Aluminum alloy surface defect identification system and method
By working together with a 3D line laser contour scanner and a central processing module, the problems of low efficiency and poor accuracy in identifying surface defects in aluminum alloys have been solved. This has enabled precise defect identification and scientific production decision support, thereby improving production efficiency and product quality.
Patent Information
- Application Number
- CN202511112385.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-08-08
- Publication Date
- 2025-11-21
AI Technical Summary
Existing technologies for identifying surface defects in aluminum alloys are inefficient and inaccurate. Manual inspection is slow and easily affected by human factors. 2D vision recognition systems have high requirements for lighting conditions and are difficult to extract small defect features, resulting in imprecise inspection results.
A 3D line laser contour scanner is used to acquire three-dimensional depth images of the aluminum alloy surface. Combined with the multi-threaded task scheduler and high-speed data cache of the central processing module, the images are scientifically analyzed by the defect identification module, and an alarm module is provided to ensure the accuracy and timeliness of the detection results.
It enables precise identification of surface defects in aluminum alloys, improves detection efficiency and accuracy, ensures product quality and production safety, reduces labor costs, and provides scientific decision support.
Smart Images
Figure CN120997176A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of defect identification technology, and in particular to a system and method for identifying defects on the surface of aluminum alloys. Background Technology
[0002] Aluminum alloys are widely used industrial materials, especially in the automotive, aerospace, and electronics industries. To achieve lightweight design and enhance structural stability and load-bearing capacity, many aluminum alloy products feature specially designed ribbed structures on their surfaces. However, during production, due to limitations in processing technology and equipment, various minor defects may still appear on the surface of aluminum alloy products, such as surface protrusions, particles, rib protrusions, scratches, and pits. These surface defects not only affect the appearance quality of the product but also, due to their contact with other components, lead to accelerated wear, causing economic losses, and in extreme cases, potentially causing major safety accidents. Therefore, accurate identification of surface defects in aluminum alloy products is crucial for ensuring product quality and safe use.
[0003] Currently, the identification of surface defects in aluminum alloys is mainly achieved through manual visual inspection or 2D visual inspection. While manual inspection offers some flexibility and intuitiveness, it has several shortcomings. Low inspection efficiency is a significant problem, as it requires personnel to inspect each product individually, resulting in slow speeds that cannot meet the demands of large-scale production. The stability of inspection results is also poor, easily influenced by factors such as the inspector's experience, fatigue, and attention, leading to missed or false positives. Furthermore, manual inspection requires substantial investment in manpower and management, increasing production costs for enterprises. In addition, 2D visual recognition systems have high requirements for lighting conditions; the intensity, angle, and uniformity of the light all affect the accuracy of the inspection results. The lack of clear surface texture information on aluminum alloys makes it difficult for 2D visual recognition systems to extract defect features, and the inability to eliminate noise interference results in imprecise defect feature extraction. In particular, its ability to detect small defects is insufficient, easily missing some minute defects, thus affecting product quality and safety. Summary of the Invention
[0004] This invention overcomes the shortcomings of the prior art and provides a method for identifying surface defects in aluminum alloys, solving the technical problems of poor performance and low efficiency in existing aluminum alloy surface defect identification methods.
[0005] To solve the above-mentioned technical problems, the present invention is achieved through the following technical solution:
[0006] An aluminum alloy surface defect identification system, comprising:
[0007] The image acquisition module acquires three-dimensional depth images of the aluminum alloy surface using a 3D line laser contour scanner;
[0008] The central processing module includes:
[0009] A multi-threaded task scheduler is used to coordinate the data flow and instruction interaction between various modules within the system;
[0010] A high-speed data cache is used to temporarily store the three-dimensional depth image and intermediate results during system processing;
[0011] The defect identification module is used to extract and identify defects in the planar detection area and the rib detection area of the aluminum alloy surface, respectively.
[0012] The analysis module performs defect analysis based on preset defect judgment thresholds for various types; if a defect exists, the sample is deemed unqualified, and further analysis is conducted to determine whether the defect is feasible for reprocessing.
[0013] The display module is used to visualize system information, including but not limited to: three-dimensional depth images of the aluminum alloy surface, identified defect location markers, and analysis results, to assist operators in making adjustments based on the central control module;
[0014] The alarm module triggers a buzzer alarm when the system determines that a sample is unqualified; the alarm must be cleared by the operator before the system can continue to perform the testing task.
[0015] Furthermore, in the image acquisition module, the deviation between the actual number of rows n of the three-dimensional depth image acquired by the 3D line laser contour scanner and the set target number of rows N is within ±100 rows.
[0016] This invention also claims protection for a method for identifying surface defects in aluminum alloys, applied to the aforementioned aluminum alloy surface defect identification system, comprising the following steps:
[0017] S1: Data Acquisition: Acquire three-dimensional depth images of the aluminum alloy surface using a 3D line laser contour scanner;
[0018] S2: Image preprocessing: Preprocess the acquired 3D depth image to obtain an effective complete surface area of the aluminum alloy;
[0019] S3: Detection Area Extraction: Extract the target area for defect identification;
[0020] S4: Defect Identification and Extraction: Identify and extract defects in the rib area and the planar area respectively.
[0021] Furthermore, in step S1, the deviation between the actual number of rows n of the acquired 3D depth image and the set target number of rows N is within ±100 rows.
[0022] Furthermore, in step S2, the preprocessing includes:
[0023] Threshold segmentation: used to segment the effective area of the aluminum alloy surface from the background;
[0024] Interpolation: Used to repair or fill missing data within a valid surface area.
[0025] Furthermore, in step S3, the extraction of the detection region includes:
[0026] Identify and process the largest planar block ROI;
[0027] Create grayscale plane processing;
[0028] Perform image difference processing;
[0029] Threshold segmentation is applied.
[0030] Furthermore, the grayscale plane creation process refers to: based on the grayscale information after the maximum ROI processing, calculating the centroid coordinates (x_c, y_c) and the slope (α, β) of the grayscale plane for that region, denoted as argmin(Σ[Z(x,y)-(ax+by+c)]). 2 Given the average gray value μ, create a grayscale plane of the same size as the original image area: G(x,y)=α(x-x_c)+β(y-y_c)+μ.
[0031] Furthermore, in step S4, the identification and extraction of defects in the rib area and the planar area specifically includes the following sub-steps:
[0032] a. Mean filtering: Mean filtering is used to calculate the mean of the image. Specifically:
[0033]
[0034] Where O(i,j) is the position of a pixel in the output image;
[0035] k is the radius of the filtering window, and the window size is (2k+1)×(2k+1);
[0036] I(i+m,j+n) is the pixel value in the neighborhood centered at (i,j) in the input image;
[0037] b. Dynamic Threshold Segmentation: A dynamic threshold segmentation method is used to obtain potential defect areas. Specifically, when LightDark = 'light': g o (i,j)≥g t (i,j)+Offset; when LightDark = 'dark': g o (i,j)≤g t(i,j)-Offset;
[0038] Where g o (i,j) represents the pixel value at the original image position, g t (i,j) represents the pixel value at the corresponding position in the processing image, and Offset represents the grayscale value. At this time, the pixels that meet the conditions will be selected as part of the target area. When LightDark = 'light', the defect is a convex shape, and when LightDark = 'dark', the defect is a pit.
[0039] c. Morphological post-processing: The possible defective areas are subjected to erosion operations, and the intersection of the erosion results and the possible defective areas is processed to perform circular and opening operations;
[0040] d. Defect screening: Defects are selected by using area as a threshold;
[0041] e. Output Results: Merge the defect detection results of the rib area and the planar area, label the defect type, and output quantitative parameters, including defect location, size and depth information.
[0042] Compared with the prior art, the beneficial effects of the present invention are:
[0043] 1. Precise Defect Identification: A 3D depth map of the aluminum alloy surface is acquired using a 3D line laser profile scanner. Compared to traditional 2D visual recognition, this provides more comprehensive and accurate surface information, including depth and shape, facilitating more precise identification of various defects. Furthermore, neighborhood-weighted interpolation is used to address image pixel anomalies, fully restoring the surface morphology.
[0044] 2. The multi-threaded task scheduler of the central processing module can effectively coordinate the data flow and instruction interaction of each module, improve the system operating efficiency, and ensure smooth collaboration among all parts. The high-speed data cache can temporarily store the original depth map and intermediate processing results, providing data support for subsequent processing and avoiding data loss or duplicate collection.
[0045] 3. Scientific Analysis and Decision Support: The analysis module analyzes the test results by setting threshold values for various types of defects, and makes scientific judgments on the results. It can not only accurately identify whether defects exist, but also further analyze whether the defects can be reprocessed, providing a strong basis for production decisions and helping to optimize production processes and improve product quality.
[0046] 4. Reliable alarm mechanism: The buzzer alarm module will promptly issue an alarm when the system detects an NG sample, reminding the operator to pay attention and handle the problem. The equipment can only continue testing after the operator clears the alarm, ensuring that defects are detected and handled in a timely manner during the production process, and guaranteeing product quality and production safety. Attached Figure Description
[0047] The accompanying drawings are provided to further illustrate the invention and are used together with the embodiments of the invention to explain the invention. They do not constitute a limitation of the invention. In the drawings:
[0048] Figure 1 This invention provides a structural diagram of an aluminum alloy surface defect identification system.
[0049] Figure 2 The flowchart illustrates a method for identifying surface defects in aluminum alloys according to the present invention. Detailed Implementation
[0050] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.
[0051] like Figure 1 As shown, an aluminum alloy surface defect identification system includes an image acquisition module, a central processing module, a defect identification module, an analysis module, a display module, and an alarm module. The image acquisition module acquires a three-dimensional depth image of the aluminum alloy surface using a 3D line laser profile scanner. In actual operation, the 3D line laser profile scanner emits laser lines onto the aluminum alloy surface and calculates the depth information of each point on the surface based on the laser reflection, thereby generating a three-dimensional depth image. To ensure the quality of the acquired image, the deviation between the actual number of rows n in the acquired three-dimensional depth image and the set target number of rows N needs to be controlled within ±100 rows. For example, if the set target number of rows N is 1000, then the actual number of rows n should be between 900 and 1100.
[0052] The central processing module includes a multi-threaded task scheduler and a high-speed data buffer. The multi-threaded task scheduler is responsible for coordinating data flow and instruction interaction between modules within the system. During system operation, each module generates a large amount of data and instructions. The multi-threaded task scheduler allocates resources appropriately to ensure that data can be accurately and timely transmitted between modules. For example, when the image acquisition module acquires a 3D depth image, the multi-threaded task scheduler quickly transmits the image data to the high-speed data buffer for temporary storage.
[0053] A high-speed data cache is used to temporarily store 3D depth images and intermediate results during system processing. Since data processing takes time, the high-speed data cache prevents data loss and ensures data integrity. For example, during the preprocessing and defect identification of 3D depth images, temporary data generated in between is stored in the high-speed data cache.
[0054] The defect identification module is used to extract and identify defects in the planar inspection area and the rib inspection area of the aluminum alloy surface, respectively. It analyzes the acquired 3D depth images to find potential defects in the planar and rib areas.
[0055] The analysis module performs defect analysis based on preset defect judgment thresholds for various types. If a defect is found, the sample is deemed unqualified, and further analysis is conducted to determine whether the defect is feasible for reprocessing. For example, some small scratches or dents may be reprocessed through grinding or repair; however, some severe protrusions or large-area defects may not be suitable for reprocessing.
[0056] The display module visualizes system information, including a 3D depth image of the aluminum alloy surface, identified defect location markers, and analysis results, assisting operators in making adjustments based on the central control module. Operators can intuitively see the condition of the aluminum alloy surface and the system's analysis results through the display module, enabling them to make appropriate decisions.
[0057] When the system determines a sample to be non-compliant, it triggers an alarm. The system can only continue the testing process after the operator confirms that the alarm has been cleared. This promptly alerts the operator to non-compliant samples, preventing defective products from proceeding to the next stage of the process.
[0058] like Figure 2 As shown, this invention also claims protection for a method for identifying surface defects in aluminum alloys, comprising the following steps:
[0059] S1: Data Acquisition: Acquire three-dimensional depth images of the aluminum alloy surface using a 3D line laser contour scanner;
[0060] S2: Image preprocessing: Preprocess the acquired 3D depth image to obtain an effective complete surface area of the aluminum alloy;
[0061] S3: Detection Area Extraction: Extract the target area for defect identification;
[0062] S4: Defect Identification and Extraction: Identify and extract defects in the rib area and the planar area respectively.
[0063] In S1, data acquisition refers to acquiring a three-dimensional depth image of the aluminum alloy surface using a 3D line laser contour scanner, while ensuring that the deviation between the actual number of rows n of the acquired three-dimensional depth image and the set target number of rows N is within ±100 rows.
[0064] Image preprocessing includes thresholding and interpolation. Thresholding segmentation is used to separate the effective area of the aluminum alloy surface from the background. Acquired images may contain background information, such as the surrounding environment and equipment. By setting an appropriate threshold, the effective area of the aluminum alloy surface can be separated from the background. For example, based on the image's grayscale values, pixels with grayscale values within a certain range are identified as effective areas of the aluminum alloy surface. Interpolation is used to repair or fill in missing data within the effective surface area. During acquisition, some data may be missing due to various reasons, such as poor laser reflection or equipment malfunction. Interpolation can estimate the missing data based on the surrounding pixel values, thus ensuring image integrity.
[0065] The threshold segmentation process described above can be used to obtain a general area of the aluminum alloy surface, specifically:
[0066] Z _valid ={z|z_ min +Δ≤z≤z_ max -Δ} (Δ is the noise margin);
[0067] The binary mask matrix M(x,y) is generated using the interpolation method described above. For invalid points M(x,y) = 0, neighborhood weighted interpolation is used: Z'(x,y) = Σ[W(i,j)·Z(x+i,y+j)] / ΣW(i,j) (the weight W(i,j) decreases exponentially with increasing distance).
[0068] Detection region extraction includes the following steps:
[0069] Identifying the largest planar area of interest (ROI): In an image of an aluminum alloy surface, identify the largest planar area as the region of interest (ROI). This can be achieved by analyzing features such as the image's grayscale values and texture.
[0070] Create a grayscale plane for processing: Based on the grayscale information after processing the maximum ROI, calculate the centroid coordinates (x_c, y_c) and the slope (α, β) of the grayscale plane for that region. 2 Given the average gray value μ, create a grayscale plane of the same size as the original image area: G(x,y)=α(x-x_c)+β(y-y_c)+μ.
[0071] Perform image difference processing: The original image is compared with the created grayscale plane to obtain a difference image between the two. This highlights the features of the aluminum alloy surface, facilitating subsequent defect identification.
[0072] Threshold segmentation is applied: Threshold segmentation is performed on the differentially processed image to further extract regions that may have defects.
[0073] Defect identification and extraction includes the following steps:
[0074] a. Mean filtering: Mean filtering is used to calculate the mean of the image. Specifically:
[0075]
[0076] Where O(i,j) is the position of a pixel in the output image;
[0077] k is the radius of the filtering window, and the window size is (2k+1)×(2k+1);
[0078] I(i+m,j+n) is the pixel value in the neighborhood centered at (i,j) in the input image;
[0079] Mean filtering can remove noise from an image, making it smoother.
[0080] b. Dynamic Threshold Segmentation: A dynamic threshold segmentation method is used to obtain potential defect areas. Specifically, when LightDark = 'light': g o (i,j)≥g t (i,j)+Offset; when LightDark = 'dark': g o (i,j)≤g t (i,j)-Offset;
[0081] Where g o (i,j) represents the pixel value at the original image position, g t (i,j) represents the pixel value at the corresponding position in the processing image, and Offset represents the grayscale value. At this time, the pixels that meet the conditions will be selected as part of the target area. When LightDark = 'light', the defect is a convex shape, and when LightDark = 'dark', the defect is a pit.
[0082] c. Morphological post-processing: The possible defective areas are subjected to erosion operation. After the intersection of the erosion operation results and the possible defective areas is processed, circular operation and opening operation are performed. The erosion operation can remove some small noise and edge burrs, and the circular operation and opening operation can further optimize the shape of the defective areas.
[0083] d. Defect screening: Defects are extracted by using area as a threshold; set an appropriate area threshold, filter out areas with an area smaller than the threshold, and only retain potential defect areas with larger areas.
[0084] e. Output Results: The defect detection results from the ribbed and planar areas are merged, the defect type is labeled, and quantitative parameters, including defect location, size, and depth information, are output. This information helps operators to more accurately understand the defect situation on the aluminum alloy surface so that appropriate measures can be taken.
[0085] Through the above specific implementation methods, accurate identification and analysis of surface defects in aluminum alloys can be achieved, thereby improving product quality and production efficiency.
[0086] Finally, it should be noted that the above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. However, any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.
Claims
1. A surface defect identification system for aluminum alloys, characterized in that, include: The image acquisition module acquires three-dimensional depth images of the aluminum alloy surface using a 3D line laser contour scanner; The central processing module includes: A multi-threaded task scheduler is used to coordinate the data flow and instruction interaction between various modules within the system; A high-speed data cache is used to temporarily store the three-dimensional depth image and intermediate results during system processing; The defect identification module is used to extract and identify defects in the planar detection area and the rib detection area of the aluminum alloy surface, respectively. The analysis module performs defect analysis based on preset defect judgment thresholds for various types; if a defect exists, the sample is deemed unqualified, and further analysis is conducted to determine whether the defect is feasible for reprocessing. The display module is used to visualize system information, including but not limited to: three-dimensional depth images of the aluminum alloy surface, identified defect location markers, and analysis results, to assist operators in making adjustments based on the central control module; The alarm module triggers a buzzer alarm when the system determines that a sample is unqualified; the alarm must be cleared by the operator before the system can continue to perform the testing task.
2. The aluminum alloy surface defect identification system according to claim 1, characterized in that, In the image acquisition module, the actual number of rows n of the three-dimensional depth image acquired by the 3D line laser contour scanner deviates from the set target number of rows N within ±100 rows.
3. A method for identifying surface defects in aluminum alloys, applied to the aluminum alloy surface defect identification system of claim 1, characterized in that, Includes the following steps: S1: Data Acquisition: Acquire three-dimensional depth images of the aluminum alloy surface using a 3D line laser contour scanner; S2: Image preprocessing: Preprocess the acquired 3D depth image to obtain an effective complete surface area of the aluminum alloy; S3: Detection Area Extraction: Extract the target area for defect identification; S4: Defect Identification and Extraction: Identify and extract defects in the rib area and the planar area respectively.
4. The method for identifying surface defects in aluminum alloys according to claim 3, characterized in that, In step S1, the deviation between the actual number of rows n of the acquired 3D depth image and the set target number of rows N is within ±100 rows.
5. The method for identifying surface defects in aluminum alloys according to claim 3, characterized in that, In step S2, the preprocessing includes: Threshold segmentation: used to segment the effective area of the aluminum alloy surface from the background; Interpolation: Used to repair or fill missing data within a valid surface area.
6. The method for identifying surface defects in aluminum alloys according to claim 3, characterized in that, In step S3, the extraction of the detection region includes: Identify and process the largest planar block ROI; Create grayscale plane processing; Perform image difference processing; Threshold segmentation is applied.
7. The method for identifying surface defects in aluminum alloys according to claim 6, characterized in that, The grayscale plane creation process refers to: based on the grayscale information after the maximum ROI processing, calculating the centroid coordinates (x_c, y_c) and the slope (α, β) of the grayscale plane for that region, which is denoted as argmin(Σ[Z(x,y)-(ax+by+c)]). 2 Given the average gray value μ, create a grayscale plane of the same size as the original image area: G(x,y)=α(x-x_c)+β(y-y_c)+μ.
8. The method for identifying surface defects in aluminum alloys according to claim 6, characterized in that, In step S4, the identification and extraction of defects in the rib area and the planar area specifically includes the following sub-steps: a. Mean filtering: Mean filtering is used to calculate the mean of the image. Specifically: Where O(i,j) is the position of a pixel in the output image; k is the radius of the filtering window, and the window size is (2k+1)×(2k+1); I(i+m,j+n) is the pixel value in the neighborhood centered at (i,j) in the input image; b. Dynamic Threshold Segmentation: A dynamic threshold segmentation method is used to obtain potential defect areas. Specifically, when LightDark = 'light': g o (i,j)≥g t (i,j)+Offset; when LightDark = 'dark': g o (i,j)≤g t (i,j)-Offset; Where g o (i,j) represents the pixel value at the original image position, g t (i,j) represents the pixel value at the corresponding position in the processed image, and Offset represents the grayscale value; At this point, pixels that meet the conditions will be selected as part of the target area. When LightDark = 'light', the defect is a bump, and when LightDark = 'dark', the defect is a pit. c. Morphological post-processing: The possible defective areas are subjected to erosion operations, and the intersection of the erosion results and the possible defective areas is processed to perform circular and opening operations; d. Defect screening: Defects are selected by using area as a threshold; e. Output Results: Merge the defect detection results of the rib area and the planar area, label the defect type, and output quantitative parameters, including defect location, size and depth information.