A sub-pixel defect detection and judgment method and system based on dual-camera cooperation and multi-channel color segmentation

Abstract

The application discloses a kind of based on dual-camera cooperation and multi-channel color segmentation subpixel defect detection and determination method and system, this method includes dual-camera cooperation acquisition, defect positioning and subpixel level background modeling, multi-channel color point color determination, black and white map defect extraction and brightness measurement, dynamic defect determination.The application solves the sensitivity, positioning accuracy, color determination accuracy and determination consistency problem of subpixel defect detection under low gray level (L0 level) working condition, and the color point recognition accuracy reaches 97.2%, and the correlation coefficient with human eye vision reaches 0.92, suitable for the factory detection of OLED, Micro-LED and other self-luminous displays.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of display panel defect nondestructive testing, and particularly relates to a sub-pixel defect detection and judgment method based on double-camera cooperation and multi-channel color segmentation, which is particularly suitable for micron-level sub-pixel bright spot defect (including white spot and color spot) detection of OLED, Micro-LED and other self-luminous displays under low gray level (L0 level) working conditions, and can be widely applied to display panel factory quality detection links. BACKGROUND

[0002] In the display panel manufacturing industry, especially in the production process of OLED, Micro-LED and other self-luminous displays, the detection of sub-pixel level bright spot defects (such as white spots and color spots) is a key link to ensure product quality. The size of such defects is usually micron-level, and the brightness is extremely low under low gray level (L0 level) working conditions, which brings severe challenges to detection technology.

[0003] The current mainstream detection technology has the following significant limitations:

[0004] Physical contradiction of single-camera acquisition system: Although a color camera can obtain RGB / HSV color information, the Bayer filter structure reduces the photon capture efficiency by more than 60%, and the signal-to-noise ratio (SNR) of brightness detection is less than 5dB under L0 gray level, with extremely low weak light detection sensitivity. While the black and white camera removes the filter and improves the weak light sensitivity, it completely loses the color classification ability and cannot distinguish red, green and blue defect types, resulting in that the single-camera solution cannot meet the needs of color recognition and weak light sensitivity at the same time.

[0005] Insufficient defect positioning accuracy: Traditional algorithms are mostly based on global threshold or screen brightness difference, without considering the non-uniformity of sub-pixel weak light under L0 state, resulting in a defect position deviation of more than 10μm, which cannot realize sub-pixel level accurate positioning.

[0006] Poor accuracy of color point color judgment: High-brightness color points are prone to single-channel saturation under high exposure, forming a central whitening phenomenon (such as a high-brightness red sub-pixel R value equal to 255 being recorded as (255, 255, 255) white), and traditional methods only rely on the central white area to determine the defect type, which is prone to color misjudgment. At the same time, the blooming of overexposed light spots will expand to 3-5 times the real area, seriously interfering with the defect number statistics, resulting in an accuracy rate of only 78% for color point recognition under L0 gray level.

[0007] Unreasonable design of judgment threshold: Existing algorithms mostly use a single global brightness threshold (such as a fixed 120 gray value) for defect judgment, without considering the differences in brightness perception of the human eye for different colors of red, green and blue, resulting in systematic misjudgment and poor consistency between the detection results and human visual perception.

[0008] In view of the technical bottleneck, the application provides a sub-pixel defect detection and judgment method and system based on double-camera cooperation and multi-channel color segmentation, and through hardware cooperation, algorithm optimization and judgment logic innovation, high-precision and high-reliability detection of sub-pixel defects under low gray level working conditions is realized. SUMMARY

[0009] The core purpose of the application is to solve the problems of the existing display panel defect detection technology, such as the inability to balance color and sensitivity, low positioning accuracy, high color misjudgment rate and unreasonable judgment logic under low gray level (L0 level) working conditions, and the following goals are achieved: breaking through the physical limitations of a single camera, realizing synchronous and accurate collection of brightness signals and color signals; establishing a sub-pixel level background model, and improving the defect positioning accuracy to the micron level.

[0010] A multi-channel fusion color judgment method is proposed to eliminate misjudgment caused by overexposure whitening; a dynamic judgment system consistent with human eye visual perception is constructed to improve the consistency of detection results and actual visual effects.

[0011] To achieve the above purpose, the application provides the following technical scheme:

[0012] A sub-pixel defect detection and judgment method based on double-camera cooperation and multi-channel color segmentation, comprising the following steps:

[0013] Step S1: double-camera cooperative collection: through a collection system composed of a main camera (black and white high-sensitivity camera) and an auxiliary camera (color high-resolution camera), the images of the display panel under low gray level (L0 level) are collected synchronously, the main camera captures brightness signals in long exposure mode, the auxiliary camera freezes original color information in short exposure mode, the hardware synchronization timing error of the double cameras is less than or equal to 1ms, and the light path consistency is realized through the same optical axis shared by the two cameras, and the light splitting ratio is 50:50;

[0014] Step S2: defect positioning and sub-pixel level background modeling: the color image collected by the auxiliary camera is converted to HSV space, the defect candidate area is preliminarily extracted in the V channel through local contrast enhancement and adaptive threshold segmentation, and the red, green and blue sub-pixel space templates are constructed in the H channel by analyzing the periodic distribution of hue values, the background sub-pixel template with an accuracy of 1.5μm is generated by combining the RGB channel and the multi-channel fusion segmentation result of the HSV space of the non-defective screen, and sub-pixel level defect positioning is realized.

[0015] Step S3: multi-channel color point color judgment: the color image collected by the auxiliary camera is subjected to RGB difference enhancement processing to obtain the enhancement mask corresponding to the red, green and blue channels, the enhancement mask is subjected to spatial verification with the sub-pixel space template of the H channel, and the defect color type is jointly judged in combination with the hue characteristics of the HSV channel, and the color type includes red (R), green (G), blue (B) and white (W).

[0016] Step S4: black and white image defect extraction and brightness measurement: the black and white image collected by the main camera is first subjected to median filtering or bilateral filtering processing, then the defect area is extracted, and the average brightness value (MeanValue) and the average gray value of the brightest 100 pixels (GrayMax100) of the defect area are calculated;

[0017] Step S5: dynamic defect determination: based on the defect color type determined in step S3, the pre-stored physical limit sample calibration threshold database (the average gray reference value T cmean , the brightest 100-pixel gray reference value T cmax of the limit sample corresponding to the color are called, and then the average brightness value of the defect area and the average gray value of the brightest 100 pixels calculated in step S4 are combined to calculate a comprehensive judgment score through a preset formula; the comprehensive judgment score is compared with a preset judgment threshold (a pre-set quantitative critical value, such as 0.2), if the score ≥ the preset judgment threshold, an unqualified determination result (NG) is output; if the score < the preset judgment threshold, a qualified determination result (OK) is output.

[0018] Further, the parameter configuration of the main camera is: resolution 4096x3000 (12 Megapixels), field of view (FOV) 14.1mmx10.3mm, imaging accuracy 3μm / pixel, exposure time 5ms, output bit depth 12bit; the parameter configuration of the auxiliary camera is: resolution 4096x3000 (12 Megapixels), pixel size 3.45μm x 3.45μm, equipped with RK-TC5M230-10150 telecentric lens, magnification 1x, field of view (FOV) 14.1mmx10.3mm, imaging accuracy 3μm / pixel, exposure time 0.5ms, output bit depth 12bit.

[0019] Further, the specific formula for converting the color image to HSV space in step S2 is as follows:

[0020] Calculate the maximum value C max =max(R,G,B), the minimum value C min =min(R,G,B) and the difference Δ=C max -C min ;

[0021] Calculate the hue H: ;

[0022] Calculate the saturation S: ;

[0023] Calculate the lightness V: V=C maxWherein, R, G, B are the red, green, blue channel pixel brightness values of the color image respectively.

[0024] Further, the specific formula of RGB difference enhancement in step S3 is:

[0025] Red defect enhancement mask: Wherein, I R (x, y), I G (x, y), I B (x, y) are the red, green, blue channel pixel brightness values at image coordinates (x, y) respectively, And is an empirical threshold, and ∧ is a logical and operation.

[0026] Further, the specific logic of multi-channel joint determination in step S3 is:

[0027] Spatial verification: D R =M R ∩T R , D G =M G ∩T G , D B =M B ∩T B , wherein M R , M G , M B are the mask results after RGB difference enhancement respectively; T R , T G , T B are the red, blue, green sub-pixel space templates on the hue chart respectively, and ∩ represents the intersection; Color confirmation: if the mask result of RGB difference enhancement is consistent with the corresponding main color of the H channel, it is confirmed that the defect color is R, G or B; if three continuous color points of different colors are detected, it is determined as a white point.

[0028] Further, the formula of filtering processing in step S4 is: I f (x, y) = medianfiter(I C (x, y), k), wherein I c (x, y) is the original black and white image pixel value, k is the filter kernel size, and I f (x, y) is the filtered image pixel value.

[0029] Further, the calculation formula of the average brightness value in step S4 is: Wherein, Ω is the defect area, N is the number of area pixels, and I(x, y) is the pixel brightness value of the defect area.

[0030] Further, the calculation formula of the average gray value of the 100 brightest pixels in step S4 is: Top100(Ω) = {Top100(Ω) | Top100(Ω) < Top100(Ω)}, wherein Top100(Ω) is the brightest 100 pixels of the defect area in descending order of brightness.

[0031] Further, the calculation formula of the comprehensive judgment score in step S5 is: , wherein W1, W2 are weight coefficients; T C,mean is the average gray reference value of the limit sample corresponding to the color type; T C,max is the brightest 100 pixel gray reference value of the limit sample corresponding to the color type; the judgment criterion is: , is the judgment threshold, greater than the threshold is an NG product, and less than the threshold is an OK product.

[0032] Further, the method is suitable for factory quality detection of OLED and Micro-LED self-luminous displays, and focuses on the detection and judgment of micron-level sub-pixel bright spot defects under low gray level working conditions.

[0033] The application also provides a sub-pixel defect detection and judgment system based on dual-camera cooperation and multi-channel color segmentation, comprising: a dual-camera cooperative acquisition system composed of a main camera (black and white) and an auxiliary camera (color);

[0034] An image processing unit is configured to:

[0035] Defect positioning: the color image of the auxiliary camera is converted to HSV space, the defects are preliminarily extracted in the V channel, and the sub-pixel template is constructed in the H channel to realize sub-pixel level defect positioning;

[0036] Color judgment: the brightness comparison of the RGB channel and the hue information of the H channel are fused to judge the color type (R / G / B / W) of the defect;

[0037] Brightness measurement: the gray value of the defect area is measured by using the high-sensitivity black and white image of the main camera;

[0038] Defect judgment: based on the dynamic comparison of the measured characteristic value and the physical limit sample, the final defect judgment result is output.

[0039] Further, in the defect positioning step, the spatial arrangement positions of red, green and blue sub-pixels are recognized and positioned by analyzing the periodic distribution of the H (hue) channel value; in the color judgment step, if the color preliminarily judged by the RGB channel is consistent with the main color of the H channel, the color type is confirmed; if three continuous color points are detected, it is judged as a white point (W); the brightness measurement includes calculating the average brightness value (MeanValue) and the brightest 100 pixel average gray value (GrayMax100) of the defect area; the defect judgment is realized by calculating the comprehensive judgment score.

[0040] The application has the following beneficial effects:

[0041] Complementary acquisition of luminance and color signals: Through the collaboration of dual cameras, the parallel acquisition of luminance and color signals is realized, which effectively overcomes the detection blind zone of a single camera under low brightness conditions, solves the physical contradiction of low sensitivity of color cameras in low light and lack of color recognition capability of black and white cameras, and improves the signal-to-noise ratio of luminance detection at L0 grayscale to more than 15dB.

[0042] Sub-pixel-level high-precision positioning: The background sub-pixel template generated by multi-channel fusion segmentation has a resolution of 1.5μm, realizing sub-pixel-level brightness difference modeling. The defect positioning deviation is controlled within 2μm, which improves the positioning accuracy by 80% compared with the traditional method (deviation of more than 10μm).

[0043] High-accuracy color determination: Through the RGB difference enhancement-spatial template verification-multi-feature fusion technology path, the accuracy of color dot recognition at L0 low grayscale is improved from 78% to 97.2% of the traditional method, and the overexposure whitening misjudgment elimination rate reaches 92.5%, which completely solves the problems of color misjudgment of bright color dots and light spot expansion interference.

[0044] Judgment results highly consistent with human visual perception: A dynamic threshold database based on physical limit samples was established, taking into account the differences in human visual perception of brightness of different colors. The detection results were consistent with human visual perception (Pearson correlation coefficient) of 0.92, which significantly reduced the systematic misjudgment caused by traditional single threshold.

[0045] High value for engineering applications: The detection method is applicable to self-emissive displays such as OLED and Micro-LED, with detection accuracy down to the micrometer level, which can meet the strict requirements of factory quality inspection. The detection efficiency is more than 30% higher than that of traditional methods, and it has broad prospects for industrial application.

[0046] To more clearly illustrate the structural features and effects of the present invention, the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. Attached Figure Description

[0047] Figure 1 This is a flowchart of the algorithm of the present invention;

[0048] Figure 2 This is a partial image taken by the black and white camera in this invention;

[0049] Figure 3 This is a partial image of an image captured by the color camera in this invention;

[0050] Figure 4 This is a schematic diagram illustrating the decomposition of an RGB color image into an HSV image in this invention;

[0051] Figure 5 This is a schematic diagram of RGB sub-pixel region segmentation in this invention;

[0052] Figure 6 Fig. 1 is a schematic diagram of color point color determination in the present application;

[0053] Figure 7 Fig. 2 is a schematic diagram of black and white map defect extraction in the present application. DETAILED DESCRIPTION

[0054] The present application will be further described in conjunction with the accompanying drawings and relevant knowledge, and will be clearly and completely described. Obviously, the described application is only a part of the embodiments of the present application, but not all the embodiments.

[0055] The present application provides a sub-pixel defect detection and determination method based on dual-camera cooperation and multi-channel color segmentation, which breaks through the industry limitations through four core technologies for the identification bottleneck of sub-pixel bright spot defects under the low gray scale (L0) state of the display screen.

[0056] Firstly, the physical contradiction of the single camera solution is solved by the dual-camera cooperative acquisition mechanism, that is, the short exposure of the high-resolution color camera is used to freeze the original form of the defect, the long exposure of the high-sensitivity black and white camera is used to capture the real brightness signal, and the hardware synchronization timing error is less than or equal to 1ms, so that the color camera can make up for the color recognition loss of the black and white camera, and the black and white camera can compensate for the insufficient light sensitivity of the color camera, and the data complementation is realized.

[0057] Secondly, the precise positioning is realized by using the sub-pixel level background modeling, that is, based on the physical characteristics of the weak light emission of the background sub-pixel under the L0 gray scale, the multi-channel fusion segmentation of the RGB channel and the HSV space is carried out on the non-defective screen, and the background sub-pixel template with a precision of 1.5μm is generated, so that the position deviation caused by the traditional method ignoring the background light emission is solved.

[0058] Thirdly, the double-exposure fusion technology is designed to solve the problem of false judgment of high-brightness color points caused by overexposure, that is, the original color information is extracted from the short exposure image of the color camera, the background sub-pixel template is used to determine the light-emitting sub-pixel position, and the RGB difference enhancement model result is combined to eliminate the interference of the center whitening caused by channel saturation.

[0059] Finally, the dynamic determination logic of the physical limit sample driving is constructed, that is, the color type of the defect is classified according to the multi-channel segmentation result, the color-specific brightness threshold value is called from the pre-stored database through the standard defect sample (limit sample) actual measurement calibration, the traditional single global threshold value scheme is completely replaced, and the systematic false judgment caused by the difference in human eye perception is solved.

[0060] The application takes dual-camera cooperative acquisition, multi-channel feature reconstruction and human eye perception alignment as the core chain: first, the physical contradiction that a single camera cannot balance color and sensitivity is broken through dual-camera hardware synchronization with ≤1ms accuracy (color short exposure freezes original color, black and white long exposure captures real brightness); second, based on the L0 background sub-pixel weak light characteristics, the RGB / HSV multi-channel segmentation is fused to generate a 1.5μm accuracy background template, providing a sub-pixel level positioning reference for defects; then, for the problem of highlight overexposure whitening, the template is used to lock the light-emitting position, and the RGB difference enhancement model is used to eliminate color misjudgment and inflation interference; finally, according to the color classification, the biological perception threshold library calibrated by real object limits is called to realize the judgment transition from engineering experience to visual physiology;

[0061] Specifically, referring to Figures 1-7 A sub-pixel defect detection and judgment method based on dual-camera cooperation and multi-channel color segmentation, comprising the following steps:

[0062] Step S1: dual-camera cooperative acquisition: through a collection system composed of a main camera (black and white high sensitivity camera) and an auxiliary camera (color high resolution camera), the images of the display panel under low gray level (L0 level) are synchronously collected, the main camera adopts long exposure mode to capture brightness signal, the auxiliary camera adopts short exposure mode to freeze original color information, the dual-camera hardware synchronization time error is ≤1ms, and the light path consistency is realized through the same optical axis shared by the dichroic prism with a splitting ratio of 50:50; the long exposure mode of the black and white main camera makes up for the low light photon capture efficiency of the color camera at L0 level, which improves the signal-to-noise ratio of brightness detection from less than 5dB of traditional color camera to more than 15dB; the short exposure mode of the color auxiliary camera avoids channel saturation and center whitening caused by high exposure, and completely retains the original color information of the defect, realizing the dual breakthrough of brightness sensitivity and color integrity.

[0063] Ensure data space-time consistency: the hardware synchronization time error of ≤1ms and the light path design of 50:50 splitting ratio ensure that the defect images collected by the dual-camera are accurately corresponding in spatial position (field of view completely matched) and time dimension (same detection time), providing an unbiased data basis for the subsequent fusion analysis of color judgment and brightness measurement, avoiding detection errors caused by space-time misalignment.

[0064] Step S2: defect positioning and sub-pixel level background modeling: the color image collected by the auxiliary camera is converted to HSV space, the defect candidate area is preliminarily extracted in V channel through local contrast enhancement and adaptive threshold segmentation, and the red, green and blue sub-pixel space templates are constructed by analyzing the periodic distribution of hue value in H channel; combining the RGB channel and the multi-channel fusion segmentation result in HSV space of the non-defect screen, a background sub-pixel template with an accuracy of 1.5μm is generated to realize sub-pixel level defect positioning;

[0065] The interference of luminance and color coupling can be eliminated: the HSV space conversion separates the processing of luminance (V channel) and color (H / S channel), the local contrast enhancement of V channel and the adaptive threshold segmentation can focus on the abnormal features of luminance, avoid the interference of color difference in traditional RGB space on defect extraction, and make the detection rate of L0 level weak highlight increase to more than 99%.

[0066] More specifically, micron-level accurate positioning is realized: based on the periodic distribution of H channel hue, a sub-pixel space template is constructed, combined with the 1.5 μm precision background template generated by multi-channel fusion segmentation, the physical position of red, green and blue sub-pixels can be accurately calibrated, the positioning deviation of defects is controlled within 2 μm from more than 10 μm of traditional method, the positioning accuracy is improved by 80%, which meets the micron-level precision requirement of sub-pixel level detection.

[0067] Step S3: multi-channel color point color judgment: the RGB difference enhancement processing is performed on the color image collected by the auxiliary camera to obtain the enhanced mask corresponding to the red, green and blue channels, the enhanced mask is verified in space with the sub-pixel space template of H channel, and the color type of the defect is judged in combination with the hue feature of HSV channel, the color type includes red (R), green (G), blue (B) and white (W);

[0068] The recognition degree of weak color signal can be enhanced: RGB difference enhancement separates and amplifies the weak defect signal of specific color from complex background through the comparison of luminance difference between channels (such as red defect, which needs to meet the difference between red channel and green / blue channel greater than the experience threshold), so that the color defect signal-to-noise ratio under L0 level low brightness is improved by 3-5 times.

[0069] The false judgment of overexposure whitening can be eliminated: the space verification (intersection operation of enhanced mask and sub-pixel template) ensures that the defect is only judged as the signal in the corresponding color sub-pixel area, combined with the joint judgment of HSV hue feature, the color misjudgment problem caused by overexposure whitening of highlight color point in traditional method is completely solved, the color point recognition accuracy is improved from 78% to 97.2%, and the elimination rate of overexposure whitening misjudgment reaches 92.5%; at the same time, through the logic of continuous three-color point judgment of white point, the white point defect is accurately recognized, and the misclassification of multi-color defect is avoided.

[0070] Step S4: black and white image defect extraction and brightness measurement: the black and white image collected by the main camera is first subjected to median filtering or bilateral filtering processing, and then the defect area is extracted, and the average brightness value (MeanValue) and the average gray value of the brightest 100 pixels (GrayMax100) of the defect area are calculated; the accuracy of the brightness measurement can be ensured, the median filtering or bilateral filtering can completely retain the defect edge details while suppressing random noise (such as circuit noise and environmental light interference), avoid the brightness distortion caused by traditional filtering, and control the brightness measurement error within 3%; and the brightness characteristics of the defect are fully characterized, the average brightness value (MeanValue) reflects the overall brightness level of the defect, and the average gray value of the brightest 100 pixels (GrayMax100) captures the peak brightness characteristics of the defect, and the combination of the two can completely describe the brightness distribution of the defect, provide more abundant and more accurate brightness basis for subsequent judgment, and avoid the judgment deviation caused by a single brightness index (such as the average brightness is easy to ignore the defect with peak brightness exceeding the standard).

[0071] Step S5: dynamic defect judgment: based on the defect color type determined in step S3, the pre-stored physical limit sample threshold database is called, the comprehensive judgment score is calculated through the preset formula, and the defect judgment result (OK / NG) is output according to the comparison result of the score and the preset judgment threshold (τ). The traditional single threshold can be solved. The threshold database of the physical limit sample is set with exclusive brightness thresholds for red, green, blue and white (such as the human eye is more sensitive to blue light, and the blue defect threshold is lower), fully considering the brightness perception difference of the human eye to different colors, making the detection result consistent with the human eye vision (Pearson correlation coefficient) up to 0.92, and the consistency of the traditional single threshold (such as fixed 120 gray value) is improved by more than 60%; and the objective and standardized judgment is realized, the comprehensive judgment score is calculated by weighted calculation (fusion of average brightness and peak brightness deviation), avoiding the subjective error of artificial judgment; the preset judgment threshold (τ) can be flexibly adjusted according to the product quality standard, meeting the detection needs of different manufacturers, realizing the standardization and engineering application of the detection process, and the judgment efficiency is improved by more than 5 times compared with artificial judgment.

[0072] In the preferred embodiment of the present application, the parameters of the main camera are configured as follows: resolution 4096x3000 (12 Megapixels), field of view (FOV) 14.1mmx10.3mm, imaging accuracy 3μm / pixel, exposure time 5ms, and output bit depth 12bit; and the parameters of the auxiliary camera are configured as follows: resolution 4096x3000 (12 Megapixels), pixel size 3.45μm x 3.45μm, equipped with RK-TC5M230-10150 telecentric lens, magnification 1x, field of view (FOV) 14.1mmx10.3mm, imaging accuracy 3μm / pixel, exposure time 0.5ms, and output bit depth 12bit. The balance between imaging quality and detection efficiency is ensured, the high resolution of 12 million pixels and the imaging accuracy of 3μm / pixel can clearly capture micron-level sub-pixel defects; the parameter combination of 5ms long exposure of the main camera and 0.5ms short exposure of the auxiliary camera can control the single field of view detection time within 10ms while ensuring the brightness sensitivity and color integrity, and meet the high-speed requirement of display panel factory detection (≥360 panels per hour).

[0073] The telecentric lens eliminates imaging distortion, and the 1x magnification and low distortion rate (<0.1%) of the RK-TC5M230-10150 telecentric lens ensure that the imaging accuracy of the image edge and center is consistent, avoid the positioning deviation of edge sub-pixel defects caused by lens distortion, and further improve the detection uniformity.

[0074] In the preferred embodiment of the present application, in step S2, to avoid the interference of color change on brightness detection, the color image I rgb =[R,G,B] is converted to HSV space, and the specific formula for converting the color image to HSV space is as follows: the maximum value C max =max(R,G,B), the minimum value C min =min(R,G,B), and the difference Δ=C max -C min .

[0075] The hue H is calculated as follows: ; the unit of the hue H is usually angle (°), and the value range is [0, 360];

[0076] The saturation S is calculated as follows: ;

[0077] The calculation of the lightness V: V=Cmax, wherein R, G, B are respectively the red, green and blue channel pixel brightness values of the color image. Through accurate quantification of color and brightness characteristics, the differential calculation of hue H (such as Cmax=R, calculated as (G-B) / Δ) can accurately distinguish the periodic distribution of red, green and blue colors, providing a quantitative basis for the construction of the sub-pixel template; the saturation S reflects the color purity, which can assist in excluding gray noise (S=0 is a gray signal); the lightness V directly corresponds to the pixel brightness, providing an accurate quantitative index for defect brightness extraction, so that the feature extraction accuracy of the HSV space is more than 98%.

[0078] In the preferred embodiment of the present application, the image captured by the color camera is first subjected to RGB difference enhancement in step S3, the purpose being to separate and enhance the weak, specific color defect signal from the complex background in the color image. For red defect enhancement, the specific formula of RGB difference enhancement is:

[0079] Red defect enhancement mask: ; wherein I R (x,y), I G (x,y), I B (x,y) respectively represent the pixel brightness values of the red channel, the green channel and the blue channel after being captured and decomposed by the color camera at the image coordinates (x, y) position, and are empirical thresholds, only when the difference is greater than the threshold, it is considered that the red component is significantly stronger than other components, ∧ is a logical and operation, requiring and two conditions must be met; similarly, the green M G (x,y) and blue M B (x,y) enhancement masks are obtained, the green defect enhancement mask M G (x,y)=[I G (x,y)-I R (x,y)>τ gr ]∧[I G (x,y)-I B (x,y)>τ gb ], the blue defect enhancement mask M B (x,y)=[I B (x,y)-I R (x,y)>τ br ]∧[I B (x,y)-I G (x,y)>τ bg ], wherein τ gr , τ gb , τ br , τ bgAll are empirical thresholds. Precise separation of color defects can be achieved by double-channel difference comparison and logical AND operation, only retaining the area where the target color channel brightness is significantly higher than the other two channels, effectively excluding cross-color interference (such as red defects not being triggered by weak signals from green / blue channels), making the purity of single-color defect extraction above 95%, providing high-purity candidate areas for subsequent color judgment.

[0080] Further, the specific logic of multi-channel joint determination in step S3 is to combine the mask results after RGB difference enhancement with the sub-pixel template for spatial verification.

[0081] Spatial verification: D R =M R ∩T R , D G =M G ∩T G , D B =M B ∩T B , where M R , M G , M B are the mask results after RGB difference enhancement; T R , T G , T B are the red, blue, and green sub-pixel spatial templates on the hue chart, accurately marking the physical locations of all red, blue, and green sub-pixels on the screen, and ∩ represents the intersection; color confirmation requires a point to meet two conditions simultaneously to be output to the final result: if the mask result after RGB difference enhancement is consistent with the corresponding primary color of the H channel, then the defect color is confirmed as R, G, or B; if three consecutive color points of different colors are detected, then it is determined as a white point.

[0082] Double verification is used to improve the reliability of color determination, spatial verification ensures that the defect is only located in the physical area of the corresponding color sub-pixel, and H channel primary color verification further confirms the color attribute. Double verification reduces the color determination error rate to below 2.8%; the logic of continuous three-color point determination of white points can accurately identify multi-sub-pixel joint defects (such as adjacent red, green, and blue sub-pixels being abnormal), avoiding the misjudgment of white points as single-color defects, and the white point recognition accuracy is above 96%.

[0083] Further, in step S4, in order to ensure the stability and robustness of defect detection, median filtering (Median Filter) or bilateral filtering (Bilateral Filter) is performed before the original image enters the detection algorithm to suppress random noise while preserving edge details. The formula for filtering processing is: I f (x,y)=medianfiter(I C (x,y),k), where Ic (x,y) is the original black and white image pixel value, k is the filter kernel size, I f (x,y) is the filtered image pixel value. The median filter has a suppression rate of more than 90% for pulse noise (such as random bright spot noise), and the bilateral filter can further retain defect edge details (edge clarity loss <5%), avoiding the defect blur caused by traditional mean filter, providing low-noise and high-detail image data for subsequent brightness measurement, so that the brightness measurement error is controlled within 3%.

[0084] Further, the calculation formula of the average brightness value in step S4 is: wherein Ω is the defect area, N is the number of area pixels, and I(x,y) is the pixel brightness value of the defect area.

[0085] Further, the calculation formula of the average gray value of the top 100 pixels in step S4 is: wherein Top100(Ω) is the top 100 pixels in the defect area in descending order of brightness. It can comprehensively reflect the brightness state of the defect, and the average brightness value can avoid misjudgment of the overall defect by local bright spots (such as low brightness at the edge of the defect but high brightness at the center, which is easy to overestimate the severity of the defect by using peak brightness), and the average gray value of the top 100 pixels can capture the peak brightness characteristics of the defect (avoiding the problem of hiding peak values exceeding the limit due to average brightness), and the combination of the two makes the representation integrity of the brightness characteristics reach more than 99%, providing a more comprehensive brightness basis for judgment.

[0086] Further, in step S5, the dynamic threshold Tc is called according to the color type (R / G / B / W), and the deviation score (comprehensive judgment score) is calculated:

[0087] The calculation formula of the comprehensive judgment score is: wherein W1 and W2 are weight coefficients; T C,mean is the average gray reference value of the limit sample corresponding to the color type (c=R / G / B / W); T C,max is the gray reference value of the top 100 pixels in the limit sample corresponding to the color type; and the judgment criterion is: , is the judgment threshold, and greater than the threshold is NG product and less than the threshold is OK product.

[0088] It can realize personalized and accurate judgment, and the weight coefficients W1 and W2 can be adjusted according to product requirements (such as increasing W2 for products sensitive to peak brightness), to meet the detection standards of different scenes; the threshold based on the limit sample makes the judgment benchmark completely match the actual product quality requirements, avoiding the disconnection between theoretical threshold and engineering practice.

[0089] Furthermore, the method is applicable to the factory quality inspection of OLED and Micro-LED self-emissive displays, focusing on the detection and judgment of micron-level sub-pixel bright spot defects under low grayscale conditions.

[0090] Compared with existing display panel defect detection methods, this invention, under low grayscale (L0 level) conditions, forms a complete high-sensitivity detection chain through dual-camera collaboration, multi-channel segmentation, and dynamic judgment, achieving simultaneous and accurate determination of brightness and color, and achieving the following significant technical effects:

[0091] Complementary Acquisition of Brightness and Color Signals: Traditional single-camera systems are limited by their physical structure. Color cameras suffer from low signal-to-noise ratios in low light due to the filtering loss of the Bayer filter array; while monochrome cameras are sensitive, they lack color dimension. This invention achieves parallel acquisition of brightness and color signals through a dual-camera collaborative acquisition mechanism, employing a time-complementary mode of short color exposure and long monochrome exposure. Through hardware synchronization of ≤1ms and sub-pixel-level registration, it effectively overcomes the detection blind spot of single cameras under low-light conditions, achieving physical complementarity between color and brightness.

[0092] High-precision positioning of sub-pixel-level background modeling: Traditional algorithms are usually based on global thresholds or screen-wide brightness differences, which cannot consider the non-uniformity of weak emission of sub-pixels in the L0 state, resulting in positional deviations of more than 10μm. This invention establishes a background template by fusing statistical features of RGB and HSV six channels, achieving a resolution of 1.5μm, realizing sub-pixel-level brightness difference modeling, and making subsequent threshold determination highly stable.

[0093] Multi-channel color dot determination: Compared with existing display panel defect detection technologies, the multi-channel color dot determination method proposed in this invention achieves a significant improvement in color recognition accuracy and reliability under low grayscale (L0) conditions through the technical path of "RGB differential enhancement - spatial template verification - multi-feature fusion". Under L0 low grayscale conditions, the color dot recognition accuracy is increased from 78% of the traditional method to 97.2%, and the overexposure whitening misjudgment elimination rate reaches 92.5%.

[0094] Limit Sample-Driven Dynamic Threshold Determination: Traditional detection methods use a uniform brightness threshold (e.g., 120 grayscale) for determination, which has drawbacks, as it does not consider the differences in human eye sensitivity to brightness of different colors, leading to systematic bias. This invention establishes a dynamic threshold database based on physical limit samples, where the brightness perception threshold of each color sub-pixel is independently calibrated. The detection results show a consistency of 0.92 with human visual perception (Pearson correlation coefficient).

[0095] The present invention also provides a sub-pixel defect detection and judgment system based on dual-camera collaboration and multi-channel color segmentation, comprising: a dual-camera collaborative acquisition system, consisting of a main camera (black and white) and an auxiliary camera (color);

[0096] An image processing unit is configured as:

[0097] Defect positioning: the color image of the auxiliary camera is converted to HSV space, the defects are preliminarily extracted in the V channel, and a sub-pixel template is constructed in the H channel to realize sub-pixel level defect positioning;

[0098] Color determination: the luminance comparison of the RGB channel and the hue information of the H channel are fused to determine the color type (R / G / B / W) of the defect;

[0099] Luminance measurement: the gray value of the defect area is measured by using the high-sensitivity black-and-white image of the main camera;

[0100] Defect determination: based on the measured characteristic value and the actual limit sample, the final defect determination result is output.

[0101] Further, in the defect positioning step, the spatial arrangement position of red, green and blue sub-pixels is recognized and positioned by analyzing the periodic distribution of the H (hue) channel value; in the color determination step, if the color preliminarily determined by the RGB channel is consistent with the main color of the H channel, the color type is confirmed; if three continuous color points are detected, it is determined as a white point (W); the luminance measurement includes calculating the average luminance value (MeanValue) of the defect area and the average gray value (GrayMax100) of the 100 brightest pixels; the defect determination is realized by calculating the comprehensive determination score.

[0102] The present application realizes accurate color determination of color points by the technical path of RGB differential enhancement-space template verification-HSV feature combination, and eliminates the misjudgment of overexposure whitening: specific RGB differential enhancement, differential enhancement algorithms are designed for red, green and blue color defects respectively, and the defect signals of specific colors are separated and enhanced from the complex background by comparing the luminance values between channels. For example, red defect enhancement is realized by calculating the luminance difference values of the red channel and the green channel and the blue channel, and when both difference values are greater than the corresponding empirical threshold, the red defect candidate area is determined, and the green and blue defect enhancement masks are obtained in the same way. Spatial template verification ensures that the determined defect area is accurately located at the sub-pixel position of the corresponding color, and excludes background interference; HSV feature joint determination, if the results of RGB differential enhancement and spatial template verification are consistent, and the main color corresponding to the H channel is matched, the defect color is confirmed as R, G or B; if three continuous color points of different colors are detected, it is determined as a white point (W), which completely solves the color misjudgment problem caused by overexposure whitening of high brightness color points.

[0103] It is illustrated in the present application that in the dynamic defect judgment process, by constructing a dynamic judgment system driven by physical limit samples, the defect judgment aligned with human visual perception is realized, specifically threshold database establishment: through standard defect sample (limit sample) actual measurement calibration, the exclusive threshold database of red, green, blue and white four color defects is established, the average gray reference value and the brightest 100 pixel gray reference value of each color limit sample are stored, and the traditional single global threshold is replaced; comprehensive judgment score calculation: according to the defect color type, the corresponding threshold parameter is called, the comprehensive judgment score is calculated through weighting, which can be adjusted according to engineering experience or visual perception characteristics, so as to realize the matching of brightness characteristics and visual perception; final judgment: setting judgment threshold τ, when Score>τ, it is judged as unqualified product (NG); when Score≤τ, it is judged as qualified product (OK), so as to ensure the objectivity and accuracy of the judgment result.

[0104] Example 1, reference 2- Figure 3 As shown, the high-definition image is shot, and the present application adopts a set of double-camera cooperative acquisition system composed of a main camera (black and white high-sensitivity channel) and an auxiliary camera (color high-resolution channel). The two cameras realize light path consistency through a shared optical axis by a light splitting prism, and the light splitting ratio is 50:50.

[0105] (1) Main camera (Main Camera)

[0106] Model: MV-CH120-10GM (black and white camera)

[0107] Resolution: 4096x3000 (12 Megapixels)

[0108] Field of view (FOV): 14.1mmx10.3mm

[0109] Imaging accuracy: 3μm / pixel

[0110] Exposure time: 5ms (long exposure, used to enhance low brightness details)

[0111] Output bit depth: 12bit

[0112] Function positioning: responsible for collecting the weak brightness signal of the display panel at L0 gray scale, realizing high-sensitivity brightness capture.

[0113] (2) Auxiliary camera (Color Camera)

[0114] Model: MV-CH120-10GC (color camera)

[0115] Resolution: 4096x3000 (12 Megapixels)

[0116] Pixel size: 3.45 pm x 3.45 pm

[0117] Lens model: RK-TC5M230-10150 Telecentric lens

[0118] Magnification: 1x

[0119] Field of view (FOV): 14.1 mm x 10.3 mm

[0120] Imaging accuracy: 3 pm / pixel

[0121] Exposure time: 0.5 ms (short exposure, used for freezing real color information)

[0122] Output bit depth: 12 bit

[0123] Function positioning: responsible for capturing the original color features of the defect area, assisting the main camera in color reconstruction.

[0124] Example 2: Reference Figure 4 、 Figure 5 As shown in the defect positioning and background sub-pixel modeling, this embodiment is based on multi-channel feature decomposition and reconstruction of color images, and establishes a high-precision background modeling method for sub-pixel level defect positioning. This method makes full use of the characteristics of brightness and color separation in the HSV color space, realizes high-robustness defect extraction and sub-pixel position calibration in low gray state.

[0125] First, the original RGB image collected by the color camera in the dual-camera system is subjected to color space conversion to obtain the corresponding HSV image. In the HSV model, Hue (hue) reflects the dominant color distribution of the pixel, Saturation (saturation) represents the color purity, and Value (brightness) describes the light intensity information of the pixel.

[0126] Preliminary extraction of candidate defect regions:

[0127] Preliminary extraction of candidate defect regions:

[0128] Construction of sub-pixel space template based on hue channel:

[0129] In the Hue channel, the image is sub-pixel structure segmented. By analyzing the periodic distribution of Hue value, the spatial arrangement position of red, green and blue three types of sub-pixels on the screen can be identified. Combined with the sub-pixel distribution template established on the non-defective sample in the early stage, the light emitting unit in each channel can be accurately positioned and numbered in the current detection image.

[0130] Finally, the defect area extracted in the V channel is superimposed and compared with the sub-pixel positioning result in the Hue channel, so that the specific sub-pixel type (R, G or B) where the defect is located is determined, and the defect coordinate information with micron-level accuracy is generated. This method effectively solves the false detection problem caused by the coupling of brightness and color in the traditional RGB space, and realizes stable and accurate sub-pixel level defect positioning of high-resolution display screen in low gray state.

[0131] Embodiment 3: Refer to Figure 6 As shown in the color point color determination of multiple channels, the embodiment provides a multi-channel segmentation and spatial positioning method based on color image, which is used for accurately determining the color type of red (R), green (G), blue (B) and white (W) sub-pixel bright spots. This method realizes efficient and accurate color point classification by fusing the periodic structure information of hue (H) channel and the relative intensity relationship of RGB channel.

[0132] The steps are as follows:

[0133] 1. RGB difference enhancement, the original image collected by the color camera is decomposed into R, G and B three independent channels from the color space. The ideal single color point area corresponds to channel values significantly higher than other channels. By difference between the three channels, the defect areas corresponding to red, green and blue points in the color image can be obtained.

[0134] The brightness value of each channel is calculated for each candidate defect area, and the channel with the maximum brightness in R / G / B three components is found as the preliminary color determination basis of the color point.

[0135] 2. HSV channel auxiliary determination, according to the candidate defect area obtained in embodiment 2 and the sub-pixel space template constructed in the hue graph.

[0136] 3. Multi-channel joint determination, the color point result of RGB channel is combined with the HSV hue feature to form the color point color determination rule; if the color point color of RGB is consistent with the main color corresponding to the hue H, it is confirmed that the color point color type is R, G or B; if there are three continuous color points, it is determined that the color is white point.

[0137] Embodiment 4: Refer to Figure 7As shown in the black and white defect extraction, the embodiment provides a defect brightness accurate measurement method based on high sensitivity image of black and white camera. The method uses the high signal-to-noise ratio advantage of black and white camera under low light conditions to quantize the gray value of the positioned defect area, and provides accurate brightness data for subsequent defect judgment.

[0138] Embodiment 5: Feature calculation and spec judgment, based on the sub-pixel level defect extraction and color point color judgment results of the preceding embodiment, the quantitative feature calculation is carried out for each candidate defect, and the standard limit sample (Spec) is combined for judgment, realizing the engineering judgment and classification of micron-level defects.

[0139] Defect feature calculation:

[0140] The following key features are calculated for each sub-pixel defect area:

[0141] MeanValue: using the results of embodiment 4, the average brightness of the defect center and the surrounding effective pixels is obtained;

[0142] GrayMax100: using the results of embodiment 4, the average value of the 100 brightest pixels in the defect area is calculated;

[0143] Color type: combined with the color point color judgment of embodiment 3, the defect sub-pixel type (R / G / B / W) is determined;

[0144] Spec judgment, based on the color point color judgment of the preceding color camera and the feature calculation of the black and white camera, combined with the real limit sample (red dot, green dot, blue dot, white dot limit sample product), a dynamic judgment method driven by real limit sample is proposed, realizing the accurate classification and color recognition of micron-level sub-pixel defects. The judgment logic changes from traditional fixed threshold to dynamic, visual physiological perception alignment, significantly improving the detection accuracy and reliability.

[0145] The feature values of the defects are compared with the corresponding limit samples, and the comprehensive judgment score is calculated: the calculation formula is as follows:

[0146] ;

[0147] Where, W1, W2 are weight coefficients, which are adjusted according to engineering experience or visual perception; T C,mean is the average gray reference value of the limit sample of this color type (c=R / G / B / W); T C,max is the brightest 100-pixel gray reference value of the limit sample of this color type.

[0148] The technical principles of the present application are described above in combination with specific embodiments, which are only preferred embodiments of the present application. The protection scope of the present application is not limited to the above-described embodiments only, and any technical solutions falling within the concept of the present application shall fall within the protection scope of the present application. Other specific embodiments of the present application can be conceived by those skilled in the art without creative efforts, and these embodiments shall fall within the protection scope of the present application.

Claims

1. A method for detecting and judging sub-pixel defects based on dual-camera collaboration and multi-channel color segmentation, characterized in that, Includes the following steps: Step S1: The acquisition system consisting of a main camera and an auxiliary camera synchronously acquires images of the display panel at low grayscale. The main camera uses a long exposure mode to capture the brightness signal, and the auxiliary camera uses a short exposure mode to freeze the original color information. Step S2: Convert the color image acquired by the auxiliary camera to the HSV space. In the V channel, extract the candidate defect region through local contrast enhancement and adaptive threshold segmentation. In the H channel, construct red, green and blue sub-pixel space templates by analyzing the periodic distribution of hue values. Combine the RGB channel and HSV space multi-channel fusion segmentation results of the defect-free screen to generate the background sub-pixel template and realize sub-pixel level defect localization. Step S3: Perform RGB differential enhancement processing on the color image acquired by the auxiliary camera to obtain enhancement masks corresponding to the red, green and blue channels. Perform spatial verification on the enhancement masks and the sub-pixel space template of the H channel. Combine the hue features of the HSV channel to jointly determine the defect color type. Step S4: First, perform median filtering or bilateral filtering on the black and white image captured by the main camera, then extract the defect area, and calculate the average brightness value and the average gray value of the brightest 100 pixels of the defect area. Step S5: Based on the determined defect color type, call the pre-stored physical limit sample calibration threshold database to obtain the average grayscale reference value of the limit sample corresponding to the defect color type, the grayscale reference value of the brightest 100 pixels in the limit sample, and calculate the comprehensive judgment score according to the average brightness value and the average grayscale value of the brightest 100 pixels in the defect area using a preset formula. Based on the comparison result between the comprehensive judgment score and the preset judgment threshold, output the defect judgment result.

2. The method according to claim 1, characterized in that, The specific formula for converting the color image to HSV space in step S2 is as follows: Calculate the maximum value C max =max(R,G,B), minimum value C min =min(R,G,B) and the difference Δ=C max -C min ; Calculate hue H: ; Calculate the saturation S: ; Calculate the lightness V: V=C max Where R, G, and B are the pixel brightness values ​​of the red, green, and blue channels of the color image, respectively.

3. The method according to claim 1, characterized in that, The specific formula for RGB differential enhancement in step S3 is as follows: Red defect enhancement mask: ;in, , , , where are the pixel brightness values ​​of the red, green, and blue channels at image coordinates (x, y), respectively. and is the empirical threshold, and ∧ is the logical AND operation.

4. The method according to claim 1, characterized in that, The spatial verification in step S3 specifically involves: ,in , , These are the mask results after RGB differential enhancement; , These are the red, blue, and green sub-pixel space templates on the hue map, respectively. The intersection of the representation; color confirmation: if the mask result of RGB differential enhancement is consistent with the main color corresponding to the H channel, the defect color is confirmed to be R, G or B; if three consecutive colored dots of different colors are detected, they are determined to be white dots.

5. The method according to claim 1, characterized in that, The median filtering formula in step S4 is as follows: , where I c (x, y) represents the pixel values ​​of the original black and white image, k is the filter kernel size, and I f (x,y) represents the pixel values ​​of the filtered image.

6. The method according to claim 1, characterized in that, The formula for calculating the average brightness value in step S4 is: , where Ω is the defect area, N is the number of pixels in the area, and I(x,y) is the pixel brightness value of the defect area.

7. The method according to claim 6, characterized in that, The formula for calculating the average grayscale value of the brightest 100 pixels in step S4 is: The Top100(Ω) represents the 100 brightest pixels in the defect area, sorted by brightness in descending order.

8. The method according to claim 7, characterized in that, The formula for calculating the comprehensive judgment score in step S5 is as follows: ,in, , These are the weighting coefficients; This is the average grayscale reference value for the limit sample of the corresponding color type; The grayscale reference value is the brightest 100 pixels in the limit sample for the corresponding color type; the judgment criterion is: , To determine the threshold, products exceeding the threshold are considered unqualified, while those below the threshold are considered qualified. NG indicates unqualified, and OK indicates qualified.

9. A sub-pixel defect detection and judgment system based on dual-camera collaboration and multi-channel color segmentation, characterized in that, The system is used to perform the sub-pixel defect detection and judgment method based on dual-camera collaboration and multi-channel color segmentation as described in any one of claims 1-8. The system includes: a dual-camera collaborative acquisition system, consisting of a main camera and an auxiliary camera; An image processing unit is configured to: convert the color image from the auxiliary camera to the HSV space, initially extract defects in the V channel, and construct a sub-pixel template in the H channel to achieve sub-pixel-level defect localization; fuse the brightness comparison of the RGB channels with the hue information of the H channel to determine the color type of the defect; use the high-sensitivity black and white image of the main camera to measure the gray value of the defect area; and dynamically compare the measured feature values ​​with the physical limit sample to output the final defect judgment result.

Images