Automatic photoelectric test and defect detection method and system for MicroLED display

Abstract

The invention provides an automatic photoelectric test and defect detection method and system for a MicroLED display. Belongs to the technical field of display detection. Belongs to the technical field of display detection. The method comprises the following steps: carrying out comprehensive scanning on the MicroLED display by utilizing a high-precision photoelectric sensor array, and collecting initial photoelectric signal data of the display; preprocessing the collected initial photoelectric signal data; standard photoelectric signal data are obtained, pixel point positioning is carried out on the obtained standard photoelectric signal data based on the layout and resolution information of the display, and the position coordinate of each pixel point is determined; and dividing the display into a plurality of sub-regions according to the display region characteristics of the display. The MicroLED display is comprehensively scanned through the high-precision photoelectric sensor array, and the automatic data acquisition and processing flow is combined, so that the precision and efficiency of the photoelectric test are remarkably improved, the manual intervention is reduced, and the consistency and reliability of the detection are improved.

Description

Technical Field

[0001] This invention proposes an automated optoelectronic testing and defect detection method and system for MicroLED displays, belonging to the field of display testing technology. Background Technology

[0002] MicroLED displays, as a novel display technology, boast numerous advantages such as high brightness, high contrast, and low power consumption, demonstrating immense application potential in the display field. However, during the production process of MicroLED displays, various optoelectronic performance defects are unavoidable due to complex manufacturing processes and material properties, such as uneven pixel brightness, color shift, dark spots, and bright spots. These defects severely affect the display's performance and product quality.

[0003] Currently, traditional optoelectronic testing and defect detection methods for MicroLED displays mainly rely on manual operation and simple testing equipment, which suffer from problems such as low detection efficiency, low accuracy, and strong subjectivity, making it difficult to meet the needs of large-scale production. Summary of the Invention

[0004] This invention provides an automated optoelectronic testing and defect detection method and system for MicroLED displays, to solve the problems mentioned in the background section above:

[0005] This invention proposes an automated optoelectronic testing and defect detection method for MicroLED displays, the method comprising:

[0006] S1. A high-precision photoelectric sensor array is used to perform a full scan of the MicroLED display to collect the initial photoelectric signal data of the display; the collected initial photoelectric signal data is preprocessed; standard photoelectric signal data is obtained; based on the layout and resolution information of the display, the obtained standard photoelectric signal data is used to locate the pixels and determine the position coordinates of each pixel; and the display is divided into multiple sub-regions according to the display area characteristics of the display.

[0007] S2. Extract standard photoelectric signal data based on the division of multiple sub-regions to obtain photoelectric characteristic data;

[0008] S3. Use environmental sensors to collect environmental data of the environment in which the MicroLED display is located in real time. Based on the collected environmental data, establish an environmental influence model between environmental factors and the photoelectric characteristics of the display. According to the environmental influence model, correct the extracted photoelectric characteristic data to eliminate the influence of environmental factors on the test results and obtain the corrected photoelectric characteristic parameters.

[0009] S4. Combine the corrected photoelectric characteristic parameters to extract features related to defects; use machine learning algorithms to classify and identify the extracted defect features; input the corrected photoelectric characteristic parameters into the trained model to detect defective pixels and defective regions in the display.

[0010] S5. Based on the defect detection results, evaluate the overall quality of the MicroLED display; for the detected defects, generate corresponding optimization suggestions based on the display's manufacturing process and material characteristics.

[0011] The present invention proposes an automated optoelectronic testing and defect detection system for MicroLED displays, comprising a memory, a processor, and a computer program stored in the memory and capable of running on the memory. The processor executes the program to implement the automated optoelectronic testing and defect detection method for MicroLED displays as described above.

[0012] Beneficial effects of this invention:

[0013] By using a high-precision photoelectric sensor array to perform a comprehensive scan of the MicroLED display, combined with an automated data acquisition and processing workflow, the accuracy and efficiency of photoelectric testing are significantly improved, manual intervention is reduced, and the consistency and reliability of testing are enhanced.

[0014] By locating pixels in standard photoelectric signal data, the position coordinates of each pixel can be accurately identified, enabling pixel-level defect detection in subsequent steps and improving the fineness and accuracy of defect identification.

[0015] The display area is divided into multiple sub-regions, which facilitates detailed analysis of the characteristics of different regions, helps to discover local anomalies or non-uniformity issues, and improves the overall detection quality.

[0016] By introducing environmental sensors and establishing an environmental impact model, the collected photoelectric characteristic parameters are corrected, effectively eliminating the influence of environmental factors such as temperature, humidity, and light on the test results, and ensuring the stability and repeatability of the test results.

[0017] Using machine learning algorithms to classify and identify defect features not only improves the intelligence level of defect detection, but also allows for continuous optimization of the model to adapt to new defect patterns, exhibiting good scalability and adaptability.

[0018] Based on defect detection, optimization suggestions can be generated by combining manufacturing processes and material characteristics, which helps to quickly locate the causes of defects, guide the improvement of production processes, and improve product yield and reliability.

[0019] The overall quality of MicroLED displays was evaluated based on defect detection results, and a complete closed-loop system was constructed, from data acquisition and analysis to defect identification and quality feedback, providing a scientific basis for product quality control. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of the method steps described in this invention;

[0021] Figure 2 for Figure 1 A detailed flowchart illustrating the implementation steps of S3. Detailed Implementation

[0022] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0023] One embodiment of the present invention, such as Figure 1 As shown, an automated optoelectronic testing and defect detection method for MicroLED displays is provided, the method comprising:

[0024] S1. A high-precision photoelectric sensor array is used to perform a full scan of the MicroLED display to collect the initial photoelectric signal data of the display; the collected initial photoelectric signal data is preprocessed; standard photoelectric signal data is obtained; based on the layout and resolution information of the display, the obtained standard photoelectric signal data is used to locate the pixels and determine the position coordinates of each pixel; and the display is divided into multiple sub-regions according to the display area characteristics of the display.

[0025] S2. Extract standard photoelectric signal data based on the division of multiple sub-regions to obtain photoelectric characteristic data;

[0026] S3. Use environmental sensors to collect environmental data of the environment in which the MicroLED display is located in real time. Based on the collected environmental data, establish an environmental influence model between environmental factors and the photoelectric characteristics of the display. According to the environmental influence model, correct the extracted photoelectric characteristic data to eliminate the influence of environmental factors on the test results and obtain the corrected photoelectric characteristic parameters.

[0027] S4. Combine the corrected photoelectric characteristic parameters to extract features related to defects; use machine learning algorithms to classify and identify the extracted defect features; input the corrected photoelectric characteristic parameters into the trained model to detect defective pixels and defective regions in the display.

[0028] S5. Based on the defect detection results, evaluate the overall quality of the MicroLED display; for the detected defects, generate corresponding optimization suggestions based on the display's manufacturing process and material characteristics.

[0029] The working principle of the above technical solution is as follows:

[0030] A high-precision photoelectric sensor array is used to perform a comprehensive scan of the MicroLED display to acquire initial photoelectric signal data for each pixel. The acquired initial photoelectric signal data is preprocessed to obtain standard photoelectric signal data. Based on the display's layout and resolution information, pixel location is performed on the standard photoelectric signal data to determine the position coordinates of each pixel.

[0031] Based on the display area characteristics, the monitor is divided into multiple sub-regions. This division facilitates more detailed analysis of the monitor because different sub-regions may differ in photoelectric characteristics, and it also facilitates subsequent data processing and defect detection for different sub-regions. Based on the previously defined sub-regions, standard photoelectric signal data is extracted to obtain photoelectric characteristic feature data. This feature data reflects the photoelectric characteristics of pixels within each sub-region, such as brightness and color, providing crucial feature information for subsequent defect detection.

[0032] Environmental sensors are used to collect real-time environmental data such as light intensity, temperature, and humidity of the environment in which the MicroLED display is located. These environmental factors affect the optoelectronic characteristics of the display, and without calibration, the test results may be inaccurate. An environmental impact model is established based on the collected environmental data to study the relationship between environmental factors and the optoelectronic characteristics of the display.

[0033] Based on the environmental impact model, the extracted photoelectric characteristic data are corrected to eliminate the influence of environmental factors on the test results, yielding corrected photoelectric characteristic parameters. These corrected parameters more accurately reflect the photoelectric characteristics of the display itself, eliminating interference from environmental factors. Combining these corrected photoelectric characteristic parameters, features related to defects are extracted. For example, for bright spot defects, features showing significantly higher brightness values ​​than surrounding pixels are extracted; for dark spot defects, features showing significantly lower brightness values ​​than surrounding pixels are extracted; and for color shift defects, features showing large color deviations are extracted, etc.

[0034] Machine learning algorithms are employed to classify and identify extracted defect features. Through extensive training samples, a model capable of accurately distinguishing between normal and defective pixels is developed. This model can determine the presence and type of defects in pixels based on input feature data. Corrected photoelectric characteristic parameters are input into the trained model to detect defective pixels and regions in the display. Based on the detection results, different types of defects are classified, such as bright spots, dark spots, color casts, and uneven brightness. Simultaneously, combined with previously determined pixel location coordinates and sub-region divisions, the location coordinates and sub-region of each defect are determined, providing detailed information for subsequent quality assessment and optimization suggestions.

[0035] Based on the defect detection results, the overall quality of the MicroLED display is evaluated. Indicators such as the number of defective pixels and the severity of defects are calculated to assess whether the display meets quality standards. These indicators directly reflect the display's quality status, providing a basis for quality control during the production process. For the detected defects, corresponding optimization suggestions are generated based on the display's manufacturing process and material characteristics. For example, for uneven brightness, adjustments to the driving circuit parameters are recommended; for color shift issues, optimization of the color correction algorithm is suggested.

[0036] The effects of the above technical solution are as follows:

[0037] By using a high-precision photoelectric sensor array for comprehensive scanning, the initial photoelectric signal data of MicroLED displays can be accurately acquired, thereby improving the accuracy of photoelectric testing.

[0038] Based on the photoelectric characteristics extracted from standard photoelectric signal data, and combined with machine learning algorithms for defect classification and identification, the system can accurately distinguish between normal pixels and defective pixels, thereby improving the sensitivity of defect detection.

[0039] By establishing an environmental impact model and correcting the collected photoelectric signal data for environmental factors, the interference of environmental factors such as light intensity, temperature, and humidity on the test results was eliminated, ensuring the reliability of the test results.

[0040] By using pixel-level localization and sub-region division, the specific location and region of each defect can be accurately determined, thus improving the spatial localization accuracy of defect detection.

[0041] By comprehensively evaluating indicators such as the number and severity of defective pixels, the overall quality of MicroLED displays can be fully assessed, enabling a more accurate determination of whether the displays meet quality standards.

[0042] Based on the defect type and test results, combined with the manufacturing process and material characteristics of the display, targeted optimization suggestions are generated, such as adjusting the driving circuit parameters and optimizing the color correction algorithm, which helps to improve quality control and optimization in the production process.

[0043] This method, based on automated photoelectric testing and machine learning algorithms, enables automated defect detection and classification, reduces manual intervention, minimizes human error, and improves production efficiency.

[0044] Through high-precision optoelectronic testing and systematic defect detection, we can ensure that MicroLED display products have higher consistency and reliability during the production process, and reduce the generation of defective products.

[0045] In one embodiment of the present invention, S1 includes:

[0046] S11. A high-precision photoelectric sensor array is used to perform a full scan of the MicroLED display to obtain initial photoelectric signal data;

[0047] S12. Filter the initial photoelectric signal data to obtain filtered photoelectric signal data, and normalize the filtered photoelectric signal data to obtain normalized photoelectric signal data.

[0048] S13. Detect and remove outliers in the obtained normalized photoelectric signal data to obtain standard photoelectric signal data. Based on the layout and resolution information of the display, locate the pixels in the standard photoelectric signal data. Determine the position coordinates of each pixel by analyzing the distribution characteristics of the signal and the physical structure of the display.

[0049] S14. Divide the display into multiple sub-regions according to the display area characteristics of the display.

[0050] The working principle of the above technical solution is as follows:

[0051] A high-precision photoelectric sensor array is used to perform a comprehensive scan of the MicroLED display, aiming to fully acquire the initial photoelectric signal data of each pixel. The sensor array collects signals row by row or column by column to ensure coverage of the entire display area. The acquired brightness and color signals are the foundational data for subsequent analysis and processing, reflecting the photoelectric characteristics of the display at different locations.

[0052] During data acquisition, real-time monitoring of the sensor's operational status is crucial. Sensor malfunctions can lead to inaccurate signal acquisition, impacting subsequent testing and detection results. Real-time monitoring allows for timely alarm activation and recording of any sensor anomalies, facilitating subsequent investigation and handling to ensure stable and reliable signal acquisition. The initial photoelectric signal data is filtered to eliminate noise and interference. The filtered photoelectric signal data is then normalized to obtain normalized data. Outliers in the normalized photoelectric signal data are detected and removed. These outliers may be due to errors during acquisition, sensor malfunctions, or other factors. They affect data accuracy and reliability; detecting and removing outliers ensures standard photoelectric signal data and improves data quality.

[0053] Based on the display's layout and resolution information, pixel location is performed on standard photoelectric signal data. By analyzing the signal distribution characteristics and the display's physical structure, the position coordinates of each pixel are determined. Image processing algorithms, such as edge detection and corner detection, can be used to assist in pixel location. Edge detection algorithms can identify the features of the display's edges, while corner detection algorithms can locate the corners of the display. Combining these algorithms allows for more precise determination of pixel positions, providing accurate spatial information for subsequent defect detection and area segmentation.

[0054] Based on the characteristics of the display area, the monitor is divided into multiple sub-regions. This division can be based on factors such as the characteristics of the displayed content and the monitor's circuit layout. For example, the monitor can be divided into multiple rectangular sub-regions, each containing a certain number of pixels. This division method facilitates more detailed analysis of the monitor, as different sub-regions may differ in photoelectric characteristics. It also facilitates subsequent data processing and defect detection for different sub-regions, improving the efficiency and accuracy of testing and inspection.

[0055] The effects of the above technical solution are as follows:

[0056] By scanning row-by-row or column-by-column with a high-precision photoelectric sensor array, the initial photoelectric signal data of the MicroLED display can be accurately obtained. Real-time monitoring of the sensor's operating status during the acquisition process allows for timely detection and alarm of sensor anomalies, ensuring data stability and reliability.

[0057] During data processing, the step of detecting and removing outliers ensures the accuracy of the final standard photoelectric signal data and avoids interference from abnormal signals with subsequent detection and analysis.

[0058] By combining the display's layout and resolution information and using image processing algorithms to locate pixels, the exact position of each pixel can be precisely determined. This provides more accurate basic data for subsequent display quality testing.

[0059] By dividing the display into multiple sub-regions, it is possible to better adapt to the needs of the displayed content and circuit layout, and to accurately locate defects or anomalies in the display area. Sub-region division also improves the ability to detect faults in localized areas, thereby enhancing the overall display effect.

[0060] Due to the refinement of signal acquisition, processing, and positioning steps, this technical solution can more accurately detect minute defects in displays, thereby improving the overall quality control and production efficiency of displays.

[0061] The entire process includes automated data acquisition, signal processing, and outlier removal, which reduces manual intervention, improves detection speed and accuracy, and lowers the risk of human error.

[0062] In one embodiment of the present invention, step S14 includes:

[0063] Based on the overall size and resolution of the display, as well as the expected detection accuracy requirements, the number and basic shape of the sub-regions are initially constructed; common display content types of the display are analyzed, and preliminary sub-region divisions are made based on the analysis results of different display content.

[0064] Based on the pixel location coordinates determined in step S13, the distribution density and regularity of the pixels are analyzed to obtain pixel analysis results; based on the pixel analysis results, the specific shape and boundary of each sub-region are determined.

[0065] The working principle of the above technical solution is as follows:

[0066] Based on the overall size and resolution of the display, as well as the expected detection accuracy requirements, the approximate number and basic shape of the sub-regions are initially constructed. This is because the size and resolution of the display determine the number and distribution density of pixels, while the required detection accuracy affects the fineness of the sub-region division. For example, when the display resolution is high and the displayed content is complex, in order to more accurately detect potential defects, it is necessary to divide the display into a larger number of sub-regions, thereby improving the fineness of the detection and ensuring that every potentially problematic area can be analyzed in detail.

[0067] The circuit layout of the display has a significant impact on the test results. Important circuit areas may generate electromagnetic interference or other factors affecting photoelectric properties. Dividing these areas separately or having them span multiple sub-regions may lead to errors in the test results due to circuit interference. Therefore, when initially dividing the sub-regions, the characteristics of the display's circuit layout should be considered to reduce the impact of circuit interference on the test results and ensure their accuracy.

[0068] The analysis focuses on common display content types, such as text, images, and videos, and preliminary sub-region division is performed based on the analysis results. Different display content has different requirements for pixel distribution and photoelectric characteristics. For areas that frequently display text, due to the relatively regular pixel distribution and high detail requirements, smaller, regularly shaped sub-regions can more accurately detect potential defects in the text display, such as bright spots and dark spots. However, for areas displaying images or videos, considering the continuity and color variation characteristics of images, larger, more flexible sub-regions can better adapt to the display needs of images and videos, reducing detection errors caused by improper sub-region division.

[0069] Based on the pixel coordinates determined in step S13, the pixel distribution density and regularity are analyzed to obtain pixel analysis results. The pixel distribution density and regularity directly affect the division of sub-regions. In areas with dense and uniform pixel distribution, appropriately expanding the sub-region range can reduce the amount of subsequent data processing, because the pixel characteristics in these areas are relatively consistent, and defects can be accurately detected without overly fine division. However, in areas with sparse pixel distribution or obvious irregular distribution, reducing the sub-region range can ensure that the pixel distribution within each sub-region is relatively consistent, improving detection accuracy. For example, the pixel distribution in the edge area of ​​the display is relatively sparse, and dividing the edge sub-region into smaller ones can more accurately detect defects in the edge area; while the pixel distribution in the central display area is dense, and appropriately expanding the sub-region area can reduce the amount of data processing while ensuring detection accuracy.

[0070] Based on pixel analysis results, the specific shape and boundaries of each sub-region are determined. Common sub-region shapes include rectangles and circles. Rectangular sub-regions are easier to calculate and locate due to their regular boundaries, making data processing and algorithm implementation relatively simple. Circular sub-regions may better reflect pixel distribution characteristics in certain special cases, such as when pixel distribution is circular or approximately circular, using circular sub-regions can better adapt to the pixel distribution. When determining boundaries, efforts should be made to match the boundaries with the pixel distribution characteristics to avoid dividing the same pixel into different sub-regions or causing overlapping sub-regions. For example, for displays with specially shaped display areas, irregular polygonal sub-regions can be used to better adapt to the displayed content, ensuring that each sub-region accurately reflects the pixel characteristics and displayed content within that area, providing accurate region division basis for subsequent defect detection.

[0071] The effects of the above technical solution are as follows:

[0072] By rationally dividing the display into sub-regions, the accuracy of detection can be improved. For example, in cases where the display has a high resolution and the displayed content is complex, refining the sub-region division can more accurately capture the details of the displayed content.

[0073] Considering the circuit layout characteristics of the display, avoiding the separate division of important circuit areas or their spanning multiple sub-areas can effectively reduce the impact of circuit interference on the test results and improve the overall reliability of the test.

[0074] A flexible sub-region division strategy is adopted for different types of display content, so that the display needs of various types of content can be effectively met.

[0075] By analyzing the distribution density and regularity of pixels, and appropriately adjusting the size of sub-regions, unnecessary computational load and data processing burden can be reduced. For example, expanding sub-regions in densely distributed pixel areas and shrinking sub-regions in sparsely or irregularly distributed areas can effectively improve detection efficiency and accuracy.

[0076] Based on the characteristics of the displayed content, select an appropriate sub-region shape so that the boundary of the sub-region can better match the distribution of pixels, thereby avoiding unnecessary errors or overlaps and improving the accuracy of detection.

[0077] In one embodiment of the present invention, S2 includes:

[0078] S21. For each pixel, calculate the average brightness value based on brightness data under different time periods or different display screens; the average brightness value is calculated using the following formula:

[0079]

[0080] Among them, L avg (p) represents the weighted average brightness value of pixel p; N represents the total number of displayed frames (N≥1); ω i This represents the importance weight of the i-th display screen (0 < ω). i ≤1); n i This represents the number of samples (n) for the i-th frame. i ≥1); L p,i,t β represents the brightness value of pixel p at time t in the i-th frame; t Let β represent the time decay factor, and β∈(0,1);

[0081] S22. Based on the calculated average brightness value, the distribution of pixel brightness in the sub-region is statistically analyzed. By comparing the brightness data of different pixels and sub-regions, pixels or regions with abnormal brightness are identified. A brightness threshold is set, and when the brightness value of a pixel exceeds the normal range, it is determined to be abnormal.

[0082] S23. Extract color feature data of each pixel and sub-region through color analysis algorithm;

[0083] S24. Based on the extracted color feature data of each pixel and sub-region, calculate the deviation between the pixel and the standard color value; and analyze the consistency and uniformity of color within the sub-region.

[0084] S25. Measure the response time of each pixel and record the time required for a pixel to switch from one brightness state to another; analyze the distribution of response times and identify pixels with excessively long or short response times; set a response time threshold, and determine an abnormal response when the response time of a pixel exceeds the normal range; at the same time, combine the brightness and color characteristics of the pixel to comprehensively determine the cause of the abnormal response.

[0085] The working principle of the above technical solution is as follows: Brightness data is collected for each pixel at different time periods or under different display conditions. Different display conditions may have different levels of importance; for example, the brightness data of some scenes used to display key information or important content has a greater impact on evaluating display performance. By employing a weighted average algorithm, different weights are assigned according to the importance of different display conditions, which can more accurately calculate the average brightness value of each pixel.

[0086] Based on the calculated average brightness value, the distribution of pixel brightness within a sub-region is statistically analyzed. By comparing the brightness data of different pixels and sub-regions, pixels or regions with abnormal brightness are identified. A brightness threshold is set; when the brightness value of a pixel exceeds the normal range, it is determined to be abnormal. This threshold setting method can quickly filter out potentially problematic pixels or regions, providing a focus area for subsequent defect detection and improving detection efficiency.

[0087] Color analysis algorithms are used to extract color feature data for each pixel and sub-region, including chromaticity coordinates and color temperature. Chromaticity coordinates accurately describe the position of a pixel or sub-region's color in the color space, while color temperature reflects the warmth or coolness of the color. Color feature data is an important indicator for evaluating the color performance of a display. Extracting and analyzing this data allows us to understand the display's color performance in different areas, providing data support for subsequent color deviation analysis and consistency assessment.

[0088] Based on the extracted color feature data of each pixel and sub-region, the deviation between the pixel and the standard color value is calculated. The standard color value can be determined according to the display's design requirements or industry standards, and it is an important basis for measuring whether the display's color performance is up to standard. Color distance measurement methods, such as Euclidean distance and Manhattan distance, are used to quantify the color deviation. These distance measurement methods can convert color differences into numerical values, facilitating quantitative analysis and comparison of color deviation, and more accurately evaluating the color performance of pixels or sub-regions.

[0089] The system analyzes the color consistency and uniformity within a sub-region and calculates the standard deviation of pixel color deviation within that region. When the standard deviation exceeds a set threshold, it is considered to have poor color consistency. The standard deviation reflects the dispersion of pixel color deviation within a sub-region; a larger standard deviation indicates poorer color consistency. Simultaneously, the spatial distribution characteristics of color are considered to avoid misjudgments caused by localized color shifts. For example, in display areas with special patterns or gradient colors, localized color shifts may be a normal design effect. By comprehensively considering the spatial distribution characteristics of color, the color consistency of sub-regions can be judged more accurately, improving the accuracy of the detection.

[0090] The response time of each pixel is measured. The switching of the displayed images should have a certain regularity and randomness to cover different display states. The time required for a pixel to switch from one brightness state to another is recorded using a high-speed data acquisition system to accurately measure the response time. The monitor's refresh rate and grayscale level affect the pixel's response time, so these factors need to be considered during the measurement process to ensure the accuracy of the results. Accurate response time measurement provides important data for evaluating the monitor's dynamic display performance.

[0091] Analyzing the distribution of response times identifies pixels with excessively long or short response times. A response time threshold is set; when a pixel's response time exceeds the normal range, it is considered an abnormal response. Simultaneously, the brightness and color characteristics of the pixels are considered to comprehensively determine the cause of the abnormal response. For example, an excessively long response time may be related to the pixel's driving circuitry, while an excessively short response time accompanied by color abnormalities may be due to problems with the display materials or circuit design. By comprehensively considering multiple factors, the cause of the abnormal response can be more accurately located, providing targeted suggestions for monitor optimization and repair.

[0092] The effects of the above technical solution are as follows:

[0093] By calculating the average brightness value of each pixel using a weighted averaging algorithm, unnecessary interference factors in the displayed image can be effectively eliminated, ensuring accurate analysis of the monitor's brightness data. Statistical analysis of brightness distribution can clearly identify potential areas of abnormal brightness on the screen, improving display quality.

[0094] By setting a brightness threshold, abnormal brightness of pixels or regions can be automatically determined, avoiding human error and quickly locating abnormal areas, thus improving detection efficiency.

[0095] By employing color analysis algorithms and combining data such as chromaticity coordinates and color temperature, a comprehensive understanding of the display screen's color performance and consistency can be achieved. By calculating the standard deviation and spatial distribution of color deviation, misjudgments caused by localized color shifts can be identified and avoided, ensuring high standards of color consistency and uniformity.

[0096] Precisely measuring pixel response time through a high-speed data acquisition system helps analyze a monitor's performance during different screen transitions. It effectively identifies pixels with abnormal response times, especially in fast-moving or complex dynamic scenes, providing data support for improving monitor response speed. By systematically detecting and analyzing key indicators such as brightness, color, and response time, accurate performance evaluation and timely adjustments can be made, ultimately improving overall monitor quality and enhancing the user experience.

[0097] The formula described above, by incorporating weights for different display elements and a time decay factor, more accurately reflects the actual brightness of each pixel under varying display content. Especially for elements containing important information, higher weights ensure a more significant impact on the final brightness assessment. This helps obtain brightness data that better reflects actual display effects.

[0098] The introduction of a time decay factor takes into account the difference in importance between long-term or slowly changing brightness data and rapidly changing brightness data. By weighted averaging of the brightness values ​​of each pixel over different time periods, the impact of short-term brightness changes that do not have long-term effects on the results can be effectively reduced, thus yielding more stable and reliable brightness assessment results.

[0099] By statistically analyzing and detecting anomalies in the brightness distribution of different pixels, pixels or areas with abnormal brightness can be accurately identified. Setting brightness thresholds and comparing brightness data from different areas and pixels helps to promptly identify potential quality problems in the monitor, such as excessively high or low brightness in certain areas, thereby improving the monitor's quality inspection and maintenance capabilities.

[0100] By using color analysis algorithms and extracting color feature data, the color characteristics of each pixel and sub-region can be further analyzed, and the deviation from the standard color value can be calculated. This provides data support for evaluating the color consistency and uniformity of the display, and helps improve the color stability of the display under different screen conditions and scenarios.

[0101] By monitoring the response time of individual pixels and combining it with brightness and color characteristics, a comprehensive assessment of a monitor's response performance can be achieved. Identifying pixels with excessively long or short response times, and analyzing the interplay of brightness and color, can help pinpoint the cause of abnormal monitor response. For example, a slow response may affect the display quality of dynamic images, while an excessively fast response may lead to color distortion.

[0102] In one embodiment of the present invention, step S23 includes:

[0103] For each pixel, the original signal value related to color is extracted from the standard photoelectric signal data obtained from S1, and the extracted original signal value is preprocessed to obtain the standard signal value.

[0104] The original color information of a single pixel of the obtained standard signal value is converted according to the color space conversion formula to obtain the converted color feature data;

[0105] The color feature data of all pixels in the sub-region are converted and preliminarily statistically analyzed to obtain preliminary statistical results. Based on the preliminary statistical results, the color distribution characteristics of the sub-region are further analyzed. For each pixel, the color temperature is calculated based on its color feature data.

[0106] The color feature data of each extracted pixel and sub-region is recorded; at the same time, these color feature data are correlated with the pixel position information obtained in S1 and the data in subsequent analysis steps to establish a complete dataset.

[0107] The working principle of the above technical solution is as follows:

[0108] The color space used for color analysis is determined, such as RGB, HSV, CIEXYZ, and CIELab. Different color spaces have different characteristics and applicable scenarios. The RGB color space is an additive color space based on the three primary colors of red, green, and blue, widely used in display devices, but it has certain limitations in terms of the intuitiveness and uniformity of color representation. The HSV color space decomposes color into three components: hue, saturation, and value, which is more in line with human intuitive color perception and facilitates the adjustment of color brightness and saturation. The CIEXYZ color space is a color space based on the characteristics of human vision, and can more accurately describe human perception of color. The CIELab color space is developed based on the CIEXYZ color space, and has better color uniformity and perceptual consistency, making it suitable for precise color matching and distance measurement.

[0109] Choose an appropriate color space based on the characteristics of MicroLED displays and subsequent analysis requirements. For example, if the focus is on the intuitive perception of color and information on brightness and saturation, the HSV color space may be more suitable, as its three components directly correspond to the main dimensions of human color perception. For precise color matching and distance measurement, such as in display color calibration and quality testing, the CIELab color space may be more advantageous, as it can more accurately reflect the differences between colors.

[0110] For each pixel, the raw signal value related to color is extracted from the standard photoelectric signal data obtained from S1. In the RGB color space, these raw signal values ​​are the signal intensity values ​​of the red, green, and blue channels. These signal intensity values ​​reflect the luminous intensity of the pixel in different color channels and are the basic data for subsequent color analysis and processing.

[0111] The extracted raw signal values ​​are preprocessed to obtain standard signal values. The raw color information of individual pixels within the obtained standard signal values ​​is then converted using color space conversion formulas to obtain the converted color feature data. Conversions between different color spaces require adherence to corresponding conversion algorithms. For example, converting from RGB to HSV color space requires calculating the hue, saturation, and value of the pixel using an RGB-to-HSV conversion algorithm. Converting to CIELab color space involves a series of complex conversion steps, including conversion from RGB to XYZ color space, and then from XYZ to Lab, ultimately yielding the pixel's chromaticity coordinates (L, a, b) and other color feature data in the CIELab color space.

[0112] Preliminary statistical analysis is performed on the converted color feature data of all pixels within the sub-region to obtain preliminary statistical results. For each color component (such as the H, S, and V components in the HSV color space, or the L, a, and b components in the CIELab color space), the average, maximum, and minimum values ​​of that component within the sub-region are calculated. These statistical values ​​provide a preliminary understanding of the approximate range and concentration trend of colors within the sub-region. For example, if the average hue within the sub-region is close to a specific value, it indicates that the sub-region as a whole exhibits a certain hue tendency, which helps to quickly grasp the color characteristics of the sub-region.

[0113] Based on the preliminary statistical results, the distribution characteristics of colors within the sub-regions are further analyzed. The variance of each color component can be calculated to measure the dispersion of that color component within the sub-region. A larger variance indicates more drastic changes in that color component within the sub-region, and a more uneven color distribution. Simultaneously, a histogram of the color components is plotted to visually demonstrate the distribution of each color component within the sub-region.

[0114] For each pixel, the color temperature is calculated based on its color feature data. Color temperature is an important parameter describing the color characteristics of a light source and can be calculated using the relationship between color coordinates and the blackbody radiation trajectory. For example, in the CIEXYZ color space, color coordinates can be converted into color temperature values ​​using relevant formulas.

[0115] For each sub-region, the average color temperature of all pixels within that sub-region is calculated to reflect its overall color temperature characteristics. Simultaneously, the color temperature distribution within the sub-region is analyzed to determine if there are significant color temperature deviations or uneven distribution. Excessive color temperature deviations can affect the realism and comfort of the displayed image; for example, when displaying the same scene, color temperature differences in different areas may cause the image to look unnatural. Color temperature analysis allows for the timely detection and resolution of color temperature-related issues, improving the monitor's color quality.

[0116] The extracted color feature data for each pixel and sub-region, including chromaticity coordinates and color temperature, are recorded. The recorded data should include the pixel or sub-region's ID, location information, and corresponding color feature value. This recorded data is crucial for subsequent analysis and management, facilitating the querying and comparison of color characteristics across different pixels and sub-regions.

[0117] Simultaneously, these color feature data are correlated with the pixel location information obtained in S1 and data such as brightness and response time from subsequent analysis steps to establish a complete dataset. Through data correlation, multiple photoelectric characteristics of pixels can be comprehensively considered, allowing for a more holistic evaluation of the display's performance and quality. For example, when analyzing display defects, combining data such as color, brightness, and response time can more accurately determine the type and cause of the defects, providing strong support for display optimization and repair.

[0118] The effects of the above technical solution are as follows:

[0119] By selecting an appropriate color space for analysis, the most suitable color space can be chosen according to specific needs, thereby ensuring the accuracy of color analysis and the precision of intuitive perception.

[0120] After converting the color signal of each pixel, the resulting color features can provide more comprehensive color data support, which helps to analyze the color performance of the display in depth.

[0121] By calculating the mean, maximum, minimum, and variance of color components, we can gain a comprehensive understanding of the color concentration trend and distribution characteristics within a sub-region, helping to determine whether the colors are uniform and whether there are any anomalies or color deviations.

[0122] Color temperature calculation helps analyze the light source color characteristics of a display, assess the uniformity of color temperature distribution, avoid color temperature deviations affecting image realism and comfort, and further optimize display effects.

[0123] By linking color feature data with pixel location information, brightness, response time, and other data to form a comprehensive dataset, it is helpful for subsequent in-depth analysis and debugging, and provides a higher quality display effect.

[0124] By accurately analyzing colors and optimizing color temperature, the color performance of the monitor can be effectively improved, making the display effect more realistic and comfortable. It is especially suitable for application scenarios with high-precision color requirements, such as image processing and video production.

[0125] One embodiment of the present invention, such as Figure 2 As shown, S3 includes:

[0126] S31. Use environmental sensors to collect environmental data of the environment where the MicroLED display is located in real time, preprocess the collected environmental data, and obtain optimized preprocessed data.

[0127] S32. Measure the photoelectric characteristic parameters of MicroLED displays under different environmental conditions, obtain environmental impact results, and establish an environmental impact model between environmental factors and display photoelectric characteristics based on the optimized preprocessed data and the obtained environmental impact results.

[0128] S33. Train the model using the collected experimental data, validate the trained environmental impact model using cross-validation, and optimize and adjust the model based on the validation results.

[0129] S34. Based on the environmental impact model, correct the extracted photoelectric characteristic data; input the collected environmental data into the environmental impact model, calculate the influence of environmental factors on the photoelectric characteristic parameters, and subtract the influence from the original photoelectric characteristic parameters to obtain the corrected photoelectric characteristic parameters.

[0130] The working principle of the above technical solution is as follows:

[0131] Environmental sensors are used to collect key environmental data such as light intensity, temperature, and humidity in the environment where the MicroLED display is located in real time. The collected environmental data is preprocessed to obtain optimized preprocessed data. The photoelectric characteristic parameters of the MicroLED display, such as brightness, colorimetry, and response time, are measured under different environmental conditions to obtain the environmental impact results. By changing environmental conditions such as light intensity, temperature, and humidity, the changes in the display's photoelectric characteristic parameters under different environments are recorded. For example, measuring the display's brightness under different temperature environments allows observation of the effect of temperature on the display's luminous efficiency, thus obtaining the environmental impact results between temperature and brightness.

[0132] Using the optimized preprocessed data and the obtained environmental impact results, an environmental impact model is established between environmental factors and the optoelectronic properties of the display. This model aims to describe how environmental factors affect the optoelectronic properties of the display, quantifying this relationship through mathematical expressions or algorithms. For example, a multiple linear regression model can be used, with environmental factors such as light intensity, temperature, and humidity as independent variables, and optoelectronic property parameters such as brightness and colorimetry of the display as dependent variables. The model parameters are determined using fitting methods such as the least squares method, thereby establishing the environmental impact model.

[0133] The model is trained using collected experimental data. This data should include environmental data and corresponding photoelectric properties of the display under different conditions. By inputting a large amount of experimental data into the model and continuously adjusting its parameters, the model can better fit the actual environmental impact. During training, the model learns the inherent relationship between environmental factors and photoelectric properties, thereby improving its predictive accuracy.

[0134] Cross-validation is used to validate the trained environmental impact model. Cross-validation is a common model evaluation method that divides experimental data into multiple subsets, using one subset as the validation set and the remaining subsets as the training set, to train and validate the model multiple times. Cross-validation assesses the model's generalization ability on different datasets, avoiding overfitting or underfitting. Based on the validation results, the model is optimized and adjusted. If the validation results show a large prediction error, it may be necessary to adjust the model's structure or parameters, increase its complexity, or improve the training algorithm to enhance its performance.

[0135] The extracted photoelectric characteristic data are corrected based on the environmental impact model. First, the collected environmental data is input into the environmental impact model to calculate the impact of environmental factors on the photoelectric characteristic parameters. The environmental impact model can predict the specific impact of environmental factors on the photoelectric characteristic parameters of the display based on current environmental conditions. For example, based on current light intensity, temperature, and humidity, the environmental impact model calculates the impact of these environmental factors on the display brightness.

[0136] Subtracting this influencing factor from the original photoelectric characteristic parameters yields the corrected photoelectric characteristic parameters. Through correction, the influence of environmental factors on the display's photoelectric characteristic parameters can be eliminated, resulting in more accurate and realistic display photoelectric characteristic data. This is of great significance for display quality assessment, performance optimization, and fault diagnosis, improving the accuracy and reliability of assessment results and providing more scientific and effective support for display research and development and production.

[0137] The effects of the above technical solution are as follows:

[0138] By collecting environmental data in real time and establishing an environmental impact model, the photoelectric characteristics of MicroLED displays under different environmental conditions can be effectively evaluated and adjusted, ensuring the stability and reliability of the displays in various real-world environments.

[0139] By applying environmental impact models to the correction of optoelectronic characteristic parameters, the influence of environmental factors on display performance can be effectively removed, thereby obtaining more accurate and consistent optoelectronic characteristics and improving display quality.

[0140] The optimized photoelectric characteristics ensure the best performance of MicroLED displays under various environmental changes, ensuring that users can obtain ideal display effects under different lighting, temperature and humidity conditions.

[0141] By introducing cross-validation and model training methods from machine learning, this approach can intelligently quantify and predict the impact of environmental factors, improve the system's adaptability, and reduce the need for human intervention and adjustments.

[0142] Real-time environmental monitoring and automatic correction systems can identify performance changes of displays in specific environments in advance, reducing display quality degradation caused by environmental changes, thereby reducing the frequency and cost of maintenance and adjustments.

[0143] By continuously optimizing the display's environmental adaptability, reducing the impact of the environment on its performance, extending the lifespan of MicroLED displays, and enhancing their long-term stability.

[0144] In one embodiment of the present invention, S32 includes:

[0145] Based on the working principle and actual application scenarios of MicroLED displays, this study analyzes the impact of environmental factors on the optoelectronic properties of the displays; based on the extracted and analyzed optoelectronic characteristics, the key optoelectronic parameters for establishing an environmental impact model are determined.

[0146] Set a range of values ​​for each key environmental variable; use orthogonal experimental design to generate multiple different combinations of environmental conditions;

[0147] Based on the generated combination of environmental conditions, the experimental environment was set up sequentially, and after running stably under each environmental condition for a period of time, the photoelectric characteristic parameters of the display were collected; the collected environmental data and photoelectric characteristic parameter data were then visualized.

[0148] Statistical methods were used to calculate the correlation between environmental variables and photoelectric characteristic parameters. Based on the characteristics of the relationship between environmental variables and photoelectric characteristic parameters, different types of mathematical models were evaluated, and the structure of the environmental impact model was selected.

[0149] Based on the selected model structure, a parameter estimation method is chosen; using the collected experimental data, the model is trained using the selected parameter estimation method to estimate the model's parameter values, and an environmental impact model between environmental factors and the photoelectric properties of the display is established.

[0150] The working principle of the above technical solution is as follows:

[0151] Starting from the working principle and practical application scenarios of MicroLED displays, this paper delves into the impact of environmental factors such as light intensity, temperature, and humidity on the optoelectronic properties of the displays. MicroLED displays consist of numerous tiny MicroLED chips, and their light-emitting process involves the recombination of electrons and holes as well as the emission of photons. Changes in temperature affect the movement of charge carriers within the chip, such as electron mobility. Increased temperature may lead to increased electron mobility, but it also affects the chip's luminous efficiency. High temperatures increase the probability of non-radiative recombination, thus reducing luminous efficiency. Regarding humidity, the display's packaging structure needs to maintain a certain degree of airtightness. High humidity environments may cause moisture to seep into the packaging, affecting electrical connection performance and potentially corroding the chips or electrodes, thereby affecting the normal operation of the display. Light intensity primarily affects the display from a visual perception perspective. Excessive ambient light reduces the display's contrast, interfering with the human eye's visual perception of the displayed content. This analysis clarifies the potential influence and mechanisms of various environmental factors on the optoelectronic properties of displays, providing a theoretical foundation for subsequent research.

[0152] Based on the extracted and analyzed photoelectric characteristics, key photoelectric parameters for establishing environmental impact models were determined, such as brightness, chromaticity coordinates, contrast ratio, and response time. Brightness is a direct reflection of the display's luminous intensity, directly affecting the clarity of the displayed content perceived by the human eye; chromaticity coordinates describe the color characteristics of the light emitted by the display, and accurate color display is crucial for image and video presentation; contrast ratio reflects the display's ability to display bright and dark areas, affecting the sense of depth and three-dimensionality of the image; response time determines the display's response speed to input signals, and a fast response time can reduce image ghosting and improve the display quality of dynamic images. These key parameters comprehensively reflect the display's photoelectric characteristics and are important indicators for assessing the impact of environmental factors on the display.

[0153] Based on the normal operating range of MicroLED displays and the extreme environmental conditions they may encounter, reasonable value ranges were set for each key environmental variable. The temperature range was set from -20℃ to 80℃, because MicroLED displays may be exposed to varying temperature environments in practical applications, from cold outdoor environments to high-temperature industrial equipment interiors. Below -20℃, the chip material may become brittle and performance may degrade; above 80℃, the chip's luminous efficiency and reliability may be severely affected. The humidity range was set from 10%RH to 90%RH, considering the packaging stability and electrical performance of the display under different humidity conditions. Excessively low humidity may cause electrostatic problems, while excessively high humidity may lead to corrosion and short circuits. Different light intensity levels were set according to actual usage scenarios, such as indoor office environments and outdoor high-light environments, to comprehensively cover the lighting conditions the display may face. By reasonably setting these value ranges, it was ensured that the experiments could simulate various real-world environmental conditions, providing comprehensive data support for model building.

[0154] Orthogonal experimental design was employed to generate multiple combinations of different environmental conditions. Orthogonal experimental design is a highly efficient multi-factor experimental design method that can significantly reduce the number of experiments while ensuring the reliability of experimental results. By rationally arranging the levels of various environmental factors through orthogonal arrays, the different levels of each factor are evenly distributed in the experiment, thus enabling a comprehensive examination of the impact of various environmental factors and their interactions on the photoelectric properties of the display.

[0155] Based on the generated combination of environmental conditions, experimental environments were set sequentially, and after a period of stable operation under each condition, the photoelectric characteristic parameters of the display were collected. Stable operation for a period of time allows the display to reach thermal and electrical equilibrium under those conditions, ensuring that the collected photoelectric characteristic parameters accurately reflect the actual performance under those conditions. Data collection under each environmental condition was repeated multiple times to reduce the impact of random errors. Random errors may be caused by factors such as the accuracy of measuring instruments and minor fluctuations in environmental conditions. By repeatedly collecting data and averaging it, the accuracy and reliability of the data can be improved. Simultaneously, the environmental variable values ​​and corresponding photoelectric characteristic parameter values ​​under each experimental condition were recorded in detail, providing complete raw data for subsequent data analysis and model building. The collected environmental data and photoelectric characteristic parameter data were visualized. Scatter plots, line graphs, and other graphs were used to intuitively display the relationship between environmental variables and photoelectric characteristic parameters.

[0156] Statistical methods were used to calculate the correlation between environmental variables and photoelectric characteristic parameters. Correlation analysis quantifies the degree of association between the two, helping to determine whether the impact of environmental variables on photoelectric characteristic parameters is significant. Based on the characteristics of the relationship between environmental variables and photoelectric characteristic parameters, the applicability of different types of mathematical models was evaluated. Linear regression models are suitable for situations where there is a linear relationship between variables; they are computationally simple and easy to understand and interpret. Multinomial regression models can fit nonlinear relationships by adding polynomial terms to capture the curve trends in the data. Neural network models have strong nonlinear fitting capabilities and are suitable for complex relationships; they can automatically learn complex patterns and rules in the data. Support vector machine models also achieve good results with small sample data; they achieve classification or regression by finding the optimal hyperplane. For example, if there is a significant nonlinear relationship between environmental variables and photoelectric characteristic parameters, and the data volume is large, a neural network model may be a better choice because it can handle complex nonlinear relationships. If the data volume is small and the relationship is relatively simple, linear regression or multinomial regression models may be more suitable because they have lower computational complexity and provide good performance even with small samples.

[0157] Based on the selected model structure, choose an appropriate parameter estimation method. For linear regression models, the least squares method can be used for parameter estimation. The goal of the least squares method is to minimize the sum of squared errors between the model's predicted values ​​and the actual observed values. The model's parameter values ​​can be obtained by solving the normal equations. For neural network models, the backpropagation algorithm is used for training and parameter tuning. The backpropagation algorithm continuously optimizes the model's performance by calculating the gradient of the loss function with respect to the parameters of each layer and updating the parameters in the opposite direction of the gradient. For support vector machine models, it is necessary to choose an appropriate kernel function and parameters. The choice of kernel function determines how the data is mapped in high-dimensional space; different kernel functions are suitable for different types of data distributions. The choice of parameters affects the model's complexity and generalization ability.

[0158] Using the collected experimental data, the model was trained using a selected parameter estimation method to estimate the model's parameter values, establishing an environmental impact model between environmental factors and the photoelectric properties of the display. Through the training process, the model learned the mapping relationship between environmental variables and photoelectric property parameters, enabling it to predict the corresponding photoelectric property parameter values ​​based on given environmental conditions. This provides an important tool for subsequent photoelectric property calibration and performance evaluation.

[0159] The effects of the above technical solution are as follows:

[0160] By systematically analyzing the impact of environmental factors such as temperature, humidity, and light intensity on the photoelectric properties of displays, we can better understand and predict the performance of displays under different environments.

[0161] By setting appropriate ranges of environmental variables and generating combinations of experimental conditions, the impact of these environmental factors on the photoelectric properties of the display can be accurately quantified, ensuring stable display performance in practical applications.

[0162] By collecting data and using statistical methods to analyze the relationship between environmental variables and photoelectric properties, an accurate environmental impact model can be established, thereby providing a theoretical basis for the design and optimization of MicroLED displays and improving their adaptability and stability in various environments.

[0163] By evaluating and selecting various mathematical models, the most suitable model can be chosen for prediction and optimization based on different environmental impact characteristics, thereby improving the accuracy and practicality of the model.

[0164] Based on the established environmental impact model, designers can optimize the design and use of MicroLED displays for specific environmental conditions, ensuring that the display performance remains optimal in various complex environments.

[0165] Through visualization and statistical analysis, the experimental results can provide data support for the research and development and production of displays, helping manufacturers to test, adjust and optimize products under different environmental conditions, and enhance the market competitiveness of their products.

[0166] In one embodiment of the present invention, S34 includes:

[0167] S341. Obtain the timestamp information of the collected environmental data and extracted photoelectric characteristic data, and merge the photoelectric characteristic data into the same time interval as the environmental data according to the timestamp to form a one-to-one corresponding data pair; integrate the matched environmental data and photoelectric characteristic data into a complete data matrix; each row represents a data record at a specific time point or under specific environmental conditions.

[0168] S342. Check whether the format of the environmental data input into the environmental impact model meets the model requirements; load the model parameters from the optimized and adjusted environmental impact model; for each key environmental variable, input its value under specific environmental conditions into the environmental impact model, and calculate the impact of the environmental variable on each photoelectric characteristic parameter.

[0169] S343. And use the environmental impact model to calculate the comprehensive impact of the combined effect of various environmental variables on the photoelectric characteristic parameters under given environmental conditions; set the impact threshold for each photoelectric characteristic parameter;

[0170] S344. Compare the calculated comprehensive impact with the set threshold, identify abnormal impacts that exceed the threshold, and mark the abnormalities; take corresponding measures to handle the identified abnormal impacts.

[0171] S345. From the extracted raw photoelectric characteristic feature data, obtain the values ​​of each photoelectric characteristic parameter corresponding to the current environmental conditions; for each photoelectric characteristic parameter, use the calculated comprehensive influence quantity to correct the original parameter value; conduct a preliminary evaluation of the corrected photoelectric characteristic parameter to check whether it is within a reasonable range and whether it meets the performance expectations of the MicroLED display.

[0172] S346. If there are obvious problems with the correction results, such as some parameters still exceeding the reasonable range after correction, the correction results will be fed back to the environmental impact model establishment process in S32 to analyze whether there are defects in the model or errors in the calculation process in S34; based on the feedback information, the model will be further optimized and adjusted.

[0173] The working principle of the above technical solution is as follows:

[0174] The collected environmental data and extracted photoelectric characteristic data typically have their own timestamps. The timestamp records the specific moment the data was collected. Grouping the photoelectric characteristic data to the same time interval as the environmental data ensures a one-to-one correspondence between the environmental and photoelectric characteristic data in the time dimension. For example, environmental data of a temperature of 25℃ and humidity of 50%RH might be collected at a certain moment, while simultaneously obtaining data showing a display brightness of 300 cd / m² at that moment. 2 The photoelectric properties data, such as chromaticity coordinates (0.31, 0.32), are used to form a data pair.

[0175] The matched environmental data and photoelectric characteristic data are integrated into a complete data matrix. Each row represents a data record at a specific time point or under specific environmental conditions, including environmental variable values ​​and corresponding photoelectric characteristic parameter values. This data matrix integration method facilitates subsequent batch processing and analysis of the data, and can intuitively display the changes in the photoelectric characteristics of the display under different environmental conditions, providing a unified data format for model input and calculation.

[0176] Before inputting environmental data into an environmental impact model, it is necessary to check whether its format meets the model's requirements. Different models have different requirements for the format of input data; for example, some models may require input data to be within a specific numerical range or data type. If the environmental data format does not meet the requirements, appropriate format conversion is necessary. For example, the string-type temperature value "25℃" should be converted to the numeric type 25 so that the model can correctly recognize and process the input data.

[0177] The optimized environmental impact model loads model parameters, which represent the mapping relationships between environmental variables and photoelectric characteristic parameters learned during model training. For each key environmental variable (such as light intensity, temperature, and humidity), its value under specific environmental conditions is input into the environmental impact model to calculate the impact of that environmental variable on each photoelectric characteristic parameter. For example, first, temperature and humidity are fixed, and only light intensity is changed; the model calculates the impact of light intensity changes on photoelectric characteristic parameters such as luminance and chromaticity coordinates. Then, light intensity and humidity are fixed, and temperature is changed to calculate the impact of temperature changes, and so on.

[0178] Because multiple environmental variables may interact, their individual effects cannot be simply summed. For example, changes in temperature and humidity can jointly affect the luminous efficiency of a display. Increased temperature may lead to a decrease in chip luminous efficiency, while increased humidity may affect the performance of packaging materials, further impacting luminous efficiency. Therefore, it is necessary to comprehensively consider the effects of all environmental variables and use an environmental impact model to calculate the combined influence of each variable on photoelectric properties under given environmental conditions. By comprehensively considering the complex relationships between environmental variables, the model can more accurately predict the actual impact of environmental factors on the photoelectric properties of a display.

[0179] Based on the performance requirements and industry standards of MicroLED displays, threshold values ​​for the impact of environmental factors are set for each photoelectric characteristic parameter. These thresholds are set considering the normal operating range and performance stability of the display. When the impact of environmental factors on photoelectric characteristic parameters exceeds the set threshold range, it may indicate that the environmental factors are having an excessive impact on the display's performance, requiring special attention. For example, if the change in brightness due to environmental factors exceeds the normal fluctuation range, it may indicate that environmental factors are significantly interfering with the brightness stability of the display, potentially affecting the display's visual effects and user experience.

[0180] The calculated comprehensive impact is compared with a set threshold to identify and mark any abnormal impact exceeding the threshold. The occurrence of abnormal impact may indicate problems in data acquisition, experimental operation, or model prediction. Marking abnormalities facilitates further in-depth analysis and handling of these issues. The causes of abnormal impact may include environmental sensor malfunctions, experimental errors, or inaccurate model predictions for certain extreme environmental conditions. For example, a malfunctioning environmental sensor may lead to inaccurate environmental data, affecting the calculation of the comprehensive impact. Appropriate measures are needed for the identified abnormal impact. If the data anomaly is caused by an environmental sensor malfunction, environmental data needs to be re-acquired; if it's due to experimental errors, the experiment needs to be repeated; if the model prediction is inaccurate, the model needs further optimization and adjustment to improve its adaptability to various environmental conditions and its predictive accuracy.

[0181] From the extracted raw photoelectric characteristic data, the values ​​of each photoelectric characteristic parameter corresponding to the current environmental conditions are obtained. For each photoelectric characteristic parameter, the original parameter value is corrected using the calculated comprehensive influence quantity. The correction formula is: Corrected parameter value = Original parameter value - Comprehensive influence quantity. This correction method can eliminate the influence of environmental factors on photoelectric characteristic parameters, resulting in more accurate and realistic photoelectric characteristic data for the display. A preliminary evaluation of the corrected photoelectric characteristic parameters is then performed to check whether they are within a reasonable range and whether they meet the performance expectations of the MicroLED display. For example, whether the corrected brightness value is within the normal display range, and whether the chromaticity coordinates meet color accuracy requirements. This preliminary evaluation can promptly identify potential problems during the correction process, ensuring that the corrected data meets the display's performance requirements.

[0182] If the calibration results have significant problems, such as certain parameters still exceeding reasonable ranges after calibration, the calibration results are fed back into the environmental impact model building process in S32. The analysis determines whether there are defects in the model, such as insufficient coverage of extreme environmental conditions during training, leading to inaccurate predictions of impact quantities under certain environmental conditions; or errors in the calculation process in S34, such as an unreasonable method for calculating the comprehensive impact quantity. Based on the feedback information, the model is further optimized and adjusted.

[0183] The effects of the above technical solution are as follows:

[0184] By matching the collected environmental data and photoelectric characteristic data according to the timestamp, it is possible to ensure one-to-one correspondence of the data and integrate them into a complete data matrix, thereby improving the efficiency of data processing.

[0185] The system automatically checks and converts the data format input into the environmental impact model, avoiding the problem that the model cannot correctly identify and process data due to incompatible formats. This ensures the standardization and compatibility of the data and improves the stability and usability of the model.

[0186] This technical solution not only considers the impact of a single environmental factor on photoelectric characteristic parameters, but also the interaction between multiple environmental variables, avoiding the errors of simple summation. This multi-dimensional comprehensive analysis can more accurately reflect the true impact of the environment on display performance.

[0187] By comparing the calculated overall impact with a preset threshold, abnormal data can be identified and marked in a timely manner. Corresponding corrective measures can be taken when problems occur, ensuring the stability and reliability of the display.

[0188] By using comprehensive influence parameters to correct photoelectric characteristic parameters and conducting preliminary evaluations of the corrected parameters, the performance of the display is ensured to consistently meet expected standards. Even when problems arise, timely model optimization through feedback mechanisms minimizes performance deviations.

[0189] The solution utilizes feedback of anomalous data to facilitate further optimization and adjustment of the model. Through continuous iteration and improvement, the accuracy and reliability of the environmental impact model can be enhanced, ultimately improving the overall performance of MicroLED displays.

[0190] In one embodiment of the present invention, S341 includes:

[0191] From the sensor data file storing environmental data and the analysis result file storing photoelectric characteristic data, locate and read the corresponding data respectively; for the environmental data and photoelectric characteristic data, extract the timestamp information recorded in them respectively, and parse and standardize the extracted timestamps;

[0192] The time acquisition intervals of environmental data and photoelectric characteristic data are analyzed to obtain the time interval analysis results between the two.

[0193] Based on the time interval analysis results, a time alignment strategy is determined and a time alignment method is selected; according to the selected time alignment strategy, the photoelectric characteristic data are merged into the same time interval as the environmental data according to the timestamp;

[0194] During the data merging process, check for any missing data; record the matched environmental data and photoelectric characteristic data; define the structure of the data matrix and determine the meaning of the rows and columns of the matrix;

[0195] Based on the matching results, the data is populated into the defined matrix structure. In the order of rows and columns of the matrix, the timestamp, environmental variable values, and photoelectric characteristic parameter values ​​are filled into the corresponding positions to construct a complete data matrix. The constructed data matrix is ​​then verified to check the accuracy and consistency of the data.

[0196] The working principle of the above technical solution is as follows:

[0197] The corresponding data is located and read from both the sensor data file storing environmental data and the analysis result file storing photoelectric characteristic data. This process is akin to precisely locating the required "goods" in two different "warehouses." The sensor data file records various parameters of the environment at different times, while the analysis result file contains characteristic data of photoelectric properties under different conditions. By using specific file paths and reading methods, the data in these two files is extracted, providing a foundation for subsequent data processing and analysis.

[0198] For environmental data and photoelectric characteristic data, the timestamp information recorded in them is extracted. A timestamp is a "marker" of the data acquisition time, recording when the data was generated. The extracted timestamps may have different formats, such as Unix timestamps (the number of seconds since January 1, 1970, 00:00:00 UTC). To facilitate subsequent data alignment and processing, the extracted timestamps need to be parsed and standardized, converting various formats into a unified standard format, such as "YYYY-MM-DDHH:MM:SS".

[0199] The time acquisition intervals of environmental data and photoelectric characteristic data are analyzed to obtain the time interval analysis results. Environmental data may have a relatively fixed acquisition interval due to the sensor's operating mode, such as once per minute. This fixed acquisition interval results in a regular distribution of environmental data over time. Photoelectric characteristic data, on the other hand, may be continuously acquired but recorded at specific time points, or the acquisition interval may be variable, resulting in a more complex temporal distribution. By analyzing the time intervals of both, we can understand the characteristics of the data in the time dimension, providing a basis for selecting an appropriate time alignment strategy.

[0200] Based on the time interval analysis results, a time alignment strategy is determined, and a time alignment method is selected. Common time alignment methods include upward alignment, downward alignment, and nearest neighbor alignment. Upward alignment merges photoelectric characteristic data to the most recent environmental data time point that is greater than or equal to its timestamp. This method ensures that photoelectric characteristic data does not "miss" corresponding environmental data. Downward alignment merges data to the most recent environmental data time point that is less than or equal to its timestamp, which may cause some photoelectric characteristic data to be mapped to environmental data prematurely. Nearest neighbor alignment finds the environmental data time point that is closest to the timestamp of the photoelectric characteristic data and merges it. This method comprehensively considers data accuracy and computational efficiency.

[0201] Following the selected time alignment strategy, photoelectric characteristic data are merged into time intervals matching the environmental data based on their timestamps. For each environmental data time point, matching photoelectric characteristic data is searched. If nearest neighbor alignment is used, efficient algorithms such as binary search can be employed to quickly locate the closest time point. The binary search algorithm, by continuously halving the search range, can quickly find the environmental data time point closest to the timestamp of the photoelectric characteristic data, improving the efficiency of data merging. Through data merging, a one-to-one correspondence is achieved between environmental data and photoelectric characteristic data in time, providing an accurate data foundation for subsequent data analysis and model building.

[0202] During data merging, it's crucial to check for missing data. For example, some environmental data points may lack corresponding photoelectric characteristic data, or certain photoelectric characteristic data may lack suitable environmental data points for matching. Missing data can be handled differently depending on the specific circumstances. If the missing data has minimal impact on subsequent analysis, it can be deleted. If the missing data is significant, interpolation methods such as linear interpolation or polynomial interpolation can be used to fill in the missing values. By handling missing data, the integrity and accuracy of the data are ensured, preventing data gaps from affecting subsequent analysis results.

[0203] Record the matched environmental data and photoelectric characteristic data, including timestamps, environmental variable values ​​(light intensity, temperature, humidity, etc.), and corresponding photoelectric characteristic parameter values ​​(brightness, chromaticity coordinates, response time, etc.). Recording can be done using data structures such as lists, dictionaries, or database tables. Lists can store data sequentially, facilitating traversal and access; dictionaries can store data in key-value pairs for quick lookup; and database tables offer better data management and query capabilities, suitable for storing and processing large-scale data.

[0204] Define the structure of the data matrix, determining the meaning of its rows and columns. Typically, each row represents a data record at a specific point in time or under specific environmental conditions, and each column represents a variable, including environmental variables and photoelectric characteristic parameters. For example, the first column could be a timestamp, the second light intensity, the third temperature, the fourth humidity, and subsequent columns could be photoelectric characteristic parameters such as brightness, chromaticity coordinates, and response time. Based on the matching results, fill the data into the defined matrix structure. Following the row and column order of the matrix, sequentially fill in the timestamp, environmental variable values, and photoelectric characteristic parameter values ​​in their respective positions to construct the complete data matrix. Constructing the data matrix makes the data more structured and standardized, facilitating subsequent data analysis and model training.

[0205] Validate the constructed data matrix to check its accuracy and consistency. For example, check if the data is within a reasonable range and if the logical relationships between different variables meet expectations. If errors or anomalies are found, return to previous steps for correction, such as re-extracting data, adjusting the time alignment strategy, or handling missing data.

[0206] The effects of the above technical solution are as follows:

[0207] By standardizing the timestamps of environmental data and optoelectronic characteristic data, and using appropriate time alignment strategies, precise alignment of these two data sources across the time dimension can be ensured. This not only facilitates efficient data fusion but also improves the accuracy of subsequent analysis.

[0208] With multiple time alignment methods such as upward alignment, downward alignment, and nearest neighbor alignment, the most suitable alignment method can be selected according to the characteristics of the specific data, adapting to different acquisition modes.

[0209] To address the issue of missing data, the solution proposes several methods, such as deleting missing data pairs or using interpolation to fill in missing values. This flexible data completion approach ensures data integrity and reduces analytical errors caused by missing data.

[0210] By designing a reasonable data structure, environmental data and optoelectronic characteristic data can be stored and accessed efficiently.

[0211] During data processing, validation and consistency checks are performed to ensure the accuracy and reasonableness of the data. If abnormal data is found, it can be corrected in a timely manner to avoid affecting subsequent analysis results due to data errors.

[0212] By using efficient algorithms such as binary search for time alignment, the computational efficiency of data matching and merging can be significantly improved, and processing time can be reduced. Especially when the amount of data is large, it can ensure the system's response speed and stability.

[0213] This scheme, through standardized timestamp format, can effectively process data from different sources or formats, enhancing the system's versatility and adaptability.

[0214] In one embodiment of the present invention, step S4 includes:

[0215] S41. Collect a large amount of photoelectric characteristic parameter data of MicroLED display, and label it to determine whether each sample is a defective pixel and the type of defect; divide the labeled data into training set, validation set and test set;

[0216] S42. Train the preset machine learning model using the training set, validate the trained model using the validation set, and optimize the model based on the validation results.

[0217] S43. Input the corrected photoelectric characteristic parameters into the trained machine learning model to detect defective pixels and defective areas in the display; the model outputs the probability of each pixel being normal or defective based on the input feature data, and when the probability exceeds the set threshold, it is determined to be a defective pixel.

[0218] S44. Based on the detection results, classify different types of defects. At the same time, combine the determined pixel position coordinates and sub-region division information to determine the position coordinates and sub-region to which each defect belongs.

[0219] The working principle of the above technical solution is as follows:

[0220] A large amount of optoelectronic characteristic parameter data for MicroLED displays was collected, covering the optoelectronic performance of each pixel under different operating states and environmental conditions. The collected data was labeled to determine whether each sample was a defective pixel and the type of defect. This labeling process is like "tagging" the data, enabling the model to learn the feature differences between normal and defective pixels. For example, for bright spot defects, the labelers would record that the pixel's brightness value is significantly higher than surrounding normal pixels; for dark spot defects, they would record that its brightness value is significantly lower than the normal range. This labeling provides supervised learning samples for model training.

[0221] The labeled data is divided into training, validation, and test sets. The training set is used during model training, allowing the model to learn features and patterns in the data. The validation set is used to evaluate the model's performance during training, helping to adjust the model's parameters and structure, and preventing overfitting. The test set is used after model training is complete to evaluate its final generalization ability, verifying the model's accuracy and reliability on unknown data. This dataset partitioning method is standard practice in machine learning, ensuring that the model has good performance and generalization ability.

[0222] A pre-defined machine learning model is trained using a training set. During training, the model continuously adjusts its parameters based on the input photoelectric characteristic parameters to learn the feature differences between normal and defective pixels. For example, the model might learn specific distribution patterns of bright and chromaticity defective pixels on parameters such as brightness and chromaticity, and alternative distribution patterns of dark and chromatic pixels on these parameters. Through a large number of training samples, the model gradually establishes a mapping relationship from input features to output results.

[0223] The trained model is validated using a validation set to evaluate its performance. During validation, the validation set data is input into the model, and the model's output is compared to the actual labeled results. Based on the validation results, the model is fine-tuned. If the model's performance on the validation set is unsatisfactory, such as low accuracy or high false positive rate, it may be necessary to adjust model parameters, such as the learning rate and regularization coefficient; or change the model's structure, such as increasing or decreasing the number of neurons in the hidden layers or changing the activation function. Through continuous optimization, the model's performance on the validation set is improved, enabling it to better generalize to new data.

[0224] The corrected photoelectric characteristic parameters are input into the trained machine learning model. These corrected parameters more accurately reflect the actual state of the display, reducing the impact of environmental factors on the data. Based on the input feature data, the model uses the features and patterns learned during training to output the probability of each pixel being normal or defective. For example, the model calculates the similarity between the input data and the features of normal and defective pixels, outputting the corresponding probability value based on the similarity.

[0225] When the probability exceeds a set threshold, the pixel is identified as a defective pixel. The threshold is set based on a comprehensive consideration of model performance and actual needs. If the threshold is set too high, some real defective pixels may be missed; if the threshold is set too low, the false positive rate may increase, misclassifying some normal pixels as defective pixels. By setting a reasonable threshold, false positives can be minimized while ensuring detection accuracy.

[0226] Based on the detection results, different types of defects are classified, such as bright spots, dark spots, color casts, and uneven brightness. Different types of defects exhibit different characteristics in their photoelectric properties, and the model has learned these differences during training. For example, bright spot defects typically exhibit abnormally high brightness values, and their chromaticity coordinates may deviate from the normal range; dark spot defects exhibit abnormally low brightness values. By analyzing the defect information and feature data output by the model, defects can be accurately classified into their corresponding types.

[0227] By combining the determined pixel location coordinates with sub-region division information, the location coordinates and sub-region to which each defect belongs are determined. Pixel location coordinates record the specific position of the defect on the display screen, while sub-region division information divides the display screen into different regions, facilitating more refined management and analysis of defects. By combining the defect location coordinates with the sub-region division information, the sub-region in which each defect is located can be accurately pinpointed, providing crucial information for subsequent defect repair and display quality assessment.

[0228] The effects of the above technical solution are as follows:

[0229] By applying machine learning models, defective pixels in MicroLED displays can be detected automatically. This not only greatly improves detection efficiency and reduces manual intervention, but also ensures the stability and consistency of detection results, offering significant advantages, especially in large-scale production.

[0230] This solution enables precise defect classification and localization by combining pixel location and sub-region division. This refined classification and localization capability ensures comprehensive inspection of every detail of the display, improving overall display quality control.

[0231] By using training sets, validation sets, and tuning mechanisms, the model can be continuously optimized based on feedback from different datasets. This allows the model to maintain high detection accuracy when faced with different types of displays and defects.

[0232] By adopting this technical solution, defect detection can be completed more quickly and accurately, thereby improving the overall efficiency of the production line. At the same time, it effectively controls product quality, avoiding quality problems caused by human oversight or errors in inspection.

[0233] Automated inspection processes reduce reliance on manual inspection, thereby lowering labor costs, especially when dealing with a large number of displays, significantly reducing labor input.

[0234] The model can detect defects in real time during actual production and provide feedback, offering timely adjustment suggestions to production line personnel. It can promptly identify potential problems, reduce rework and scrap rates, and further improve production efficiency.

[0235] In one embodiment of the present invention, step S5 includes:

[0236] S51. Based on the defect detection results, calculate the defect index; compare the calculated defect index with the preset quality standard to evaluate whether the display meets the quality requirements; the defect index is obtained in the following way:

[0237] DII=ξ·D density +ζ·S severity +γ·W position

[0238] Among them, ξ+ζ+γ=3; 0.2≤ξ≤0.5; 0.3≤ζ≤0.6; 0.1≤γ≤0.3;

[0239] Among them, the defect density index D density Obtain it using the following formula:

[0240]

[0241] Where, N defect Indicates the total number of defective pixels; A total Indicates the total screen area; A max The maximum area of ​​a single defect is represented by λ; λ represents the area magnification factor (0.3-0.7).

[0242] Among them, the defect severity score S severity Obtain it using the following formula:

[0243]

[0244] Among them, w t Indicates the defect type weight (bright spot = 0.1, dark spot = 0.8, chromatic spot = 0.6); A ref Indicates the reference defect area;

[0245] Among them, the position weight coefficient W position Obtain it using the following formula:

[0246]

[0247] Where, d i This represents the distance from the defect center to the critical area (such as the logo area); k represents the attenuation coefficient; H represents the normalization coefficient.

[0248] S52. For the detected defects, analyze the causes of the defects in combination with the manufacturing process and material characteristics of the display; based on the results of the defect cause analysis, generate corresponding optimization suggestions;

[0249] S53. Integrate test results, defect information, evaluation results, and optimization suggestions into a detailed test report; output the generated test report in electronic or paper form and archive it.

[0250] The working principle of the above technical solution is as follows:

[0251] Based on the defect detection results, defect indicators such as the number of defective pixels and the severity of the defects are calculated. The number of defective pixels is relatively straightforward; by counting the total number of pixels identified as defects, one can gain a preliminary understanding of the distribution scale of defects in the display. However, assessing the severity of defects is more complex, requiring a comprehensive consideration of factors such as the type and size of the defect, and its impact on display quality. For example, the severity of a bright spot defect may be related to its size (i.e., the degree of abnormal pixel brightness) and its position on the screen (whether it is in a critical display area); the severity of a color cast defect can be quantified based on the magnitude of the color cast (the degree to which the chromaticity coordinates deviate from the normal range) and its impact on color display accuracy. This quantitative assessment allows for a more accurate description of the characteristics and impact of defects.

[0252] The calculated defect indicators are compared with preset quality standards to assess whether the monitor meets the quality requirements. Quality standards are the basis for measuring whether a monitor is up to standard, and are usually determined by industry specifications, company standards, or customer requirements. For example, quality standards might stipulate that the number of bright spots on a monitor screen of a certain area must not exceed a certain threshold, or the color shift must not exceed a certain range. By comparing the actual defect indicators with the quality standards, it can be determined whether the monitor meets the quality requirements, providing a basis for subsequent decision-making.

[0253] For the detected defects, we conduct an in-depth analysis of their causes, taking into account the monitor's manufacturing process and material characteristics. The monitor's manufacturing process involves multiple stages, such as pixel manufacturing, driver circuit design and installation, and color correction algorithm implementation; material characteristics include the performance of luminescent materials and the characteristics of driver circuit components. Different types and locations of defects may be caused by different reasons. For example, uneven brightness may be due to improper driver circuit parameter settings, leading to uneven current distribution among different pixels, thus causing brightness differences; it could also be due to inconsistent pixel manufacturing processes, resulting in different luminous efficiencies among different pixels. Color shift issues may be due to inaccurate color correction algorithms, failing to accurately convert the input signal into the correct color display; or it could be due to differences in luminescent material characteristics, causing deviations in the emitted color of different pixels. By comprehensively analyzing the manufacturing process and material characteristics, we can accurately pinpoint the root cause of the defects.

[0254] Based on the results of the defect cause analysis, corresponding optimization suggestions are generated. These suggestions aim to address the root causes of defects and improve display quality. For example, for uneven brightness, it is recommended to adjust the parameters of the drive circuit and optimize the current distribution among pixels to achieve more uniform brightness across different pixels; or to improve the manufacturing process of the pixels to enhance the consistency of luminous efficiency. For color shift issues, it is recommended to optimize the color correction algorithm to more accurately correct color deviations; or to screen and match luminescent materials to ensure that the emitted color of different pixels meets the requirements. By generating targeted optimization suggestions, specific directions and measures are provided for improving display quality.

[0255] The test results, defect information, evaluation results, and optimization suggestions are integrated into a detailed test report. The test report is a summary and record of the entire testing process and should include basic information about the monitor (such as model and specifications), test environment information (such as test temperature, humidity, and lighting conditions), test methods (such as the testing equipment and procedures used), test data (such as the number, location, and severity of defective pixels), a defect distribution map (visually showing the distribution of defects on the monitor screen), quality evaluation conclusions (whether it meets quality requirements), and optimization suggestions. By integrating this information, the quality status and existing problems of the monitor can be comprehensively and accurately reflected.

[0256] The generated test reports will be output in either electronic or paper format and archived. Electronic documents offer advantages in terms of ease of storage, transmission, and sharing, and can be quickly delivered to relevant personnel via the network; paper documents, on the other hand, are intuitive and stable, suitable for long-term preservation and retrieval. Archive management ensures the security and traceability of test reports, providing crucial evidence for subsequent quality traceability, product improvement, and customer feedback.

[0257] The effects of the above technical solution are as follows:

[0258] By calculating indicators such as the number and severity of defective pixels and comparing them with preset quality standards, the quality of a display can be accurately assessed. This assessment method reduces errors from manual inspection and provides a more scientific and objective quality control standard.

[0259] This solution combines the manufacturing process and material properties of the display to conduct an in-depth analysis of the causes of defects. This analysis not only pinpoints the source of the problem but also provides a clear direction for subsequent optimization.

[0260] Based on the defect root cause analysis, the generated optimization suggestions are highly targeted. For example, optimization suggestions for uneven brightness can be specific to adjusting current distribution, while optimization for color shift issues can start from algorithm adjustments and material selection. This refined suggestion can help manufacturers effectively solve practical problems and improve production efficiency and product consistency.

[0261] By promptly detecting, analyzing, and providing optimization suggestions, this solution helps manufacturers quickly locate and fix defects, reducing rework and waste caused by quality issues. This not only improves production efficiency but also reduces overall production costs.

[0262] By generating and archiving detailed test reports, manufacturers can clearly record the results and optimization suggestions for each test. This information management not only improves the transparency of the testing process but also provides strong support for subsequent quality tracking and improvement.

[0263] Long-term use of this solution can provide monitor manufacturers with a continuous quality monitoring mechanism, helping them to continuously optimize processes during production and enhance product competitiveness.

[0264] The above formula, by comprehensively considering three indicators—defect density, defect severity, and defect location—can provide a more comprehensive and objective assessment of monitor quality. Each indicator involves weighting multiple factors during calculation, thus accurately reflecting different aspects of defects, rather than simply counting the number or area of ​​defects. The defect density calculation formula, by considering the ratio of the total number of defective pixels to the total screen area, as well as the impact of the largest single defect area on the overall area, can reflect the overall defect distribution of the monitor, helping to determine the overall quality of the monitor.

[0265] The calculation of defect severity quantifies the impact of defects by considering factors such as defect type (bright spots, dark spots, color cast, etc.), defect size, and color variation, providing quantitative scores for different types of defects. This detailed assessment makes the detection results more accurate and provides data support for subsequent quality improvement. Through location weighting coefficients, the location of defects can be assigned weight, especially for defects in critical areas of the display (such as the logo area). This allows display quality assessment to focus more on areas that have a significant impact on user experience or display effect, particularly defects located in key visual areas.

[0266] Based on the results of defect root cause analysis, corresponding optimization suggestions can be generated. This not only helps improve product quality but also allows for targeted improvements to production processes or material properties, thereby reducing the occurrence of future defects.

[0267] This mathematical model makes the defect assessment process more standardized and systematic, providing display manufacturers and testing organizations with a unified assessment standard. This helps to promote consistent quality testing methods throughout the industry, improving the efficiency and transparency of quality management.

[0268] According to one embodiment of the present invention, an automated optoelectronic testing and defect detection system for a MicroLED display includes a memory, a processor, and a computer program stored in the memory and executable on the memory. The processor executes the program to implement the automated optoelectronic testing and defect detection method for a MicroLED display as described above.

[0269] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. An automated photoelectric testing and defect detection method for MicroLED displays, characterized in that, The method includes: S1. A high-precision photoelectric sensor array is used to perform a full scan of the MicroLED display to collect the initial photoelectric signal data of the display; the collected initial photoelectric signal data is preprocessed; standard photoelectric signal data is obtained; based on the layout and resolution information of the display, the obtained standard photoelectric signal data is used to locate the pixels and determine the position coordinates of each pixel; and the display is divided into multiple sub-regions according to the display area characteristics of the display. S2. Extract standard photoelectric signal data based on the division of multiple sub-regions to obtain photoelectric characteristic data; S3. Use environmental sensors to collect environmental data of the environment in which the MicroLED display is located in real time. Based on the collected environmental data, establish an environmental influence model between environmental factors and the photoelectric characteristics of the display. According to the environmental influence model, correct the extracted photoelectric characteristic data to eliminate the influence of environmental factors on the test results and obtain the corrected photoelectric characteristic parameters. S4. Combine the corrected photoelectric characteristic parameters to extract features related to defects; use machine learning algorithms to classify and identify the extracted defect features; input the corrected photoelectric characteristic parameters into the trained model to detect defective pixels and defective regions in the display. S5. Based on the defect detection results, evaluate the overall quality of the MicroLED display; for the detected defects, generate corresponding optimization suggestions based on the display's manufacturing process and material characteristics.

2. The automated photoelectric testing and defect detection method for MicroLED displays according to claim 1, characterized in that, S1 includes: S11. A high-precision photoelectric sensor array is used to perform a full scan of the MicroLED display to obtain initial photoelectric signal data; S12. Filter the initial photoelectric signal data to obtain filtered photoelectric signal data, and normalize the filtered photoelectric signal data to obtain normalized photoelectric signal data. S13. Detect and remove outliers in the obtained normalized photoelectric signal data to obtain standard photoelectric signal data. Based on the layout and resolution information of the display, locate the pixels in the standard photoelectric signal data. Determine the position coordinates of each pixel by analyzing the distribution characteristics of the signal and the physical structure of the display. S14. Divide the display into multiple sub-regions according to the display area characteristics of the display.

3. The automated photoelectric testing and defect detection method for MicroLED displays according to claim 1, characterized in that, The S2 includes: S21. For each pixel, calculate the average brightness value based on its brightness data under different time periods or different display screens. S22. Based on the calculated average brightness value, the distribution of pixel brightness within the sub-region is statistically analyzed; by comparing the brightness data of different pixels and sub-regions, pixels or regions with abnormal brightness are identified; a brightness threshold is set, and when the brightness value of a pixel exceeds the normal range, it is determined to be abnormal brightness. S23. Extract color feature data of each pixel and sub-region through color analysis algorithm; S24. Based on the extracted color feature data of each pixel and sub-region, calculate the deviation between the pixel and the standard color value; and analyze the consistency and uniformity of color within the sub-region. S25. Measure the response time of each pixel and record the time required for a pixel to switch from one brightness state to another; analyze the distribution of response times and identify pixels with excessively long or short response times; set a response time threshold, and determine an abnormal response when the response time of a pixel exceeds the normal range; at the same time, combine the brightness and color characteristics of the pixel to comprehensively determine the cause of the abnormal response.

4. The automated photoelectric testing and defect detection method for MicroLED displays according to claim 1, characterized in that, The S3 includes: S31. Use environmental sensors to collect environmental data of the environment where the MicroLED display is located in real time, preprocess the collected environmental data, and obtain optimized preprocessed data. S32. Measure the photoelectric characteristic parameters of MicroLED displays under different environmental conditions, obtain environmental impact results, and establish an environmental impact model between environmental factors and display photoelectric characteristics based on the optimized preprocessed data and the obtained environmental impact results. S33. Train the model using the collected experimental data, validate the trained environmental impact model using cross-validation, and optimize and adjust the model based on the validation results. S34. Based on the environmental impact model, correct the extracted photoelectric characteristic data; input the collected environmental data into the environmental impact model, calculate the influence of environmental factors on the photoelectric characteristic parameters, and subtract the influence from the original photoelectric characteristic parameters to obtain the corrected photoelectric characteristic parameters.

5. The automated photoelectric testing and defect detection method for MicroLED displays according to claim 4, characterized in that, S32 includes: Based on the working principle and actual application scenarios of MicroLED displays, this study analyzes the impact of environmental factors on the optoelectronic properties of the displays; based on the extracted and analyzed optoelectronic characteristics, the key optoelectronic parameters for establishing an environmental impact model are determined. Set a range of values ​​for each key environmental variable; use orthogonal experimental design to generate multiple different combinations of environmental conditions; Based on the generated combination of environmental conditions, the experimental environment was set up sequentially, and after running stably under each environmental condition for a period of time, the photoelectric characteristic parameters of the display were collected; the collected environmental data and photoelectric characteristic parameter data were then visualized. Statistical methods were used to calculate the correlation between environmental variables and photoelectric characteristic parameters. Based on the characteristics of the relationship between environmental variables and photoelectric characteristic parameters, different types of mathematical models were evaluated, and the structure of the environmental impact model was selected. Based on the selected model structure, a parameter estimation method is chosen; using the collected experimental data, the model is trained using the selected parameter estimation method to estimate the model's parameter values, and an environmental impact model between environmental factors and the photoelectric properties of the display is established.

6. The automated photoelectric testing and defect detection method for MicroLED displays according to claim 4, characterized in that, S34 includes: S341. Obtain the timestamp information of the collected environmental data and extracted photoelectric characteristic data, and merge the photoelectric characteristic data into the same time interval as the environmental data according to the timestamp to form a one-to-one corresponding data pair; integrate the matched environmental data and photoelectric characteristic data into a complete data matrix; each row represents a data record at a specific time point or under specific environmental conditions. S342. Check whether the format of the environmental data input into the environmental impact model meets the model requirements; load the model parameters from the optimized and adjusted environmental impact model; for each key environmental variable, input its value under specific environmental conditions into the environmental impact model, and calculate the impact of the environmental variable on each photoelectric characteristic parameter. S343. And use the environmental impact model to calculate the comprehensive impact of the combined effect of various environmental variables on the photoelectric characteristic parameters under given environmental conditions; set the impact threshold for each photoelectric characteristic parameter; S344. Compare the calculated comprehensive impact with the set threshold, identify abnormal impacts that exceed the threshold, and mark the abnormalities; take corresponding measures to handle the identified abnormal impacts. S345. From the extracted raw photoelectric characteristic feature data, obtain the values ​​of each photoelectric characteristic parameter corresponding to the current environmental conditions; for each photoelectric characteristic parameter, use the calculated comprehensive influence quantity to correct the original parameter value; conduct a preliminary evaluation of the corrected photoelectric characteristic parameter to check whether it is within a reasonable range and whether it meets the performance expectations of the MicroLED display. S346. If there are obvious problems with the correction results, such as some parameters still exceeding the reasonable range after correction, the correction results will be fed back to the environmental impact model establishment process in S32 to analyze whether there are defects in the model or whether there are errors in the calculation process in S34; based on the feedback information, the model will be further optimized and adjusted.

7. The automated optoelectronic testing and defect detection method for MicroLED displays according to claim 6, characterized in that, S341 includes: From the sensor data file storing environmental data and the analysis result file storing photoelectric characteristic data, locate and read the corresponding data respectively; for the environmental data and photoelectric characteristic data, extract the timestamp information recorded in them respectively, and parse and standardize the extracted timestamps; The time acquisition intervals of environmental data and photoelectric characteristic data are analyzed to obtain the time interval analysis results between the two. Based on the time interval analysis results, a time alignment strategy is determined and a time alignment method is selected; according to the selected time alignment strategy, the photoelectric characteristic data are merged into the same time interval as the environmental data according to the timestamp; During the data merging process, check for any missing data; record the matched environmental data and photoelectric characteristic data; define the structure of the data matrix and determine the meaning of the rows and columns of the matrix; Based on the matching results, the data is populated into the defined matrix structure. In the order of rows and columns of the matrix, the timestamp, environmental variable values, and photoelectric characteristic parameter values ​​are filled into the corresponding positions to construct a complete data matrix. The constructed data matrix is ​​then verified to check the accuracy and consistency of the data.

8. The automated optoelectronic testing and defect detection method for MicroLED displays according to claim 1, characterized in that, The S4 includes: S41. Collect a large amount of photoelectric characteristic parameter data of MicroLED display, and label it to determine whether each sample is a defective pixel and the type of defect; divide the labeled data into training set, validation set and test set; S42. Train the preset machine learning model using the training set, validate the trained model using the validation set, and optimize the model based on the validation results. S43. Input the corrected photoelectric characteristic parameters into the trained machine learning model to detect defective pixels and defective areas in the display; the model outputs the probability of each pixel being normal or defective based on the input feature data, and when the probability exceeds the set threshold, it is determined to be a defective pixel. S44. Based on the detection results, classify different types of defects. At the same time, combine the determined pixel position coordinates and sub-region division information to determine the position coordinates and sub-region to which each defect belongs.

9. The automated photoelectric testing and defect detection method for MicroLED displays according to claim 1, characterized in that, The S5 includes: S51. Based on the defect detection results, calculate the defect index; compare the calculated defect index with the preset quality standard to evaluate whether the display meets the quality requirements. S52. For the detected defects, analyze the causes of the defects in combination with the manufacturing process and material characteristics of the display; based on the results of the defect cause analysis, generate corresponding optimization suggestions; S53. Integrate test results, defect information, evaluation results, and optimization suggestions into a detailed test report; output the generated test report in electronic or paper form and archive it.

10. An automated optoelectronic testing and defect detection system for MicroLED displays, characterized in that, The method includes a memory, a processor, and a computer program stored on and executable on the memory, wherein the processor executes the program to implement an automated optoelectronic testing and defect detection method for a MicroLED display as described in any one of claims 1-9.

Images