Automatic detection system and method for shield machine tool wear

Abstract

The present application relates to the technical field of shield machine tool detection, and discloses a shield machine tool wear automatic detection system and method, which obtains a shield machine tool image collected by a camera component, and performs image preprocessing, noise suppression, tool region segmentation and region of interest extraction to obtain a tool target ROI image, wherein the tool region segmentation adopts image segmentation based on grayscale threshold, and the grayscale threshold is determined by the tool geometric features and tool texture features of each sliding window area. This method can improve the accuracy of image segmentation and the accuracy of tool geometric features and tool texture features. Finally, the tool wear feature extraction is performed on the tool target ROI image and input into the tool wear intelligent detector based on the support vector regression model to obtain the detection result. In this way, the automatic evaluation of the wear degree is realized, human intervention is reduced, and the detection efficiency and accuracy are improved.

Description

Technical Field

[0001] The present application relates to the technical field of shield machine tool detection technology, and more specifically, to a shield machine tool wear automatic detection system and method. Background Art

[0002] Shield machines are critical equipment in modern tunnel excavation projects, and the wear of their cutters directly impacts construction efficiency and safety. Traditional methods for detecting cutter wear in shield machines rely primarily on manual inspection or periodic replacement. This method is not only labor-intensive and resource-intensive, but also makes it difficult to accurately assess the actual degree of cutter wear, leading to excessive maintenance and delayed repairs. Furthermore, manual inspections in harsh working environments can pose a threat to worker safety.

[0003] With technological advancements, some sensor-based automated detection solutions have begun to be applied to shield machine tool wear monitoring. However, these solutions often require the installation of additional sensors on the tool, which not only increases system complexity and cost but can also affect monitoring effectiveness due to sensor durability issues. Furthermore, these sensors often only provide indirect wear information, such as temperature changes or vibration frequency, and cannot directly reflect the actual wear condition of the tool surface.

[0004] Advances in image processing technology offer new solutions to these challenges. By capturing tool images with a camera and analyzing tool wear using image processing algorithms, non-contact, real-time monitoring can be achieved. However, existing image processing technology still faces numerous challenges in practical application. For example, factors such as dust and varying lighting in the shield machine's operating environment can significantly impact image quality, reducing the accuracy of image analysis.

[0005] Therefore, an optimized automatic detection scheme for shield machine tool wear is expected. Summary of the Invention

[0006] To address the above-mentioned technical problems, the present application is proposed. The embodiments of the present application provide a shield machine tool wear automatic detection system and method, which achieves automated assessment of wear levels, reduces human intervention, improves detection efficiency and accuracy, and can also promptly detect tool wear conditions, avoiding safety hazards caused by delayed maintenance, thereby greatly improving the safety and efficiency of shield machine construction.

[0007] According to one aspect of the present application, a method for automatically detecting shield machine tool wear is provided, comprising: acquiring a shield machine tool image captured by a camera assembly; performing image preprocessing and noise suppression on the shield machine tool image to obtain an enhanced shield machine tool image; performing tool region segmentation and region of interest extraction on the enhanced shield machine tool image to obtain a tool target ROI image, wherein the tool region segmentation adopts image segmentation based on grayscale threshold, and the grayscale threshold is determined by the tool geometric features and tool texture features of each sliding window area; performing tool wear feature extraction on the tool target ROI image to obtain a tool wear feature; and inputting the tool wear feature into a tool wear intelligent detector based on a support vector regression model to obtain a detection result, wherein the detection result is a wear degree grade label.

[0008] In the above-mentioned automatic detection method for shield machine tool wear, the shield machine tool image is subjected to image preprocessing and noise suppression to obtain an enhanced shield machine tool image, including: performing median filtering on the shield machine tool image to obtain a filtered shield machine tool image; performing contrast enhancement on the filtered shield machine tool image to obtain a contrast enhanced shield machine tool image; and performing image defogging on the contrast enhanced shield machine tool image to obtain the enhanced shield machine tool image.

[0009] In the above-mentioned shield machine tool wear automatic detection method, the camera assembly includes a camera, an annular LED light source arranged around the camera and an adjustable bracket, and the camera and the annular LED light source are fixed to the adjustable bracket.

[0010] In the above-mentioned shield machine tool wear automatic detection method, the annular LED light source has a narrow spectrum, and its central wavelength is close to 450nm.

[0011] In the above-mentioned automatic detection method for shield machine tool wear, the enhanced shield machine tool image is subjected to tool region segmentation and region of interest extraction to obtain a tool target ROI image, including: performing image segmentation based on grayscale threshold on the enhanced shield machine tool image to obtain a tool region; performing morphological operations on the tool region to obtain an optimized tool region; and using a template matching algorithm to locate the blade region of the optimized tool region and cropping it to obtain the tool target ROI image.

[0012] In the above-mentioned automatic detection method for shield machine tool wear, the enhanced shield machine tool image is subjected to image segmentation based on grayscale threshold to obtain the tool area, including: step 311: converting the enhanced shield machine tool image into a grayscale image to obtain a grayscale tool image; step 312: performing sliding window sampling on the grayscale tool image and determining the grayscale threshold corresponding to the sliding window area based on the image grayscale value distribution within the sliding window; step 313: based on the grayscale threshold corresponding to the sliding window area, performing image binary segmentation on the sliding window area of the grayscale tool image to obtain the tool sliding window segmentation area; step 314: repeating steps 312 and 313 until all areas of the grayscale tool image are segmented by image binary to obtain the tool area.

[0013] In the above-mentioned shield machine tool wear automatic detection method, tool wear feature extraction is performed on the tool target ROI image to obtain tool wear features, including: calculating tool geometric features based on the Canny edge detection algorithm; calculating tool texture features based on the grayscale co-occurrence matrix; and merging the tool geometric features and the tool texture features to obtain the tool wear features.

[0014] In the above-mentioned automatic detection method for shield machine tool wear, sliding window sampling is performed on the grayscale tool image and the grayscale threshold corresponding to the sliding window area is determined based on the image grayscale value distribution within the sliding window, including: calculating the initial grayscale threshold corresponding to the sliding window area; calculating the tool geometric features and tool texture features of the sliding window area; based on the tool geometric features and tool texture features of the sliding window area, optimizing the initial grayscale value initial grayscale threshold to obtain the grayscale threshold corresponding to the sliding window area.

[0015] In the above-mentioned shield machine tool wear automatic detection method, based on the tool geometric characteristics and tool texture characteristics of the sliding window area, the initial grayscale value initial grayscale threshold is optimized to obtain the grayscale threshold corresponding to the sliding window area, including: according to the initial grayscale threshold To calculate the causal relationship between tool geometry and tool texture features, the tool geometry and tool texture features are normalized and expressed as: ;in, Indicates the Tool geometry features, Represents the total number of tool geometric features, and are the mean and standard deviation of the tool geometry characteristics, Indicates the Tool texture features, represents the total number of tool texture features, and are the mean and standard deviation of tool texture features, represents the normalized tool geometry, Represents the normalized tool texture features.

[0016] Based on the normalized tool geometry features and the normalized tool texture features, the initial grayscale threshold Optimize, expressed as: ;in, is the tuning coefficient, represents a natural constant, Represents the grayscale threshold corresponding to the sliding window area.

[0017] According to another aspect of the present application, a shield machine tool wear automatic detection system is also provided, which is used to execute the above-mentioned shield machine tool wear automatic detection method, including: a shield machine tool image acquisition module, used to acquire the shield machine tool image collected by the camera component; a shield machine tool image preprocessing module, used to perform image preprocessing and noise suppression on the shield machine tool image to obtain an enhanced shield machine tool image; a tool target extraction module, used to perform tool area segmentation and region of interest extraction on the enhanced shield machine tool image to obtain a tool target ROI image, wherein the tool area segmentation adopts image segmentation based on grayscale threshold, and the grayscale threshold is determined by the tool geometric features and tool texture features of each sliding window area; a tool wear feature extraction module, used to extract tool wear features on the tool target ROI image to obtain tool wear features; a tool wear intelligent detection module, used to input the tool wear features into a tool wear intelligent detector based on a support vector regression model to obtain a detection result, and the detection result is a wear degree level label.

[0018] Compared with the prior art, the shield machine tool wear automatic detection system and method provided by this application obtains the shield machine tool image collected by the camera component, and performs image preprocessing, noise suppression, tool region segmentation and region of interest extraction to obtain the tool target ROI image. Among them, the tool region segmentation adopts image segmentation based on grayscale threshold. The grayscale threshold is determined by the tool geometric features and tool texture features of each sliding window area. This method can improve the accuracy of image segmentation and the accuracy of geometric features and texture features. Finally, the tool wear feature extraction is performed on the tool target ROI image and input into the tool wear intelligent detector based on the support vector regression model to obtain the detection result. In this way, the automatic assessment of the wear degree is achieved, human intervention is reduced, and the detection efficiency and accuracy are improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The above and other purposes, features, and advantages of the present application will become more apparent through a more detailed description of the embodiments of the present application in conjunction with the accompanying drawings. The accompanying drawings are intended to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the drawings, the same reference numerals generally represent the same components or steps.

[0020] Figure 1 Schematic flow chart of a method for automatically detecting tool wear of a shield machine according to an embodiment of the present application.

[0021] Figure 2 This is a schematic flowchart of step 2 in the method for automatically detecting tool wear of a shield machine according to an embodiment of the present application.

[0022] Figure 3 This is a schematic flowchart of step 3 in the shield machine tool wear automatic detection method according to an embodiment of the present application.

[0023] Figure 4 This is a schematic flowchart of step 31 in the shield machine tool wear automatic detection method according to an embodiment of the present application.

[0024] Figure 5 This is a schematic flowchart of step 4 in the method for automatically detecting tool wear of a shield machine according to an embodiment of the present application.

[0025] Figure 6 This is a schematic block diagram of a shield machine tool wear automatic detection system according to an embodiment of the present application. DETAILED DESCRIPTION

[0026] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described herein.

[0027] Figure 1 FIG. 1 is a schematic flow chart of a shield machine tool wear automatic detection method according to an embodiment of the present application. Figure 1As shown, the automatic detection method for shield machine tool wear includes: step 1: acquiring a shield machine tool image captured by a camera component; step 2: performing image preprocessing and noise suppression on the shield machine tool image to obtain an enhanced shield machine tool image; step 3: performing tool region segmentation and region of interest extraction on the enhanced shield machine tool image to obtain a tool target ROI image; step 4: performing tool wear feature extraction on the tool target ROI image to obtain a tool wear feature; step 5: inputting the tool wear feature into a tool wear intelligent detector based on a support vector regression model to obtain a detection result, and the detection result is a wear degree grade label.

[0028] Specifically, in step 1, an image of the shield machine's cutterhead is acquired using a camera assembly. In one embodiment, the camera assembly includes a camera, a ring-shaped LED light source surrounding the camera, and an adjustable bracket, with the camera and the ring-shaped LED light source secured to the adjustable bracket. The ring-shaped LED light source has a narrow spectrum, with a central wavelength approaching 450nm. Specifically, the camera assembly is first installed in a suitable position within the shield machine to ensure that the camera can clearly capture the cutterhead and its attached cutterhead. The adjustable bracket allows the camera's angle and height to be adjusted based on the specific structure of the shield machine and the layout of the cutterheads, thereby achieving an optimal viewing angle. Furthermore, the ring-shaped LED light source design not only provides uniform and concentrated illumination, but also, due to its narrow spectrum (central wavelength approaching 450nm), effectively reduces the impact of ambient light on image quality. This is particularly important in low-light or uneven lighting conditions common in underground tunnel construction environments. The specific wavelength of the light source helps improve image contrast, making the cutter edges more distinct, which facilitates region segmentation and feature extraction in subsequent image processing steps.

[0029] In one specific embodiment, the camera assembly is mounted in a stable position in front of the cutterhead, ensuring that the camera continuously captures the working status of the cutterhead during its rotation. Given that dust and vibration generated by shield machine operation may interfere with image quality, the use of a ring-shaped LED light source with narrow spectrum characteristics is crucial. This light source not only provides sufficient brightness but also penetrates dust particles to a certain extent, reducing scattering effects, thereby ensuring the clarity and contrast of the captured image. Furthermore, the light source's blue spectrum (central wavelength near 450nm) has good reflectivity on metal surfaces, helping to highlight the contours and details of the cutterhead and facilitating subsequent high-precision image analysis.

[0030] Specifically, in step 2, the shield machine tool image undergoes image preprocessing and noise suppression to obtain an enhanced shield machine tool image. It should be understood that in the actual operating environment of a shield machine, the original acquired tool image is often severely disturbed by the large amount of dust generated during tunnel excavation, low lighting conditions, and complex geological environments. These factors can lead to image quality degradation, such as image blur, insufficient contrast, and increased noise, which seriously affect the accuracy of subsequent image analysis. Therefore, in order to improve the accuracy and reliability of tool wear detection, the original image must be preprocessed.

[0031] In one embodiment, Figure 2 As shown, in step 2, the shield machine tool image is subjected to image preprocessing and noise suppression to obtain an enhanced shield machine tool image, including: step 21: performing median filtering on the shield machine tool image to obtain a filtered shield machine tool image; step 22: performing contrast enhancement on the filtered shield machine tool image to obtain a contrast enhanced shield machine tool image; step 23: performing image dehazing on the contrast enhanced shield machine tool image to obtain the enhanced shield machine tool image.

[0032] In one specific embodiment, in step 21, the shield machine tool image is first subjected to median filtering to obtain a filtered shield machine tool image. Median filtering is a nonlinear filtering method that removes noise while preserving edge information by replacing the grayscale value of each pixel with the median of the grayscale values of all pixels in its neighborhood. This is particularly important for maintaining the clarity of the tool outline.

[0033] In step 22, after median filtering, random noise in the image is significantly reduced, but insufficient contrast may still exist, especially in low-light conditions. Contrast enhancement techniques are used to adjust the image's brightness and contrast. Contrast enhancement can be implemented using methods such as histogram equalization or adaptive contrast enhancement algorithms. By adjusting the image's grayscale distribution, details of the tool and its surrounding area are more prominent, facilitating subsequent accurate analysis. In one specific embodiment, adaptive histogram equalization (CLAHE) is used. Compared to traditional histogram equalization, CLAHE divides the image into blocks and performs histogram equalization on each block, avoiding the over-enhancement and noise amplification that can occur with global histogram equalization. This method first divides the image into multiple non-overlapping blocks, then calculates and equalizes the histogram of each block. Finally, bilinear interpolation is used to smooth the transition areas between adjacent blocks, ensuring a consistent and natural appearance for the entire image.

[0034] In step 23, considering the possible presence of fog or other factors affecting visibility in the tunnel, image dehazing is also required. Dehazing not only restores obscured details but also improves the overall visual quality of the image, laying the foundation for further precise analysis. In one specific embodiment, a dehazing algorithm based on dark channel prior theory is employed. This algorithm assumes that in a fog-free image, most local regions have at least one color channel with very low intensity values. Based on this assumption, the fog-free image can be restored by estimating the atmospheric light value and transmittance map. Specifically, the dark channel map of the input image is first calculated, and then the atmospheric light value is estimated based on the dark channel map. Next, the initial transmittance map is optimized using a guided filter to reduce noise and preserve edge information. Finally, the fog-free image is restored based on the atmospheric scattering model. This method is particularly suitable for light fog or smoke interference that may occur in closed environments such as underground tunnels, and can significantly improve image clarity and visibility.

[0035] Specifically, in step 3, the enhanced shield machine tool image undergoes tool region segmentation and region of interest (ROI) extraction to obtain a tool target ROI image. It should be understood that shield machines typically operate in complex and harsh underground environments, and the acquired images may contain a significant amount of background noise, shadows, and other interfering factors, which can severely impact the accuracy of subsequent tool wear feature extraction and analysis. Therefore, effective tool region segmentation and region of interest (ROI) extraction techniques can accurately separate the tool and its key components (such as the cutting edge) from the complex background, providing a reliable data foundation for subsequent precise analysis.

[0036] In one embodiment, Figure 3 As shown, in step 3, the enhanced shield machine tool image is subjected to tool region segmentation and region of interest extraction to obtain a tool target ROI image, including: step 31: performing image segmentation based on grayscale threshold on the enhanced shield machine tool image to obtain a tool region; step 32: performing morphological operations on the tool region to obtain an optimized tool region; step 33: using a template matching algorithm to locate the blade region of the optimized tool region and cropping it to obtain the tool target ROI image.

[0037] In one embodiment, Figure 4As shown, in step 31, the enhanced shield machine tool image is segmented based on the grayscale threshold to obtain the tool area, including: step 311: converting the enhanced shield machine tool image into a grayscale image to obtain a grayscale tool image; step 312: performing sliding window sampling on the grayscale tool image and determining the grayscale threshold corresponding to the sliding window area based on the image grayscale value distribution in the sliding window; step 313: based on the grayscale threshold corresponding to the sliding window area, performing image binarization segmentation on the sliding window area of the grayscale tool image to obtain the tool sliding window segmentation area; step 314: repeating steps 312 and 313 until all areas of the grayscale tool image are segmented by image binarization to obtain the tool area.

[0038] Specifically, in step 31, the grayscale tool image is sampled using a sliding window, and a corresponding grayscale threshold is determined based on the distribution of the image's grayscale values within each sliding window. This local characteristic-based approach can adapt to image variations under varying lighting conditions and improve segmentation accuracy. For each sliding window region, an appropriate grayscale threshold is set based on the grayscale distribution within it. For example, the mean or median of the grayscale values within the sliding window region can be selected as the grayscale threshold. This threshold is then used to perform binary segmentation on the region. Specifically, pixels with values above the threshold are designated as the foreground (i.e., the tool portion), while pixels with values below the threshold are designated as the background. This process is repeated until all regions of the grayscale tool image are segmented, resulting in a preliminary tool region.

[0039] In step 32, in order to further improve the clarity and integrity of the tool area, morphological operations such as expansion and erosion can be performed on the preliminary segmentation results to remove small noise points and fill the holes in the tool contour. The purpose of doing so is to make the tool area more continuous and smooth, and to reduce errors caused by image noise or incomplete segmentation. In a specific embodiment, the morphological operation uses a method that combines opening and closing operations to optimize the tool area. The opening operation removes small noise points by first corroding and then expanding, while the closing operation fills the holes in the tool contour by first expanding and then corroding. The effective combination of these two operations can not only remove noise, but also repair the broken or discontinuous tool contour caused by image segmentation, making the tool area more complete and smooth. In addition, the size and shape of the structural element can be adjusted according to the specific situation to better adapt to tools of different sizes and shapes.

[0040] In step 33, a template matching algorithm is used to precisely locate the blade region, the key part of the tool. Template matching is a search method based on image similarity. By comparing a predefined blade template with the segmented tool region, the best matching location is found as the blade region. In one specific embodiment, the template matching algorithm includes designing a template representing a typical blade shape, which should cover as many possible blade shapes as possible. The template is then moved pixel by pixel within the optimized tool region, and the similarity between the template and the current sub-image is calculated (for example, using a normalized cross-correlation coefficient). The location with the highest similarity is considered the best match, and the blade region can be precisely located based on this. To further improve matching accuracy, the tool region can be scaled and rotated before template matching to ensure that the template and the actual blade shape are as consistent as possible. Finally, the blade region is cropped based on the positioning results to obtain the final tool target ROI image.

[0041] Specifically, in step 4, tool wear feature extraction is performed on the tool target ROI image to obtain tool wear characteristics. It should be understood that a single type of feature often fails to fully describe the wear condition of a tool. For example, relying solely on geometric features may not accurately distinguish different types of wear (such as uniform wear and localized severe wear), while relying solely on texture features may overlook changes in the tool structure. Therefore, combining geometric and texture features can provide richer and more accurate information.

[0042] In one embodiment, Figure 5 As shown, in step 4, tool wear feature extraction is performed on the tool target ROI image to obtain tool wear features, including: step 41: calculating tool geometric features based on the Canny edge detection algorithm; step 42: calculating tool texture features based on the gray level co-occurrence matrix; step 43: merging the tool geometric features and the tool texture features to obtain the tool wear features.

[0043] Specifically, in step 41, the tool geometric features are calculated based on the Canny edge detection algorithm. Canny edge detection is a multi-stage optimized edge detection method that finally outputs an edge image through steps such as smoothing the image with Gaussian filtering, calculating the image gradient amplitude and direction, non-maximum suppression, and double threshold screening. In tool wear detection, Canny edge detection can effectively identify subtle changes in the tool contour and its surface, such as cracks, wear areas, etc. In a specific embodiment, in a typical shield machine tool wear detection scenario, the tool target ROI image obtained after preliminary processing is input into the Canny edge detection algorithm, and the high and low threshold parameters are adjusted to ensure that the tool contour and wear marks can be clearly depicted. These edge information reflects the geometric shape of the tool and its changes, and provides an important reference basis for subsequent wear degree assessment.

[0044] In step 42, the tool texture features are calculated based on the gray-level co-occurrence matrix (GLCM). The gray-level co-occurrence matrix is a statistical method used to describe the spatial dependencies or texture features between pixels in an image. By calculating the co-occurrence matrix at different directions and distances and extracting statistics such as contrast, correlation, energy, and entropy from it, the texture characteristics of the tool surface can be fully reflected. In practical applications, the gray-level co-occurrence matrix is calculated for different parts of the tool target ROI image, and the corresponding texture features are extracted. For example, for the tool edge area, the surface roughness may increase due to long-term use, and this change can be reflected through certain statistics in the gray-level co-occurrence matrix. Specifically, by analyzing the gray-level co-occurrence matrix of the tool edge area, the texture features of the area can be obtained, thereby helping to identify the degree and pattern of wear.

[0045] In step 43, the tool geometric features and tool texture features are combined to form a comprehensive tool wear signature. This process is more than a simple feature superposition; rather, it effectively integrates geometric and texture features to construct a multi-dimensional feature vector that more comprehensively describes the tool wear state. In one specific embodiment, geometric features obtained from Canny edge detection (such as edge length and curvature) are combined with texture features (such as contrast and energy) calculated from the gray-level co-occurrence matrix to generate a multi-dimensional feature vector. This feature vector can more comprehensively reflect the overall wear of the tool, effectively capturing both surface wear and structural damage.

[0046] Specifically, in step 5, the tool wear characteristics are input into the tool wear intelligent detector based on the support vector regression model to obtain a detection result, which is a wear degree grade label. It should be understood that traditional tool wear assessment methods often rely on empirical judgment or simple physical measurements, and it is difficult to provide accurate and consistent results. The intelligent detection method based on the support vector regression model can make full use of modern computing technology and big data advantages. By learning and analyzing a large amount of historical data, a complex mapping relationship between tool wear characteristics and wear degree is established. This method not only improves the accuracy of the detection results, but also enhances the robustness and adaptability of the system, enabling it to operate stably in various complex engineering environments and meet the needs of actual construction. In addition, such a design can also reduce maintenance costs because it reduces the occurrence of misjudgments or repeated detections due to human judgment errors.

[0047] In a specific embodiment, an intelligent tool wear detector based on a support vector regression (SVR) model is first constructed. Support vector regression is a supervised learning method particularly well-suited for small sample sizes, nonlinearity, and high-dimensional pattern recognition. These characteristics make it well-suited for tool wear detection in complex engineering environments. During the preparatory phase, a large amount of tool wear data must be collected as a training set. This data should cover tools with varying degrees of wear and their corresponding geometric and textural features. The SVR model is then trained using this data, and model parameters are adjusted until optimal performance is achieved. During training, model hyperparameters, such as the penalty coefficient C and kernel function parameter γ, can be optimized through techniques such as cross-validation to ensure good generalization. Once tool wear features are input into the SVR model, the model analyzes them based on the rules learned during training and outputs a detection result representing the tool wear grade. This detection result can be a continuous variable representing the tool wear grade, or a discrete wear grade label, such as slight wear, moderate wear, or severe wear.

[0048] In particular, considering that image binarization segmentation based on sliding window regions may result in segmentation errors due to the blurred edge smoothness of grayscale values within the sliding window region and the sensitivity of thresholded texture transitions to unevenness, which in turn further affects the calculation of geometric and texture features, in a preferred embodiment, a preset grayscale threshold is optimized based on the tool geometric and texture features of each sliding window region. In other words, the grayscale threshold is determined based on the tool geometric and texture features of each sliding window region, thereby improving the accuracy of grayscale threshold-based image segmentation based on image geometric and texture features.

[0049] In one embodiment, sliding window sampling is performed on the grayscale tool image and a grayscale threshold corresponding to the sliding window area is determined based on the image grayscale value distribution in the sliding window, including: calculating an initial grayscale threshold corresponding to the sliding window area, wherein the initial grayscale threshold is the mean or median of the grayscale values in the sliding window area; calculating tool geometric features and tool texture features in the sliding window area, wherein the tool geometric features in the sliding window area are calculated based on a Canny edge detection algorithm, and the tool texture features in the sliding window area are calculated based on a gray level co-occurrence matrix; based on the tool geometric features and tool texture features in the sliding window area, the initial grayscale value and initial grayscale threshold are optimized to obtain the grayscale threshold corresponding to the sliding window area.

[0050] In a specific embodiment, based on the tool geometric features and tool texture features of the sliding window area, the initial grayscale value initial grayscale threshold is optimized to obtain the grayscale threshold corresponding to the sliding window area, including: Here, assuming that the initial grayscale threshold in the predetermined sliding window area is , and the tool geometric features corresponding to the predetermined sliding window area are expressed as , and the tool texture feature corresponding to the predetermined sliding window area is expressed as , then first according to the initial gray threshold The causal relationship between the calculation of tool geometric features and tool texture features is calculated by normalizing the tool geometric features and tool texture features to express them as follows: ;in, Indicates the Tool geometry features, Represents the total number of tool geometric features, and are the mean and standard deviation of the tool geometry characteristics, Indicates the Tool texture features, represents the total number of tool texture features, and are the mean and standard deviation of tool texture features, represents the normalized tool geometry, Represents the normalized tool texture features.

[0051] Based on the normalized tool geometry features and the normalized tool texture features, the initial grayscale threshold Optimize, expressed as: ;in, is a tuning coefficient used to adjust the influence of the normalized tool geometry and texture features on the initial grayscale threshold. For example, , represents a natural constant, Represents the grayscale threshold corresponding to the sliding window area.

[0052] In this way, when the grayscale value blurs the binary segmentation boundary due to the smoothness of the geometric edge, and the local texture transition unevenness in the thresholded segmentation area disturbs the grayscale distribution on both sides of the edge, the feature robustness of the segmentation threshold can be improved by the tool geometric feature threshold correlation and tool texture feature threshold density constraint, and the image segmentation accuracy as well as the accuracy of the tool geometric features and tool texture features can be simultaneously improved.

[0053] In summary, the automatic detection method for shield machine tool wear provided in this application uses a camera assembly to capture shield machine tool images and performs a series of preprocessing steps (including median filtering, contrast enhancement, and dehazing). This significantly improves image quality, making subsequent analysis more accurate and reliable. This method effectively overcomes the impact of factors such as dust and lighting changes in the working environment on image acquisition, ensuring the high quality of the data source. Furthermore, during the tool region segmentation stage, this solution utilizes a grayscale threshold-based image segmentation method, where the grayscale threshold is determined based on precalculated geometric and texture features within each sliding window region. This strategy not only enhances the ability to identify tool regions in complex backgrounds but also improves the accuracy of tool region segmentation, facilitating more accurate positioning and extraction of tool target ROI images. Furthermore, the Canny edge detection algorithm and gray-level co-occurrence matrix are used to separately calculate the tool's geometric and texture features, and these are combined to form the tool wear signature. This comprehensive approach comprehensively reflects the wear condition of the tool surface and is more accurate than single-feature analysis. Finally, the extracted wear signature is input into a support vector regression model for intelligent detection, resulting in a wear level label. This method realizes the automated assessment of wear degree, reduces human intervention, improves detection efficiency and accuracy, and can also promptly detect tool wear conditions and avoid safety hazards caused by delayed maintenance, thereby greatly improving the safety and efficiency of shield machine construction.

[0054] This application also provides a shield machine tool wear automatic detection system, such as Figure 6As shown, the shield machine tool wear automatic detection system 600 includes: a shield machine tool image acquisition module 610, which is used to acquire the shield machine tool image collected by the camera component; a shield machine tool image preprocessing module 620, which is used to perform image preprocessing and noise suppression on the shield machine tool image to obtain an enhanced shield machine tool image; a tool target extraction module 630, which is used to perform tool area segmentation and region of interest extraction on the enhanced shield machine tool image to obtain a tool target ROI image, wherein the tool area segmentation adopts image segmentation based on grayscale threshold, and the grayscale threshold is determined by the tool geometric features and tool texture features of each sliding window area; a tool wear feature extraction module 640, which is used to extract tool wear features from the tool target ROI image to obtain tool wear features; and a tool wear intelligent detection module 650, which is used to input the tool wear features into a tool wear intelligent detector based on a support vector regression model to obtain a detection result, and the detection result is a wear degree grade label.

[0055] The basic principles of the present application have been described above in conjunction with specific embodiments. However, it should be noted that the advantages, strengths, and effects mentioned in this application are merely illustrative and not restrictive, and it should not be assumed that these advantages, strengths, and effects are required of each embodiment of this application. In addition, the specific details disclosed above are merely illustrative and facilitating understanding, and are not restrictive. The above details do not limit this application to necessarily being implemented using the above specific details.

[0056] The block diagrams of the devices, devices, equipment, and systems involved in this application are merely illustrative examples and are not intended to require or imply that they must be connected, arranged, or configured in the manner shown in the block diagrams. As will be appreciated by those skilled in the art, these devices, devices, equipment, and systems can be connected, arranged, or configured in any manner. Words such as "include," "comprise," "have," and the like are open-ended words, meaning "including but not limited to," and can be used interchangeably therewith. The words "or" and "and" used herein refer to the words "and / or" and can be used interchangeably therewith, unless the context clearly indicates otherwise. The word "such as" used herein refers to the phrase "such as but not limited to," and can be used interchangeably therewith.

[0057] It should also be noted that in the apparatus, device, and method of the present application, each component or each step can be decomposed and / or recombined, and such decomposition and / or recombination should be regarded as equivalent solutions of the present application.

[0058] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use the present application. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein may be applied to other aspects without departing from the scope of the present application. Therefore, the present application is not intended to be limited to the aspects shown herein, but rather to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0059] The above description has been provided for the purpose of illustration and description. Furthermore, this description is not intended to limit the embodiments of the present application to the forms disclosed herein. Although a number of example aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, alterations, additions, and sub-combinations thereof.

Claims

1. A shield machine tool wear automatic detection method, characterized in that: include: Acquire shield machine tool images captured by the camera assembly; performing image preprocessing and noise suppression on the shield machine tool image to obtain an enhanced shield machine tool image; Performing tool region segmentation and region of interest extraction on the enhanced shield machine tool image to obtain a tool target ROI image, wherein the tool region segmentation adopts image segmentation based on grayscale threshold, and the grayscale threshold is determined by the tool geometric features and tool texture features of each sliding window area; performing tool wear feature extraction on the tool target ROI image to obtain a tool wear feature; inputting the tool wear feature into a tool wear intelligent detector based on a support vector regression model to obtain a detection result, wherein the detection result is a wear degree grade label; Sliding window sampling is performed on the grayscale tool image and a grayscale threshold corresponding to the sliding window area is determined based on the image grayscale value distribution within the sliding window, including: Calculate the initial grayscale threshold corresponding to the sliding window area; Calculating tool geometric features and tool texture features in the sliding window area; Based on the tool geometric features and tool texture features of the sliding window area, the initial grayscale threshold is optimized to obtain a grayscale threshold corresponding to the sliding window area, including: To calculate the causal relationship between tool geometry and tool texture features, the tool geometry and tool texture features are normalized and expressed as: ;in, Indicates the Tool geometry features, Represents the total number of tool geometric features, and are the mean and standard deviation of the tool geometry characteristics, Indicates the Tool texture features, represents the total number of tool texture features, and are the mean and standard deviation of tool texture features, represents the normalized tool geometry, Represents the normalized tool texture features; Based on the normalized tool geometric features and the normalized tool texture features, the initial grayscale threshold Optimize, expressed as: ;in, is the tuning coefficient, represents a natural constant, Represents the grayscale threshold corresponding to the sliding window area.

2. The shield machine tool wear automatic detection method according to claim 1, characterized in that: The shield machine tool image is subjected to image preprocessing and noise suppression to obtain an enhanced shield machine tool image, including: performing median filtering on the shield machine tool image to obtain a filtered shield machine tool image; performing contrast enhancement on the filtered shield machine tool image to obtain a contrast enhanced shield machine tool image; and performing image defogging on the contrast enhanced shield machine tool image to obtain the enhanced shield machine tool image.

3. The shield machine tool wear automatic detection method according to claim 1, characterized in that: The camera assembly includes a camera, an annular LED light source arranged around the camera, and an adjustable bracket. The camera and the annular LED light source are fixed to the adjustable bracket.

4. The shield machine tool wear automatic detection method according to claim 3, characterized in that: The annular LED light source has a narrow spectrum, with a central wavelength close to 450 nm.

5. The shield machine tool wear automatic detection method according to claim 4, characterized in that: The enhanced shield machine tool image is subjected to tool region segmentation and region of interest extraction to obtain a tool target ROI image, comprising: performing image segmentation based on grayscale threshold on the enhanced shield machine tool image to obtain a tool region; performing morphological operations on the tool region to obtain an optimized tool region; and locating the blade region of the optimized tool region using a template matching algorithm and performing cropping to obtain the tool target ROI image.

6. The shield machine tool wear automatic detection method according to claim 5, characterized in that: The enhanced shield machine tool image is subjected to image segmentation based on grayscale threshold to obtain a tool area, comprising: step 311: converting the enhanced shield machine tool image into a grayscale image to obtain a grayscale tool image; step 312: performing sliding window sampling on the grayscale tool image and determining a grayscale threshold corresponding to the sliding window area based on the image grayscale value distribution within the sliding window; step 313: performing image binarization segmentation on the sliding window area of the grayscale tool image based on the grayscale threshold corresponding to the sliding window area to obtain a tool sliding window segmentation area; step 314: repeating steps 312 and 313 until all areas of the grayscale tool image are segmented by image binarization to obtain the tool area.

7. The shield machine tool wear automatic detection method according to claim 6, characterized in that: Tool wear feature extraction is performed on the tool target ROI image to obtain tool wear features, including: calculating tool geometric features based on a Canny edge detection algorithm; calculating tool texture features based on a gray level co-occurrence matrix; and merging the tool geometric features and the tool texture features to obtain the tool wear features.

8. A shield machine tool wear automatic detection system, used to execute the shield machine tool wear automatic detection method according to any one of claims 1 to 7, characterized in that: include: A shield machine tool image acquisition module is used to acquire a shield machine tool image captured by a camera assembly; a shield machine tool image preprocessing module is used to perform image preprocessing and noise suppression on the shield machine tool image to obtain an enhanced shield machine tool image; a tool target extraction module is used to perform tool region segmentation and region of interest extraction on the enhanced shield machine tool image to obtain a tool target ROI image, wherein the tool region segmentation adopts image segmentation based on grayscale threshold, and the grayscale threshold is determined by the tool geometric features and tool texture features of each sliding window area; a tool wear feature extraction module is used to extract tool wear features on the tool target ROI image to obtain tool wear features; a tool wear intelligent detection module is used to input the tool wear features into a tool wear intelligent detector based on a support vector regression model to obtain a detection result, and the detection result is a wear degree grade label.

Images