A display gamut calibration method and device, computer equipment and medium

Abstract

The application discloses a display gamut calibration method and device, computer equipment and medium, the method is in situ collected screen spectrum through spectral sensor array and combines environmental spectrum to carry out compression reconstruction and environmental light offset, and then obtains gamut calibration result through dynamic gamut mapping and driving signal iterative feedback. The application integrates spectral sensor array in the display pixel gap to obtain original screen spectrum data in situ, and synchronously collects environmental light spectrum data with the help of environmental light spectrum sensor, combines screen surface reflectivity to calculate environmental light offset of reconstructed screen spectrum data, accurately peels off the environmental light reflection component from the actual screen radiation spectrum, and then realizes calibration through dynamic gamut mapping and iterative feedback of driving signal, solves the problems that the existing gamut calibration scheme is difficult to peel off environmental light interference in situ and dynamic calibration precision is limited, and achieves the technical effects of realizing full gamut, high precision and dynamic real-time calibration in a complex and changeable lighting environment.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a display color gamut calibration method, apparatus, computer equipment, and medium. Background Technology

[0002] With the continuous development of display technology, monitors are increasingly widely used in color-sensitive industries such as medical imaging diagnosis, professional photography and printing, and high-end film and television production. These applications place high demands on the color gamut accuracy of monitors, as the accuracy of color reproduction directly affects the quality of work in image interpretation, printing proofing, and post-production. To ensure that monitors maintain stable color gamut performance during long-term use, appropriate color gamut calibration methods are required to adjust the monitors.

[0003] Existing display color gamut calibration methods mainly include the following implementation forms: First, a scheme that uses an external spectrophotometer or colorimeter for periodic calibration. This scheme involves attaching an external measuring instrument to the surface of the display screen for measurement and manually adjusting the display parameters based on the measurement results. Second, a scheme that integrates a single color sensor on the bezel or back of the display screen. This scheme uses the single color sensor to acquire the mixed light signal after reflection or refraction by the panel, thereby providing feedback on the display status. Third, an ambient light compensation scheme based on software simulation. This scheme estimates the ambient light parameters using a single color temperature sensor and then performs software-level compensation on the display output based on a preset algorithm model.

[0004] However, existing display color gamut calibration methods still have technical shortcomings. Existing solutions cannot obtain the true spectral information of each area of ​​the screen in situ during the operation of the display. Furthermore, the handling of ambient light interference usually relies on assumed models or rough estimates, making it difficult to accurately separate the ambient light reflection component from the actual radiation spectrum of the screen. This limits the accuracy of calibration results in complex and variable lighting environments, and fails to meet the needs of color-sensitive applications for full color gamut, high precision, and dynamic real-time calibration. Summary of the Invention

[0005] This invention provides a display color gamut calibration method, apparatus, computer equipment, and medium, aiming to solve the technical problems of existing color gamut calibration schemes, such as difficulty in in-situ removal of ambient light interference and limited dynamic calibration accuracy.

[0006] In a first aspect, embodiments of the present invention provide a display color gamut calibration method, comprising: The display acquires raw screen spectral data by integrating a spectral sensor array into the pixel gaps of the display and simultaneously acquires ambient light spectral data around the display by using an ambient light spectral sensor. The original screen spectral data is compressed and encoded according to a preset spectral feature basis vector library to obtain the spectral compression coefficient result; The spectral compression coefficient results are reconstructed based on the spectral feature basis vector library to obtain reconstructed screen spectral data. Based on the preset screen surface reflectivity, ambient light cancellation calculation is performed on the reconstructed screen spectral data using the ambient light spectral data to obtain clean screen spectral data; Dynamic color gamut mapping is performed on the pure screen spectral data according to the preset target color gamut to obtain color gamut correction matrix data; Based on preset triggering conditions, a driving signal is generated using the color gamut correction matrix data and iteratively fed back to obtain the color gamut calibration result.

[0007] Secondly, embodiments of the present invention provide a display color gamut calibration device, comprising: The data acquisition unit is used to acquire raw screen spectral data of the display through a spectral sensor array integrated into the pixel gap of the display and to synchronously acquire ambient light spectral data around the display through an ambient light spectral sensor. The data compression unit is used to compress and encode the original screen spectral data according to a preset spectral feature basis vector library to obtain the spectral compression coefficient result; The spectral reconstruction unit is used to reconstruct the spectral compression coefficient results based on the spectral feature basis vector library to obtain reconstructed screen spectral data. The data calculation unit is used to perform ambient light cancellation calculation on the reconstructed screen spectral data based on the preset screen surface reflectance and the ambient light spectral data to obtain clean screen spectral data. The data correction unit performs dynamic color gamut mapping on the pure screen spectral data according to the preset target color gamut to obtain color gamut correction matrix data; The data output unit is used to generate a driving signal based on the color gamut correction matrix data according to preset trigger conditions and iteratively feed back to obtain the color gamut calibration result.

[0008] Thirdly, embodiments of the present invention provide a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the display color gamut calibration method of the first aspect.

[0009] Fourthly, embodiments of the present invention provide a computer-readable storage medium, wherein a computer program is stored on the computer-readable storage medium, and the computer program, when executed by a processor, implements the display color gamut calibration method of the first aspect.

[0010] This invention provides a display color gamut calibration method, comprising: acquiring raw screen spectral data of the display through a spectral sensor array integrated into the pixel gaps of the display and simultaneously acquiring ambient light spectral data around the display through an ambient light spectral sensor; compressing and encoding the raw screen spectral data according to a preset spectral feature basis vector library to obtain a spectral compression coefficient result; reconstructing the spectral compression coefficient result according to the spectral feature basis vector library to obtain reconstructed screen spectral data; performing ambient light cancellation calculation on the reconstructed screen spectral data based on a preset screen surface reflectance using the ambient light spectral data to obtain clean screen spectral data; dynamically mapping the clean screen spectral data according to a preset target color gamut to obtain color gamut correction matrix data; and generating a driving signal using the color gamut correction matrix data based on preset trigger conditions and iteratively feeding it back to obtain a color gamut calibration result. This invention integrates a spectral sensor array into the pixel gaps of a display to acquire raw screen spectral data of each region in situ. It also synchronously acquires ambient light spectral data using an ambient light spectral sensor. Combined with a preset screen surface reflectivity, it performs ambient light cancellation calculations on the reconstructed screen spectral data, accurately removing the ambient light reflection component from the actual radiation spectrum of the screen. Then, it achieves closed-loop calibration through dynamic color gamut mapping and iterative feedback of the driving signal. This solves the technical problems of existing color gamut calibration schemes, such as difficulty in removing ambient light interference in situ and limited dynamic calibration accuracy. It achieves the technical effect of full color gamut, high precision, and dynamic real-time calibration in complex and ever-changing lighting environments.

[0011] This invention also provides a display color gamut calibration device, a computer device, and a storage medium, which have the same beneficial effects as described above. Attached Figure Description

[0012] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0013] Figure 1 A schematic flowchart of a display color gamut calibration method provided in an embodiment of the present invention; Figure 2 This is a schematic block diagram of a display color gamut calibration device provided in an embodiment of the present invention.

[0014] Explanation of reference numerals in the attached figures: 200. Display color gamut calibration device; 201. Data acquisition unit; 202. Data compression unit; 203. Spectral reconstruction unit; 204. Data calculation unit; 205. Data correction unit; 206. Data output unit. Detailed Implementation

[0015] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. 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.

[0016] It should be understood that, when used in this specification and the appended claims, the terms "comprising" and "including" indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0017] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0018] It should also be further understood that the term "and / or" as used in this specification and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0019] Please see below. Figure 1 , Figure 1 The flowchart of a display color gamut calibration method provided in an embodiment of the present invention specifically includes steps S101 to S106.

[0020] S101. Acquire raw screen spectral data of the display by means of a spectral sensor array integrated into the pixel gap of the display and synchronously acquire ambient light spectral data around the display by means of an ambient light spectral sensor. S102. Compress and encode the original screen spectral data according to the preset spectral feature basis vector library to obtain the spectral compression coefficient result; S103. Reconstruct the spectral compression coefficient result based on the spectral feature basis vector library to obtain reconstructed screen spectral data; S104. Based on the preset screen surface reflectivity, perform ambient light cancellation calculation on the reconstructed screen spectral data using the ambient light spectral data to obtain clean screen spectral data. S105. Dynamically map the pure screen spectral data according to the preset target color gamut to obtain color gamut correction matrix data. S106. Based on the preset triggering conditions, a driving signal is generated using the color gamut correction matrix data and iteratively fed back to obtain the color gamut calibration result.

[0021] In step S101, by embedding the spectral sensor array into the pixel gaps of the display, the spectral sensor array can directly acquire the luminescence information of each area of ​​the screen during normal operation of the display, thereby obtaining the original screen spectral data in situ. Simultaneously, the ambient light spectral sensor is positioned towards the user's normal viewing direction to collect the illumination distribution of the environment in which the display is located within the same time window as the spectral sensor array, thereby obtaining the ambient light spectral data. The spectral sensor array and the ambient light spectral sensor complete their acquisition operations under unified synchronization signal control, ensuring a strict temporal correspondence between the original screen spectral data and the ambient light spectral data, providing the foundational data for subsequent ambient light cancellation calculations.

[0022] In one embodiment, step S101 includes: When a synchronous acquisition task is received, the original screen spectral data and ambient light spectral data frames are acquired respectively through the spectral sensor array and the ambient light spectral sensor within the same time window. The ambient light spectral data frame is marked as an environmental reference value in the header of the original screen spectral data.

[0023] In this embodiment, when a synchronous acquisition task is received, the original screen spectral data and ambient light spectral data frames are acquired by the spectral sensor array and the ambient light spectral sensor respectively within the same time window. During this process, the synchronous acquisition task is issued by the main processor after determining that the calibration trigger condition is met, and is simultaneously sent to the spectral sensor array and the ambient light spectral sensor with a unified hardware synchronization signal. The hardware synchronization signal strictly constrains the acquisition start time at the timing level, enabling the spectral sensor array and the ambient light spectral sensor to complete their respective acquisition actions within the same time window. This time window is an extremely short period of microseconds, sufficient to ensure that the physical lighting conditions corresponding to the two acquisitions are essentially the same. Within this time window, the spectral sensor array, integrated into the pixel gaps of the display, samples the emission from various areas of the screen to obtain the original screen spectral data; simultaneously, the ambient light spectral sensor samples the viewing environment around the display to obtain the ambient light spectral data frames. With the help of the above-mentioned timing synchronization mechanism, the original screen spectral data and the ambient light spectral data frame can physically correspond to the same instantaneous ambient light conditions, thus laying a timing foundation for subsequent accurate ambient light cancellation calculations.

[0024] Furthermore, the ambient light spectral data frame is marked as an environmental reference value in the header of the original screen spectral data. During this process, each time the original screen spectral data acquired by the spectral sensor array is transmitted to the main processor, the synchronously acquired ambient light spectral data frame is written into the data packet header as an environmental reference value. This establishes a one-to-one correspondence between the original screen spectral data and the ambient light spectral data frame at the data frame format level. With this header marking mechanism, even if the viewing environment is dynamically changing, each set of original screen spectral data can find a strictly corresponding environmental reference value, avoiding data mismatch problems caused by ambient light fluctuations, thus ultimately obtaining ambient light spectral data with clear environmental correlation attributes.

[0025] Furthermore, the hardware aspect is described in more detail. The spectral sensor array is connected via a high-speed serial peripheral interface bus or a low-voltage differential serial bus to obtain the array connection relationship. In this process, the spectral sensor array adopts an on-chip spectrometer design concept, that is, it is deeply integrated with the display panel using semiconductor technology. At the manufacturing level, it can employ miniature Fourier transform spectrometer chips or miniature grating spectrometer chips based on Micro-Electro-Mechanical Systems (MEMS) or Complementary Metal-Oxide-Semiconductor (CMOS) technology. The individual size of these chips can be controlled within 200μm × 200μm, and the thickness less than 100μm, thus enabling them to be embedded in the pixel gaps of modern high screen-to-body ratio displays without significantly affecting the effective area of ​​the display area or causing obvious visual dark spots. In terms of layout, the spectral sensor array is arranged in a periodic grid pattern or a non-uniform key area reinforcement pattern. Taking an 8K resolution (7680×4320) display as an example, one spectral sensor can be placed at the center of every 320×320 pixel group, thus forming an array of 27 rows × 15 columns, totaling 405 sensing nodes. Regarding integration location, the spectral sensor array is precisely mounted on the thin-film transistor (TFT) glass substrate of the display panel, below the color filter and polarizer. For organic light-emitting diode (OLED) displays, it is integrated under the encapsulation layer, enabling the spectral sensor array to directly receive the light generated by the excitation of sub-pixels (R, G, B), avoiding refraction, reflection, and crosstalk from the upper optical film layer, and achieving in-situ measurement. A miniature silicon-based microlens array can also be integrated above each spectral sensor unit to enhance light collection efficiency and precisely constrain the field of view to its corresponding local pixel area, preventing interference from adjacent pixel light. Based on the above hardware structure, the spectral sensor array is connected to the main processor through a shared high-speed serial peripheral interface (SPI) bus or a customized low-voltage differential signaling (LVDS) bus, thereby constructing the array connection relationship that characterizes the data path between the spectral sensor array and the main processor.

[0026] Furthermore, based on the array connection relationship, each element is individually addressed, configured, and read data using a unique logical address to obtain the addressing configuration. During this process, each spectral sensor unit in the spectral sensor array is assigned a unique logical address. The main processor uses the aforementioned high-speed serial peripheral interface bus or low-voltage differential serial bus to individually address, configure operating parameters, and read spectral data from each spectral sensor unit. This method significantly reduces the number of required data lines, thereby achieving high integration. The resulting addressing configuration serves as the basis for subsequent operating mode scheduling.

[0027] Furthermore, based on the addressing configuration, a global snapshot mode, a region scan mode, or a low-power monitoring mode is selected to obtain the operating mode instruction. In this process, the global snapshot mode refers to all spectral sensor units being synchronously triggered under unified signal control to capture global spectral data for an entire frame of the screen for comprehensive calibration; the region scan mode refers to activating only the spectral sensor units corresponding to a specific area of ​​the screen (e.g., the area the user is focusing on viewing) for high-frequency, low-latency local color tracking and calibration; the low-power monitoring mode refers to putting most spectral sensor units into a sleep state, with only a few sampling at extremely low frequencies to monitor for significant changes in the screen or environment, thereby determining whether to wake up the entire system. The operating mode instruction is obtained by selecting from these three modes based on current accuracy requirements and power constraints, enabling the system to initiate global measurement when high accuracy is needed and maintain low-power operation during normal operation.

[0028] Furthermore, synchronous acquisition is initiated according to the aforementioned working mode instructions, resulting in a synchronous acquisition task. During this process, the main processor simultaneously sends a unified hardware synchronization signal to both the spectral sensor array and the ambient light spectral sensor, enabling them to perform their respective acquisition actions within the same microsecond-level time window, thus forming the synchronous acquisition task. The hardware synchronization signal strictly constrains the acquisition start time in terms of timing, providing a fundamental prerequisite for subsequent accurate ambient light cancellation calculations.

[0029] Furthermore, based on the ambient light spectral sensor, the data is collected within the time window according to the synchronous acquisition task, resulting in raw screen spectral data and ambient light spectral data frames. In this process, the ambient light spectral sensor serves as a reference for screen spectral measurement, aiming to accurately capture real viewing environment information. A small-package, high dynamic range (HDR) miniature spectral sensor chip (such as the AS7341 chip from ams OSRAM or a similar custom chip) can be selected, with a spectral response range covering the visible light band from 380nm to 780nm and possessing at least 16 independent spectral channels. In terms of physical installation and optical design, the ambient light spectrum sensor is independently installed inside the upper bezel of the display (or the extremely narrow notch or forehead area of ​​a full-screen device), with its photosensitive surface facing the user's normal viewing direction. A miniature diffuser and an infrared cut-off filter are positioned in front of the ambient light spectrum sensor. The miniature diffuser eliminates the polarization characteristics of the incident light and provides a cosine response to ensure measurement accuracy at different incident angles. The infrared cut-off filter eliminates interference from non-visible light. The field of view of the ambient light spectrum sensor is designed to be a wide field of view of ±60° to fully capture ambient light information within the hemisphere in front of the user. Under the unified scheduling of the synchronous acquisition task, the spectrum sensor array completes the acquisition of screen emission within the time window, obtaining the raw screen spectrum data; the ambient light spectrum sensor completes the acquisition of ambient light around the display within the same time window, obtaining the ambient light spectrum data frame.

[0030] Finally, the ambient light spectral data frame is marked as an environmental reference value in the header of the original screen spectral data to obtain the ambient light spectral data. During this process, each original screen spectral data frame acquired by the spectral sensor array has its synchronously acquired ambient light spectral data frame written into its data packet header, ensuring that each set of original screen spectral data can find a strictly corresponding environmental reference value. Thanks to this header marking mechanism, even if the viewing environment is dynamically changing, a stable data association can be formed between the original screen spectral data and the ambient light spectral data frame, ultimately yielding the ambient light spectral data.

[0031] In step S102, the spectral feature basis vector library is a set of feature basis vectors extracted from a large number of spectral samples during the offline stage. This set of feature basis vectors can characterize the main variation patterns of the display spectrum and the ambient light spectrum. This step projects the original screen spectral data into the low-dimensional space spanned by the spectral feature basis vector library, thereby approximating the original spectral data with a small number of weight coefficients to obtain the spectral compression coefficient result. Through the above compression encoding process, the originally high-dimensional spectral data is converted into a coefficient sequence with significantly reduced dimensionality, which helps to reduce data transmission load and processing latency.

[0032] In one embodiment, step S102 includes: Construct a spectral database, obtain the average spectrum from the spectral database, and then calculate the covariance matrix of the spectral database based on the average spectrum; The covariance matrix is ​​subjected to eigenvalue decomposition to obtain the eigenvalues ​​and corresponding eigenvectors of the covariance matrix. Select the eigenvectors corresponding to the first preset number of largest eigenvalues ​​from the eigenvalues ​​to obtain the spectral feature basis vector library; The average spectrum is subtracted from each original spectral vector in the original screen spectral data to obtain the centered spectrum. The weighting coefficients are obtained by projecting the centered spectrum onto the spectral feature basis vector library and calculating the weighting coefficients. The weighting coefficients and their corresponding sensor identifiers are packaged and transmitted together to obtain the spectral compression coefficient results.

[0033] In this embodiment, a spectral database is constructed, and the average spectrum of the spectral database is calculated. Then, the covariance matrix of the spectral database is calculated based on the average spectrum. During this process, the construction of the spectral database covers various spectral forms that may occur in display scenarios. Its sample sources include a massive amount of screen spectral samples generated at different aging stages by different types of display panels (e.g., planar switching liquid crystal displays (LCD-IPS), vertical alignment liquid crystal displays (LCD-VA), organic light-emitting diode displays (OLED), and mini-LED displays). Simultaneously, it incorporates environmental spectral samples corresponding to various typical ambient light sources (e.g., D65 standard daylight, D50 standard daylight, A light source (incandescent light), and various light-emitting diode (LED) light sources and fluorescent lamp light sources), thus forming a large-scale spectral database. ,in Represents the first in the database Each spectral sample is a column vector characterizing the spectral power distribution in the 380 nm to 780 nm wavelength band. This represents the total number of spectral samples in the spectral database. Based on the spectral database, the average spectrum is obtained by performing an arithmetic mean calculation on all spectral samples according to their corresponding wavelength positions. Then, based on the average spectrum, the covariance matrix of the spectral database is calculated using the following expression: ; in, This represents the covariance matrix of the spectral database. Represents the first in the database One spectral sample, This represents the average spectrum obtained from all spectral samples. Indicates the total number of spectral samples, superscript This represents the matrix transpose operation. The covariance matrix characterizes the statistical distribution of each spectral sample in the spectral database relative to the average spectrum, laying the mathematical foundation for subsequent feature extraction.

[0034] Furthermore, the covariance matrix is ​​subjected to eigenvalue decomposition to obtain the eigenvalues ​​and corresponding eigenvectors of the covariance matrix. During this process, the covariance matrix... By applying the eigenvalue decomposition operation of Principal Component Analysis (PCA), a set of eigenvalues ​​and corresponding eigenvectors are obtained. The eigenvalues ​​reflect the energy concentration of the spectral database along the corresponding eigenvector directions; the larger the eigenvalue, the more spectral variation information is contained in that direction. The eigenvectors corresponding to the eigenvalues ​​characterize the main directions of spectral variations in the spectral database.

[0035] The eigenvectors corresponding to the first preset number of largest eigenvalues ​​among the eigenvalues ​​are selected to obtain the spectral feature basis vector library. In this process, all eigenvalues ​​obtained in the previous process are sorted in descending order of numerical value, and the first... The eigenvectors corresponding to the largest eigenvalues The spectral feature basis vector library is formed by concatenating columns. ,in That is, the preset quantity. Indicates the first 1 feature vector. The preset number Dimensions much smaller than the original spectrum, for example, can be taken The original spectral dimension was 81, thus achieving significant dimensionality reduction of the features while preserving the main spectral variation information. At this point, the offline training phase ended, and the obtained average spectrum... With the spectral feature basis vector library It is stored in the system and serves as the common basis for subsequent online compression and reconstruction operations.

[0036] Further, in the online compression stage, the average spectrum is subtracted from each original spectral vector in the original screen spectral data to obtain the centered spectrum. During this process, for each original spectral vector collected by each spectral sensor unit in the original screen spectral data... This is compared with the average spectrum obtained during the offline training phase. The centered spectrum is obtained by subtracting the corresponding wavelength positions, and its calculation expression is as follows: ; in, This represents the centered spectrum obtained after centered processing. This represents the raw spectral vector currently acquired in real time. This represents the average spectrum obtained during the offline training phase. The purpose of centering is to shift the original spectral vector to a coordinate system with the average spectrum as the origin, thereby eliminating the influence of the DC component on subsequent projection calculations and enabling the coefficients obtained from the projection to more accurately reflect the main variation patterns of the spectrum.

[0037] Furthermore, the centered spectrum is projected onto the spectral feature basis vector library for calculation to obtain weight coefficients. During this process, the spectral feature basis vector library... The transpose matrix and the centered spectrum Multiply by the weighting coefficients as follows: ; in, Indicates by A column vector composed of weight coefficients The centralized spectrum indicates that the centralized spectrum is in the first... Projection components along the directions of each eigenvector, This represents the transpose matrix of the spectral feature basis vector library. This represents the centered spectrum obtained from the previous processing step. The weighting coefficients are the compact representation of the centered spectrum in the low-dimensional space spanned by the spectral feature basis vector library, which can approximate the main variation patterns of the original spectrum with a data scale much smaller than the dimension of the original spectrum.

[0038] Finally, the weighting coefficients and their corresponding sensor identifiers are packaged and transmitted together to obtain the spectral compression coefficient result. During this process, each set of weighting coefficients... After generation, each data point is bound to the sensor identifier (Sensor ID) of the spectral sensor unit that acquired the corresponding original spectral vector. The data is then packaged according to a preset data frame format and uploaded to the main processor via the data path to obtain the spectral compression coefficient result. By leveraging the binding relationship between the sensor identifier and the weighting coefficients, the main processor can clearly identify the screen space location corresponding to each set of weighting coefficients, facilitating accurate subsequent reconstruction processing. Compared to directly transmitting the original 81 floating-point spectral data points, this process only requires transmitting k weighting coefficients and the sensor identifier, reducing the data volume by approximately 90%. This significantly reduces the load on the internal bus and processing latency, providing a feasible data throughput guarantee for high-frequency real-time calibration in resource-constrained embedded systems.

[0039] In step S103, the same spectral feature basis vector library as used in the compression encoding stage is used to perform a reverse calculation on the spectral compression coefficient result, restoring it to a complete spectral curve, thereby obtaining the reconstructed screen spectral data. Since the spectral feature basis vector library has captured the main components of the display spectrum, the reconstruction process can achieve a balance between data compression and accuracy preservation, ensuring that the reconstructed screen spectral data and the original screen spectral data maintain a high degree of consistency at the chromaticity calculation level, and can be used as input data for subsequent ambient light cancellation calculations.

[0040] In one embodiment, step S103 includes: The spectral compression coefficient results are analyzed to obtain the spectral weighting coefficients; Based on the spectral feature basis vector library, a linear combination operation is performed on the spectral weight coefficients to obtain the spectral combination result; The combined spectral result is added to the average spectrum at corresponding positions to obtain the reconstructed screen spectral data of the corresponding sensor.

[0041] In this embodiment, it should be noted that this step can be regarded as the reverse operation of the online compression process in step S102, which relies on the same spectral feature basis vector library at the receiving end (e.g., the main processor) as at the compression end. and the average spectrum It restores the compressed, transmitted data into a complete spectral curve that can be used for precise calculations.

[0042] The spectral compression coefficient result is parsed to obtain the spectral weighting coefficient. In this process, after receiving the spectral compression coefficient result uploaded from the data path, the main processor unpacks the data frame according to a pre-agreed data frame format, separating the sensor identifier (Sensor ID) from the weighting coefficient. The sensor identifier is used to determine the spatial location of the specific spectral sensor unit corresponding to this set of data on the display. The weighting coefficient is then extracted as the spectral weighting coefficient, denoted as... .in, This represents a column vector consisting of k weight coefficients. The value represents the j-th projection component calculated by the compression end, k represents the number of eigenvectors contained in the spectral feature basis vector library, and the superscript T indicates matrix transpose. The spectral weight coefficients carry all the projection information of the original spectrum in the low-dimensional space spanned by the spectral feature basis vector library and are the direct input for subsequent reconstruction operations.

[0043] Furthermore, a linear combination operation is performed on the spectral weight coefficients based on the aforementioned spectral feature basis vector library to obtain the spectral combination result. In this process, the spectral feature basis vector library stored at the receiving end, which is completely identical to that at the compression end, is utilized. The linear combination operation is performed by multiplying each weight component of the spectral weighting coefficients with the corresponding eigenvector in the spectral feature basis vector library one by one and summing them at their corresponding positions, thereby obtaining the spectral combination result. The linear combination operation can be characterized by the following expression: ; in, This represents the spectral combination result obtained from linear combination operations. This represents the spectral feature basis vector library. This represents the column vector formed by the spectral weighting coefficients. This represents the j-th eigenvector in the spectral feature basis vector library. The weight coefficient corresponding to the j-th eigenvector is represented by , and k represents the total number of eigenvectors participating in the linear combination. The spectral combination result physically corresponds to the deviation component of the original spectrum from the average spectrum, that is, the approximate restored value corresponding to the projection of the centered spectrum into the low-dimensional subspace in step S102, which itself does not yet constitute a complete screen spectral curve.

[0044] Finally, the spectral combination result is added to the average spectrum at corresponding positions to obtain the reconstructed screen spectral data of the corresponding sensor. In this process, the spectral combination result obtained in the previous process is... The average spectrum pre-stored during the offline training phase The DC component of the original spectral vector before compression is recovered by summing the corresponding positions of the spectral wavelengths, thus obtaining the complete reconstructed screen spectral data. The calculation expression is as follows: ; in, This refers to the reconstructed screen spectral data. This represents the average spectrum obtained during the offline training phase. This represents the spectral feature basis vector library. This represents the spectral weighting coefficient obtained from the spectral compression coefficient result. Combined with the spatial location information determined by the sensor identifier at the beginning of the processing, each set of reconstructed screen spectral data can correspond one-to-one with the corresponding spectral sensor unit, thereby forming a screen spectral map with complete spatial distribution attributes on the main processor side.

[0045] Experiments have verified that, for both the display's own radiation spectrum and the ambient light spectrum, when the number of feature vectors selected in the spectral feature basis vector library is k=5, the reconstructed screen spectral data obtained from the above processing procedure... Compared with the original spectral vector before compression The color difference calculated under the uniform color space (CIE 1976 LAB) published by the International Commission on Illumination in 1976. It can be stably controlled within 0.8. Among them, The Euclidean distance, representing the chromaticity coordinates between two spectra under the CIE 1976 LAB colorimetric system, is a standard quantitative indicator for measuring the difference in perceived color. A smaller value indicates a closer similarity between the two colors. The aforementioned color difference level provides sufficient chromaticity accuracy for most color-sensitive professional applications, thus achieving high-fidelity reconstruction while significantly reducing data transmission volume. This provides a reliable spectral data foundation for subsequent ambient light cancellation calculations and color gamut mapping processing.

[0046] In step S104, the screen surface reflectivity is an inherent physical property of the display, which can be pre-measured and stored in the system during the factory calibration stage through a corresponding calibration procedure. The ambient light spectral data is converted based on the screen surface reflectivity to obtain the ambient light component generated by screen surface reflection; then, the ambient light component is removed from the reconstructed screen spectral data, thereby accurately separating the spectral information of the screen's own radiation from the mixed spectrum to obtain the pure screen spectral data. This step achieves ambient light stripping based on a physical optics model, avoiding the accuracy loss caused by traditional solutions based on color temperature estimation or assumed models.

[0047] In one embodiment, step S104 includes: The calibration measurement spectrum is obtained, and the calibration measurement spectrum is calculated with the preset ambient light spectrum to obtain the screen surface reflectivity; The ambient light reflectance component is obtained by multiplying the screen surface reflectance with the ambient light spectral data. The screen spectral data is obtained by calculating the difference between the reconstructed screen spectral data and the ambient light reflection component.

[0048] In this embodiment, this step is one of the key processing steps in the entire color gamut calibration method. Its core purpose is to accurately separate the screen's own emission spectrum from the mixed spectrum corresponding to the reconstructed screen spectral data. This processing is based on a well-defined physical optical model, which assumes that the mixed spectrum obtained at the screen measurement point consists of two parts: the screen's own emission spectrum and the ambient light component reflected from the screen surface. This can be expressed as the following relationship: ; in, This represents the wavelength variable corresponding to the spectrum. This indicates the mixed spectrum obtained at the screen measurement point, corresponding to the reconstructed screen spectral data output in step S103; This represents the screen's own emission spectrum, corresponding to the screen spectral data ultimately required in this step; Represents the reflectivity of the screen surface, which is an inherent physical property of the display. This represents the ambient light spectrum, corresponding to the ambient light spectrum data obtained in step S101.

[0049] A calibration measurement spectrum is acquired, and this spectrum is then compared with a preset ambient light spectrum to calculate the screen surface reflectance. During this process, the screen surface reflectance... The data is obtained beforehand during the factory calibration phase of the monitor. The specific calibration process involves placing the monitor to be calibrated in a darkroom environment. The purpose of this darkroom environment is to eliminate stray light interference with the calibration results. In this darkroom environment, the monitor is controlled to display a pure black image, so that the monitor's own light emission components... During the calibration phase, the value is approximately zero; simultaneously, an ambient light beam with a known spectral distribution is used to illuminate the screen surface of the display, and this known spectrum is the preset ambient light spectrum. Under the above conditions, the measuring device collects the spectrum reflected from the screen surface, and the collected spectrum is the calibration measurement spectrum. Since the screen's own light emission component is zero at this time, the aforementioned physical optics model, under calibration conditions, is modified into the following relationship: ; in, This indicates the calibration measurement spectrum collected in the darkroom environment. This represents the screen surface reflectivity to be determined. This represents a preset, known ambient light spectrum used for calibration. Based on the aforementioned degradation model, the screen surface reflectivity can be accurately obtained by dividing the calibrated measurement spectrum by the preset ambient light spectrum at the corresponding wavelength position. The calculation expression is as follows: ; The meanings of the variables are the same as above. The obtained screen surface reflectivity... The inherent physical parameters of the display are stored in the system and can be repeatedly retrieved in subsequent operation phases, eliminating the need to measure them again during each calibration.

[0050] Furthermore, the screen surface reflectivity is multiplied by the ambient light spectral data to obtain the ambient light reflectance component. During this process, the ambient light spectrum corresponding to the current moment is extracted from the packet header position of the ambient light spectral data output in step S101. Then compare it with the screen surface reflectivity pre-stored in the system. The ambient light reflection component is obtained by performing multiplication operations point by point according to the corresponding position of the spectral wavelength. The calculation expression is as follows: ; in, This represents the ambient light reflection component. This represents the screen surface reflectivity obtained from factory calibration. This refers to the ambient light spectrum obtained in step S101, which is strictly synchronized with the reconstructed screen spectral data. The ambient light reflection component physically corresponds to the portion of energy distribution in the mixed spectrum that is reflected from the screen surface. Its shape is influenced by both the external ambient spectrum and the reflective properties of the screen surface material.

[0051] Finally, the difference between the reconstructed screen spectral data and the ambient light reflectance component is calculated to obtain the screen spectral data. In this process, based on the aforementioned physical optics model, the reconstructed screen spectral data output in step S103 is... The ambient light reflection component obtained in the previous processing step By performing a difference operation on the corresponding positions of the spectral wavelengths, the energy contribution corresponding to the ambient light reflection component is extracted from the mixed spectrum, thus obtaining the screen spectral data corresponding to the screen's own emission. The calculation expression is as follows: ; in, This refers to the screen spectral data obtained after ambient light cancellation calculation, which is also known as the pure screen spectral data. This represents the reconstructed screen spectral data output from step S103; This represents the ambient light reflection component. The obtained screen spectral data has eliminated the interference energy introduced by ambient light reflection through the screen surface, retaining only the spectral information corresponding to the screen's own emission, and can be used as input data for subsequent dynamic color gamut mapping processing.

[0052] By employing the above processing steps, this embodiment counteracts the impact of ambient light at the physical source level. Compared with traditional schemes that rely on color temperature estimation and software simulation compensation, this method is based on the inherent physical property of screen surface reflectivity and the real quantitative data obtained through synchronous acquisition of ambient light spectral data. It does not require any hypothetical models regarding the distribution of ambient light, thus offering significant improvement in accuracy. In particular, it can maintain a stable and reliable ambient light stripping effect, especially in viewing environments composed of complex non-standard light sources.

[0053] In step S105, the target color gamut is the color space standard that the display should conform to after calibration, which can be represented, for example, by a corresponding color space transformation matrix. The pure screen spectral data is converted into parameters representing the actual color gamut state of the current display through chromaticity calculations. These parameters, along with the parameters corresponding to the target color gamut, are then used as inputs and solved through a pre-trained mapping model to obtain the color gamut correction matrix data that can adjust the current color gamut of the display to the target color gamut. The color gamut correction matrix data reflects the transformation relationships required to correct the display's color output.

[0054] In one embodiment, step S105 includes: The tristimulus values ​​of the measurement point are obtained by performing integral transformation calculation on the spectral data of the pure screen using a preset chromaticity function; The local color transformation matrix is ​​obtained by performing multiple linear regression calculations on the tristimulus values ​​of the measurement points and the corresponding three primary color driving values. Extract the transformation matrix of the target color gamut and combine it with the local color transformation matrix to obtain the model input sequence data; The input sequence data of the model is input into a preset feedforward neural network model, and the output sequence elements are generated. Extract the sequence elements and recombine the matrix to obtain the color gamut correction matrix data.

[0055] In this embodiment, the tristimulus values ​​of the measurement point are obtained by integrating and transforming the spectral data of the clean screen using a preset chromaticity function. In this process, the preset chromaticity function adopts the CIE 1931 Standard Colorimetric Observer Functions, promulgated by the International Commission on Illumination in 1931. This function consists of three wavelength-dependent curves. The components represent the standard physiological response characteristics of the human eye to different wavelengths of visible light in three independent color channels. The pure screen spectral data output from step S104... By multiplying the values ​​by the three curves mentioned above at their corresponding wavelengths within the visible light band and then integrating, the tristimulus values ​​(X, Y, Z) of the measurement point in the CIE 1931 XYZ Color Space can be obtained. The calculation expression is as follows: ; ; ; in, This represents the wavelength variable corresponding to the spectrum, with the integration limits of 380nm to 780nm corresponding to the visible light band. This represents the clean screen spectral data output by step S104; These represent the three response curves of the CIE 1931 standard colorimetric observer function; X, Y, and Z are the three components of the obtained tristimulus values ​​at the measurement point; K represents the normalization constant corresponding to the integral. These tristimulus values ​​at the measurement point physically quantify the human eye's color perception induced by the screen's radiation spectrum into a set of three-dimensional values, thereby mapping the pure screen spectral data from the spectral domain to the color perception domain, facilitating further analysis of subsequent color parameters.

[0056] Furthermore, a multivariate linear regression calculation is performed on the tristimulus values ​​of the measurement point and the corresponding primary color driving values ​​to obtain a local color transformation matrix. In this process, the primary color driving values ​​refer to the driving signal values ​​of the red, green, and blue primary color channels applied to the display at the current measurement point, denoted as... The tristimulus values ​​(X, Y, Z) at the measurement points are used as response variables, and the three primary color driving values ​​are used as driving variables. As explanatory variables, a multiple linear regression method is applied to fit the data and solve for the transformation matrix that maps the three primary color driving values ​​to the tristimulus values ​​of the measurement points. This transformation matrix is ​​the local color transformation matrix that characterizes the current local color gamut state of the screen. Mathematically, it forms a 3x3 real number matrix, and the color space mapping relationship it represents is as follows: ; in, This represents the local color transformation matrix obtained by fitting multiple linear regression, which is also the RGB-to-XYZ transformation matrix under the current screen state; (X, Y, Z) represents the primary color driving values ​​at the current measurement point; (X, Y, Z) represents the tristimulus values ​​at the measurement point corresponding to the primary color driving values. The local color transformation matrix... It fully characterizes the actual color gamut state of the current screen at the measurement point, and the specific values ​​of its matrix elements reflect the color reproduction characteristics of the display under the current aging level and environmental conditions.

[0057] Furthermore, the transformation matrix of the target color gamut is extracted and combined with the local color transformation matrix to obtain the model input sequence data. In this process, the target color gamut refers to the standard color space (e.g., sRGB, Adobe RGB, or DCI-P3, etc.) that the display is required to achieve in this calibration. The transformation matrix of the target color gamut is the standard RGB-to-XYZ transformation matrix defined by its specification for that standard color space. The transformation matrix of the target color gamut is extracted from the standard color gamut parameters pre-stored in the system. and its matrix elements are compared with the local color transformation matrix obtained in the previous processing step. The matrix elements are flattened into one-dimensional vectors in either row-major or column-major order. and Each matrix is ​​3x3. The flattening operation yields two one-dimensional vectors of length 9. These two vectors are then concatenated end-to-end to form a single one-dimensional vector of length 18, which serves as the model input sequence data. This model input sequence data simultaneously carries both the current actual color gamut state of the screen and the target color gamut reference state, providing complete input conditions for subsequent model inference.

[0058] Furthermore, the input sequence data of the model is input into a preset feedforward neural network model, which outputs sequence elements. In this process, the feedforward neural network model is a lightweight network structure, with its input layer containing 18 neurons, corresponding to the length of the input sequence data of the model, each receiving the local color transformation matrix. Transformation matrix with the target color gamut The flattened matrix yields all 18 matrix elements; its output layer contains 9 neurons, corresponding to the 9 matrix elements of the color gamut correction matrix to be obtained. Before the system is put into use, the feedforward neural network model undergoes offline supervised training using massive amounts of screen color gamut variation data and the corresponding optimal correction solutions as training samples. The loss function used in the training phase is a weighted combination of two indices: corrected color difference and uniformity, thus guiding the network to simultaneously consider the two core performance dimensions of color reproduction accuracy and display uniformity during parameter optimization. In the real-time calibration phase, the feedforward neural network model performs a forward propagation operation for each set of input sequence data, instantly outputting 9 values ​​from its output layer; these 9 values ​​constitute the sequence elements. By employing the above-mentioned AI-based solution method, this process avoids the problems of traditional iterative algorithms potentially getting stuck in local optima or experiencing slow convergence, thereby ensuring both the speed and quality of the solution.

[0059] Finally, the sequence elements are extracted and the matrix is ​​reshaped to obtain the color gamut correction matrix data. In this process, the nine sequence elements output by the feedforward neural network model from the previous process are reshaped in the reverse order of the flattening operation, i.e., the nine one-dimensional sequence elements are sequentially filled into the corresponding positions of a 3x3 matrix, thereby obtaining the color gamut correction matrix data. The color gamut correction matrix data completely characterizes the local color transformation matrix. The transformation matrix that maps the current screen's actual color gamut to the target color gamut. The specific function of all the color transformation relationships required for the corresponding target color gamut is to ensure that, when subsequently applied to image signals, the three primary color driving values ​​output by the display satisfy the following transformation relationships: ; in, This represents the color gamut correction matrix data obtained in this step; This represents the three primary color driving values ​​originally carried by the input image signal; This represents the output primary color driving values ​​after correction using the color gamut correction matrix data. With the help of this color gamut correction matrix data, the system can accurately map the display's color gamut to the range defined by the target color gamut at the pixel level, thus providing a decision-making basis for subsequent driving signal generation and iterative feedback processing.

[0060] In step S106, the triggering conditions may include various events such as changes in ambient light, device wake-up, timed refresh, or external calibration commands. When the system determines that the triggering conditions are met, the color gamut correction matrix data is loaded into the display's drive path to perform real-time correction on the image signal, thereby generating a corresponding drive signal. Simultaneously, the actual display effect after correction is measured synchronously, and the measurement result is compared with the reference value corresponding to the target color gamut. Based on the comparison result, a new round of iterative correction process is initiated until the display effect reaches the preset accuracy requirements, thus obtaining the color gamut calibration result. Through the above iterative feedback mechanism, this method can continuously maintain the stability and accuracy of color output during display operation.

[0061] In one embodiment, step S106 includes: The color gamut correction matrix data and the corresponding pixels are subjected to time-series dithering to obtain the verification image data. The standard color blocks in the verification screen data are measured synchronously to obtain the actual verification spectral data; The actual verification spectral data are converted into color values ​​and compared and calculated to obtain color difference and uniformity evaluation index data; The color difference and uniformity evaluation index data are verified and rolled back to obtain the color gamut calibration results.

[0062] In this embodiment, before initiating this step, the preset trigger conditions are first determined. These preset trigger conditions may include any one or more of the following events: First, an ambient light event, emitted by the ambient light spectrum sensor, triggered when the detected change in ambient light spectrum or intensity exceeds a set threshold (e.g., greater than 10%); second, a power-on / wake-up event, automatically triggering a quick calibration when the display is woken from sleep mode; third, a timer event, automatically triggering a refresh calibration every 4 hours of continuous operation by the system's built-in watchdog timer to compensate for long-term thermal drift of the panel; and fourth, a user / host command, responding to a forced calibration command issued by the user or computer host. Only when the preset trigger conditions are met is the corresponding spectrum sensor array and processor woken up as needed to enter the execution flow of this step, thereby achieving a balance between high performance and low power consumption while ensuring calibration accuracy.

[0063] First, the color gamut correction matrix data and the corresponding pixels are subjected to timing dithering to obtain verification image data. During this process, the color gamut correction matrix data output in step S105... The data is loaded into the hardware logic registers integrated in the display timing controller (T-CON) or application-specific gate array (FPGA) / application-specific integrated circuit (ASIC) for each pixel in the image to be displayed. Pixel-level matrix multiplication is performed based on the following expression: ; in, This represents the color gamut correction matrix data. This represents the original three primary color driving values ​​of the pixel. This represents the high-precision primary color driving values ​​obtained after pixel matrix operations. Since the primary color driving values ​​obtained from the above matrix operations are usually presented in high-precision floating-point form, they need to be quantized to the bit depth supported by the display driver (e.g., 10-bit). During the quantization process, in order to maintain smooth color gradation transitions and avoid color banding even when quantization errors are unavoidable, this processing introduces a temporal dithering algorithm. This algorithm distributes the quantization error according to a preset rule across the spatiotemporal domain composed of time and spatial dimensions, making the quantization error more visually averaged. Based on this, the system does not wait for the arrival of a natural image but actively applies the color gamut correction matrix data to a set of pre-generated verification pattern pixels. After processing by the pixel-level matrix operations and the temporal dithering algorithm, the data is output to the display panel, thereby generating the verification image data. The verification screen data corresponds to a screen containing a series of standard color blocks (e.g., pure red, pure green, pure blue, white, and multiple intermediate colors) covering the key areas of the target color gamut. These standard color blocks are used to subsequently verify the actual display effect produced by the current color gamut correction matrix data.

[0064] Furthermore, the standard color blocks in the verification screen data are simultaneously measured to obtain actual verification spectral data. During this process, as the verification screen data is driven to the display panel and actually displayed, a synchronous trigger signal strictly corresponding to the display time of the verification screen is sent to the spectral sensor array. This causes the spectral sensor array to perform a second spectral acquisition of the standard color blocks currently displayed on the screen, thereby obtaining the corrected actual display spectrum, which is recorded as the actual verification spectral data. .in, This represents the wavelength variable corresponding to the spectrum. This represents the actual spectral power distribution corresponding to each of the standard color patches, collected by the spectral sensor array after the current color gamut correction matrix data has been practically applied. The actual verification spectral data is an objective physical measure of the current correction effect, and its acquisition sequence strictly corresponds to the verification screen data, thereby ensuring the effectiveness of subsequent evaluations.

[0065] Furthermore, the actual verification spectral data is converted into chromaticity values ​​and compared and calculated to obtain color difference and uniformity evaluation index data. During this process, the actual verification spectral data... After being integrated using the CIE 1931 standard chromaticity observer function, the colorimetric values ​​are further mapped to the CIE 1976 LAB chromaticity system. These chromaticity values ​​are then compared with the theoretical target chromaticity values ​​corresponding to each standard color patch within the target color gamut. Based on this, the system further statistically obtains three types of color difference and uniformity evaluation index data. The first is the average color difference, defined as the average of the color differences of all measured color patches, used to reflect the overall accuracy level of color reproduction after correction. The second is the maximum color difference, defined as the maximum value among all measured color patches, used to reflect the degree of color deviation in the worst-case scenario. The third is the uniformity difference, defined as the chromaticity difference between the corresponding chromaticity values ​​when displaying the same standard color patch in different areas, used to reflect the color consistency of the screen in spatial dimensions. These color difference and uniformity evaluation index data will serve as the objective basis for decisions in the next processing step.

[0066] Finally, the color difference and uniformity evaluation index data are verified and rolled back to obtain the color gamut calibration result. During this process, the system selects one of the following three processing paths according to the current correction effect reflected by the color difference and uniformity evaluation index data, following a preset decision logic: The first path is successful calibration. When the average color difference is less than 2.0 and the maximum color difference is less than 3.5 (the above thresholds are configurable), the system determines that the calibration is successful, marks the currently applied color gamut correction matrix data as valid and stores it, and then enters a low-power monitoring state to wait for the next preset trigger condition to be met; The second approach requires fine-tuning. When the color difference and uniformity evaluation index data do not fully meet the above threshold requirements, but are very close (e.g., the average color difference is less than 3.0), the system initiates a micro-iteration process. The aforementioned feedforward neural network model uses the color gamut state corresponding to the current actual verification spectral data as a new input to resolve a small incremental correction matrix. This matrix is ​​then combined with the original color gamut correction matrix data. The new color gamut correction matrix data is used as the basis for the next round of verification and re-enters the aforementioned processing steps. The micro-iteration process typically enables the color difference and uniformity evaluation index data to meet the above threshold requirements within 1 to 2 iterations, thereby achieving rapid convergence. The third path is calibration failure. When the color difference and uniformity evaluation index data still fail to meet the above threshold requirements after a preset number of iterations (e.g., 3 times) of the micro-iteration process, or when the color difference and uniformity evaluation index data reflect a very large color difference after the initial calibration, the system determines that the calibration is in an abnormal state, triggers the rollback mechanism, rolls back the current color gamut correction matrix data to the most recently marked valid color gamut correction matrix data, and sends an error diagnosis report to the upper-layer application through the host interface, indicating that the current display may need hardware maintenance or deep calibration.

[0067] Based on the judgment conclusions generated by the three processing paths described above and the final applied color gamut correction matrix data, the color gamut calibration result output in this step is constituted. With the help of the aforementioned verification and rollback mechanisms, the system possesses self-verification and continuous optimization capabilities, enabling it to cope with various complex situations that may occur during long-term use of the monitor, such as panel aging and drastic environmental changes, thereby ensuring the stability and reliability of the monitor's color output during long-term use.

[0068] In summary, this invention integrates a spectral sensor array into the pixel gaps of a display as an on-chip spectrometer, enabling in-situ measurement of the screen's own emission. This avoids the mixed light interference introduced by multiple reflections and refractions through optical films in traditional external instruments or bezel-integrated solutions, ensuring high physical fidelity of the screen spectral data during acquisition. Simultaneously, by using a pre-defined spectral feature basis vector library to compress and encode the original screen spectral data, the high-dimensional spectral data is transformed into a small number of weight coefficients and sensor identifiers for transmission. This significantly reduces the load and processing latency of the internal bus, providing a feasible guarantee of data throughput for high-frequency real-time calibration in resource-constrained embedded systems. Furthermore, during the reconstruction phase, the color difference between the reconstructed and original screen spectral data can be stably controlled at a low level, balancing data compression and reconstruction accuracy.

[0069] Furthermore, relying on the pre-calibrated screen surface reflectivity and synchronously acquired ambient light spectral data, the ambient light reflection component is accurately stripped at the physical optics model level. Compared with traditional solutions that rely on color temperature estimation and software simulation compensation, this method has a significant improvement in accuracy, especially in viewing environments with complex non-standard light sources, where it can maintain a stable and reliable ambient light stripping effect. In addition, by using a pre-set feedforward neural network model to jointly solve the local color conversion matrix and the target color gamut conversion matrix, the method can intelligently and quickly output the optimal color gamut correction matrix data, avoiding the problem that traditional iterative algorithms may get stuck in local optima or converge slowly.

[0070] Finally, by simultaneously verifying the actual display effect corresponding to the color gamut correction matrix data and evaluating color difference and uniformity, and in conjunction with micro-iteration and rollback mechanisms, an intelligent closed loop of measurement, decision-making, execution, and verification is constructed. This enables the system to have self-verification and continuous optimization capabilities, and can cope with complex situations such as panel aging and drastic environmental changes that may occur during long-term use of the display. In summary, the embodiments of the present invention fundamentally solve the technical problems of existing color gamut calibration schemes, such as the difficulty in removing ambient light interference in situ and the limited accuracy of dynamic calibration. It achieves the technical effect of full color gamut, high precision, and dynamic real-time calibration in complex and ever-changing lighting environments, providing reliable color reproduction assurance for color-sensitive industries.

[0071] Combination Figure 2 As shown, Figure 2 This is a schematic block diagram of a display color gamut calibration device provided in an embodiment of the present invention. The display color gamut calibration device 200 includes: The data acquisition unit 201 is used to acquire raw screen spectral data of the display through a spectral sensor array integrated into the pixel gap of the display and to synchronously acquire ambient light spectral data around the display through an ambient light spectral sensor. The data compression unit 202 is used to compress and encode the original screen spectral data according to a preset spectral feature basis vector library to obtain the spectral compression coefficient result; The spectral reconstruction unit 203 is used to reconstruct the spectral compression coefficient result based on the spectral feature basis vector library to obtain reconstructed screen spectral data; The data calculation unit 204 is used to perform ambient light cancellation calculation on the reconstructed screen spectral data based on the preset screen surface reflectivity and the ambient light spectral data to obtain pure screen spectral data. Data correction unit 205 is used to perform dynamic color gamut mapping on the pure screen spectral data according to the preset target color gamut to obtain color gamut correction matrix data; The data output unit 206 is used to generate a driving signal based on the color gamut correction matrix data according to a preset trigger condition and iteratively feed back to obtain the color gamut calibration result.

[0072] In this embodiment, the data acquisition unit 201 acquires the original screen spectral data of the display through a spectral sensor array integrated into the pixel gap of the display and simultaneously acquires the ambient light spectral data around the display through an ambient light spectral sensor; the data compression unit 202 compresses and encodes the original screen spectral data according to a preset spectral feature basis vector library to obtain a spectral compression coefficient result; the spectral reconstruction unit 203 reconstructs the spectral compression coefficient result according to the spectral feature basis vector library to obtain reconstructed screen spectral data; the data calculation unit 204 performs ambient light cancellation calculation on the reconstructed screen spectral data based on a preset screen surface reflectance using the ambient light spectral data to obtain clean screen spectral data; the data correction unit 205 performs dynamic color gamut mapping on the clean screen spectral data according to a preset target color gamut to obtain color gamut correction matrix data; and the data output unit 206 generates a driving signal using the color gamut correction matrix data based on a preset trigger condition and iteratively feeds back to obtain a color gamut calibration result.

[0073] In one embodiment, the data acquisition unit 201 is specifically used for: When a synchronous acquisition task is received, the original screen spectral data and ambient light spectral data frames are acquired respectively through the spectral sensor array and the ambient light spectral sensor within the same time window. The ambient light spectral data frame is marked as an environmental reference value in the header of the original screen spectral data.

[0074] In one embodiment, the data compression unit 202 is specifically used for: Construct a spectral database, obtain the average spectrum from the spectral database, and then calculate the covariance matrix of the spectral database based on the average spectrum; The covariance matrix is ​​subjected to eigenvalue decomposition to obtain the eigenvalues ​​and corresponding eigenvectors of the covariance matrix. Select the eigenvectors corresponding to the first preset number of largest eigenvalues ​​from the eigenvalues ​​to obtain the spectral feature basis vector library; The average spectrum is subtracted from each original spectral vector in the original screen spectral data to obtain the centered spectrum. The weighting coefficients are obtained by projecting the centered spectrum onto the spectral feature basis vector library and calculating the weighting coefficients. The weighting coefficients and their corresponding sensor identifiers are packaged and transmitted together to obtain the spectral compression coefficient results.

[0075] In one embodiment, the spectral reconstruction unit 203 is specifically used for: The spectral compression coefficient results are analyzed to obtain the spectral weighting coefficients; Based on the spectral feature basis vector library, a linear combination operation is performed on the spectral weight coefficients to obtain the spectral combination result; The combined spectral result is added to the average spectrum at corresponding positions to obtain the reconstructed screen spectral data of the corresponding sensor.

[0076] In one embodiment, the data calculation unit 204 is specifically used for: The calibration measurement spectrum is obtained, and the calibration measurement spectrum is calculated with the preset ambient light spectrum to obtain the screen surface reflectivity; The ambient light reflectance component is obtained by multiplying the screen surface reflectance with the ambient light spectral data. The screen spectral data is obtained by calculating the difference between the reconstructed screen spectral data and the ambient light reflection component.

[0077] In one embodiment, the data correction unit 205 is specifically used for: The tristimulus values ​​of the measurement point are obtained by performing integral transformation calculation on the spectral data of the pure screen using a preset chromaticity function; The local color transformation matrix is ​​obtained by performing multiple linear regression calculations on the tristimulus values ​​of the measurement points and the corresponding three primary color driving values. Extract the transformation matrix of the target color gamut and combine it with the local color transformation matrix to obtain the model input sequence data; The input sequence data of the model is input into a preset feedforward neural network model, and the output sequence elements are generated. Extract the sequence elements and recombine the matrix to obtain the color gamut correction matrix data.

[0078] In one embodiment, the data output unit 206 is specifically used for: The color gamut correction matrix data and the corresponding pixels are subjected to time-series dithering to obtain the verification image data. The standard color blocks in the verification screen data are measured synchronously to obtain the actual verification spectral data; The actual verification spectral data are converted into color values ​​and compared and calculated to obtain color difference and uniformity evaluation index data; The color difference and uniformity evaluation index data are verified and rolled back to obtain the color gamut calibration results.

[0079] Since the embodiments of the apparatus and the embodiments of the method correspond to each other, please refer to the description of the embodiments of the method for the embodiments of the apparatus, which will not be repeated here.

[0080] This invention also provides a computer-readable storage medium storing a computer program thereon, which, when executed, can perform the steps provided in the above embodiments. The storage medium may include various media capable of storing program code, such as a USB flash drive, a portable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0081] This invention also provides a computer device, which may include a memory and a processor. The memory stores a computer program, and when the processor calls the computer program in the memory, it can implement the steps provided in the above embodiments. Of course, the computer device may also include various network interfaces, a power supply, a graphics card, etc., to utilize the graphics card's performance to operate the model, such as for inference and training.

[0082] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the systems disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the descriptions are relatively simple; relevant parts can be referred to in the method section. It should be noted that those skilled in the art can make various improvements and modifications to this application without departing from the principles of this application, and these improvements and modifications also fall within the protection scope of the claims of this application.

[0083] It should also be noted that, in this specification, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover a 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. Without further limitations, 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 said element.

Claims

1. A method for calibrating the color gamut of a display, characterized in that, include: The display acquires raw screen spectral data by integrating a spectral sensor array into the pixel gaps of the display and simultaneously acquires ambient light spectral data around the display by using an ambient light spectral sensor. The original screen spectral data is compressed and encoded according to a preset spectral feature basis vector library to obtain the spectral compression coefficient result; The spectral compression coefficient results are reconstructed based on the spectral feature basis vector library to obtain reconstructed screen spectral data. Based on the preset screen surface reflectivity, ambient light cancellation calculation is performed on the reconstructed screen spectral data using the ambient light spectral data to obtain clean screen spectral data; Dynamic color gamut mapping is performed on the pure screen spectral data according to the preset target color gamut to obtain color gamut correction matrix data; Based on preset triggering conditions, a driving signal is generated using the color gamut correction matrix data and iteratively fed back to obtain the color gamut calibration result. The step of dynamically mapping the pure screen spectral data according to a preset target color gamut to obtain color gamut correction matrix data includes: performing integral transformation calculation on the pure screen spectral data using a preset chromaticity function to obtain tristimulus values ​​at measurement points; performing multivariate linear regression calculation on the tristimulus values ​​at measurement points and the corresponding primary color driving values ​​to obtain a local color transformation matrix; extracting the transformation matrix of the target color gamut and combining it with the local color transformation matrix to obtain model input sequence data; inputting the model input sequence data into a preset feedforward neural network model to output sequence elements; extracting the sequence elements and performing matrix recombination to obtain color gamut correction matrix data.

2. The display color gamut calibration method according to claim 1, characterized in that, The acquisition of raw screen spectral data of the display through a spectral sensor array integrated into the pixel gaps of the display and the synchronous acquisition of ambient light spectral data around the display through an ambient light spectral sensor include: When a synchronous acquisition task is received, the original screen spectral data and ambient light spectral data frames are acquired respectively through the spectral sensor array and the ambient light spectral sensor within the same time window. The ambient light spectral data frame is marked as an environmental reference value in the header of the original screen spectral data.

3. The display color gamut calibration method according to claim 1, characterized in that, The step of compressing and encoding the original screen spectral data according to a preset spectral feature basis vector library to obtain spectral compression coefficient results includes: Construct a spectral database, obtain the average spectrum from the spectral database, and then calculate the covariance matrix of the spectral database based on the average spectrum; The covariance matrix is ​​subjected to eigenvalue decomposition to obtain the eigenvalues ​​and corresponding eigenvectors of the covariance matrix. Select the eigenvectors corresponding to the first preset number of largest eigenvalues ​​from the eigenvalues ​​to obtain the spectral feature basis vector library; The average spectrum is subtracted from each original spectral vector in the original screen spectral data to obtain the centered spectrum. The weighting coefficients are obtained by projecting the centered spectrum onto the spectral feature basis vector library and calculating the weighting coefficients. The weighting coefficients and their corresponding sensor identifiers are packaged and transmitted together to obtain the spectral compression coefficient results.

4. The display color gamut calibration method according to claim 3, characterized in that, The step of reconstructing the spectral compression coefficient result based on the spectral feature basis vector library to obtain reconstructed screen spectral data includes: The spectral compression coefficient results are analyzed to obtain the spectral weighting coefficients; Based on the spectral feature basis vector library, a linear combination operation is performed on the spectral weight coefficients to obtain the spectral combination result; The combined spectral result is added to the average spectrum at corresponding positions to obtain the reconstructed screen spectral data of the corresponding sensor.

5. The display color gamut calibration method according to claim 1, characterized in that, The process involves calculating ambient light cancellation on the reconstructed screen spectral data based on a preset screen surface reflectance, using the ambient light spectral data to obtain clean screen spectral data, including: The calibration measurement spectrum is obtained, and the calibration measurement spectrum is calculated with the preset ambient light spectrum to obtain the screen surface reflectivity; The ambient light reflectance component is obtained by multiplying the screen surface reflectance with the ambient light spectral data. The screen spectral data is obtained by calculating the difference between the reconstructed screen spectral data and the ambient light reflection component.

6. The display color gamut calibration method according to claim 1, characterized in that, The process of generating a driving signal based on the color gamut correction matrix data using preset trigger conditions and iteratively feeding back to obtain the color gamut calibration result includes: The color gamut correction matrix data and the corresponding pixels are subjected to time-series dithering to obtain the verification image data. The standard color blocks in the verification screen data are measured synchronously to obtain the actual verification spectral data; The actual verification spectral data are converted into color values ​​and compared and calculated to obtain color difference and uniformity evaluation index data; The color difference and uniformity evaluation index data are verified and rolled back to obtain the color gamut calibration results.

7. A display color gamut calibration device, characterized in that, include: The data acquisition unit is used to acquire raw screen spectral data of the display through a spectral sensor array integrated into the pixel gap of the display and to synchronously acquire ambient light spectral data around the display through an ambient light spectral sensor. The data compression unit is used to compress and encode the original screen spectral data according to a preset spectral feature basis vector library to obtain the spectral compression coefficient result; The spectral reconstruction unit is used to reconstruct the spectral compression coefficient results based on the spectral feature basis vector library to obtain reconstructed screen spectral data. The data calculation unit is used to perform ambient light cancellation calculation on the reconstructed screen spectral data based on the preset screen surface reflectance and the ambient light spectral data to obtain clean screen spectral data. The data correction unit is used to perform dynamic color gamut mapping on the pure screen spectral data according to the preset target color gamut to obtain color gamut correction matrix data; The data output unit is used to generate a driving signal based on the color gamut correction matrix data according to a preset trigger condition and iteratively feed it back to obtain the color gamut calibration result; The data correction unit is specifically used to perform integral transformation calculation on the pure screen spectral data using a preset chromaticity function to obtain the tristimulus value of the measurement point; perform multivariate linear regression calculation on the tristimulus value of the measurement point and the corresponding three primary color driving values ​​to obtain the local color transformation matrix; extract the transformation matrix of the target color gamut and combine it with the local color transformation matrix to obtain the model input sequence data; input the model input sequence data into a preset feedforward neural network model to output sequence elements; Extract the sequence elements and recombine the matrix to obtain the color gamut correction matrix data.

8. A computer device, characterized in that, It includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the display color gamut calibration method as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the display color gamut calibration method as described in any one of claims 1 to 6.

Images