A wheel hub quality detection method and system

Abstract

The application relates to the fields of image processing and industrial detection, and discloses a wheel hub quality detection method and system, wherein the method comprises the following steps: collecting a wheel hub coating image through an industrial camera; introducing a U-Net network combined with an attention mechanism to perform bubble area segmentation on the image; extracting a coating peeling area by using a ResNet network combined with an edge enhancement method; and outputting a wheel hub coating quality evaluation report. Compared with the prior art which mainly relies on manual visual inspection or traditional image processing methods to detect defects of the uniformity of a wheel hub surface, especially under uneven illumination or complex texture conditions, it is difficult to realize high-precision automatic detection of the uniformity of the wheel hub coating. Since the application adopts a multi-source defect fusion mode, a fusion mechanism for bubbles, peeling and thickness error points is constructed, thereby avoiding the problems of boundary misjudgment, misjudgment and inability to quantitatively output in the traditional method, and the accuracy and automation degree of the wheel hub coating quality detection are improved.

Description

Technical Field

[0001] This invention belongs to the field of image processing and industrial inspection technology, and in particular relates to a method and system for inspecting wheel hub quality. Background Technology

[0002] Currently, the quality inspection of wheel hub coatings mainly relies on manual visual inspection and traditional image processing methods, such as edge detection and threshold segmentation techniques. However, traditional methods have many shortcomings. For example, visual inspection depends on manual operation, which is not only subject to human error, but also slow and unsuitable for large-scale production. Traditional image processing methods often rely on fixed thresholds or edge detection algorithms, which have high rates of false positives and false negatives when dealing with complex coating backgrounds, irregular coating peeling, micro-bubbles, and uneven thickness. Existing technologies cannot fully meet the needs of high-precision, high-efficiency, and automated coating defect detection, especially in cases with complex coating textures, uneven lighting, or micro-defects (such as bubbles, peeling areas, and thickness error points). Traditional image processing methods still struggle to achieve accurate coating uniformity analysis and defect detection. Therefore, there is an urgent need for a high-precision, automated, and comprehensive wheel hub coating quality inspection method that can still achieve high precision, automation, and comprehensiveness under conditions of complex coating textures, lighting variations, and the presence of micro-defects, in order to improve the efficiency, accuracy, and automation of quality control in industrial production. Summary of the Invention

[0003] To address the aforementioned technical shortcomings, the purpose of this invention is to propose a wheel hub quality inspection method. This method aims to solve the technical problem that existing technologies often rely on manual visual inspection or traditional image processing methods to detect defects in the uniformity of wheel hub surfaces, especially under conditions of uneven lighting or complex textures, making it difficult to achieve high-precision automatic detection of wheel hub coating uniformity.

[0004] To solve the above-mentioned technical problems, the present invention adopts the following technical solution: The present invention provides a method for detecting wheel hub quality.

[0005] The wheel hub quality testing method includes:

[0006] Step S10: Acquire a first wheel hub coating image using an industrial camera, and perform noise reduction and illumination compensation processing on the first wheel hub coating image to obtain a second wheel hub coating image;

[0007] Step S20: Introduce a U-Net network combined with an attention mechanism to segment the second wheel hub coating image to obtain a first wheel hub segmentation image; extract the bubble region image based on the first wheel hub segmentation image using an adaptive thresholding method;

[0008] Step S30: Introduce a deep residual network combined with an edge enhancement method to segment the second wheel hub coating image to obtain a second wheel hub segmentation image; extract the peeling area image based on the second wheel hub segmentation image and a gradient-based edge detection method.

[0009] Step S40: Obtain coating surface height data by scanning the wheel hub with a laser, and obtain coating surface texture data from the second wheel hub segmentation image; combine the coating surface height data and coating surface texture data, and use a depth regression network to predict coating thickness error points to obtain a coating thickness error image;

[0010] Step S50: Perform a comprehensive analysis of the uniformity of the wheel hub coating based on the bubble area image, peeling area image, and coating thickness error image, and generate and output a wheel hub coating quality report.

[0011] Preferably, in step S10, the step of performing denoising and illumination compensation processing on the first wheel hub coating image to obtain the second wheel hub coating image specifically includes: performing denoising processing on the first wheel hub coating image using bilateral filtering; and performing illumination compensation processing on the first wheel hub coating image using adaptive histogram equalization.

[0012] Preferably, in step S20, the step of introducing a U-Net network combined with an attention mechanism to segment the second hub coating image to obtain a first hub segmentation image; and the step of extracting the bubble region image using an adaptive thresholding method based on the first hub segmentation image specifically includes: introducing a U-Net network and adding an attention mechanism to the U-Net network; training the U-Net network with the attention mechanism using a cross-entropy loss function to obtain a pre-trained U-Net network; inputting the second hub coating image into the pre-trained U-Net network to extract the first hub segmentation image including bubble position features, bubble shape features, and bubble size features; calculating an adaptive threshold within a local window based on the first hub segmentation image using an adaptive thresholding method; and extracting the bubble region image based on the adaptive threshold within the local window.

[0013] Preferably, in step S20, the formula used for the cross-entropy loss function is: Among them, L seg Let y be the cross-entropy loss function, and y be the true label of the bubble region. The bubble region labels predicted by the U-Net network; the adaptive thresholding method uses the following formula: Where T(x,y) is the adaptive threshold within the local window containing pixel (x,y), used to determine whether the current pixel belongs to the bubble region; N is the total number of pixels in the local window; i is the horizontal offset relative to the center pixel; j is the vertical offset relative to the center pixel; and k is the radius of the local window.

[0014] Preferably, in step S30, a deep residual network combined with an edge enhancement method is introduced to segment the second wheel hub coating image to obtain a second wheel hub segmentation image; the step of extracting the peeling region image based on the second wheel hub segmentation image and a gradient-based edge detection method specifically includes:

[0015] Step S301: Use a ResNet network to encode the features of the second wheel hub coating image to obtain multi-scale deep texture feature maps;

[0016] Step S302: Enhance the multi-scale deep texture feature map using the first-order and second-order gradient information of the image to obtain the edge enhancement feature map of the peeling area;

[0017] Step S303: Input the edge enhancement feature map of the peeled area into the lightweight segmentation network head FCN, and output the second wheel hub segmentation image;

[0018] Step S304: Extract the peeling area image based on the second wheel hub segmentation image and the gradient-based edge detection method.

[0019] Preferably, step S40, which integrates coating surface height data and coating surface texture data, and uses a depth regression network to predict coating thickness error points to obtain a coating thickness error image, specifically includes:

[0020] By combining coating surface height data and coating surface texture data, a high-dimensional fused input data pair is constructed.

[0021] The high-dimensional fused input data is fed into a deep regression network to obtain the mapping relationship features between texture changes and height errors. Based on the mapping relationship features, the thickness error of each pixel in the second wheel hub coating image is predicted and calculated.

[0022] A coating thickness error image is obtained by spatially mapping the thickness error of each pixel.

[0023] Preferably, step S50, which involves performing a comprehensive analysis of the wheel hub coating uniformity based on the bubble area image, the peeling area image, and the coating thickness error image, and generating and outputting a wheel hub coating quality report, specifically includes:

[0024] Step S501: Align the bubble region image output in step S20, the peeling region image output in step S30, and the thickness error image output in step S40, and generate a quality analysis matrix by superimposing logical masks.

[0025] Step S502: Calculate the wheel hub coating evaluation index based on the quality analysis matrix. The wheel hub coating evaluation index includes the total defect area ratio, the maximum thickness error point, and the local anomaly density.

[0026] Step S503: Match the quality level according to the wheel hub coating evaluation index and generate a quality rating result;

[0027] Step S504: Output the final wheel coating quality report based on the quality rating results. The wheel coating quality report includes the total defect area percentage, the maximum thickness error point, the local abnormal density, the quality rating results, the inspection time, and the wheel number.

[0028] The present invention also provides a wheel hub quality inspection system comprising:

[0029] The image acquisition and processing module is used to acquire a first wheel hub coating image through an industrial camera, and to perform noise reduction and illumination compensation processing on the first wheel hub coating image to obtain a second wheel hub coating image.

[0030] The bubble region image extraction module is used to introduce a U-Net network combined with an attention mechanism to segment the second wheel hub coating image to obtain a first wheel hub segmentation image; and to extract the bubble region image based on the first wheel hub segmentation image using an adaptive thresholding method.

[0031] The peeling region image extraction module is used to introduce a deep residual network combined with an edge enhancement method to segment the second wheel hub coating image, thereby obtaining a second wheel hub segmentation image; and to extract the peeling region image based on the second wheel hub segmentation image and a gradient-based edge detection method.

[0032] The coating thickness error image analysis module is used to obtain coating surface height data by scanning the wheel hub with a laser, and to obtain coating surface texture data from the second wheel hub segmentation image; by combining the coating surface height data and coating surface texture data, and using a depth regression network to predict coating thickness error points, a coating thickness error image is obtained.

[0033] The wheel hub coating quality report generation module is used to perform a comprehensive analysis of the uniformity of the wheel hub coating based on images of bubble areas, peeling areas, and coating thickness errors, and to generate and output a wheel hub coating quality report.

[0034] The present invention also provides a computer program product, including a wheel hub quality inspection program, which, when executed by a processor, implements the wheel hub quality inspection method.

[0035] The beneficial effects of this invention are as follows: This invention employs deep learning technology based on image recognition, combining U-Net and ResNet networks, to accurately segment regions of bubbles, peeling, and coating thickness errors. This avoids the common misjudgments and missed detections in traditional methods when detecting complex backgrounds, uneven lighting, or minute defects. Through automated image recognition, manual intervention is reduced, improving the accuracy and efficiency of coating defect detection and meeting the high-precision requirements for coating quality inspection in modern industry.

[0036] This invention, by fusing image recognition technology with laser scanning height data, can not only analyze minute defects on the coating surface but also accurately identify problems such as uneven thickness and peeling, thereby achieving multi-dimensional and comprehensive coating quality inspection. This inspection method based on the fusion of image and physical data overcomes the limitations of traditional single image processing methods, improves the robustness and adaptability of the inspection, and is suitable for coating quality control under various coating complexities and different production environments. Attached Figure Description

[0037] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0038] Figure 1 This is a flowchart illustrating the first embodiment of a wheel hub quality inspection method according to the present invention.

[0039] Figure 2 This is a schematic diagram of the thermal representation of the coating thickness error image provided by the wheel hub quality inspection method of the present invention.

[0040] Figure 3 This is a schematic diagram of the equipment for a wheel hub quality inspection method according to the present invention. Detailed Implementation

[0041] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0042] Example 1: As Figure 1 The diagram shown is a flowchart of the first embodiment of the wheel hub quality inspection method of the present invention, which presents the first embodiment of the wheel hub quality inspection method of the present invention.

[0043] In the first embodiment, the wheel hub quality inspection method includes:

[0044] Step S10: Acquire a first wheel hub coating image using an industrial camera, and perform noise reduction and illumination compensation processing on the first wheel hub coating image to obtain a second wheel hub coating image;

[0045] It should be noted that in step S10, the step of performing denoising and illumination compensation processing on the first wheel hub coating image to obtain the second wheel hub coating image specifically includes: performing denoising processing on the first wheel hub coating image using bilateral filtering; and performing illumination compensation processing on the first wheel hub coating image using adaptive histogram equalization.

[0046] Understandably, in this processing step, bilateral filtering is used to denoise the coating image. Bilateral filtering, by considering spatial distance and pixel similarity within the image, effectively removes noise while preserving image edges and details, making it particularly suitable for processing subtle noise and textures in wheel hub coating images. To address image brightness differences caused by uneven illumination, an adaptive histogram equalization method is employed. This method adaptively enhances the brightness of local areas, ensuring the coating image still displays sufficient detail even under uneven illumination, especially small defects on the wheel hub surface (such as bubbles and peeling).

[0047] It should be understood that, compared to traditional global histogram equalization which may lead to over-enhancement of the image or even the appearance of overly bright or dark areas, bilateral filtering can automatically adjust the contrast of the coating image under different lighting conditions. In particular, when there are local reflections or lighting differences on the coating surface, it can improve the visibility of small defects in the image.

[0048] For example, in the experiment, when the illumination of the first-round coating image was uneven, bilateral filtering significantly reduced salt-and-pepper noise (such as black or white spots in the image) while preserving the details of the coating edges. After illumination compensation, the coating details in low-light areas were significantly enhanced, and the contrast of the bubble area increased from 30% to 80%, resulting in a significant improvement in subsequent bubble extraction and a 15% reduction in the false detection rate.

[0049] Step S20: Introduce a U-Net network combined with an attention mechanism to segment the second wheel hub coating image to obtain a first wheel hub segmentation image; extract the bubble region image based on the first wheel hub segmentation image using an adaptive thresholding method;

[0050] It should be noted that in step S20, the formula used for the cross-entropy loss function is: Among them, L seg Let y be the cross-entropy loss function, and y be the true label of the bubble region. The bubble region labels predicted by the U-Net network; the adaptive thresholding method uses the following formula: Where T(x,y) is the adaptive threshold within the local window containing pixel (x,y), used to determine whether the current pixel belongs to the bubble region; N is the total number of pixels in the local window; i is the horizontal offset relative to the center pixel; j is the vertical offset relative to the center pixel; and k is the radius of the local window.

[0051] Understandably, the U-Net network, with its encoder-decoder structure and attention mechanism, can more effectively focus on regions with localized anomalous texture changes in an image, thereby enhancing the response to bubble regions, suppressing interference from non-target regions, and improving image recognition accuracy. The use of cross-entropy loss during training encourages the model to approximate the true bubble boundary labels as closely as possible, ensuring the model learns accurate segmentation strategies. Compared to a fixed global threshold, the adaptive thresholding method dynamically calculates the threshold based on the average brightness of the local area surrounding each pixel, thus stably extracting bubble regions even when faced with uneven coating illumination or background texture variations.

[0052] For example, in a batch of industrial wheel hub image tests, the traditional fixed threshold method showed significant missed detections in dimly lit areas, with a false detection rate as high as 22%. However, the method of this invention first performs preliminary segmentation using a U-Net network and utilizes an attention mechanism to enhance the response to small bubble regions, accurately identifying bubbles with a diameter of less than 2mm in the test set. Subsequently, an adaptive thresholding method is used to further refine the boundary extraction, ultimately reducing the false detection rate to 5.7% and the missed detection rate to 3.2%. Furthermore, through optimization of the cross-entropy loss function, the convergence speed of the training process is improved by 28%, achieving a segmentation accuracy of 85.6% IoU (Intersection over Union) in the test set.

[0053] Step S30: Introduce a deep residual network combined with an edge enhancement method to segment the second wheel hub coating image to obtain a second wheel hub segmentation image; extract the peeling area image based on the second wheel hub segmentation image and a gradient-based edge detection method.

[0054] It should be noted that in step S30, the step of introducing a deep residual network combined with an edge enhancement method to segment the second wheel hub coating image to obtain a second wheel hub segmented image; and the step of extracting the peeling region image based on the second wheel hub segmented image using a gradient-based edge detection method, specifically includes:

[0055] Step S301: Use a ResNet network to encode the features of the second wheel hub coating image to obtain multi-scale deep texture feature maps;

[0056] Step S302: Enhance the multi-scale deep texture feature map using the first-order and second-order gradient information of the image to obtain the edge enhancement feature map of the peeling area;

[0057] Step S303: Input the edge enhancement feature map of the peeled area into the lightweight segmentation network head FCN, and output the second wheel hub segmentation image;

[0058] Step S304: Extract the peeling area image based on the second wheel hub segmentation image and the gradient-based edge detection method.

[0059] It should be understood that when extracting image features, the ResNet network, through its residual connection structure, can effectively alleviate the gradient vanishing problem of deep networks, thereby extracting richer and more discernible coating texture features. Especially when the edges of the peeled area are blurred and the texture is discontinuous, ResNet can preserve the high-order information of the coating structure, providing a more discriminative deep representation for subsequent edge enhancement and segmentation. Simultaneously, introducing first-order gradients (such as Sobel edges) and second-order gradients (such as Laplacian gradients) to significantly enhance deep features can further improve the model's sensitivity to peeled boundaries and enhance boundary localization accuracy.

[0060] For example, in a certain test sample, the boundaries of the peeling area in the original wheel hub image were blurred, and the transition to the normal coating texture was slow. When using the traditional Sobel operator for edge detection, only a portion of the peeling area was extracted, resulting in a false negative rate as high as 35%. Using the method of this invention, after extracting deep texture features through a ResNet network, the data is then enhanced by gradients and input into an FCN decoder for segmentation. This allows for clear extraction of the peeling area. Finally, after combining gradient edge correction, the average edge quality of the peeling area is improved by 21.4%, and the false negative rate is reduced to 6.3%. Furthermore, the edge enhancement module also demonstrates stronger robustness in peeling extraction in highly reflective areas, reducing the false positive rate in bright areas by more than 18%, significantly improving the overall accuracy and stability of defect identification.

[0061] Step S40: Obtain coating surface height data by scanning the wheel hub with a laser, and obtain coating surface texture data from the second wheel hub segmentation image; combine the coating surface height data and coating surface texture data, and use a depth regression network to predict coating thickness error points to obtain a coating thickness error image;

[0062] It should be noted that step S40, which integrates coating surface height data and coating surface texture data, and uses a deep regression network to predict coating thickness error points to obtain a coating thickness error image, specifically includes: integrating coating surface height data and coating surface texture data to construct a high-dimensional fused input data pair; inputting the high-dimensional fused input data pair into a deep regression network to obtain the mapping relationship features between texture changes and height errors, and predicting and calculating the thickness error of each pixel in the second wheel hub coating image based on the mapping relationship features; and spatially mapping the thickness error of each pixel to obtain a coating thickness error image.

[0063] Understandably, minute fluctuations in coating thickness are often difficult to accurately identify using only texture images. However, relying solely on laser height data lacks a contextual understanding of the coating texture's changing trends. Therefore, jointly modeling the texture information of the image with the actual height data helps to construct a nonlinear mapping relationship between texture changes and thickness fluctuations, enabling more accurate prediction of thickness error points. Furthermore, employing deep regression networks allows for the learning of complex mappings between high-dimensional inputs and regression labels, outperforming traditional linear fitting or local filtering methods, especially when dealing with thickness data exhibiting continuity, trends, and contextual dependencies.

[0064] It should be understood that the thickness error prediction in this invention does not simply make a direct deviation judgment on the height, but rather, based on the synergistic representation characteristics of texture features and height information, it learns the nonlinear relationship between the combined features and thickness error through a deep network, making it applicable to various complex structures and irregularly varied coating surfaces. Furthermore, as... Figure 2 As shown, a thermal map representation of a coating thickness error image is presented. By outputting the error value at the pixel level and spatially mapping it into an image, a thickness error image can be directly generated, achieving high-resolution visualization of thickness anomalies, which is suitable for subsequent automatic diagnosis and zoning evaluation.

[0065] For example, on a typical wheel hub sample, the standard thickness range measured by laser scanning is 0.18–0.22 mm. Traditional visual estimation methods can only identify error areas larger than ±0.05 mm. However, the method of this invention, by fusing image and height data to train a regression network, achieved an average error of 0.008 mm on the test set, improving accuracy by over 60%. In one set of data, the predicted thickness error image clearly identified three local thickness anomalies with diameters less than 3 mm, none of which were captured by traditional methods. The location of the maximum value in the error map deviated from the actual measured error point by less than 1.2 pixels, verifying the advantages of this method in spatial positioning accuracy and anomaly detection capability.

[0066] Step S50: Perform a comprehensive analysis of the uniformity of the wheel hub coating based on the bubble area image, peeling area image, and coating thickness error image, and generate and output a wheel hub coating quality report.

[0067] It should be noted that step S50, which involves performing a comprehensive analysis of the wheel hub coating uniformity based on the bubble area image, peeling area image, and coating thickness error image, and generating and outputting a wheel hub coating quality report, specifically includes:

[0068] Step S501: Align the bubble region image output in step S20, the peeling region image output in step S30, and the thickness error image output in step S40, and generate a quality analysis matrix by superimposing logical masks.

[0069] Step S502: Calculate the wheel hub coating evaluation index based on the quality analysis matrix. The wheel hub coating evaluation index includes the total defect area ratio, the maximum thickness error point, and the local anomaly density.

[0070] Step S503: Match the quality level according to the wheel hub coating evaluation index and generate a quality rating result;

[0071] Step S504: Output the final wheel coating quality report based on the quality rating results. The wheel coating quality report includes the total defect area percentage, the maximum thickness error point, the local abnormal density, the quality rating results, the inspection time, and the wheel number.

[0072] Understandably, by fusing and analyzing images of three types of defects (bubbles, peeling, and thickness error), the quality analysis matrix constructed in this invention achieves a unified spatial representation of coating uniformity defects. Compared with traditional itemized detection methods, this matrix-based analysis approach enables pixel-level comparison and comprehensive evaluation, providing a high-resolution, visualized data foundation for subsequent index calculation and grade matching. The total defect area ratio in the quality assessment index quantifies the overall defect level of the coating, the maximum thickness error point reflects local structural safety hazards, and the local anomaly density reflects the trend of defect concentration within a small area. The combination of these multi-dimensional indicators can comprehensively characterize the quality status of coating uniformity.

[0073] It should be understood that the evaluation logic adopted in this invention not only possesses quantitative characteristics but also interpretability and scalability. Various evaluation indicators are independent yet complementary, forming a multi-dimensional comprehensive analysis model. Furthermore, the quality grade matching component can flexibly set standards (such as national standards, enterprise standards, or industry grading) according to user needs and supports adaptive judgment through rule trees or lightweight machine learning models (such as KNN and XGBoost), exhibiting good engineering deployability. Simultaneously, the output coating quality report is structured and standardized, capable of visually displaying defect areas in image format and also connecting to an enterprise quality management system (QMS) through data format, achieving closed-loop quality control from inspection to rating to traceability.

[0074] For example, in actual testing, after analyzing a batch of wheel hubs for bubbles, peeling, and thickness error, the quality analysis matrix identified an effective defect area covering approximately 3.8% of the total area. Of this, bubble areas accounted for 1.2%, peeling areas for 0.8%, and areas with thickness errors greater than ±0.02mm for 1.8%. Based on the established rules, this wheel hub was rated as "B (Good)". An automatic wheel hub coating quality report was generated, including: total defect area percentage: 3.8%, maximum thickness error: 0.037mm, local abnormal density: 2.1%, quality rating: B, testing time: 2025-04-10 10:32:48, wheel hub number: WZ20250410-015. This report was exported in both PDF and JSON formats and automatically pushed to the downstream production management system, achieving traceability and visualization of the quality inspection results.

[0075] Example 2: Furthermore, the present invention provides a wheel hub quality inspection system that employs a wheel hub quality inspection method described in the above embodiments, thereby solving a technical problem related to wheel hub quality inspection. Compared with the prior art, the beneficial effects of the wheel hub quality inspection system provided by the present invention are the same as those of the wheel hub quality inspection method described in the above embodiments, and other technical features of the wheel hub quality inspection system are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0076] Example 3: This invention provides a wheel hub quality testing device, please refer to... Figure 3A wheel hub quality inspection device includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, which, when executed by the at least one processor, enable the at least one processor to perform a wheel hub quality inspection method as described in Embodiment 1 above. The wheel hub quality inspection device in this embodiment may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital radio receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Description), PMPs (Portable Media Players), in-vehicle terminals (e.g., in-vehicle navigation terminals), and fixed terminals such as digital TVs and desktop computers. This wheel hub quality inspection device is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this invention. A wheel hub quality inspection device may include a processing device 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 1002 or a program loaded from a storage device 1003 into a random access memory (RAM) 1004. RAM 1004 also stores various programs and data required for the operation of a wheel hub quality inspection device. Processing unit 1001, ROM 1002, and RAM 1004 are interconnected via bus 1005. Input / output (I / O) interface 1006 is also connected to the bus. Typically, the following systems can be connected to I / O interface 1006: input devices 1007 including, for example, touchscreens, touchpads, keyboards, mice, image sensors, microphones, accelerometers, gyroscopes, etc.; output devices 1008 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 1003 including, for example, magnetic tapes, hard disks, etc.; and communication devices 1009. Communication device 1009 allows a wheel hub quality inspection device to communicate wirelessly or wiredly with other devices to exchange data. Although a wheel hub quality inspection device with various systems is shown in the figure, it should be understood that it is not required to implement or possess all the systems shown. More or fewer systems can be implemented alternatively.

[0077] Example 4: This invention also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the wheel hub quality inspection method described above. The computer program product provided by this invention can solve a technical problem related to wheel hub quality inspection. Compared with the prior art, the beneficial effects of the computer program product provided by this invention are the same as those of the wheel hub quality inspection method provided in the above embodiments, and will not be repeated here.

[0078] In particular, according to the embodiments disclosed in this invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this invention include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from storage device 1003, or installed from ROM 1002. When the computer program is executed by processing device 1001, it performs the functions defined in the methods of the embodiments disclosed in this invention.

[0079] It should be understood that the various parts disclosed in this invention can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics may be combined in any suitable manner in one or more embodiments or examples.

[0080] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for inspecting wheel hub quality, characterized in that, The methods include: Step S10: Acquire a first wheel hub coating image using an industrial camera, and perform noise reduction and illumination compensation processing on the first wheel hub coating image to obtain a second wheel hub coating image; Step S20: Introduce a U-Net network combined with an attention mechanism to segment the second wheel hub coating image to obtain a first wheel hub segmentation image; extract the bubble region image based on the first wheel hub segmentation image using an adaptive thresholding method; Step S30: Introduce a deep residual network combined with an edge enhancement method to segment the second wheel hub coating image to obtain a second wheel hub segmentation image; extract the peeling area image based on the second wheel hub segmentation image and a gradient-based edge detection method. Step S40: Obtain coating surface height data by scanning the wheel hub with a laser, and obtain coating surface texture data from the second wheel hub segmentation image; combine the coating surface height data and coating surface texture data, and use a depth regression network to predict coating thickness error points to obtain a coating thickness error image; Step S50: Perform a comprehensive analysis of the uniformity of the wheel hub coating based on the bubble area image, peeling area image, and coating thickness error image, and generate and output a wheel hub coating quality report.

2. The wheel hub quality inspection method as described in claim 1, characterized in that, Step S10, which involves denoising and illumination compensation of the first wheel hub coating image to obtain the second wheel hub coating image, specifically includes: denoising the first wheel hub coating image using bilateral filtering; and illumination compensation of the first wheel hub coating image using adaptive histogram equalization.

3. The wheel hub quality inspection method as described in claim 1, characterized in that, In step S20, the U-Net network combined with an attention mechanism is introduced to segment the second hub coating image to obtain a first hub segmentation image. The step of extracting the bubble region image using an adaptive thresholding method based on the first hub segmentation image specifically includes: introducing a U-Net network and adding an attention mechanism to the U-Net network; training the U-Net network with the attention mechanism using a cross-entropy loss function to obtain a pre-trained U-Net network; inputting the second hub coating image into the pre-trained U-Net network to extract the first hub segmentation image including bubble position features, bubble shape features, and bubble size features; calculating the adaptive threshold within a local window based on the first hub segmentation image using an adaptive thresholding method; and extracting the bubble region image based on the adaptive threshold within the local window.

4. The wheel hub quality inspection method as described in claim 3, characterized in that, In step S20, the formula used for the cross-entropy loss function is: Among them, L seg Let y be the cross-entropy loss function, and y be the true label of the bubble region. The bubble region labels predicted by the U-Net network; the adaptive thresholding method uses the following formula: Where T(x,y) is the adaptive threshold within the local window containing pixel (x,y), used to determine whether the current pixel belongs to the bubble region; N is the total number of pixels in the local window; i is the horizontal offset relative to the center pixel; j is the vertical offset relative to the center pixel; and k is the radius of the local window.

5. The wheel hub quality inspection method as described in claim 1, characterized in that, In step S30, a deep residual network is introduced in combination with an edge enhancement method to segment the second wheel hub coating image, thereby obtaining a segmented image of the second wheel hub. The steps for extracting the peeling region image based on the second wheel hub segmentation image and gradient-based edge detection include: Step S301: Use a ResNet network to encode the features of the second wheel hub coating image to obtain multi-scale deep texture feature maps; Step S302: Enhance the multi-scale deep texture feature map using the first-order and second-order gradient information of the image to obtain the edge enhancement feature map of the peeling area; Step S303: Input the edge enhancement feature map of the peeled area into the lightweight segmentation network head FCN, and output the second wheel hub segmentation image; Step S304: Extract the peeling area image based on the second wheel hub segmentation image and the gradient-based edge detection method.

6. The wheel hub quality inspection method as described in claim 1, characterized in that, Step S40, which integrates coating surface height data and coating surface texture data, and uses a depth regression network to predict coating thickness error points to obtain a coating thickness error image, specifically includes: By combining coating surface height data and coating surface texture data, a high-dimensional fused input data pair is constructed. The high-dimensional fused input data is fed into a deep regression network to obtain the mapping relationship features between texture changes and height errors. Based on the mapping relationship features, the thickness error of each pixel in the second wheel hub coating image is predicted and calculated. A coating thickness error image is obtained by spatially mapping the thickness error of each pixel.

7. The wheel hub quality inspection method as described in claim 1, characterized in that, Step S50, which involves performing a comprehensive analysis of the wheel hub coating uniformity based on the bubble area image, peeling area image, and coating thickness error image, and generating and outputting a wheel hub coating quality report, specifically includes: Step S501: Align the bubble region image output in step S20, the peeling region image output in step S30, and the thickness error image output in step S40, and generate a quality analysis matrix by superimposing logical masks. Step S502: Calculate the wheel hub coating evaluation index based on the quality analysis matrix. The wheel hub coating evaluation index includes the total defect area ratio, the maximum thickness error point, and the local anomaly density. Step S503: Match the quality level according to the wheel hub coating evaluation index and generate a quality rating result; Step S504: Output the final wheel coating quality report based on the quality rating results. The wheel coating quality report includes the total defect area percentage, the maximum thickness error point, the local abnormal density, the quality rating results, the inspection time, and the wheel number.

8. A wheel hub quality inspection system, applied to the wheel hub quality inspection method according to any one of claims 1-7, characterized in that, The wheel hub quality inspection system includes: The image acquisition and processing module is used to acquire a first wheel hub coating image through an industrial camera, and to perform noise reduction and illumination compensation processing on the first wheel hub coating image to obtain a second wheel hub coating image. The bubble region image extraction module is used to introduce a U-Net network combined with an attention mechanism to segment the second wheel hub coating image to obtain a first wheel hub segmentation image; and to extract the bubble region image based on the first wheel hub segmentation image using an adaptive thresholding method. The peeling region image extraction module is used to introduce a deep residual network combined with an edge enhancement method to segment the second wheel hub coating image, thereby obtaining a second wheel hub segmentation image; and to extract the peeling region image based on the second wheel hub segmentation image and a gradient-based edge detection method. The coating thickness error image analysis module is used to obtain coating surface height data by scanning the wheel hub with a laser, and to obtain coating surface texture data from the second wheel hub segmentation image; by combining the coating surface height data and coating surface texture data, and using a depth regression network to predict coating thickness error points, a coating thickness error image is obtained. The wheel hub coating quality report generation module is used to perform a comprehensive analysis of the uniformity of the wheel hub coating based on images of bubble areas, peeling areas, and coating thickness errors, and to generate and output a wheel hub coating quality report.

9. A wheel hub quality inspection device, characterized in that, The wheel hub quality inspection device includes: a memory, a processor, and a wheel hub quality inspection program stored in the memory and executable on the processor. When the wheel hub quality inspection program is executed by the processor, it implements a wheel hub quality inspection method according to any one of claims 1 to 7.

10. A computer program product, characterized in that, The computer program product includes a wheel hub quality inspection program, which, when executed by a processor, implements a wheel hub quality inspection method according to any one of claims 1 to 7.

Images