A backlight module light leakage measuring method and system based on machine learning

Abstract

The present application relates to the technical field of liquid crystal display backlight module detection, and discloses a backlight module light leakage determination method and system based on machine learning. The backlight module light leakage determination method based on machine learning comprises the following steps: collecting backlight module optical characteristic matrices; screening suspicious light leakage areas according to adaptive threshold segmentation of ambient light; implementing local contrast enhancement on the suspicious light leakage areas, combining multi-scale pyramid analysis to strengthen weak light leakage signals; quantifying micro-area light leakage structure complexity by means of box dimension calculation and multi-fractal spectrum analysis, and constructing a fractal feature vector; establishing a light leakage and defect mapping based on a machine learning correlation model, and outputting light leakage details and defect types after scoring. The present application combines fractal geometry theory, image enhancement technology and machine learning algorithms, realizes accurate quantification of light leakage morphology and automatic recognition of defect types, improves detection accuracy, reduces false positive rate, reduces missed detection rate, and improves detection sensitivity.

Description

Technical Field

[0001] This invention relates to the field of liquid crystal display backlight module testing technology, and more specifically, to a machine learning-based method and system for measuring backlight module light leakage. Background Technology

[0002] As a key component of LCD monitors, the backlight module's quality directly affects the overall display effect. During the manufacturing process, light leakage may occur due to material defects or poor assembly. These light leakage areas appear as bright spots or lines when the screen is black, severely impacting the user experience. Therefore, accurately detecting backlight module light leakage is an important task in the quality control process of LCD monitor manufacturing.

[0003] Existing backlight module light leakage detection technologies mainly fall into two categories: manual visual inspection and traditional machine vision methods. Manual visual inspection relies on the experience and judgment of inspectors, resulting in low efficiency, strong subjectivity, inconsistent judgment standards, and susceptibility to fatigue. Traditional machine vision methods primarily identify light leakage areas through fixed threshold segmentation or simple edge detection algorithms. While effective in standard laboratory environments, these methods have limitations in the complex lighting environments of actual production lines. Specifically, existing technologies face the following main problems: under complex ambient lighting conditions, uneven illumination, significant reflection and scattering interference lead to high false alarm rates for fixed threshold segmentation methods; weak light leakage signals in small areas are difficult for traditional detection algorithms to extract effectively, resulting in missed detections; the complex and varied morphology of light leakage is insufficient to be accurately described by traditional geometric features, making it difficult to accurately correlate with material defect types; the lack of adaptive detection parameter adjustment mechanisms leads to poor adaptability to different product specifications; and the lack of application of intelligent algorithms such as machine learning prevents automated defect identification and classification.

[0004] Therefore, there is an urgent need for a backlight module light leakage measurement method and system that can combine machine learning technology, has high detection accuracy, and can automatically identify defect types, so as to improve the quality control level of LCD displays. Summary of the Invention

[0005] This invention provides a method and system for measuring backlight module light leakage based on machine learning, which solves the technical problems in related technologies such as interference from complex lighting environments, difficulty in identifying weak light leakage, accurate quantification of irregular shapes, and automatic identification of defect types.

[0006] This invention discloses a machine learning-based method for measuring backlight module light leakage, comprising the following steps:

[0007] Collect multi-dimensional optical feature information of the backlight module and construct the original feature matrix;

[0008] Based on ambient light conditions, an adaptive threshold algorithm is used to initially segment the backlight module and screen out suspicious light leakage areas.

[0009] A local contrast enhancement algorithm is applied to the screened suspicious areas, combined with multi-scale pyramid analysis technology, to enhance the characteristics of weak light leakage signals;

[0010] For the enhanced micro-area light leakage morphology, box dimension calculation and multifractal spectrum analysis are applied to quantify the structural complexity of the light leakage region and construct fractal feature vectors.

[0011] A correlation model between fractal feature vectors and optical properties is constructed based on machine learning algorithms. The mapping relationship between light leakage morphology and material defect type is established. The location, range, severity and defect type of light leakage are calculated and output through light leakage severity scoring.

[0012] Furthermore, the original feature matrix is ​​constructed as follows:

[0013] F(x,y)=[B(x,y),C(x,y),GLCM(x,y),E(x,y)];

[0014] Where B(x,y) represents the brightness feature, C(x,y) represents the color feature, GLCM(x,y) represents the texture feature, and E(x,y) represents the edge feature.

[0015] Furthermore, the adaptive threshold algorithm is calculated using the following formula:

[0016] T(x,y)=μ(x,y)+k·σ(x,y);

[0017] Where (x, y) represents the coordinates, T(x, y) represents the dynamic threshold, μ(x, y) is the mean of the local region, σ(x, y) is the standard deviation of the local region, and k is the threshold coefficient, which is automatically adjusted according to the ambient light intensity.

[0018] Furthermore, the formula for enhancing the selected suspicious areas using the local contrast enhancement algorithm is as follows:

[0019]

[0020] Where I′(x,y) is the enhanced image, I(x,y) is the original image, and μ local (x,y) is the local mean, σ local (x, y) is the local standard deviation, α contrast ε is the enhancement coefficient. small To prevent small constants from being divided by zero, M(x, y) is a map marking suspicious regions.

[0021] Furthermore, the multi-scale pyramid analysis technique includes constructing a Gaussian pyramid and a Laplacian pyramid for the enhanced image, and generating the final enhanced image by weighted fusion of multi-scale features, wherein the weight coefficients are adaptively adjusted according to the significance of light leakage features at different scales.

[0022] Furthermore, the box dimension calculation is achieved through the following steps:

[0023] Different scales of grids are applied to cover the light leakage area, with the grid size decreasing geometrically.

[0024] Calculate the number of grids required to cover the light leakage area for each grid size;

[0025] Plot a double logarithmic coordinate graph, with the horizontal axis being the reciprocal of the logarithm of the grid size and the vertical axis being the logarithm of the number of grid cells;

[0026] The slope is obtained by linear regression and used as the fractal dimension.

[0027] Furthermore, the correlation model between the fractal features and optical properties is implemented using a random forest classifier structure. This model includes an input layer, a set of decision trees, and a voting integration layer, which can distinguish between different types of defects.

[0028] Furthermore, the fractal feature vector is constructed as follows:

[0029] DF=[D,f(α min ), f(α) max ), α0, Δα, CN];

[0030] Where D is the box dimension, f(α) min ) and f(α) max ) are the function values ​​corresponding to the minimum and maximum values ​​of the multifractal spectrum, respectively. α0 is the value of α when f(α) reaches its maximum value, representing the dominant scaling characteristic of the light leakage region. Δα is the range of values ​​for α, representing the multifractal intensity of the light leakage region. CN is the connectivity number, representing the topological characteristics of the light leakage region.

[0031] Furthermore, the formula for calculating the severity score of light leakage is as follows:

[0032] S=w1·I norm +w2·A norm +w3·D norm +w4·L norm ;

[0033] Where S represents the severity score of light leakage, and I norm A represents the normalized leakage intensity. norm D represents the normalized leakage area. norm L represents the normalized fractal dimension.norm The normalized position factor is the distance between the light leakage area and the center of the screen. w1, w2, w3, and w4 are the first, second, third, and fourth weighting coefficients, used to adjust the degree of influence of different factors on the final score.

[0034] This invention provides a machine learning-based backlight module light leakage measurement system for performing the aforementioned machine learning-based backlight module light leakage measurement method, comprising:

[0035] The image acquisition and feature extraction module is used to acquire multi-dimensional optical feature information of the backlight module and construct the original feature matrix.

[0036] The adaptive segmentation module is used to perform preliminary segmentation of the backlight module based on ambient light conditions and an adaptive threshold algorithm to screen out suspicious light leakage areas.

[0037] The contrast enhancement module is used to apply a local contrast enhancement algorithm to the screened suspicious areas, combined with multi-scale pyramid analysis technology, to enhance the characteristics of weak light leakage signals.

[0038] The fractal feature analysis module is used to quantify the structural complexity of the light leakage region and construct fractal feature vectors by applying box dimension calculation and multifractal spectrum analysis to the enhanced micro-area light leakage morphology.

[0039] The machine learning defect assessment module is used to build a correlation model between fractal feature vectors and optical properties based on machine learning algorithms, establish a mapping relationship between light leakage morphology and material defect type, and calculate and output the location, range, severity and defect type of light leakage through light leakage severity scoring.

[0040] The beneficial effects of this invention are:

[0041] This invention solves the problem of backlight module leakage detection under complex ambient light conditions by combining multi-dimensional optical feature analysis, dynamic threshold adaptive segmentation, local contrast enhancement, fractal dimension feature quantization, and machine learning algorithms, achieving the following beneficial effects: Through multi-dimensional optical feature acquisition and dynamic threshold adaptive segmentation technology, it achieves adaptation to complex ambient light conditions, improving detection accuracy, reducing false alarm rate, and enhancing the robustness of the detection system; The use of local contrast enhancement algorithms and multi-scale pyramid analysis technology effectively enhances weak leakage signals, improves detection sensitivity, and solves the problem of traditional methods failing to identify slight leakage in small areas; The introduction of fractal geometry theory for precise quantification of leakage morphology overcomes the limitations of traditional geometric feature descriptions in analyzing irregular leakage morphology, reducing the false detection rate; The correlation model between fractal features and optical properties constructed based on machine learning algorithms achieves automatic mapping between leakage morphology and material defect types, providing a direct basis for production process optimization; The entire detection process achieves seamless integration of large-area detection and micro-area fine analysis, improving detection efficiency while reducing computational resource requirements while ensuring high-precision detection. Attached Figure Description

[0042] Figure 1 This is a flowchart of a backlight module light leakage measurement method based on machine learning according to the present invention. Detailed Implementation

[0043] The subject matter described herein will now be discussed with reference to exemplary embodiments. It should be understood that these embodiments are discussed only to enable those skilled in the art to better understand and implement the subject matter described herein, and changes may be made to the function and arrangement of the elements discussed without departing from the scope of this specification. Various processes or components may be omitted, substituted, or added as needed in the examples. Furthermore, some features described in the examples may be combined in other examples.

[0044] At least one embodiment of the present invention discloses a machine learning-based method for measuring backlight module light leakage, such as... Figure 1 As shown, it includes the following steps:

[0045] Step 1: Collect multi-dimensional optical feature information of the backlight module and construct the original feature matrix;

[0046] Step 1.1, Image Acquisition;

[0047] I. Acquire raw images of the backlight module using a high-precision CCD camera under controlled light source conditions raw (x, y), where x and y represent the coordinates in the image.

[0048] Step 1.2, Feature extraction;

[0049] Extracting multi-dimensional optical features from the original image, including:

[0050] Brightness features: Extract the brightness value B(x, y) of the pixel to reflect the light intensity distribution of the backlight module;

[0051] Color characteristics: Analyze the component values ​​C(x,y) of the RGB or HSV color space to capture the light leakage characteristics of different wavelengths of light;

[0052] Texture features: Calculate the gray-level co-occurrence matrix GLCM(x, y) of the local region to characterize the texture properties of the light leakage region;

[0053] Edge features: Operators such as Sobel or Canny are used to extract edge gradients E(x, y) to identify light leakage boundaries.

[0054] Step 1.3, Feature matrix construction;

[0055] The extracted features from each dimension are integrated into the original feature matrix F(x, y), which is calculated using the following formula:

[0056] F(x,y)=[B(x,y),C(x,y),GLCM(x,y),E(x,y)].

[0057] Where B(x,y) represents the brightness value at coordinates (x,y), C(x,y) represents the color feature vector, GLCM(x,y) represents the texture feature vector, and E(x,y) represents the edge feature vector.

[0058] Step 2: Based on the ambient light conditions, use an adaptive threshold algorithm to perform preliminary segmentation of the backlight module and screen out suspicious light leakage areas;

[0059] Step 2.1, Calculation of local statistics;

[0060] Calculate the local mean and standard deviation for each region of the feature matrix F(x, y):

[0061]

[0062] Where μ(x, y) represents the mean of the local region centered at coordinates (x, y), w and h represent the width and height of the calculation window, respectively, and F(i, j) represents the value of the feature matrix at coordinates (i, j);

[0063]

[0064] Where σ(x,y) represents the standard deviation of the local region centered at coordinates (x,y), and is used to measure the degree of dispersion of the eigenvalues.

[0065] Step 2.2, Adaptive threshold generation;

[0066] Calculate the dynamic threshold T(x, y) based on the ambient light conditions:

[0067] T(x,y)=μ(x,y)+k·σ(x,y);

[0068] Where T(x, y) represents the adaptive threshold at coordinates (x, y), and k is the threshold coefficient, which is automatically adjusted according to the ambient light intensity and is usually between 0.5 and 2.0.

[0069] Step 2.3, preliminary segmentation of the suspicious area;

[0070] The feature matrix is ​​segmented using a threshold to generate a binary map of suspected light leakage regions, M(x, y).

[0071]

[0072] Where M(x, y) represents the binarization result at coordinates (x, y), a value of 1 indicates that the point is marked as a suspected light leak area, and a value of 0 indicates that the point is a non-light leak area.

[0073] Step 3: Apply a local contrast enhancement algorithm to the selected suspicious areas, combined with multi-scale pyramid analysis technology, to enhance the characteristics of weak light leakage signals;

[0074] Step 3.1, Calculation of local statistical properties;

[0075] For each suspected light leakage area, calculate its local mean and local standard deviation:

[0076]

[0077] Where, μ local (x, y) represents the mean value of the suspicious light leakage area within the local window centered at coordinates (x, y), W(x, y) is the local window centered at (x, y), I(i, j) represents the value of the original image at coordinates (i, j), M(i, j) represents the value of the binarized marker image at coordinates (i, j), and N is the number of pixels marked as suspicious areas within the window;

[0078]

[0079] Where, σ local (x, y) represents the standard deviation of the suspected light leakage area within the local window centered at coordinates (x, y), used to measure the degree of change in pixel values ​​within the local area.

[0080] Step 3.2, contrast enhancement;

[0081] The suspicious area is enhanced using a local contrast enhancement algorithm:

[0082]

[0083] Where I′(x, y) represents the value of the enhanced image at coordinates (x, y), and α contrast ε is the enhancement factor, which controls the degree of contrast enhancement. small To prevent small constants from being divided by zero, M(x, y) ensures that enhancement is only applied to suspected light leakage areas.

[0084] Step 3.3, Multi-scale pyramid analysis;

[0085] Construct a Gaussian pyramid {G0, G1, ..., G...} for the enhanced image n} and the Laplace Pyramid {L0, L1, ..., L n};

[0086] G i+1 =DownSample(G i );

[0087] L i =G i -UpSample(G i+1 );

[0088] Where G0 represents the original image of the enhanced Gaussian pyramid, G1, G... i G n L1, L2, and Ln represent the Gaussian pyramid images at levels l, i, and n, respectively. L0 represents the original image of the enhanced Laplacian pyramid, and L1 and L2 represent the images of the same level. i L n Let represent the Laplacian pyramid images at layers l, i, and n, respectively, where n is the layer number of the Gaussian pyramid and the Laplacian pyramid. DownSample and UpSample represent the downsampling and upsampling operations of the image, respectively.

[0089] The specific implementation method of multi-scale pyramid analysis technique is as follows:

[0090] The downsampling operation (DownSample) uses a 5×5 Gaussian kernel for filtering and then takes one pixel for every two pixels, reducing the image size to half its original size.

[0091]

[0092] Where w(i,j) is the Gaussian weight matrix, G(x,y) represents the value of the downsampled image at coordinates (x,y), and I(x+i,y+j) represents the value of the original image at coordinates (x+i,y+j).

[0093] The upsampling operation UpSample doubles the size of the image by inserting zero values, and then applies the same 5×5 Gaussian kernel for interpolation.

[0094]

[0095] Where U(x, y) represents the value of the upsampled image at coordinates (x, y), and G... i+1 This represents the next level of the Gaussian pyramid image. and This represents the corresponding coordinates in the next layer of the image.

[0096] In practical applications, the number of pyramid layers, n, is typically set to 3 to 5 layers, adaptively adjusted based on the size of the backlight module and the expected range of light leakage characteristics. For large-size LCD panels, a 5-layer pyramid is used to capture light leakage characteristics at different scales, from millimeters to centimeters; for small and medium-sized panels, a 3-layer pyramid is sufficient to meet the detection requirements.

[0097] Step 3.4, Feature Enhancement Fusion;

[0098] By weighted fusion of multi-scale features, the final enhanced image I is generated. enhanced :

[0099]

[0100] Among them, w i L represents the weighting coefficients of the i-th layer of the Laplacian pyramid, which are adaptively adjusted based on the significance of light leakage characteristics at different scales. i This represents the image of the i-th layer of the Laplace pyramid.

[0101] Step 4: Apply box dimension calculation and multifractal spectrum analysis to the enhanced micro-area leakage morphology to quantify the structural complexity of the leakage region and construct fractal feature vectors.

[0102] Step 4.1, segmentation and marking of light leakage areas;

[0103] For enhanced image I enhanced Adaptive threshold segmentation is applied to obtain an accurate binary image B of the light leakage region. binary (x,y) and mark the connected components.

[0104] Step 4.2, box dimension calculation;

[0105] For each marked area of ​​light leakage, calculate its fractal dimension D using box counting:

[0106]

[0107] Where D represents the fractal dimension of the light leakage region, ε represents the side length of the square covering the light leakage region, N(ε) represents the number of squares with side length ε required to cover the light leakage region, and lim ε→0 This represents the limit as the side length of the square approaches zero. In practical calculations, it is approximated by the slope of a linear regression.

[0108] logN(ε)≈D·log(1 / ε)+C;

[0109] The actual steps for calculating the box dimension are as follows:

[0110] Image B of the binarized light leakage region binary (x, y) is covered by a series of meshes of different scales, with the mesh size ε decreasing geometrically:

[0111] ε i =ε0·2 -i ;

[0112] Where ε0 represents the initial mesh size, ε i This represents the size of the i-th grid, where i = 0, 1, 2, ..., k-1, and k is the total number of grid sizes, typically ranging from 8 to 12.

[0113] For each grid size ε i Calculate the number of grids N(ε) required to cover the light leakage area. i );

[0114] Plot a double logarithmic coordinate graph, with the x-axis being log(1 / ε). i ), with the ordinate being logN(ε) i );

[0115] The slope of the line is the fractal dimension D when the least squares method is used to fit the line.

[0116] In practical applications of light leakage detection in LCD backlight modules, the fractal dimensions of different types of light leakage vary:

[0117] Light leakage caused by bubbles: The fractal dimension D is usually between 1.0 and 1.3, which is close to the Euclidean geometry of a circle;

[0118] Light leakage caused by indentation: The fractal dimension D is usually between 1.3 and 1.6, and it appears as a slightly irregular star-shaped structure;

[0119] Light leakage caused by material fracture: The fractal dimension D is usually between 1.6 and 1.8, exhibiting a highly irregular branching structure;

[0120] Light leakage caused by uneven penetration of optical adhesive: The fractal dimension D is usually between 1.4 and 1.7, but its multifractal spectrum width Δα is significantly larger than other types.

[0121] By accurately quantifying fractal dimension and its multifractal properties, automatic classification of light leakage patterns can be achieved without the need for human experience-based judgment, thereby improving the intelligence level and reliability of the detection system.

[0122] Step 4.3, multifractal spectrum analysis;

[0123] Calculate the multifractal spectrum f(α) of the light leakage region:

[0124] f(α)=τ*(q)=q·α(q)-τ(q)

[0125] In this context, f(α) represents the multifractal spectrum, τ(q) is the quality index, representing the scaling characteristics of the leakage region at different q values, α(q) is the singularity index, representing the local scaling characteristics of the leakage region at different locations, and q is the moment trap index, used to adjust the weights for regions of different densities. The multifractal spectrum can more comprehensively characterize the complex structural properties of the leakage region.

[0126] Step 4.4, constructing fractal feature vectors;

[0127] Construct a fractal feature vector (DF) that includes box dimension, multifractal spectrum feature points, and topological features:

[0128] DF=[D,f(α min ), f(α) max ), α0, Δα, CN];

[0129] Where D is the box dimension, f(α) min ) and f(α) max ) are the function values ​​corresponding to the minimum and maximum values ​​of the multifractal spectrum, respectively. α0 is the value of α when f(α) reaches its maximum value, representing the dominant scaling characteristic of the light leakage region. Δα is the range of values ​​for α, representing the multifractal intensity of the light leakage region. CN is the connectivity number, representing the topological characteristics of the light leakage region.

[0130] Step 5: Construct a correlation model between fractal feature vectors and optical properties based on machine learning algorithms, establish a mapping relationship between light leakage morphology and material defect type, and calculate and output the location, range, severity and defect type of light leakage through light leakage severity scoring;

[0131] Step 5.1, Feature Fusion;

[0132] By combining the fractal feature vector DF with the optical feature matrix F(x,y), a comprehensive description vector CF for light leakage features is constructed:

[0133] CF = [DF, F avg F var F max [A, P];

[0134] Where CF represents the comprehensive description vector of light leakage features, DF is the fractal feature vector, and F avg F var F max Let A represent the average, variance, and maximum value of the optical characteristics within the light leakage region, respectively, and let P represent the light leakage area and the light leakage perimeter.

[0135] Step 5.2, Defect type mapping model construction;

[0136] Based on a pre-labeled database of typical defects, a mapping model M between light leakage characteristics and material defect types is constructed. defect The calculation formula is as follows:

[0137] Type defect =M defect (CF);

[0138] Among them, Type defect M represents the predicted defect type. defect This represents a mapping model that maps the light leakage feature comprehensive description vector CF to the corresponding defect type.

[0139] Defect Type Mapping Model M defect Implemented using a random forest classifier structure, this model comprises the following components:

[0140] Input layer: Receives the feature vector CF as input;

[0141] Decision tree set: consists of multiple decision trees {T1, T2, T...} i ,…,T K Composed of}, where T1, T2, T i T K These represent the 1st, 2nd, i, and Kth decision trees, respectively, where K is the number of decision trees. Each decision tree is trained based on a different subset of the feature vectors.

[0142] Voting Integration Layer: Summarizes the classification results of all decision trees and uses a majority voting strategy to determine the final defect type.

[0143] The model is trained using the standard random forest training method, and is constructed through the following steps:

[0144] For training dataset D train Perform random sampling to generate K subsets {D1, D2, D...} i D K}, where D1, D2, D i D K These are represented as the 1st, 2nd, i, and Kth subsets, respectively, where K is the number of decision trees and subsets.

[0145] For each subset D i Training a decision tree T i ;

[0146] When a node splits, each decision tree randomly selects m features (m < M) from the M features of the feature vector CF as splitting feature candidates.

[0147] Select the optimal splitting feature and threshold based on information gain or Gini coefficient;

[0148] Repeat steps 3 and 4 until all decision trees are constructed.

[0149] In practical applications, this model has been specifically optimized for typical defect types in LCD backlight modules, and can accurately distinguish the following defect types:

[0150] Bubble defects (characterized by regular circular light leakage and low fractal dimension); indentation defects (characterized by irregular star-shaped light leakage and medium fractal dimension); material fracture (characterized by linear light leakage and high fractal dimension); uneven penetration of optical adhesive (characterized by blurred boundary light leakage and broad multifractal spectrum).

[0151] Step 5.3, scoring the severity of light leakage;

[0152] Calculate the severity score S of light leakage based on its location, extent, intensity, and morphological characteristics:

[0153] S=w1·I norm +w2·A norm +w3·D norm +w4·L norm

[0154] Where S represents the severity score of light leakage, and I norm A represents the normalized leakage intensity. norm D represents the normalized leakage area. norm L represents the normalized fractal dimension. norm The normalized position factor is the distance between the light leakage area and the center of the screen. w1, w2, w3, and w4 are the first, second, third, and fourth weighting coefficients, used to adjust the degree of influence of different factors on the final score.

[0155] Step 5.4, output the evaluation results;

[0156] Generate a light leakage detection and evaluation report, including the following information: light leakage location coordinates and distribution map; light leakage range and area statistics; severity score and classification; inferred possible defect types and confidence levels; fractal feature analysis results of light leakage morphology.

[0157] This implementation method achieves high-precision measurement of backlight module light leakage by organically combining core technologies such as multi-dimensional optical feature analysis, dynamic threshold adaptive segmentation, local contrast enhancement, fractal dimension quantification, and machine learning algorithms. Furthermore, the fractal dimension analysis method combined with machine learning algorithms provides a precise means of quantifying the morphological characteristics of light leakage, overcoming the limitations of traditional geometric features in characterizing irregular light leakage morphologies. It also enables correlation analysis between light leakage morphology and material defect types, allowing direct inference of possible material defect types from light leakage phenomena, providing a direct basis for production process optimization. Notably, this technology seamlessly integrates large-area detection and micro-area fine analysis within the same detection process, improving detection efficiency.

[0158] In one embodiment of the present invention, an example of the aforementioned machine learning-based backlight module light leakage measurement method is provided:

[0159] Application Scenario Description: A certain LCD monitor manufacturer faces the following problems in the quality inspection of 55-inch TV backlight modules: the lighting conditions on the production line are complex and variable, and ambient light interferes with the detection of backlight leakage; traditional inspection methods have a low detection rate for small light leakage areas, resulting in some product quality problems being discovered only at the user end; manual inspection methods are highly subjective, have poor consistency in inspection results, and cannot accurately determine the type of defect.

[0160] The backlight module light leakage measurement method of this embodiment is used to solve the problem. The specific implementation process is as follows:

[0161] The experimental equipment and parameter configurations are shown in the table below:

[0162] Device / Parameter Category Specific configuration illustrate Image acquisition equipment 50-megapixel industrial camera Resolution 8192×5460 Light source system Uniform LED surface light source Color temperature 6500K±200K Processing platform Industrial PC Intel i7 processor, 32GB RAM Dynamic threshold coefficient k Variable range 0.8-1.5 Automatically adjusts based on ambient light intensity Contrast enhancement factor α 1.2 Experimental optimization parameters Multi-scale pyramid layers 4 floors Suitable for 55-inch backlight modules

[0163] The original backlight module image acquired in this example is 8192×5460 pixels. The extracted feature dimensions and their data ranges are shown in the table below:

[0164] Feature type Data representation Numerical range Feature Dimension Brightness feature B(x, y) Single channel grayscale value 0-255 1 Color feature C(x, y) HSV Color Space H: 0-360, S: 0-1, V: 0-1 3 Texture feature GLCM(x,y) Energy, contrast, correlation, entropy Normalized to 0-1 4 Edge features E(x, y) Gradient magnitude and direction Amplitude: 0-255, Direction: 0-360 2

[0165] In an actual test, the system detected a suspicious light leakage signal in the lower right corner of the backlight module. The example values ​​of the luminance components of the local region (10×10 pixels) of its feature matrix F(x,y) are as follows:

[0166]

[0167] The data shows that there is a significant brightness anomaly in this area, with the brightness value in the central area being significantly higher than that in the surrounding areas, which initially matches the characteristic pattern of backlight leakage.

[0168] Applying the dynamic threshold adaptive segmentation algorithm to the above area, the local mean μ(x,y) is calculated to be 221.5, the local standard deviation σ(x,y) is 14.3, and the threshold coefficient k after automatic correction of ambient brightness is 1.2. Therefore, the dynamic threshold T(x,y) is calculated to be 238.7.

[0169] Applying this threshold for segmentation, the binarized suspected light leakage region label matrix M(x, y) is obtained as follows:

[0170]

[0171] It is evident that the system successfully identified areas of suspected light leakage, forming a preliminary light leakage area marking matrix.

[0172] A local contrast enhancement algorithm is applied to the marked suspicious areas to calculate the local mean μ. local (x, y) is 243.2 and the local standard deviation σ local (x, y) is 5.8, and the enhancement coefficient α contrast It is 1.2, a small constant ε small It is 0.001.

[0173] The original image was enhanced by applying a contrast enhancement algorithm, which effectively improved the contrast between the light leakage area and the background. The brightness value of the center point was increased from 250 to 258.3, with an enhancement effect of about 3.3%, making the weak light leakage characteristics more obvious.

[0174] Next, the system constructed a four-layer Gaussian Laplacian pyramid and performed multi-scale analysis on the enhanced image. The third-layer pyramid image revealed faint light leak regions that were not detected in the original segmentation. After feature enhancement and fusion, the area of ​​the finally labeled light leak regions increased by 23%, capturing a more complete light leak morphology.

[0175] Fractal dimension analysis was performed on the finally determined light leakage region, and box counting was conducted using eight different grid scales (from 4 pixels to 512 pixels, scaled in powers of 2). The counting results are shown in the table below:

[0176] Grid size ε (pixels) The number of grid cells N(e) required for coverage log(1 / ε) Log N(ε) 4 783 5.64 6.66 8 215 4.64 5.37 16 64 3.64 4.16 32 21 2.64 3.04 64 8 1.64 2.08 128 3 0.64 1.10 256 1 -0.36 0 512 1 -1.36 0

[0177] Linear fitting of the above data using the least squares method yielded a slope D = 1.56, indicating that the fractal dimension of the light leakage region is 1.56. Simultaneously, the multifractal spectrum feature points were calculated, obtaining f(α) min The value is 0.92, and f(α) max ) is 0.88, α0 is 1.62, Δ α The value is 0.7, and the connectivity number CN is 1.

[0178] Based on the fractal feature vector [1.56,0.92,0.88,1.62,0.7,1] and the previously obtained optical features, a comprehensive analysis determines that the light leakage area is most likely caused by an indentation defect.

[0179] The system calculated a light leakage severity score S of 68 (out of 100) based on the fused feature vector CF, classifying it as moderate light leakage, located in the lower right region, with a leakage area of ​​approximately 12.6 mm². 2 The light leakage defect was identified as an indentation defect with a confidence level of 94.2%. This determination was automatically generated by a machine learning model, requiring no human intervention or experience-based judgment.

[0180] The test results were compared with the actual disassembly and inspection results, confirming that the backlight module did indeed have a slight indentation defect, verifying the accuracy of the system's judgment.

[0181] Technical effectiveness verification:

[0182] 1000 backlight modules randomly selected from the production line were tested, and the performance indicators of this method and the traditional testing method were compared, as shown in the table below:

[0183] Detection methods Accuracy (%) False alarm rate (%) False negative rate (%) <![CDATA[Sensitivity (cd / m 2 )]]> Traditional methods 82.3 12.4 18.6 0.05 This method 98.7 7.2 3.4 0.01 Improvement effect +16.4 -5.2 -15.2 +80％

[0184] As can be seen, this method improves the detection accuracy while reducing the false alarm rate and false negative rate, and increases the detection sensitivity by 80%.

[0185] Defect type identification was performed on 100 light leakage samples with known defect types, and the identification accuracy of each type of defect was statistically analyzed, as shown in the table below:

[0186] Defect types Sample size Correctly identify numbers Recognition accuracy (%) Class A: Bubble Defects 28 27 96.4 Class B: Indentation defects 35 33 94.3 Category C: Material Fracture 22 21 95.5 Category D: Uneven penetration of optical adhesive 15 14 93.3 total 100 95 95.0

[0187] The overall accuracy of defect type identification reached 95.0%, and the identification accuracy of various types of defects was above 93%, which verified the defect classification ability of this method based on machine learning.

[0188] To further improve system performance, transfer learning techniques can be employed, utilizing a pre-trained deep learning model as a feature extractor, combined with the fractal feature analysis method of this invention, to construct a more powerful light leakage detection system. Experiments show that this hybrid method can improve the defect identification accuracy to over 97.5%, especially performing exceptionally well when the number of samples is limited.

[0189] The above examples demonstrate that this method can effectively solve the technical problems faced by backlight module light leakage detection in actual production environments.

Claims

1. A method for measuring backlight leakage in a backlight module based on machine learning, characterized in that, Includes the following steps: Collect multi-dimensional optical feature information of the backlight module and construct the original feature matrix; Based on ambient light conditions, an adaptive threshold algorithm is used to initially segment the backlight module and screen out suspicious light leakage areas. A local contrast enhancement algorithm is applied to the screened suspicious areas, combined with multi-scale pyramid analysis technology, to enhance the characteristics of weak light leakage signals; For the enhanced micro-area light leakage morphology, box dimension calculation and multifractal spectrum analysis are applied to quantify the structural complexity of the light leakage region and construct fractal feature vectors. The formula for enhancing the selected suspicious areas using the local contrast enhancement algorithm is as follows: ； in, For the enhanced image, For the original image, It is a local mean. For local standard deviation, To enhance the coefficient, To prevent small constants from being divided by zero, Mark suspicious areas on the map; A correlation model between fractal feature vectors and optical properties is constructed based on machine learning algorithms. The mapping relationship between light leakage morphology and material defect type is established. The location, range, severity and defect type of light leakage are calculated and output through light leakage severity scoring. The fractal eigenvectors are constructed as follows: ； in, Let the box dimension be , and These are the function values ​​corresponding to the minimum and maximum values ​​of the multifractal spectrum, respectively. for When taking the maximum value The value represents the dominant scaling characteristic of the light leakage region. for The range of values ​​for represents the multifractal intensity of the light leakage region. is the connectivity number, representing the topological characteristics of the light leakage region; The formula for calculating the severity of light leakage is: ； in, The score indicates the severity of light leakage. This represents the normalized leakage intensity. This represents the normalized leakage area. The normalized fractal dimension, The normalized position factor is the distance between the light leakage area and the center of the screen. , , , These are the first, second, third, and fourth weighting coefficients, used to adjust the degree of influence of different factors on the final score.

2. The method for measuring backlight module light leakage based on machine learning according to claim 1, characterized in that, The original feature matrix is ​​constructed as follows: ； in, Indicates brightness characteristics, Indicates color characteristics, Represents texture features, Indicates edge features.

3. The method according to claim 1, characterized in that, The adaptive threshold algorithm has the following calculation formula: ； in, Represents coordinates, Indicates dynamic threshold. The mean of the local region. The standard deviation of the local region. This is the threshold coefficient, which is automatically adjusted based on the ambient light intensity.

4. The method for measuring backlight module light leakage based on machine learning according to claim 1, characterized in that, The multi-scale pyramid analysis technique includes constructing a Gaussian pyramid and a Laplacian pyramid for the enhanced image, and generating the final enhanced image by weighted fusion of multi-scale features, wherein the weight coefficients are adaptively adjusted according to the significance of light leakage features at different scales.

5. The method for measuring backlight module light leakage based on machine learning according to claim 1, characterized in that, The box dimension calculation is performed through the following steps: Different scales of grids are applied to cover the light leakage area, with the grid size decreasing geometrically. Calculate the number of grids required to cover the light leakage area for each grid size; Plot a double logarithmic coordinate graph, with the horizontal axis being the reciprocal of the logarithm of the grid size and the vertical axis being the logarithm of the number of grid cells; The slope is obtained by linear regression and used as the fractal dimension.

6. The method for measuring backlight module light leakage based on machine learning according to claim 1, characterized in that, The correlation model between fractal features and optical properties is implemented using a random forest classifier structure. This model includes an input layer, a set of decision trees, and a voting integration layer, which can distinguish between different types of defects.

7. A machine learning-based backlight module light leakage measurement system, used to execute the machine learning-based backlight module light leakage measurement method according to any one of claims 1-6, characterized in that, include: The image acquisition and feature extraction module is used to acquire multi-dimensional optical feature information of the backlight module and construct the original feature matrix. The adaptive segmentation module is used to perform preliminary segmentation of the backlight module based on ambient light conditions and an adaptive threshold algorithm to screen out suspicious light leakage areas. The contrast enhancement module is used to apply a local contrast enhancement algorithm to the screened suspicious areas, combined with multi-scale pyramid analysis technology, to enhance the characteristics of weak light leakage signals. The fractal feature analysis module is used to quantify the structural complexity of the light leakage region and construct fractal feature vectors by applying box dimension calculation and multifractal spectrum analysis to the enhanced micro-area light leakage morphology. The machine learning defect assessment module is used to build a correlation model between fractal feature vectors and optical properties based on machine learning algorithms, establish a mapping relationship between light leakage morphology and material defect type, and calculate and output the location, range, severity and defect type of light leakage through light leakage severity scoring.

Images