Quantitative testing method for ghost shadow of silicon-based OLED micro display screen

Abstract

The invention relates to the technical field of testing of OLED (Organic Light Emitting Diode) display screens, in particular to a residual shadow quantitative testing method of a silicon-based OLED micro display screen, which is characterized in that patterns with different shapes and different contrast ratios are used as testing pictures to acquire gray values of pixels of pure color pictures before and after testing; the method comprises the following steps of: identifying and dividing regions with image differences in a test picture, sequentially carrying out Gaussian smoothing filtering and binarization processing, carrying out Laplace boundary extraction on binarization data information, and comparing the binarization data of different regions, so as to obtain a quantization value of the ghost shadow degree of the test display screen. The method maximizes the difference between the foreground and the background and the average gray scale, and can objectively, accurately and effectively evaluate the image ghosting performance of the display image.

Description

Technical Field

[0001] This invention relates to the field of OLED display testing technology, and in particular to a method for quantifying image retention in silicon-based OLED microdisplays. Background Technology

[0002] Silicon-based OLED is a new type of display that combines OLED display technology with silicon-based CMOS circuitry. With its excellent characteristics such as self-emission, thinness, impact resistance, wide viewing angle, short response time, good high and low temperature characteristics, high luminous efficiency, high pixel density, low power consumption, high integration, and portability, it can be well-suited for use in near-eye display devices such as VR and AR. However, due to the lifespan and degradation of OLED device structures, under fixed brightness and fixed image display scenarios, after a period of time, a severe image retention phenomenon, or ghosting, can be observed visually, affecting product performance and limiting its wider application.

[0003] Currently, most display manufacturers assess the degree of image retention through visual observation. However, due to individual differences, this method often results in highly subjective and inconsistent judgments, making it difficult to effectively and accurately determine the display quality and hindering targeted improvements to processes or components. Furthermore, the small size of micro-displays, typically within one inch, significantly increases the difficulty of visual observation, and monochrome displays (such as green displays) are highly distracting, making it impossible to visually assess image retention time in a timely manner. Summary of the Invention

[0004] The main technical problem solved by this invention is to provide a silicon-based OLED microdisplay image retention quantification detection system and method, which can use image retention test images of different types of displays to quantify the severity of different retentions, thereby achieving an objective, accurate and effective evaluation of the image retention performance.

[0005] To address the aforementioned technical problems, this invention proposes a method for quantifying image retention based on image processing. The method uses patterns of different shapes and contrasts as test images. An industrial camera is used to capture solid-color images of the display screen before and after the test. The grayscale values ​​of pixels in the solid-color images before and after the test are collected. Regions in the test images where differences exist are identified and divided. Then, Gaussian smoothing filtering and binarization are performed sequentially. The binarized data is then subjected to Laplacian boundary extraction, and the binarized data from different regions are compared to obtain the quantified value of the degree of image retention on the test display screen.

[0006] Specifically, the Gaussian smoothing filter method involves performing a two-dimensional Gaussian kernel convolution on the grayscale values ​​of the acquired image pixels to remove random noise and retain boundary information in a weighted averaging manner, thereby obtaining a smooth image required for dividing image regions and calculating the mode of grayscale, thus suppressing random noise while retaining boundary information.

[0007] Specifically, the binarization method involves extracting the mode of grayscale values ​​from the grayscale information collected in each region of the image: ; Where M0 represents the mode of grayscale values; U represents the upper limit of the group containing the mode; Δ1 represents the difference in the number of pixels between the group containing the mode (i.e., the grayscale range where the pixel appears most frequently) and its adjacent lower grayscale group; Δ2 represents the difference in the number of pixels between the group containing the mode and its adjacent higher grayscale group; and d represents the group interval of the group containing the mode.

[0008] Next, calculate the arithmetic mean of the gray values ​​within the region: ; in, μ This represents the arithmetic mean of the grayscale values ​​within the region. α Represents the grayscale value of a single pixel within the region; n This represents the number of pixels contained within the region, n = 1, 2, 3, 4, 5…∞; The mode Mo is compared with the arithmetic mean μ to identify pixels with gray levels greater than or less than the mode within the region. The image information of this region is then binarized. The mode is used as the binarization threshold. Regions with gray levels higher than or equal to the mode are set to 255, and these regions are designated as the foreground. Regions with gray levels lower than the mode are set to 0, and these regions are designated as the background. The binarized data is then subjected to Laplacian boundary extraction. The percentage of foreground pixels relative to the total number of pixels in the region is calculated and denoted as P0. The percentage difference φ between a single region and its surrounding regions is calculated, and this difference represents the degree of image retention in that region. The arithmetic mean of the image retention differences across the entire display is then calculated to obtain the image retention degree value for the entire screen.

[0009] Furthermore, the method for extracting the Laplacian boundary from binarized data is as follows: ; in, This represents a data set of the image after Laplacian boundary sharpening; This represents the data set of the input image, i.e., the data set after binarization. Indicates the center coefficient of the mask; — An n-order Laplacian operator, where n is related to the number of pixels in a single region; Furthermore, for a single region f i (x, y), i = 0, 1, 2…, are segmented into a single region f using an arbitrary segmentation threshold T. i The pixels in (x, y) are divided into foreground and background. The percentage of foreground pixels to the total number of pixels in the region is denoted as P, the mean gray level is µ0, the background percentage is Q, and the mean gray level is μ1. The value of T ranges from 0 ≤ T ≤ 255. The total average gray level of a single region is denoted as µ. Let the number of pixels in a single region of the image be M×N, the number of foreground pixels in the region be denoted as N0, and the number of background pixels be denoted as N1, then we have: The proportions of foreground and background pixels relative to a single area, P and Q: P = N0 / M × N; Q = N1 / M × N; N0 + N1 = M × N; P + Q = 1; Arithmetic mean of pixels in a single region, µ: µ = Pµ0 + Qµ1; The difference φ between a single area and its surrounding areas: ; n is the number of neighboring regions surrounding a single region, n = 1, 2, 3...∞; The overall difference value φ of the entire acquired image m : ; q represents the regions of varying contrast contained in the entire image, where q = 1, 2, 3…∞; Calculate the foreground pixel ratio of each region, then compare the foreground pixel ratio of a single region with the foreground pixel ratio of its surrounding regions to calculate the regional difference value. Finally, the foreground proportion difference value of the entire image is calculated by accumulating the difference values ​​of individual regions. This value represents the current depth of the image's afterimage.

[0010] The testing method of this invention divides an image into foreground and background parts by selecting an appropriate segmentation threshold, maximizing the difference between the foreground and background and the average grayscale. This quantifies the severity of different afterimages, enabling an objective, accurate, and effective evaluation of the afterimage performance of the displayed image. This evaluation mechanism is closer to the human eye's perception of afterimages and has a wide range of applications. Attached Figure Description

[0011] Figure 1 The flowchart shows the implementation process of the image retention quantification test method for silicon-based OLED microdisplays.

[0012] Figure 2 Image boundary effect diagram extracted by Laplacian boundary sharpening algorithm.

[0013] Figure 3 This is the result of binarization. Detailed Implementation

[0014] Example 1: This invention proposes a method for quantifying image retention in silicon-based OLED microdisplays, employing an industrial-grade CMOS digital area array camera with a resolution of 5 megapixels. To quickly and conveniently measure the initial brightness of the display and determine the camera exposure time, a CS150 colorimeter is used for brightness measurement. This model of luminance meter has a brightness accuracy of ≤±2% and a brightness repeatability of ≤±0.2% at a 1° test field angle. The embodiments of this invention are further described in detail below with reference to the accompanying drawings.

[0015] like Figure 1 As shown, the method for determining the quantitative evaluation of afterimages provided in this embodiment of the invention includes the following steps: 1) Before performing the image retention test on the display screen, display a monochrome image on the display screen and test the initial brightness value of the display screen at this time.

[0016] 2) Adjust the camera exposure time to ensure that the grayscale value of the image captured by the camera in real time is consistent with the initial brightness value of the display screen, with a deviation of ±5 cd / m². 2 , capture the solid color image I0 of the current display screen.

[0017] 3) Perform a ghosting test on the display screen using high-contrast images of different colors, and use the Laplacian algorithm to extract the boundaries of different color regions, confirm the specific number of regions, and complete the region boundary division. After 10 minutes of continuous display testing, switch the display screen to a solid color image and capture the current display image I1.

[0018] 4) Using the method of extracting the mode of the image, extract the mode value in different regions, and use the mode as the threshold to segment the display images collected before and after the test, calculate whether different individual regions are macroscopically darker or brighter, use the same threshold for each pixel in the image, set the gray value of the pixels in the image to 0 if it is less than the threshold, and set the gray value of the pixels greater than the threshold to 255, and convert the image into a binary image.

[0019] 5) Calculate the percentage of brightness difference in this area, and finally calculate the percentage of brightness difference in the entire image, which is the afterimage quantization value.

Claims

1. A method for quantifying image retention based on image processing, characterized in that... This method uses patterns of different shapes and contrasts as test images. An industrial camera is used to capture solid color images of the display screen before and after the test. The grayscale values ​​of the pixels in the solid color images before and after the test are collected. The regions in the test images with differences are identified and divided. Then, Gaussian smoothing filtering and binarization processing are performed in sequence. The binarized data information is extracted using Laplacian boundary extraction and the binarized data of different regions are compared to obtain the quantification value of the afterimage degree of the test display screen.

2. The image processing-based ghosting quantization test method as described in claim 1, characterized in that... The specific method of Gaussian smoothing filtering is to perform two-dimensional Gaussian kernel convolution on the gray values ​​of the acquired image pixels, remove random noise and retain boundary information in a weighted average manner, thereby obtaining a smooth image required for dividing image regions and calculating the mode of gray values, thus suppressing random noise while retaining boundary information.

3. The image processing-based ghosting quantization test method as described in claim 1, characterized in that... The binarization method described above extracts the mode of grayscale values ​​from the grayscale information collected in each region of the image. ； Where M0 represents the mode of grayscale values; U represents the upper limit of the group containing the mode; Δ1 represents the difference in the number of pixels between the group containing the mode (i.e., the grayscale range where the pixel appears most frequently) and its adjacent lower grayscale group; Δ2 represents the difference in the number of pixels between the group containing the mode and its adjacent higher grayscale group; and d represents the group interval of the group containing the mode. Next, calculate the arithmetic mean of the gray values ​​within the region: ； in, μ This represents the arithmetic mean of the grayscale values ​​within the region. α Represents the grayscale value of a single pixel within the region; n This represents the number of pixels contained within the region, n = 1, 2, 3, 4, 5…∞; The mode Mo is compared with the arithmetic mean μ to identify pixels with gray values ​​greater than or less than the mode within the region. The image information of this region is then binarized. The mode is used as the binarization threshold. Regions with gray values ​​higher than or equal to the mode are set to 255, and these regions are designated as the foreground. Regions with gray values ​​lower than the mode are set to 0, and these regions are designated as the background. The binarized data is then subjected to Laplacian boundary extraction. The percentage of foreground pixels relative to the total number of pixels in the region is calculated and denoted as P0. The percentage difference φ between a single region and its surrounding regions is calculated, and this difference represents the degree of image retention in that region. The arithmetic mean of the image retention differences across the entire display is then calculated to obtain the image retention degree value for the entire screen.

4. The image processing-based ghosting quantization test method as described in claim 1, characterized in that... The method for extracting the Laplacian boundary from the binarized data information is as follows: ； in, This represents a data set of the image after Laplacian boundary sharpening; This represents the data set of the input image, i.e., the data set after binarization. Indicates the center coefficient of the mask; — An n-order Laplacian operator, where n is related to the number of pixels in a single region; For a single region f i (x, y), i = 0, 1, 2…, are segmented into a single region f using an arbitrary segmentation threshold T. i The pixels in (x, y) are divided into foreground and background. The percentage of foreground pixels to the total number of pixels in the region is denoted as P, the mean gray level is µ0, the background percentage is Q, and the mean gray level is μ1. The value of T ranges from 0 ≤ T ≤ 255. The total average gray level of a single region is denoted as µ. Let the number of pixels in a single region of the image be M×N, the number of foreground pixels in the region be denoted as N0, and the number of background pixels be denoted as N1, then we have: The proportions of foreground and background pixels relative to a single area, P and Q: P = N0 / M × N; Q = N1 / M × N; N0 + N1 = M × N; P + Q = 1; Arithmetic mean of pixels in a single region, µ: µ = Pµ0 + Qµ1; The difference φ between a single area and its surrounding areas: ； n is the number of neighboring regions surrounding a single region, n = 1, 2, 3...∞; The overall difference value φ of the entire acquired image m : ； q represents the regions of varying contrast contained in the entire image, where q = 1, 2, 3…∞; Calculate the foreground pixel ratio of each region, then compare the foreground pixel ratio of a single region with the foreground pixel ratio of its surrounding regions to calculate the regional difference value. Finally, the foreground proportion difference value of the entire image is calculated by accumulating the difference values ​​of individual regions. This value represents the current depth of the image's afterimage.

Images