Dynamic feedback-based organic electroluminescence uniformity detection method and system

Abstract

The embodiment of the application provides a kind of based on dynamic feedback's organic electroluminescence uniformity detection method and system, belong to organic electroluminescence uniformity detection field;The method comprises after initializing detection system, and calibrating reference baseline, acquisition standard plate reference image;Design first sparse sampling template;First sampling is carried out using first sparse sampling template and reconstruction algorithm, and the reconstruction full-width image of organic electroluminescence device under each waveband is obtained, and first light emitting uniformity evaluation is carried out;Mark abnormal light emitting area, dynamically adjust sampling strategy, using second sparse sampling template carries out second sampling, updates the reconstruction full-width image under each waveband, and carries out second light emitting uniformity evaluation, and exports organic electroluminescence uniformity detection report.The application is improved by intelligent sparse sampling and closed-loop feedback technology Organic electroluminescence uniformity detection efficiency and accurately locate defect area, provide efficient solution for production quality control.

Description

Technical Field

[0001] This invention relates to the field of organic electroluminescence uniformity detection technology, and specifically to an organic electroluminescence uniformity detection method and system based on dynamic feedback. Background Technology

[0002] The uniformity of light emission in organic electroluminescent devices (OLEDs) is a core indicator determining display quality and device lifespan. Current mainstream detection technologies primarily rely on high-resolution area array imaging systems or point-scan spectrometers. Area array imaging captures the full-width brightness distribution using CCD / CMOS sensors, achieving complete spatial coverage, but is limited by the efficiency of massive data processing and crosstalk issues in multispectral channels. Point-scan technology, while possessing high spectral accuracy, suffers from the time consumption of mechanical displacement, making it difficult to meet online detection requirements. In recent years, compressed sensing theory has been introduced into this field, reducing data dimensionality through sparse sampling, for example, using static masks to reconstruct spectral encoding in a single sampling. However, existing compressed sensing schemes generally employ fixed sampling rate templates, decoupling the sampling point distribution from local image features. This leads to the loss of details in areas with high information entropy, while low-complexity areas exhibit sampling redundancy. Summary of the Invention

[0003] The purpose of this invention is to provide a method and system for detecting the uniformity of organic electroluminescence based on dynamic feedback. This method is used to dynamically detect the brightness, color, and spectral uniformity of organic light-emitting devices. Through intelligent sparse sampling and closed-loop feedback technology, it significantly improves the detection efficiency of organic electroluminescence uniformity and accurately locates defect areas, providing an efficient solution for production quality control.

[0004] To achieve the above objectives, embodiments of the present invention provide a method for detecting the uniformity of organic electroluminescence based on dynamic feedback, comprising: after initializing the detection system and calibrating a reference baseline, acquiring a standard plate reference image; calculating the entropy distribution used to measure the complexity and information richness of the image region based on the standard plate reference image, and designing a first sparse sampling template; performing a first sampling using the first sparse sampling template and a reconstruction algorithm to obtain a reconstructed full-frame image of the organic electroluminescent device in each band, and performing a first emission uniformity evaluation based on the reconstructed full-frame image in each band; marking abnormal emission regions based on the first emission uniformity evaluation result, dynamically adjusting the sampling strategy, and generating a second sparse sampling template; performing a second sampling using the second sparse sampling template to update the reconstructed full-frame image in each band, performing a second emission uniformity evaluation, and outputting an organic electroluminescence uniformity detection report.

[0005] Optionally, the calibration reference baseline includes: performing camera spatial response calibration, mask coding mode calibration, and spectral response calibration; and integrating the results of the camera spatial response calibration, the mask coding mode calibration, and the spectral response calibration through mathematical modeling and data fusion algorithms to obtain a spatial-spectral calibration matrix.

[0006] Optionally, the step of calculating the entropy distribution for measuring the complexity and information richness of an image region based on the standard plate reference image, and designing a first sparse sampling template, includes: dividing the standard plate reference image into regions according to a preset grid size; extracting image data of each band corresponding to the multispectral filter group for each divided region; converting the image data of each band into a two-dimensional matrix; and calculating the entropy value of each band in the divided region for measuring the complexity and information richness of the image region.

[0007] Optionally, the step of calculating the entropy distribution for measuring the complexity and information richness of image regions based on the standard plate reference image and designing the first sparse sampling template further includes: setting the checkerboard size parameter, generating a basic sampling template based on the arrangement rules of the checkerboard; calculating the sampling probability of the divided region based on the entropy value; and adding random sampling points to the basic sampling template using a random algorithm based on the sampling probability to obtain the first sparse sampling template.

[0008] Optionally, the step of using the first sparse sampling template and reconstruction algorithm to perform the first sampling to obtain the reconstructed full-frame image of the organic electroluminescent device in each band includes: driving the organic electroluminescent device to emit light, using the first sparse sampling template to perform multispectral coded sampling to obtain a multispectral coded image; preprocessing the multispectral coded image to construct an observation matrix; initializing the estimated value of the original image signal, and using the reconstruction algorithm to obtain the reconstructed full-frame image of each band.

[0009] Optionally, the first emission uniformity assessment includes: calculating the luminance standard deviation of the reconstructed full-frame image for each band; for RGB band images, calculating the chromaticity coordinates of each pixel, and calculating the chromaticity deviation of the reconstructed full-frame image based on the chromaticity coordinates of each pixel; calculating the spectral deviation index of the reconstructed full-frame image; and performing the first emission uniformity assessment based on the luminance standard deviation, the chromaticity deviation, and the spectral deviation index to obtain a first emission uniformity assessment report.

[0010] Optionally, the step of marking abnormal luminescence regions, dynamically adjusting the sampling strategy, and generating a second sparse sampling template based on the first luminescence uniformity evaluation result includes: setting a luminance standard deviation threshold, a chromaticity deviation threshold, and a spectral deviation index threshold for the organic electroluminescent device; comparing the uniformity index of each region in the first luminescence uniformity evaluation report with the luminance standard deviation threshold, the chromaticity deviation threshold, and the spectral deviation index threshold; marking regions exceeding any one of the luminance standard deviation threshold, the chromaticity deviation threshold, and the spectral deviation index threshold as abnormal luminescence regions; and recording the location information, band information, and out-of-tolerance value of each abnormal luminescence region.

[0011] Optionally, the step of marking abnormal emission regions, dynamically adjusting the sampling strategy, and generating a second sparse sampling template based on the first emission uniformity evaluation result further includes: performing encrypted sampling on the abnormal emission regions; adjusting the sampling rate, generating a new overall mask pattern based on the adjusted sampling rate, and optimizing the new overall mask pattern; checking whether the optimized new overall mask pattern meets the requirements, and if it does, using the new overall mask pattern as the second sparse sampling template.

[0012] Optionally, the step of using the second sparse sampling template to perform a second sampling, updating the reconstructed full-frame image under each band, and performing a second emission uniformity evaluation to output an organic electroluminescence uniformity detection report includes: fusing the data collected in the second sampling with the data collected in the first sampling; updating the reconstructed full-frame image under each band using a compressed sensing algorithm; comparing the brightness standard deviation, chromaticity deviation, and spectral deviation index of the second sampling with those of the first sampling, and determining whether convergence has occurred. If convergence has occurred, the iteration of the compressed sensing algorithm is stopped, and an organic electroluminescence uniformity detection report is output.

[0013] On the other hand, the present invention provides an organic electroluminescence uniformity detection system based on dynamic feedback, for implementing an organic electroluminescence uniformity detection method based on dynamic feedback. The system includes a control module, the control module including a memory, a processor and a computer program stored in the memory and executable on the processor, the processor executing the computer program to implement the organic electroluminescence uniformity detection method based on dynamic feedback.

[0014] The above technical solution significantly improves detection accuracy and efficiency through entropy-driven dynamic feedback control and spatial-spectral fusion calibration technology. It dynamically generates sparse sampling templates based on image entropy distribution (focusing on highly complex regions in the first round and intensively sampling abnormally luminous regions in the second round), increasing the detection rate of micro-area defects and reducing the false negative rate. Secondly, it integrates a three-dimensional evaluation system of brightness standard deviation, chromaticity deviation, and spectral deviation index to comprehensively capture problems such as uneven brightness, color shift, and spectral distortion. Simultaneously, it uses compressed sensing algorithms to iteratively update and reconstruct the entire image, reducing the amount of sampling data and shortening detection time while ensuring the accuracy of the full-image reconstruction.

[0015] Other features and advantages of the embodiments of the present invention will be described in detail in the following detailed description section. Attached Figure Description

[0016] The accompanying drawings are provided to further illustrate embodiments of the present invention and form part of the specification. They are used together with the following detailed description to explain the embodiments of the present invention, but do not constitute a limitation thereof. In the drawings: Figure 1 This is a flowchart of the organic electroluminescence uniformity detection process based on dynamic feedback.

[0017] Figure 2 This is a flowchart for generating the first luminescence uniformity evaluation report. Detailed Implementation

[0018] The following is in conjunction with the appendix Figure 1 - Appendix Figure 2 The specific implementation methods of the embodiments of the present invention will be described in detail below. It should be understood that the specific implementation methods described herein are only for illustrating and explaining the embodiments of the present invention, and are not intended to limit the embodiments of the present invention.

[0019] It should be noted that the acquisition, transmission, storage, use, and processing of data in the technical solution of this application all comply with the relevant provisions of national laws and regulations. In the embodiments of this application, certain existing industry solutions such as software, components, and models may be mentioned. These should be considered exemplary, intended only to illustrate the feasibility of implementing the technical solution of this application, and do not imply that the applicant has already used or necessarily used such solutions.

[0020] In the process of realizing this invention, the inventors of this application discovered that traditional methods use fixed sampling templates (such as uniform grids), which cannot dynamically adjust the distribution of sampling points according to the image content (complexity, information content). This leads to oversampling in simple areas, wasting resources, and undersampling in complex or critical areas, affecting reconstruction accuracy and detection efficiency. Furthermore, a single sampling evaluation makes it difficult to accurately locate and quantify abnormal luminescence areas (such as uneven brightness, color shift, spectral anomalies). Moreover, existing technologies lack a feedback mechanism for secondary focusing detection of abnormal luminescence areas, which easily leads to the omission of detailed defects or incomplete evaluation. Existing technologies rely on single sparse sampling to reconstruct the entire image. When the sampling rate is insufficient or the template design is unreasonable, the quality of the reconstructed image (especially abnormal luminescence areas) is difficult to guarantee, affecting the accuracy of uniformity evaluation (brightness, chromaticity, spectrum).

[0021] Example 1 Reference Figures 1-2 This is the first embodiment of the present invention, which provides a method for detecting the uniformity of organic electroluminescence based on dynamic feedback, comprising: S100: After initializing the detection system and calibrating the reference baseline, acquire a reference image of the standard plate.

[0022] Specifically, select a high-resolution, high-frame-rate, high-speed area array camera to ensure it can capture subtle changes in the organic electroluminescent device under different emission states. Mount the camera on a stable bracket, adjust its position and angle so that the lens is aligned with the center area of ​​the inspection stage, and ensure that the camera's optical axis is perpendicular to the inspection plane to avoid image distortion caused by viewing angle issues.

[0023] Furthermore, a programmable optical mask (such as a DMD, Digital Micromirror Device) is installed at the front of the camera lens. Ensure the DMD's pixel array matches the camera's resolution and can accurately modulate light according to a preset encoding pattern. Connect the DMD's control cable to the corresponding interface on the computer and install the DMD's driver and control software to flexibly control the switching of its mask pattern. Install a multispectral filter group (RGB+NIR / UV). Fix the filter group to the front of the camera lens, ensuring each filter fits tightly against the lens and that there is no light leakage or positional shift when switching between different filters. Simultaneously, calibrate the filter group, recording parameters such as the center wavelength and bandwidth of each filter to accurately distinguish different wavelength bands of light signals in subsequent data processing.

[0024] Furthermore, the organic electroluminescent device to be tested is carefully placed on the testing stage. The testing stage should have good flatness and stability to ensure that the organic electroluminescent device will not shift or vibrate during the testing process. Adjust the position of the organic electroluminescent device so that its emitting surface is in close contact with the plane of the testing stage and is perpendicular to the optical axis of the camera to ensure that light can enter the camera lens evenly.

[0025] Furthermore, based on the electrical characteristics and luminescence requirements of the organic electroluminescent device, a suitable driving power supply is selected, ensuring that its output voltage and current range meet the normal operating requirements of the device. High-precision wires are used to connect the power supply to the device's electrodes to avoid current fluctuations or signal interference caused by poor contact. Simultaneously, the control interface of the driving power supply is connected to a computer so that the power supply's output parameters can be precisely controlled via a program during the detection process.

[0026] Furthermore, a reference baseline is calibrated, which includes: camera spatial response calibration, mask coding mode calibration, and spectral response calibration; through mathematical modeling and data fusion algorithms, the results of camera spatial response calibration, mask coding mode calibration, and spectral response calibration are integrated to obtain a spatial-spectral calibration matrix.

[0027] Specifically, camera spatial response calibration is performed by placing a uniformly emitting reference plate on the testing stage. This reference plate should possess known spatial emission uniformity, with its brightness and chromaticity evenly distributed in space and matching the emission wavelength range of the device under test (DUT). The power supply to the reference plate is connected, and its brightness is adjusted to be close to the brightness of the DUT during normal operation. Multiple images of the reference plate are captured using a high-speed area-scan camera, maintaining consistent camera exposure time, gain, and other parameters for each shot to obtain multiple reference images. These image data are transferred to a computer, and image processing software is used to analyze the images, calculating the camera's pixel response values ​​at different locations, including brightness and chromaticity responses. By statistically analyzing these response values, a spatial response model of the camera is established, recording various parameters in the model, such as the pixel response non-uniformity coefficient and spatial correlation, to provide a basis for subsequent image correction.

[0028] Furthermore, mask coding pattern calibration is performed, and the DMD is controlled to display a series of known mask coding patterns. At the same time, the positions and parameters of the camera and reference board are kept unchanged, and the corresponding coded images are captured.

[0029] Furthermore, spectral response calibration is performed, driving each filter in the multispectral filter group to sequentially switch in front of the camera lens. After each switch, an image of the reference board is captured, and the spectral transmittance curve corresponding to each filter is recorded. Simultaneously, a high-precision spectrometer is used to measure the actual spectral emission curves of the reference board under different filters.

[0030] Furthermore, by combining image data captured by the camera with measurement results from a spectral analyzer, a spectral response model of the camera under different filters is established. The response sensitivity of each pixel in different wavelength bands, as well as parameters such as the crosstalk coefficient between different filters, are calculated. Using this calibration data, optical signals in different wavelength bands can be accurately separated and reconstructed in subsequent multispectral reconstruction processes, improving the accuracy of spectral detection.

[0031] Furthermore, the camera spatial response calibration results, mask coding mode calibration results, and spectral response calibration results are comprehensively analyzed. Through mathematical modeling and data fusion algorithms, these calibration data from different dimensions are integrated to generate a unified spatial-spectral calibration matrix. This calibration matrix can describe the comprehensive response characteristics of the camera under different spatial locations, different mask coding modes, and different spectral bands. In subsequent sparse sampling template design and image reconstruction processes, this calibration matrix can be used to accurately correct and compensate for the acquired sparse data, thereby improving the quality of the reconstructed full-frame image and the accuracy of uniformity detection.

[0032] Furthermore, a rigorously calibrated standard uniformly emitting reference plate is selected, whose brightness and spectral characteristics are highly uniform in space and similar to the luminescence characteristics of the organic electroluminescent device under test. The reference plate is placed on the testing stage and its driving power supply is connected, with the power supply parameters adjusted to achieve stable brightness. Multispectral sampling of the reference plate is performed using a high-speed area array camera, a programmable optical mask, and a multispectral filter array. A series of standard reference plate images in different spectral bands are captured according to a predetermined mask encoding mode and filter switching sequence. It is ensured that the camera's exposure time, gain, and other parameters remain stable during sampling and are consistent with the parameters used in the calibration process to guarantee data consistency and comparability.

[0033] Preferably, the scheme employs a high-speed area array camera, a programmable optical mask (DMD), and a multispectral filter array, which can improve spatial, spectral, and temporal resolution and ensure the capture of subtle luminescence changes in organic electroluminescent devices. Through triple calibration of spatial response, mask encoding mode, and spectral response, systematic errors are eliminated, and a unified spatial-spectral calibration matrix is ​​generated, providing a precise correction basis for subsequent sparse sampling and image reconstruction. A uniform luminescence reference plate is used for calibration to ensure consistent detection benchmarks and avoid the impact of environmental or equipment fluctuations on subsequent results.

[0034] S200: Based on the standard board reference image, calculate the entropy distribution used to measure the complexity and information richness of the image region, and design the first sparse sampling template.

[0035] Furthermore, the standard plate reference image is divided into regions according to a preset grid size; for each region, image data of each band corresponding to the multispectral filter group is extracted; the image data of each band is converted into a two-dimensional matrix; the entropy value of each band in the region is calculated to measure the complexity and information richness of the image region. The checkerboard grid size parameter is set, and a basic sampling template is generated based on the checkerboard grid arrangement rules; the sampling probability of the region is calculated based on the entropy value; based on the sampling probability, random sampling points are added to the basic sampling template using a random algorithm to obtain the first sparse sampling template.

[0036] Specifically, the calibrated reference image is uniformly divided according to a predetermined grid size. For example, the image is divided into M×N non-overlapping small regions. The size of each region can be determined according to actual needs, generally to contain enough pixels to reflect the region's characteristics, such as 16×16 pixels or 32×32 pixels. Ensure that all regions cover the entire image area without gaps or overlaps. For each divided region, image data for each band corresponding to the multispectral filter group (e.g., RGB+NIR / UV, Red Green Blue+Near-Infrared / Ultraviolet, Red Green Blue+Near-Infrared+Ultraviolet) is extracted. The image data for each band is converted into a two-dimensional matrix for subsequent calculations.

[0037] Furthermore, for each region and each band, the entropy value of the entropy distribution, which measures the complexity and information richness of the image region, is calculated. The entropy value is calculated as follows:

[0038]

[0039] Where H is the information entropy, an indicator that measures the complexity and information richness of an image region, and L represents the intensity level, that is, the total number of possible values ​​a pixel can take in that band. Indicates the first in this region The probability of a pixel with a brightness of band i. Indicates that the region is in the first The band and brightness (intensity) are equal to the number of pixels i, where N represents the total number of pixels in the region, and i represents the intensity level index, from 0 to L. 1. For each possible pixel brightness value (or quantization intensity value), L represents the intensity level, that is, the total number of possible values ​​a pixel can take in this band. It represents the band index (such as red, green, and blue for visible light, or multispectral channels such as near-infrared and ultraviolet), used to distinguish the entropy calculation of different spectral channels.

[0040] Furthermore, the calculated entropy values ​​for each region and band are organized into a matrix, and their distribution is recorded to obtain the entropy value distribution, which provides a basis for subsequent template design.

[0041] Furthermore, based on the camera resolution and detection requirements, the size parameters of the checkerboard pattern for the programmable mask are determined. For example, each square of the checkerboard is set to S×S pixels, where S can be 8, 16, etc., depending on the required sampling density and the degree of image detail preservation. A basic sampling template is generated according to the checkerboard pattern arrangement. In the basic sampling template, black squares represent sampling points, and white squares represent non-sampling points. The checkerboard template is characterized by a uniform spatial distribution of sampling points, ensuring that the basic structural information of the image is sampled. Simultaneously, the checkerboard template possesses a certain regularity and symmetry, facilitating subsequent image reconstruction processing.

[0042] Furthermore, for each region, the sampling probability is determined based on the previously calculated information entropy value. Regions with higher information entropy values ​​indicate richer information or more complex variations, thus requiring a higher sampling probability. The formula for calculating the sampling probability is as follows:

[0043] in, It is the sampling probability of the j-th region. R is the information entropy value of region j, and R is the total number of regions.

[0044] Furthermore, based on the sampling probability of each region, a certain number of random sampling points are added to the unsampled regions using a random algorithm (such as accept / reject sampling) on ​​top of the checkerboard template to obtain the first sparse sampling template. These random sampling points will be used to supplement the detailed information that the checkerboard template fails to cover, especially in complex regions with high information entropy.

[0045] Preferably, the overall sampling rate is controlled within the range of 10%-20%, which can be achieved by adjusting the density of the checkerboard pattern and the number of random sampling points. For example, for a 1024×1024 pixel image, if the target sampling rate is 15%, the total number of sampling points is approximately 153,600. Based on the distribution of the checkerboard pattern and random sampling points, the number of sampling points in each region is reasonably allocated to ensure that the overall sampling rate requirement is met.

[0046] Preferably, the generated first sparse sampling template is optimized and smoothed to avoid reconstruction difficulties caused by excessively sparse or dense sampling point distribution. Image processing techniques and optimization algorithms, such as morphological operations (dilation, erosion, etc.), can be used to adjust the distribution of sampling points, or optimization algorithms such as simulated annealing and genetic algorithms can be used to fine-tune the positions of sampling points, improving template quality while meeting sampling rate requirements. The connectivity and consistency of the template are checked to ensure that the connections between sampling points form a relatively reasonable sampling network, which is beneficial for subsequent image reconstruction algorithms to accurately recover complete image information.

[0047] Preferably, by calculating the entropy distribution of multispectral images, the complexity and information richness of regions are quantified, sampling resources are dynamically allocated to avoid redundant sampling in uniform regions and focus on high-information regions; the basic checkerboard template ensures structural integrity, and the superposition of random sampling points enhances flexibility, balancing efficiency and detail preservation, reducing the total sampling rate to 10%-20%; the distribution of sampling points is optimized through morphological or genetic algorithms to improve the robustness of the reconstruction algorithm and reduce artifacts and reconstruction errors.

[0048] S300: Using the first sparse sampling template and reconstruction algorithm, perform the first sampling to obtain the reconstructed full-frame image of the organic electroluminescent device in each band, and perform the first emission uniformity evaluation based on the reconstructed full-frame image in each band.

[0049] Furthermore, the organic electroluminescent device is driven to emit light, and multispectral coded sampling is performed using the first sparse sampling template to obtain a multispectral coded image; the multispectral coded image is preprocessed to construct an observation matrix; the estimated value of the original image signal is initialized, and a reconstruction algorithm is used to obtain the reconstructed full-frame image of each band.

[0050] Specifically, based on the electrical characteristics of the organic electroluminescent device (OLED) under test, a computer-controlled adjustable driving power supply sends precise driving signals to the OLED, causing it to emit light with a stable current or voltage. Simultaneously, the duration and frequency of the emission are set to ensure the emission process matches the camera's exposure time for clear image acquisition. While the OLED emits light, a high-speed area array camera is activated for sampling. The camera's exposure time is precisely controlled to take a snapshot sample during the device's emission period. The exposure time must be short enough to avoid motion blur (if the device exhibits dynamic changes) and to increase sampling speed, while also being long enough to obtain sufficient light signal. This is typically set in the millisecond range, with the specific value determined based on the device's luminous intensity and the camera's sensitivity.

[0051] Furthermore, to ensure the accuracy and consistency of the sampled data, a synchronous triggering technique is employed, ensuring that the drive power supply and camera startup signals are triggered by the same synchronous signal source. This guarantees that the device's emission state is completely synchronized with the camera's sampling time during each sampling, avoiding sampling errors caused by timing deviations.

[0052] Furthermore, using a first sparse sampling template, the DMD is controlled to display the corresponding mask pattern. The DMD modulates the light according to the mask pattern, allowing light corresponding to the sampling points to pass through while blocking light from non-sampling points. In this way, light from different wavelengths originating from the organic electroluminescent device is encoded by the DMD, forming an coded aliased light signal. The camera's multispectral filter group performs spectral splitting on the encoded light. Each filter only allows light of a specific wavelength to pass through, thus achieving simultaneous acquisition of light signals of different wavelengths. For example, the RGB filter corresponds to the red, green, and blue visible light bands, respectively; the NIR filter corresponds to the near-infrared band; and the UV filter corresponds to the ultraviolet band. After passing through the corresponding filter, the light of each wavelength is projected onto the camera's image sensor, resulting in a multispectral coded image.

[0053] Furthermore, spectral coded image data is extracted from the spectral coded image, and preliminary processing is performed on the spectral coded image data, including noise removal and bias correction. The pre-processed coded image data of each band is then normalized to ensure that the pixel values ​​are distributed within the same dynamic range (e.g., between 0 and 1), so that subsequent reconstruction algorithms can better process the data.

[0054] Furthermore, an observation matrix is ​​constructed, which combines the mask coding matrix and the spectral response matrix of the multispectral filter set. The mask coding matrix is ​​determined by the mask pattern of the DMD, representing the sampling method of spatial information of the image during the sampling process; the spectral response matrix is ​​determined by the spectral transmittance of the filter set, representing the separation and response characteristics of light in different wavelength bands. Combining these two matrices yields an observation matrix that comprehensively describes the multispectral compressed sampling process.

[0055] Initialize the original image (a complete, unsampled, true multispectral image of the organic electroluminescent device to be reconstructed in each band) signal estimate, typically by initializing it as a zero matrix. Set the parameters of the reconstruction algorithm, such as the maximum number of iterations and sparsity. Sparsity refers to the number of non-zero coefficients in the sparse representation domain of the image, which can be estimated based on experience or prior knowledge. Execute the Cross-Band Coupled Orthogonal Matching Pursuit (OMP) algorithm. In each iteration, calculate the orthogonal matching degree between the residual and the observation matrix, and select the atom (i.e., the column vector in the observation matrix) with the highest correlation to the residual to add it to the selection set. Then, use the atoms in the selection set to solve the least squares problem and update the estimate of the reconstructed full-frame image. Repeat this process until the maximum number of iterations is reached or the residual is less than a set threshold. In this way, the full-frame image of each band is gradually recovered, resulting in the reconstructed full-frame image of each band.

[0056] Furthermore, post-processing is performed on the reconstructed full-frame images for each band, including artifact removal and smoothing, to improve image quality and readability. Artifacts are image distortion phenomena caused by limitations in compression sampling and reconstruction algorithms, and can be removed using methods such as wavelet transform.

[0057] Furthermore, the standard deviation of brightness of the reconstructed full-frame image for each band is calculated; for RGB band images, the chromaticity coordinates of each pixel are calculated, and the chromaticity deviation of the reconstructed full-frame image is calculated based on the chromaticity coordinates of each pixel; the spectral deviation index of the reconstructed full-frame image is calculated; and the first emission uniformity assessment is performed based on the standard deviation of brightness, chromaticity deviation, and spectral deviation index to obtain the first emission uniformity assessment report.

[0058] Specifically, the standard deviation of brightness in the reconstructed full-frame image for each band is calculated. The standard deviation reflects the dispersion of image brightness; a smaller standard deviation indicates more uniform brightness. The formula for calculating the standard deviation of brightness is:

[0059] in, This represents the standard deviation of brightness, where M and N are the number of rows and columns of the image, respectively. It is the brightness value of the pixel in the i-th row and j-th column. is the average brightness of the image, i is the row index of the image, ranging from 1 to M, and j is the column index of the image, ranging from 1 to N.

[0060] Furthermore, for RGB band images, the chromaticity coordinates of each pixel are calculated (e.g., CIE 1976 UCS chromaticity coordinates), and then the deviation of the entire chromaticity coordinates of the reconstructed full-frame image for each band is calculated. Chromaticity deviation can be measured using the average chromaticity distance, and the formula for calculating chromaticity deviation is as follows:

[0061] in, It's a color deviation. and Let be the chromaticity coordinates of the pixel in the i-th row and j-th column. and is the average chromaticity coordinate of the entire image, M and N are the number of rows and columns of the image, respectively, i is the row index of the image, ranging from 1 to M, and j is the column index of the image, ranging from 1 to N.

[0062] Furthermore, the spectral reflectance of the sample at different wavelengths is compared with that of the standard uniformly emitting reference plate to calculate the spectral deviation index (SDI). The formula for calculating SDI can refer to relevant spectral similarity evaluation indicators, such as spectral angle mapping (SAM) or root mean square error. For example, the formula for calculating SDI using root mean square error is as follows:

[0063] in, Indicates the spectral deviation index. Indicates the number of bands. This represents the spectral reflectance of the device under test in the b-th band. This represents the spectral reflectance of the standard reference plate in band b, where b is the band index.

[0064] Furthermore, the calculated uniformity indicators, such as luminance standard deviation, chromaticity deviation, and spectral deviation index, are compiled into a table to record the uniformity of each band and the overall image. Simultaneously, corresponding charts (such as histograms and line graphs) are drawn to visually display the distribution and trends of these indicators, facilitating visual analysis of uniformity and yielding the first luminescence uniformity assessment report.

[0065] Preferably, synchronous triggering technology is adopted to ensure that the device emission and camera sampling are strictly synchronized, eliminating timing deviations and improving data consistency; cross-band coupled OMP algorithm is adopted to utilize the correlation between multiple spectra to jointly reconstruct images of each band, improve reconstruction accuracy and spectral consistency, and evaluate the uniformity of organic electroluminescence in multiple dimensions based on brightness standard, color difference deviation, and spectral deviation index.

[0066] S400: Based on the first emission uniformity evaluation result, mark the abnormal emission areas, dynamically adjust the sampling strategy, and generate a second sparse sampling template.

[0067] Furthermore, thresholds for luminance standard deviation, chromaticity deviation, and spectral deviation index of organic electroluminescent devices are set respectively; the uniformity index of each region in the first luminescence uniformity evaluation report is compared with the thresholds for luminance standard deviation, chromaticity deviation, and spectral deviation index respectively; regions exceeding any of the thresholds for luminance standard deviation, chromaticity deviation, and spectral deviation index are marked as abnormal luminescence regions; the location information, band information, and out-of-tolerance values ​​of each abnormal luminescence region are recorded.

[0068] Specifically, based on the characteristics of the tested organic electroluminescent devices and industry standards, thresholds for luminance standard deviation, chromaticity deviation, and spectral deviation index are set. These thresholds are used to determine whether the luminous uniformity of the organic electroluminescent devices is within acceptable limits. The uniformity indicators of each region in the first luminous uniformity assessment report are compared one by one with the set thresholds, and regions exceeding any threshold are identified as abnormal areas. For each abnormal area, its location information (such as row and column coordinate range), band information (in which bands the abnormality occurs), and specific out-of-tolerance values ​​(uniformity indicator deviation exceeding the threshold value) are recorded. Morphological closing operations are then performed on the abnormal areas to merge them, providing detailed data support for subsequent feedback decisions.

[0069] Preferably, if the brightness standard deviation, chromaticity deviation, and spectral deviation index are all within the threshold range, the device is considered to have good luminous uniformity. Dynamic feedback control then issues a command to reduce the sampling rate in the next round of sampling, thereby accelerating the detection speed and improving efficiency. Simultaneously, a minimum sampling rate limit can be set according to actual needs to avoid insufficient detection accuracy due to an excessively low sampling rate. If the brightness standard deviation, chromaticity deviation, or spectral deviation index exceeds the threshold, a localized encrypted sampling process is initiated. Further analysis of which regions and bands exhibit out-of-tolerance indicators provides guidance for localized encrypted sampling.

[0070] Furthermore, the abnormal luminescent areas are sampled more densely; the sampling rate is adjusted, and a new overall mask pattern is generated based on the adjusted sampling rate. The new overall mask pattern is then optimized; the new overall mask pattern is checked to see if it meets the requirements. If it does, the new overall mask pattern is used as the second sparse sampling template.

[0071] Specifically, for identified anomalous areas, the need for localized refined sampling is determined based on their deviation (uniformity index exceeding the allowable threshold deviation) and the importance of their location. For example, for areas with severe deviations (such as spectral deviation index exceeding twice the threshold) or anomalous areas located in critical display areas of the device (such as the central area of ​​the display screen), localized refined sampling is prioritized. The sampling rate for localized refined sampling is determined based on the distribution and size of the anomalous areas. Generally, the sampling rate should be higher than the initial sampling rate to obtain more detailed local image information and accurately assess the luminescence characteristics of the anomalous areas. For areas with particularly severe deviations, the sampling rate can be further increased, but the limitations of data volume and reconstruction time must be considered to avoid excessive system load due to an excessively high sampling rate.

[0072] Preferably, in addition to localized intensified sampling in abnormal areas, the sampling rate can be appropriately reduced for areas with good overall uniformity to decrease data volume and reconstruction computation, thereby improving detection efficiency. For example, the sampling rate in normal areas can be reduced from the initial 10%-20% to around 5%. Simultaneously, it must be ensured that the data after reducing the sampling rate is still sufficient to reflect the basic luminescent characteristics of the area, avoiding missed detections or misjudgments due to excessively low sampling rates.

[0073] Preferably, to achieve a smooth transition and optimized distribution of the sampling rate, a gradual sampling rate adjustment strategy can be adopted. At the boundary between the abnormal and normal regions, the sampling rate is gradually increased to avoid a decrease in the quality of the reconstructed full-frame image caused by abrupt changes in the sampling rate. For example, the sampling rate in the boundary region can be set to an intermediate value between the initial sampling rate and the encrypted sampling rate, such as 15%.

[0074] Furthermore, based on the determined local encrypted sampling region and sampling rate, the mask pattern is redesigned. Building upon the first sparse sampling template, the mask region corresponding to the anomaly area is encrypted. This can be achieved by increasing the sampling point density, such as reducing the size of the checkerboard grid or increasing the number of entropy-weighted random sampling points, making the sampling point distribution in the encrypted region denser. Simultaneously, it is crucial to ensure that the mask pattern in the encrypted sampling region maintains a certain degree of coherence and consistency with the mask patterns of the surrounding normal sampling regions, facilitating subsequent image reconstruction. Edge transition processing techniques can be employed to gradually merge the mask patterns of the encrypted and normal regions at the edges, avoiding artifacts in the reconstructed full-image due to abrupt changes in the mask pattern. For normal regions with good overall uniformity, the sampling rate of their mask pattern is adjusted. Based on the initial mask pattern, the sampling density in this region is reduced by decreasing the number of sampling points or increasing the sampling point interval. However, the uniformity and representativeness of the sampling point distribution must be ensured to avoid image reconstruction errors caused by uneven sampling point distribution.

[0075] Preferably, optimization algorithms (such as particle swarm optimization) can be used to optimize the mask pattern in the normal region, so as to reduce the sampling rate while preserving the main structural information and features of the image as much as possible, thereby improving sampling efficiency and reconstruction quality.

[0076] Furthermore, the locally encrypted sampling mask and the basic sampling mask of the normal region are fused to generate a new overall mask pattern. During the fusion process, care must be taken to handle the transition between the encrypted and normal regions to ensure the smoothness and consistency of the entire mask pattern. The new overall mask pattern is then optimized, including removing isolated sampling points, filling sampling holes, and adjusting the uniformity of sampling point distribution. Mathematical morphology operations (such as opening and closing operations) or graph theory-based optimization algorithms (such as graph cut algorithms) can be used to improve the quality and effectiveness of the new overall mask pattern.

[0077] Furthermore, check whether the optimized new overall mask pattern meets the overall sampling rate requirements and whether it matches the camera resolution and the DMD pixel array. If any problems are found, make timely adjustments and corrections to ensure that the new overall mask pattern can be applied in the actual sampling process; if the new overall mask pattern meets the overall sampling rate requirements, then use the new overall mask pattern as the second sparse sampling template.

[0078] Preferably, based on threshold comparison, abnormal luminescence regions are accurately identified, and their location, band, and out-of-tolerance values ​​are recorded to provide data support for feedback. Furthermore, adaptive sampling rate adjustment is implemented to downsample normal regions, reducing data volume and computational load, while increasing sampling density in abnormal luminescence regions to enhance detail capture capabilities. A gradual transition strategy is implemented to smoothly adjust the sampling rate at the boundary between normal and abnormal luminescence regions, avoiding reconstruction artifacts. By fusing locally densely sampled masks and basic sampling masks in normal regions, and using algorithms such as particle swarm optimization to optimize the mask pattern in normal regions, both efficiency and reconstruction quality are balanced.

[0079] S500: Use the second sparse sampling template to perform a second sampling, update the reconstructed full-frame image under each band, and perform a second emission uniformity evaluation, outputting an organic electroluminescence uniformity detection report.

[0080] Furthermore, the data collected in the second sampling is fused with the data collected in the first sampling; the reconstructed full-frame image in each band is updated using a compressed sensing algorithm; the brightness standard deviation, chromaticity deviation, and spectral deviation index of the second sampling and the first sampling are compared, and it is determined whether convergence has occurred. If convergence has occurred, the iteration of the compressed sensing algorithm is stopped, and an organic electroluminescence uniformity detection report is output.

[0081] Specifically, the data from the second sampling is fused with the data from the first sampling to form a new observation matrix. A compressed sensing algorithm, such as Online Orthogonal Matching Pursuit (OMP), is used to rapidly update and reconstruct the full-frame image for each band based on the initial full-frame reconstruction. The changes in luminance standard deviation, chromaticity deviation, and spectral deviation index between the second and first reconstructed full-frame images are calculated, and the convergence criteria are checked. If convergence is achieved, the reconstruction process terminates, and a final organic electroluminescence uniformity detection report is output. If convergence fails, a new encrypted sampling mask is automatically generated based on the current image residual region, and a third round of sparse sampling is performed. This data fusion, image update, and convergence check process is repeated until the set convergence criteria are met or the maximum number of iterations is reached. Once the reconstructed full-frame image meets the convergence criteria, the uniformity index of the organic electroluminescence device across the entire frame is calculated based on the final reconstructed full-frame image, and an organic electroluminescence uniformity detection report is automatically generated and output. The test report includes, but is not limited to: brightness distribution map, chromaticity distribution map, heat map overlay display, brightness standard deviation, chromaticity deviation, spectral deviation index, non-uniform area annotation results, test conclusions, and a summary of image reconstruction process parameters.

[0082] Preferably, incremental update reconstruction is used, which rapidly updates and reconstructs the full image by fusing multiple rounds of sampling data and the online OMP algorithm, thereby improving efficiency by avoiding repeated full calculations. If convergence is not achieved, a new mask is automatically generated for iterative detection until the accuracy requirements are met, realizing the intelligence and automation of the detection process and achieving closed-loop feedback optimization. The automatically generated and output organic electroluminescence uniformity detection report includes a brightness distribution map, a chromaticity distribution map, a thermal map overlay display, brightness standard deviation, chromaticity deviation, spectral deviation index, non-uniformity region annotation results, detection conclusions, and a summary of image reconstruction process parameters, providing comprehensive luminescence uniformity detection data and quality analysis of the device's organic electroluminescence.

[0083] The present invention also provides an organic electroluminescence uniformity detection system based on dynamic feedback, for implementing an organic electroluminescence uniformity detection method based on dynamic feedback. The system includes a control module, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor. The processor executes the computer program to implement the organic electroluminescence uniformity detection method based on dynamic feedback.

[0084] This invention provides a storage medium storing a program that, when executed by a processor, implements a method for detecting the uniformity of organic electroluminescence based on dynamic feedback.

[0085] This invention provides a processor for running a program, wherein the program executes an organic electroluminescence uniformity detection method based on dynamic feedback.

[0086] This invention provides a device including a processor, a memory, and a program stored in the memory and executable on the processor. When the processor executes the program, it implements a method for detecting the uniformity of organic electroluminescence based on dynamic feedback. The device described herein can be a server, PC, PAD, mobile phone, etc.

[0087] This application also provides a computer program product that, when executed on a data processing device, is suitable for performing a dynamic feedback-based organic electroluminescence uniformity detection method.

[0088] Those skilled in the art will understand that embodiments of this application can provide methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0089] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0090] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0091] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0092] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0093] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0094] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0095] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0096] The above are merely embodiments of this application and are not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A method for detecting the uniformity of organic electroluminescence based on dynamic feedback, characterized in that, include: After initializing the detection system and calibrating the reference baseline, a reference image of the standard plate is acquired; Based on the standard plate reference image, calculate the entropy distribution used to measure the complexity and information richness of the image region, and design a first sparse sampling template; Using the first sparse sampling template and reconstruction algorithm, the first sampling is performed to obtain the reconstructed full-frame image of the organic electroluminescent device in each band. Based on the reconstructed full-frame image in each band, the first emission uniformity evaluation is performed. Based on the first emission uniformity evaluation result, abnormal emission areas are marked, the sampling strategy is dynamically adjusted, and a second sparse sampling template is generated. The second sparse sampling template is used to perform a second sampling, update the reconstructed full-frame image under each band, and perform a second emission uniformity evaluation, outputting an organic electroluminescence uniformity detection report.

2. The method for detecting the uniformity of organic electroluminescence based on dynamic feedback according to claim 1, characterized in that, The calibration reference baseline includes: Perform camera spatial response calibration, mask coding mode calibration, and spectral response calibration; By integrating the results of the camera spatial response calibration, the mask coding mode calibration, and the spectral response calibration through mathematical modeling and data fusion algorithms, a spatial-spectral calibration matrix is ​​obtained.

3. The method for detecting the uniformity of organic electroluminescence based on dynamic feedback according to claim 1, characterized in that, The step of calculating the entropy distribution, used to measure the complexity and information richness of image regions, based on the standard plate reference image, and designing a first sparse sampling template includes: The standard plate reference image is divided into regions according to a preset grid size; For each divided region, image data of each band corresponding to the multispectral filter group is extracted respectively; The image data of each band is converted into a two-dimensional matrix; Calculate the entropy value of each band in the divided region, which is used to measure the complexity and information richness of the image region.

4. The method for detecting the uniformity of organic electroluminescence based on dynamic feedback according to claim 3, characterized in that, The step of calculating the entropy distribution, used to measure the complexity and information richness of image regions, based on the standard plate reference image, and designing the first sparse sampling template, further includes: Set the chessboard grid size parameters, and generate a basic sampling template based on the chessboard grid arrangement rules; Calculate the sampling probability of the divided region based on the entropy value; Based on the sampling probability, random sampling points are added to the basic sampling template using a random algorithm to obtain the first sparse sampling template.

5. The method for detecting the uniformity of organic electroluminescence based on dynamic feedback according to claim 1, characterized in that, The first sampling, using the first sparse sampling template and reconstruction algorithm, yields a reconstructed full-frame image of the organic electroluminescent device in each band, including: Drive the organic electroluminescent device to emit light, and use the first sparse sampling template to perform multispectral coding sampling to obtain a multispectral coded image; The multispectral coded image is preprocessed to construct an observation matrix; The estimated values ​​of the original image signal are initialized, and the reconstruction algorithm is used to obtain the reconstructed full-frame image of each band.

6. The method for detecting the uniformity of organic electroluminescence based on dynamic feedback according to claim 1, characterized in that, The first luminescence uniformity assessment includes: Calculate the standard deviation of brightness for the reconstructed full-frame image in each band; For an RGB band image, the chromaticity coordinates of each pixel are calculated, and the chromaticity deviation of the reconstructed full image is calculated based on the chromaticity coordinates of each pixel. Calculate the spectral deviation index of the reconstructed full-frame image; Based on the brightness standard deviation, the chromaticity deviation, and the spectral deviation index, a first luminous uniformity assessment is performed to obtain a first luminous uniformity assessment report.

7. The method for detecting the uniformity of organic electroluminescence based on dynamic feedback according to claim 1, characterized in that, The step of marking abnormal emission regions, dynamically adjusting the sampling strategy, and generating a second sparse sampling template based on the first emission uniformity evaluation result includes: Set the threshold values ​​for brightness standard deviation, chromaticity deviation, and spectral deviation index for organic electroluminescent devices respectively; The uniformity index of each region in the first luminescence uniformity evaluation report is compared with the luminance standard deviation threshold, the chromaticity deviation threshold, and the spectral deviation index threshold, respectively. Regions exceeding any of the following thresholds—the luminance standard deviation threshold, the chromaticity deviation threshold, and the spectral deviation index threshold—are marked as abnormal luminous regions. Record the location information, band information, and out-of-tolerance values ​​of each abnormal luminescence region.

8. The method for detecting the uniformity of organic electroluminescence based on dynamic feedback according to claim 1, characterized in that, The step of marking abnormal emission regions, dynamically adjusting the sampling strategy, and generating a second sparse sampling template based on the first emission uniformity evaluation result further includes: Encrypted sampling is performed on the abnormally emitting area; Adjust the sampling rate, generate a new overall mask pattern based on the adjusted sampling rate, and optimize the new overall mask pattern. Check whether the new overall mask pattern after optimization meets the requirements. If it does, use the new overall mask pattern as the second sparse sampling template.

9. The method for detecting the uniformity of organic electroluminescence based on dynamic feedback according to claim 1, characterized in that, The process involves using the second sparse sampling template for a second sampling, updating the reconstructed full-frame image for each band, performing a second emission uniformity assessment, and outputting an organic electroluminescence uniformity detection report, including: The data collected in the second sampling is fused with the data collected in the first sampling. The reconstructed full-frame image for each band is updated using a compressed sensing algorithm; The standard deviation of brightness, chromaticity deviation and spectral deviation index of the second sampling are compared with those of the first sampling, and it is determined whether they converge. If they converge, the iteration of the compressed sensing algorithm is stopped and an organic electroluminescence uniformity detection report is output.

10. An organic electroluminescence uniformity detection system based on dynamic feedback, characterized in that, The system includes a control module, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor. The processor executes the computer program to implement the organic electroluminescence uniformity detection method based on dynamic feedback according to any one of claims 1-9.

Images