Display screen Mura defect detection method

Abstract

The invention provides a display screen Mura defect detection method, and the method comprises the steps: carrying out the image collection of the surface of a display screen through a camera, and obtaining original image data; carrying out texture feature enhancement on the original image by adopting an improved LoG operator; an adaptive threshold segmentation method is used to process the image after the textural features are enhanced, and a Mura defect area is extracted; calculating the severity of the Mura defect in combination with an area weighted quantization model; and performing visibility scoring on the Mura defect by using an HVS (Human Visual System) model. The problems that in the prior art, Mura defect detection of the display screen is insufficient in precision, low in efficiency and the like are effectively solved, and the detection accuracy and efficiency are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of screen defect detection, and particularly relates to a display screen Mura defect detection method. BACKGROUND

[0002] With the rapid development of science and technology, display screens are applied in various industries. For example, electronic products such as mobile phones are generally equipped with OLED screens. OLED screens have been widely used in the field of smart phones due to their self-luminous, high contrast, wide viewing angle, thinness and other significant advantages, and have become one of the important factors for consumers to choose and purchase mobile phones.

[0003] However, in the production and manufacturing process of OLED screens, Mura defects, i.e. areas with uneven brightness on the screen surface, are prone to occur due to factors such as material properties and process level. Although this defect may not be easily detected by the naked eye in some cases, it will seriously affect the display effect of the screen under certain display pictures, reduce the visual experience of users, and thus affect the market competitiveness of the product.

[0004] At present, in the detection of Mura defects of display screens, traditional manual visual detection methods are still used by some enterprises. This method mainly relies on the naked eye observation and experience judgment of the detection personnel, and has many disadvantages. On the one hand, the subjective factors of the detection personnel have a great influence on the detection results, and the judgment standards of different detection personnel are difficult to unify, resulting in a lack of consistency and accuracy of the detection results; on the other hand, the manual detection efficiency is very low, which is difficult to meet the modern large-scale and high-efficiency production demand, and for some subtle Mura defects, the human eye is difficult to accurately identify, and the missed detection situation is easy to occur.

[0005] Automatic detection technology based on ordinary image processing algorithms is also developing, but due to the small difference between the gray scale change of Mura defects and the normal screen texture gray scale, the existing algorithms face great difficulties in distinguishing the defect area and the normal area, resulting in insufficient detection accuracy and high misjudgment rate. For example, when dealing with small Mura defects in a complex texture background, the algorithm may misjudge the normal texture fluctuation as a defect, or fail to identify the real defect area, which cannot meet the strict requirements of high-quality production of mobile phone OLED screens. SUMMARY

[0006] Therefore, it is necessary for the present application to provide a display screen Mura defect detection method to effectively overcome the problems of insufficient detection accuracy and low efficiency in the prior art for detecting Mura defects of display screens, and to improve the accuracy and efficiency of detection.

[0007] The display screen Mura defect detection method provided by the present application comprises:

[0008] The surface of the display screen is imaged by a camera to obtain raw image data;

[0009] The raw image is enhanced in texture features by using an improved LoG operator;

[0010] The image after the enhanced texture features is processed by using an adaptive threshold segmentation method to extract a Mura defect region;

[0011] The severity of the Mura defect is calculated by combining an area-weighted quantization model;

[0012] The Mura defect is scored in visibility by using a human visual system (HVS) model.

[0013] Further, the calculation formula of the LoG operator is:

[0014]

[0015] where I(x, y) is the raw image data, and sigma is a scale parameter.

[0016] Further, the formula for dynamically adjusting sigma according to the local image variance of the current detection region is sigma = sigma0*(1 + alpha*var(I RoI )), where sigma0 is a reference scale, alpha is a sensitivity coefficient, I RoI is the current detection region, and var(I RoI ) is the local image variance of the current detection region (RoI).

[0017] Further, the formula for adaptive threshold segmentation is:

[0018] T(x, y) = mu(x, y) + k*sigma(x, y), k e [1.5, 3.0];

[0019] where mu(x, y) and sigma(x, y) are the local mean and standard deviation, respectively.

[0020] Further, when the adaptive threshold segmentation method is run to process the image after the enhanced texture features, the image is divided into m x n local windows, and m, n >= 20, and the threshold T k is independently calculated for each window.

[0021] Further, when the adaptive threshold segmentation method is run to process the image after the enhanced texture features, a continuous threshold field is generated by bilinear interpolation, and based on the threshold field, the gray value of each pixel in the image is compared with the threshold value at the corresponding position, and if the pixel gray value is greater than the threshold value, it is determined that the pixel belongs to the defect region, and otherwise it is a normal region.

[0022] Further, the calculation formula of the area-weighted quantification model is:

[0023]

[0024] Wherein, A i is the area of each defect area, I i is the brightness value of each pixel in the defect area, is the average brightness of the entire screen.

[0025] Further, the calculation formula of the visibility score is:

[0026]

[0027] Wherein, d j is the distance from the defect to the center of the screen, w j is the brightness sensitivity weight coefficient, Wherein, I j represents the brightness of the defect area, I bg is the background brightness, and β is the adjustment parameter. The spatial position weight d0 = λ·max(W, H), W and H are the screen width and height, and λ is the spatial position weight coefficient.

[0028] The present application effectively overcomes the problems of insufficient detection precision and low efficiency in the prior art by using innovative image processing algorithms and scientific evaluation models based on the human visual system, realizes high-precision detection, accurate quantitative evaluation and reasonable grading management of the Mura defects of the mobile phone screen, improves the accuracy and efficiency of detection, meets the strict quality control requirements in the screen production process, and improves product quality and market competitiveness. BRIEF DESCRIPTION OF DRAWINGS

[0029] Figure 1 The method flowchart of the present application. DETAILED DESCRIPTION

[0030] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all.

[0031] As Figure 1 shown, the present application provides a display screen Mura defect detection method, which comprises the following steps

[0032] S10: image acquisition is performed on the surface of the display screen by using a camera to obtain original image data.

[0033] Specifically, a 5000-megapixel global shutter CMOS camera is installed on a fixed support directly above the screen transmission track, and the position and angle of the camera are adjusted so that the lens is vertically aligned with the center of the screen.

[0034] The coaxial illumination system is turned on, the wavelength range of the coaxial illumination system is set to 450-650 nm, and the uniformity is greater than 95%, which ensures that the screen surface is uniformly illuminated and avoids reflection or shadows due to uneven illumination. The illumination wavelength is set to 550 nm, and the illumination intensity is adjusted to achieve a uniformity of 98%. When the screen is transmitted to the shooting position, the camera shoots at a frequency of 10 frames per second, and the high-quality image data I(x, y) obtained is transmitted in real time to the processing module computer through a high-speed data line, providing a reliable data basis for subsequent defect detection.

[0035] When shooting the screen, an automatic focusing algorithm can be combined to ensure that the shooting image always remains clear, thereby obtaining high-quality screen image data and providing a good basis for subsequent defect detection and analysis.

[0036] S20: The original image is subjected to texture feature enhancement using an improved LoG operator.

[0037] This step is processed by the processing module, which has a GPU-accelerated LoG convolution unit. The GPU has strong parallel computing capability, supports real-time adjustment of the scale parameter σ of the LoG operator, and can quickly complete LoG convolution operations, efficiently enhancing the micron-level texture features of the screen image.

[0038] The calculation formula of the LoG operator is:

[0039]

[0040] where I(x, y) is the original image data, and σ is the scale parameter.

[0041] The value of σ ranges from 0.5 to 2.0 μm.

[0042] According to the current screen model, the reference scale σ0 and the sensitivity coefficient α are obtained from the screen model parameter database, for example, σ0 = 1.0 μm and α = 0.5.

[0043] The scale parameter σ is dynamically adjusted according to the local image variance of the current detection area, and the formula is σ = σ0·(1 + α·var(I RoI )), where I RoI is the current detection area, and var(I RoI ) is the local image variance of the current detection area (RoI), reflecting the texture complexity or gray level fluctuation of the area.

[0044] The image can be divided into 100x100 pixel detection regions I ROI Then the local image variance var(I RoI ) of each region is calculated, the σ value is dynamically adjusted according to the adjustment formula σ = σ0·(1 + α·var(I RoI ), and then the image is convolved with the LoG operator formula to enhance the texture features of Mura defects.

[0045] Through this dynamic adjustment mechanism, the LoG operator can better adapt to the texture features of different regions of the screen, significantly improve the detection sensitivity of the edges of Mura defects in complex backgrounds, effectively enhance the texture features of Mura defects, make the defects more prominent in the image, while suppressing the interference of normal textures, reducing misjudgment, and facilitating defect identification and analysis.

[0046] S30: Using an adaptive threshold segmentation method to process the image with enhanced texture features to extract the Mura defect region.

[0047] Specifically, the formula for adaptive threshold segmentation is:

[0048] T(x, y) = μ(x, y) + k·σ(x, y), k ∈ [1.5, 3.0];

[0049] Where μ(x, y) and σ(x, y) are the local mean and standard deviation, respectively.

[0050] The image with enhanced texture can be divided into m x n local windows, m, n ≥ 20 (such as 30 x 30), the local mean μ(x, y) and standard deviation σ(x, y) of each window are calculated, and then the window threshold T k is calculated according to the formula T(x, y) = μ(x, y) + k·σ(x, y) (k can be 2.0).

[0051] When running the adaptive threshold segmentation method to process the image with enhanced texture features, a continuous threshold field is generated by bilinear interpolation, based on which the gray value of each pixel in the image is compared with the threshold value at the corresponding position. If the pixel gray value is greater than the threshold value, it is determined that the pixel belongs to the defect region, otherwise it is a normal region.

[0052] This adaptive threshold segmentation method fully considers the gray distribution characteristics of different regions of the screen, and can effectively distinguish between defect regions and normal regions, improving the accuracy and reliability of defect extraction.

[0053] S40: Calculate the severity of Mura defects in combination with the area-weighted quantization model.

[0054] Specifically, the calculation formula of the area-weighted quantization model is:

[0055]

[0056] wherein A i is the area of each defect region, I i is the brightness value of each pixel in the defect region, and is the average brightness of the entire screen.

[0057] The specific model comprehensively considers two key factors of brightness difference and area size of the defect region, and can accurately evaluate the severity of the Mura defect through scientific calculation, thereby providing a quantitative basis for objective evaluation of the defect, making the severity of different defects comparable, and helping production personnel to more intuitively understand the defect situation.

[0058] S50: using the human visual system (HVS) model, the visibility score of the Mura defect is calculated.

[0059] Specifically, the calculation formula of the visibility score is as follows:

[0060]

[0061] wherein d j is the distance from the defect to the center of the screen, w j is a weight coefficient, and β is the brightness sensitivity weight contained in the HVS model. wherein I j represents the brightness of the defect region, I bg is the background brightness, β is an adjustment parameter (which can be 0.2), the spatial position weight d0=λ·max(W,H), W and H are the width and height of the screen, and λ is a spatial position weight coefficient (which can be 0.3).

[0062] The Mura defect is divided into three levels of A / B / C, wherein the A level is a serious defect that must be returned for repair. The grading alarm program grades the defect according to the preset rules (S vis ≥ 0.8 is A level, 0.5≤S vis ≤ 0.8 is B level, and S vis ≤ 0.5 is C level).

[0063] When the A level defect is detected, an audible and visual alarm is sounded, and the defect information is sent to the repair station through the network. The repair personnel mark and remove the defective screen for repair. The B and C level defect information is stored in the database, and the quality management personnel regularly use data analysis software for statistical analysis, optimize the evaporation process parameters, improve the packaging environment, and generate a quality report to assist enterprise decision-making.

[0064] The model considers brightness sensitivity and spatial position weight and the like from the perspective of human visual perception, can simulate the human visual perception process of the Mura defect, more accurately evaluates the actual influence of the defect on the user visual experience, and thus realizes objective and quantitative evaluation of the Mura defect, and provides a reference basis more in line with actual use scenarios for product quality evaluation.

[0065] The above merely describes preferred embodiments of the present application but should not be used to restrict the present application, and any modification, equivalent replacement, improvement and the like within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for detecting Mura defects in a display screen, characterized in that: The method comprises: Using a camera to capture images of the surface of the display screen to obtain original image data; The improved LoG operator is used to enhance the texture features of the original image; The adaptive threshold segmentation method is used to process the image after texture feature enhancement to extract the mura defect area; Calculate the severity of mura defects using an area-weighted quantification model; Mura defects are scored for visibility using the Human Visual System (HVS) model.

2. The display screen Mura defect detection method according to claim 1, characterized in that: The calculation formula of the LoG operator is: Among them, I(x, y) is the original image data and σ is the scale parameter.

3. The display screen Mura defect detection method according to claim 2, characterized in that: The formula for dynamically adjusting σ based on the local image variance of the current detection area is σ=σ0·(1+α·var(I RoI )), where σ0 is the reference scale, α is the sensitivity coefficient, I RoI is the current detection area, var(I RoI ) is the local image variance of the current detection region (RoI).

4. The display screen Mura defect detection method according to claim 1, characterized in that: The formula for adaptive threshold segmentation is: T(x,y)=μ(x,y)+k·σ(x,y),k∈[1.5,3.0]; where μ(x,y) and σ(x,y) are the local mean and standard deviation, respectively.

5. The display screen Mura defect detection method according to claim 4, characterized in that: When running the adaptive threshold segmentation method to process the image after enhancing texture features, the image is divided into m×n local windows, and m,n≥20, and the threshold T is calculated independently for each window. k .

6. The display screen Mura defect detection method according to claim 5, characterized in that: When running the adaptive threshold segmentation method to process the image after enhancing texture features, a continuous threshold field is generated by bilinear interpolation. Based on this threshold field, the grayscale value of each pixel in the image is compared with the threshold at the corresponding position. If the grayscale value of the pixel is greater than the threshold, it is determined that the pixel belongs to the defect area, otherwise it is a normal area.

7. The display screen Mura defect detection method according to claim 1, characterized in that: The calculation formula of the area-weighted quantification model is: Among them, A i is the area of ​​each defect region, I i is the brightness value of each pixel in the defect area, is the average brightness of the entire screen.

8. The display screen Mura defect detection method according to claim 7, characterized in that: The formula for calculating the viewability score is: Among them, d j is the distance from the defect to the center of the screen, w j is the brightness sensitivity weight coefficient, Among them, I j Represents the brightness of the defect area, I bg is the background brightness, β is the adjustment parameter, the spatial position weight d0 = λ·max(W, H), W, H are the screen width and height, and λ is the spatial position weight coefficient.

Images