step S4, after optimizing the calibration parameter mapping, based on the distribution gradient of the pixel value compensation amount in the pixel array, the feature pixel points with prominent compensation value changes are preferentially selected to inject test pulses. The pulse amplitude and pulse width are determined according to the dispersion of the voltage distribution in the region, so that the collected response curve can reflect the real changes of liquid crystal molecules under local extreme driving conditions. After the built-in sampling circuit captures the voltage response curve of the feature pixel, it compares it with the pre-stored standard response curve at each time node, extracts the voltage difference between the two in the response stage, and forms the response difference quantity. Based on the correspondence between voltage and transmittance in the characteristic matrix, the response difference quantity is inverted into transmittance compensation quantity in interval by interval. On this basis, the real-time correction quantity mapping map is obtained by combining the response delay characteristics of the time series to correct the timing of the compensation quantity.

8. The adaptive calibration method for LCD screen illumination parameters according to claim 7, characterized in that: S5 includes: in the digital image processing unit, weighted fusion of the real-time correction amount mapping map and the pixel value compensation amount, wherein the weighting coefficient is dynamically adjusted according to the statistical characteristics of the backlight unit aging coefficient and the error distribution map; the dynamic adjustment process first calculates the basic weight value based on the aging coefficient, then extracts the mean of the brightness error and the standard deviation of the chromaticity error in the error distribution map as statistical features, and combines the basic weight value and the statistical features through nonlinear combination to generate the adjusted weighting coefficient, wherein the nonlinear combination introduces the product term of the aging coefficient and the error statistical features to amplify the interaction between the two, and dynamically scales the weight ratio of the real-time correction amount mapping map and the pixel value compensation amount according to the sign and magnitude of the product term; after weighted fusion, a calibrated calibration parameter mapping map is generated to complete adaptive calibration, thereby giving priority to the real-time correction amount mapping map when the backlight aging is severe and the error distribution is dispersed, and retaining the dominant role of the pixel value compensation amount when the error is concentrated.

Images