Mini LED lamp panel needle mark defect detection method and system

Abstract

The invention discloses a Mini LED lamp panel needle mark defect detection method and system, and the method comprises the following steps: S1, carrying out the preprocessing of a Mini LED lamp panel surface image, suppressing high-frequency noise, and enhancing the defect features; s2, extracting defect features by adopting a self-adaptive convolution adjacency analysis method: S21, dynamically adjusting the shape and size of a convolution kernel through a deformable self-adaptive convolution kernel, and matching the local form of the needle mark defect; and S22, carrying out correlation analysis on the pixel points and neighborhoods thereof, and eliminating isolated noise interference. According to the method, defect forms are dynamically matched through the self-adaptive convolution kernel, isolated noise is eliminated through adjacency analysis, defects of different sizes are covered through multi-scale fusion, wavelet filtering and a parallel computing architecture are combined, the detection precision and the anti-jamming capability are remarkably improved, stable recognition of the tiny needle marks in the complex environment is achieved, meanwhile, high-speed real-time processing is guaranteed, and the detection accuracy and the anti-jamming capability are improved. The method is suitable for diversified industrial scenes and material requirements, reduces the false detection and omission ratio, and promotes the intelligent upgrading of precision manufacturing quality inspection.

Description

Technical Field

[0001] The present invention relates to the field of optical detection technology, and in particular to a method and system for detecting needle mark defects on a MiniLED lamp board. Background Art

[0002] In the manufacturing of MiniLED light panels, accurate detection of needle mark defects is crucial to product quality.

[0003] Traditional inspection methods rely on image processing with fixed convolution kernels or manual visual inspection. These fixed convolution kernels are difficult to adapt to tiny needle marks of varying shapes, leading to missed detections. Manual visual inspection is inefficient and susceptible to fatigue. Existing algorithms are also sensitive to noise such as lighting changes and metal reflections, resulting in high false positive rates. They struggle to meet the real-time demands of high-speed production lines and lack adaptability across materials and environments, severely limiting detection efficiency and reliability.

[0004] To address the above issues, we have introduced a MiniLED light board needle mark defect detection method and system. Summary of the Invention

[0005] The present invention discloses a MiniLED lamp board needle mark defect detection method and system thereof, aiming to solve the technical problems in the background technology.

[0006] In order to achieve the above object, the present invention adopts the following technical solutions: A method for detecting needle mark defects on a MiniLED light board includes the following steps: S1. Preprocess the surface image of the MiniLED light board to suppress high-frequency noise and enhance defect features; S2. Adaptive convolution adjacency analysis method is used to extract defect features, including: S21, dynamically adjust the shape and size of the convolution kernel through the deformable adaptive convolution kernel to match the local morphology of the needle mark defect; S22, performing correlation analysis on the pixel points and their neighborhoods to eliminate isolated noise interference; S3, perform multi-scale analysis on the extracted features to identify needle mark defects of different sizes; S4, optimize the defect contour by combining morphological processing, and extract the defect area by adaptive threshold segmentation; S5. Output defect location and classification results.

[0007] In a preferred solution, the dynamic adjustment of the deformable adaptive convolution kernel includes generating convolution kernel parameters that match the needle mark defect morphology based on the gradient distribution and texture characteristics of the local area of ​​the image.

[0008] In a preferred solution, the neighborhood correlation analysis includes calculating the grayscale difference and spatial correlation between the target pixel and the neighborhood pixels, constructing an adjacency matrix, and screening out continuous defect areas.

[0009] In a preferred solution, the preprocessing step includes using a wavelet filtering method to separate the high-frequency components of the image from the low-frequency background, thereby enhancing the detailed features of the needle mark defect.

[0010] In a preferred solution, the multi-scale analysis includes synchronously performing adaptive convolutional adjacency analysis on image layers with different resolutions, and fusing the detection results of each layer to cover defects of different sizes.

[0011] In a preferred solution, the adaptive threshold segmentation includes dynamically adjusting the segmentation threshold according to the contrast of the local area to ensure the integrity of the defect contour under uneven lighting conditions.

[0012] In a preferred embodiment, an efficient computational optimization step is also included: Adaptive convolution adjacency analysis, wavelet filtering, and multi-scale analysis modules are deployed in a parallel computing architecture to achieve real-time processing.

[0013] A MiniLED light board needle mark defect detection system, comprising: Image acquisition module, including industrial camera and uniform lighting source, used to obtain the surface image of the light board; The detection algorithm module according to any one of claims 1 to 7, running on an embedded hardware platform; The result output module marks the defect location and generates a test report.

[0014] In a preferred solution, the image acquisition module further includes: Polarization filter component, used to eliminate reflection interference on the light board surface; The light source adaptive adjustment unit dynamically adjusts lighting parameters according to the ambient light intensity.

[0015] The present invention provides a method and system for detecting needle mark defects on a MiniLED light board, which has the following advantages: This patent uses adaptive convolution kernels to dynamically match defect morphology, adjacency analysis to eliminate isolated noise, and multi-scale fusion to cover defects of different sizes. It combines wavelet filtering with parallel computing architecture to significantly improve detection accuracy and anti-interference capabilities, achieve stable identification of tiny needle marks in complex environments, and ensure high-speed real-time processing. It adapts to diverse industrial scenarios and material requirements, reduces false detection and missed detection rates, and promotes the upgrade of precision manufacturing quality inspection to intelligent ones. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1This is a schematic diagram of the overall steps of a MiniLED lamp board needle mark defect detection method proposed in the present invention.

[0017] Figure 2 This is a structural schematic diagram of a MiniLED light board needle mark defect detection system proposed in the present invention. DETAILED DESCRIPTION

[0018] The technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all of the embodiments. The components of the embodiments of the present application generally described and marked in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the application for protection, but merely represents the selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without making creative work fall within the scope of protection of the present application.

[0019] The present invention discloses a method and system for detecting needle mark defects in a MiniLED lamp board.

[0020] Reference Figure 1 and Figure 2 As shown, a method for detecting needle mark defects on a MiniLED light board includes the following steps: S1. Preprocess the surface image of the MiniLED light board to suppress high-frequency noise and enhance defect features; S2. Adaptive convolution adjacency analysis method is used to extract defect features, including: S21, dynamically adjust the shape and size of the convolution kernel through the deformable adaptive convolution kernel to match the local morphology of the needle mark defect; S22, performing correlation analysis on the pixel points and their neighborhoods to eliminate isolated noise interference; S3, perform multi-scale analysis on the extracted features to identify needle mark defects of different sizes; S4, optimize the defect contour by combining morphological processing, and extract the defect area by adaptive threshold segmentation; S5. Output defect location and classification results; When processing the surface image of the MiniLEDX board, a wavelet filtering method is used to decompose the local frequency and low-frequency components of the image. By retaining high-frequency details and adding low-frequency background noise, the visibility of the needle mark defect is enhanced. Subsequently, the defect features are extracted through the adaptive convolution adjacency analysis method (ACAA): the deformable adaptive convolution kernel dynamically adjusts the aspect ratio and rotation angle of the convolution kernel according to the gradient direction and texture complexity of the local area to match the curved shape of the needle mark; the neighborhood correlation analysis calculates the grayscale difference and spatial distance between the target pixel and the surrounding pixels to construct the adjacency relationship matrix. (in is the pixel grayscale, is the pixel spacing), continuous regions with a correlation above a preset threshold are selected as candidate defects. To perform multi-scale analysis on the extracted features, ACAA is repeatedly performed on the downsampled image layer, fusing inspection results at different resolutions to cover needle marks ranging from micrometers to millimeters. Finally, morphological closing operations are used to fill the internal voids of the defect, and local contrast adaptive threshold segmentation is combined to extract the complete outline. The defect coordinates and type classification are then output.

[0021] Reference Figure 1 and Figure 2 As shown, in a preferred embodiment, the dynamic adjustment of the deformable adaptive convolution kernel includes generating convolution kernel parameters that match the needle mark defect morphology according to the gradient distribution and texture characteristics of the local area of ​​the image; The dynamic adjustment of the deformable adaptive convolution kernel is achieved by the following steps: First, the gradient amplitude of the local area of ​​the image is calculated and main direction , adjust the rotation angle of the convolution kernel according to the gradient direction so that the long axis of the kernel is aligned along the extension direction of the needle mark; secondly, calculate the complexity of the local texture (such as grayscale variance). If the texture is complex, reduce the size of the convolution kernel to focus on the details, otherwise expand the kernel size to cover the long straight defects. Convolution kernel parameters Determined jointly by gradient and texture features: ,Var), where is the preset mapping function.

[0022] Reference Figure 1 and Figure 2 As shown, in a preferred embodiment, the neighborhood correlation analysis includes calculating the grayscale difference and spatial correlation between the target pixel and the neighboring pixels, constructing an adjacency matrix, and screening out continuous defect areas; In neighborhood correlation analysis, the target pixel is calculated Grayscale difference with its 8 neighboring pixels , and based on pixel spacing Calculating spatial association weights . Construct the adjacency matrix Then, set the correlation threshold ,when When the two defects are identified as the same defect area, isolated noise points (such as random noise points) are filtered out and the continuous needle mark area is retained.

[0023] Reference Figure 1 and Figure 2 As shown, in a preferred embodiment, the preprocessing step includes using a wavelet filtering method to separate the high-frequency components of the image from the low-frequency background, thereby enhancing the detail features of the needle mark defect; In the preprocessing step, a two-dimensional discrete wavelet transform (DWT) is used to decompose the image into high-frequency sub-bands (detail components) and low-frequency sub-bands (approximation components). Nonlinear enhancement is performed on the high-frequency sub-bands to improve the sharpness of the needle mark edges. The low-frequency subbands are smoothed and filtered to suppress background texture interference. The enhanced subbands are reconstructed into preprocessed images through inverse wavelet transform.

[0024] Reference Figure 1 and Figure 2 As shown, in a preferred embodiment, the multi-scale analysis includes performing adaptive convolutional adjacency analysis simultaneously on image layers of different resolutions, fusing the detection results of each layer to cover defects of different sizes; Multi-scale analysis is achieved by constructing an image pyramid: the original image is downsampled to generate multi-resolution image layers, and adaptive convolutional adjacency analysis is performed independently on each layer to detect needle mark defects of different sizes. After the detection results are upsampled to align with the original resolution, a weighted fusion strategy is used to combine the results from each layer, ensuring that small defects are captured in the high-resolution layers and large defects are identified in the low-resolution layers.

[0025] Reference Figure 1 and Figure 2 As shown, in a preferred embodiment, the adaptive threshold segmentation includes dynamically adjusting the segmentation threshold according to the contrast of the local area to ensure the integrity of the defect contour under uneven lighting conditions; Adaptive threshold segmentation calculates dynamic thresholds based on the grayscale distribution within the local window ,in is the window mean, is the standard deviation, For high contrast areas (such as needle marks), automatically reduce For low contrast areas (such as light falloff areas), increase Avoid missed detection.

[0026] Reference Figure 1 and Figure 2 As shown, in a preferred embodiment, it also includes an efficient calculation optimization step: Deploy adaptive convolution adjacency analysis, wavelet filtering, and multi-scale analysis modules on a parallel computing architecture to achieve real-time processing; Adaptive threshold segmentation calculates dynamic thresholds based on the grayscale distribution within the local window ,in is the window mean, is the standard deviation, For high contrast areas (such as needle marks), automatically reduce For low contrast areas (such as light falloff areas), increase Avoid missed detection.

[0027] A MiniLED light board needle mark defect detection system, comprising: Image acquisition module, including industrial camera and uniform lighting source, used to obtain the surface image of the light board; The detection algorithm module according to any one of claims 1 to 7, running on an embedded hardware platform; Result output module, marking defect locations and generating inspection reports; The convolution operations in adaptive convolution adjacency analysis are deployed on the FPGA's programmable logic units, utilizing pipeline parallel acceleration. Adjacency matrix construction and wavelet filtering are processed in parallel by multi-core CPUs. Multi-scale analysis tasks are assigned to the GPU's CUDA cores, increasing throughput through thread-block-level parallelism. Data is exchanged between modules via a high-speed bus, enabling end-to-end real-time processing.

[0028] The image acquisition module also includes: Polarization filter component, used to eliminate reflection interference on the light board surface; The light source adaptive adjustment unit dynamically adjusts lighting parameters according to the ambient light intensity.

[0029] The inspection system consists of an industrial area scan camera, a coaxial illumination source, and an embedded processing platform. The camera captures images of the panel surface at a high frame rate, while the coaxial illumination source provides shadowless illumination to reduce reflection interference. The embedded platform, equipped with an ACAA algorithm module, performs real-time processing on input images. Inspection results are transmitted to a host computer via Gigabit Ethernet, generating a visual report containing defect location, size, and confidence level.

[0030] The above description is merely a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. The replacement may be a replacement of a portion of a structure, device, or method step, or it may be a complete technical solution. Any equivalent replacement or modification based on the technical solution and inventive concept of the present invention shall be covered by the scope of protection of the present invention.

Claims

1. A method for detecting needle mark defects on MiniLED light boards, characterized in that: The following steps are involved: S1. Preprocess the surface image of the MiniLED light board to suppress high-frequency noise and enhance defect features; S2. Adaptive convolution adjacency analysis method is used to extract defect features, including: S21, dynamically adjust the shape and size of the convolution kernel through the deformable adaptive convolution kernel to match the local morphology of the needle mark defect; S22, performing correlation analysis on the pixel points and their neighborhoods to eliminate isolated noise interference; S3, perform multi-scale analysis on the extracted features to identify needle mark defects of different sizes; S4, optimize the defect contour by combining morphological processing, and extract the defect area by adaptive threshold segmentation; S5. Output defect location and classification results.

2. The method for detecting needle mark defects on a MiniLED light board according to claim 1, wherein: The dynamic adjustment of the deformable adaptive convolution kernel includes generating convolution kernel parameters that match the needle mark defect morphology according to the gradient distribution and texture characteristics of the local area of ​​the image.

3. The method for detecting needle mark defects on a MiniLED light board according to claim 1, wherein: The neighborhood correlation analysis includes calculating the grayscale difference and spatial correlation between the target pixel and the neighborhood pixels, constructing an adjacency matrix, and screening out continuous defect areas.

4. The method for detecting needle mark defects on a MiniLED light board according to claim 1, wherein: The preprocessing step includes using a wavelet filtering method to separate the high-frequency components of the image from the low-frequency background, thereby enhancing the detail features of the needle mark defect.

5. The method for detecting needle mark defects on a MiniLED light board according to claim 1, wherein: The multi-scale analysis includes synchronously performing adaptive convolutional adjacency analysis on image layers of different resolutions and fusing the detection results of each layer to cover defects of different sizes.

6. The method for detecting needle mark defects on a MiniLED light board according to claim 1, wherein: The adaptive threshold segmentation includes dynamically adjusting the segmentation threshold according to the contrast of the local area to ensure the integrity of the defect contour under uneven lighting conditions.

7. The method for detecting needle mark defects on a MiniLED light board according to claim 1, wherein: Also included are computationally efficient optimization steps: Adaptive convolution adjacency analysis, wavelet filtering, and multi-scale analysis modules are deployed in a parallel computing architecture to achieve real-time processing.

8. A MiniLED light board needle mark defect detection system, characterized in that: include: Image acquisition module, including industrial camera and uniform lighting source, used to obtain the surface image of the light board; The detection algorithm module according to any one of claims 1 to 7, running on an embedded hardware platform; The result output module marks the defect location and generates a test report.

9. A MiniLED light board needle mark defect detection system, characterized in that: The image acquisition module also includes: Polarization filter component, used to eliminate reflection interference on the light board surface; The light source adaptive adjustment unit dynamically adjusts lighting parameters according to the ambient light intensity.

Images