A method and system for detecting appearance defects in bridge rubber bearings

Abstract

This invention belongs to the field of image data processing technology and relates to a method and system for detecting appearance defects in bridge rubber bearings. The method includes: acquiring a side image of the bearing and correcting it to a frontal projection image; extracting the tilt angle, left and right side height, vertical compression ratio, and texture entropy reflecting the degree of wrinkling on the rubber surface of the bearing from the frontal projection image; constructing an asymmetric distortion index and load-mode coupling anomaly degree; comparing the asymmetric distortion index and load-mode coupling anomaly degree with preset safety thresholds respectively, and determining whether the rubber bearing has defects and the type of defects based on the comparison results. This invention effectively reduces the false alarm rate and achieves accurate judgment of the bearing's health status by establishing a physical mapping relationship between macroscopic geometric deformation and microscopic surface texture.

Description

Technical Field

[0001] This invention belongs to the field of image data processing technology, and specifically relates to a method and system for detecting appearance defects in bridge rubber bearings. Background Technology

[0002] As a key component connecting the superstructure and substructure of a bridge, bridge rubber bearings bear the important responsibility of transmitting vertical loads and accommodating horizontal displacements caused by beam rotation and thermal expansion and contraction. During long-term use, due to environmental erosion and reciprocating loads, bearings are prone to defects such as aging and cracking, internal voids caused by the separation of steel plates and rubber, excessive shear deformation, or installation misalignment. Failure to detect these defects in a timely manner will directly threaten the overall safety of the bridge structure.

[0003] In existing technologies, mainstream visual inspection techniques primarily rely on edge detection algorithms to extract the support contour and then calculate the tilt angle of the contour to determine if the support is abnormal. However, this detection method based on a single geometric shape has significant technical blind spots in practical engineering. On the one hand, this type of method cannot effectively distinguish between normal shear deformation and installation misalignment, because supports under normal functional shear deformation caused by temperature changes are usually parallelogram-shaped, while installation misalignment often leads to trapezoidal distortion, and the two can easily be confused based solely on angle thresholds. On the other hand, existing technologies ignore the physical relationship between shape and stress, usually treating the support as a rigid body for measurement, failing to utilize the surface wavy wrinkle texture generated by the hyperelasticity of rubber materials. This limitation makes it difficult to detect hidden defects such as voids or internal fractures, because although these defects manifest as geometric compression, the surface lacks corresponding stress wrinkle features, resulting in missed detection.

[0004] Therefore, how to solve the problem that existing technologies cannot effectively distinguish between normal functional deformations and structural defects of branch seats through appearance images, and how to detect hidden physical defects, has become an urgent technical challenge. Summary of the Invention

[0005] The purpose of this invention is to propose a method and system for detecting appearance defects of bridge rubber bearings, in order to solve the technical problems of existing technologies that cannot effectively distinguish between normal functional deformation and structural defects of rubber bearings based solely on geometric contours, and that it is difficult to detect hidden physical defects such as internal voids.

[0006] To address the above problems, the present invention proposes a technical solution for detecting appearance defects in bridge rubber bearings:

[0007] A method for detecting appearance defects in bridge rubber bearings includes the following steps:

[0008] S1. Acquire a grayscale image of the side of the rubber bearing, extract the region of interest and perform perspective transformation correction on the grayscale image to obtain a frontal projection image;

[0009] S2. Perform feature extraction on the frontal projection image to obtain the tilt angle between the left and right edges of the support, the left and right heights, the vertical compression ratio, and the texture entropy reflecting the degree of wrinkling on the rubber surface.

[0010] S3. Calculate the asymmetric distortion index based on the tilt angle of the left and right edges, the left height and the right height, and calculate the load morphological coupling anomaly based on the texture entropy and the vertical compression ratio.

[0011] S4. Compare the asymmetric distortion index with a preset distortion judgment threshold, and compare the load morphology coupling anomaly degree with a preset anomaly degree safety threshold; determine whether the rubber bearing has defects and the type of defects based on the comparison results.

[0012] This invention establishes a physical mapping relationship between the geometric shape and surface texture of rubber bearings. It can not only effectively distinguish between normal functional deformation caused by thermal expansion and contraction and abnormal deformation caused by installation misalignment, but also identify abnormal states where the geometric height is compressed but the surface lacks stress wrinkles due to internal voids. This solves the technical problem that existing technologies that rely solely on contour judgment are prone to missed detections.

[0013] Further, step S1 involves extracting the region of interest and performing perspective transformation correction on the side grayscale image, including:

[0014] Gaussian filtering was applied to the acquired side grayscale image, and adaptive threshold segmentation was performed using the maximum inter-class variance method to separate the black rubber support area from the background.

[0015] Identify the edge lines of the upper and lower steel plates in the rubber bearing area, construct a perspective transformation matrix, and correct the side grayscale image into a frontal projection image.

[0016] Image preprocessing and perspective transformation eliminate the effects of environmental noise and geometric distortion caused by shooting angle, ensuring the objectivity and accuracy of subsequent geometric parameter measurements.

[0017] Furthermore, step S2 specifically includes:

[0018] The left and right vertical edges of the support, as well as the horizontal edges of the upper and lower steel plates, are extracted from the frontal projection image using an edge detection operator.

[0019] The angles between the left and right vertical edges and the vertical line are calculated using the least squares line fitting method, and are used as the left tilt angle and the right tilt angle, respectively.

[0020] The vertical pixel distance between the upper and lower endpoints of the left vertical edge is calculated as the left height, the vertical pixel distance between the upper and lower endpoints of the right vertical edge is calculated as the right height, the average of the left height and the right height is calculated as the average pixel height, and the ratio of the actual physical height corresponding to the average pixel height to the pre-stored initial factory height is used as the vertical compression ratio.

[0021] A rectangular sub-block is extracted from the central region of the front-view projected image to construct a gray-level co-occurrence matrix. The entropy value of the gray-level co-occurrence matrix is ​​calculated and normalized to obtain the texture entropy.

[0022] Furthermore, the formula for calculating the asymmetric distortion index is as follows:

[0023]

[0024] In the formula, Indicates the asymmetric distortion index; and These represent the left tilt angle and the right tilt angle, respectively. and These represent the left side height and the right side height, respectively. This indicates the pixel width of the bottom edge of the support; This represents the geometric sensitivity coefficient.

[0025] By constructing an asymmetric distortion index, mathematical logic is used to distinguish between parallelogram deformation and trapezoidal or irregular deformation, overcoming the frequent false alarms caused by relying solely on angle thresholds in existing technologies, and achieving sensitive detection of defects such as installation misalignment.

[0026] Furthermore, the calculation formula for the load morphological coupling anomaly degree is as follows:

[0027]

[0028] In the formula, Indicates the degree of load-mode coupling anomaly; This represents the normalized texture entropy; This represents the theoretical reference texture entropy calculated based on the current vertical compression ratio; Represents the zero safety constant; Indicates the distortion weighting coefficient; is the asymmetric distortion index.

[0029] By constructing a load-shape coupling anomaly degree, the physical consistency between the geometric compression characteristics and surface texture characteristics of rubber bearings was analyzed in depth. Even if the appearance of the bearing seems normal, as long as the texture characteristics representing its stress state do not match the current deformation state, it can be sensitively detected, thereby achieving accurate detection of hidden defects such as internal voids and material aging.

[0030] Furthermore, the theoretical reference texture entropy The calculation formula is as follows:

[0031]

[0032] In the formula, This indicates the vertical compression ratio; Indicates the material texture coefficient; This represents the nonlinear response factor.

[0033] Furthermore, the acquisition of the side grayscale image of the rubber bearing includes:

[0034] The rubber bearing was photographed using a camera installed on the side of the cap beam, along with a low-angle auxiliary light source.

[0035] The side grayscale image is obtained by enhancing the shadow contrast of the rubber surface wrinkles using the side low-angle auxiliary light source.

[0036] Furthermore, the construction of the gray-level co-occurrence matrix includes:

[0037] In the rectangular sub-block, the direction of the grayscale co-occurrence matrix is ​​set to 0 degrees to capture the texture features of the rubber bulging and wrinkling on the side after being compressed.

[0038] Furthermore, determining whether the rubber bearing has defects and the type of defects based on the comparison results includes:

[0039] If the load-mode coupling anomaly degree is greater than a preset anomaly degree safety threshold, or the asymmetric distortion index is greater than a distortion determination threshold, then the support is determined to have a defect; based on this, the defect type is determined:

[0040] If the asymmetric distortion index is greater than the distortion determination threshold, the defect type is determined to be installation misalignment or structural distortion.

[0041] If the asymmetric distortion index is less than or equal to the distortion determination threshold and the load morphology coupling anomaly degree is greater than the preset anomaly degree safety threshold, the defect type is determined to be internal void or material aging.

[0042] The technical solution of the bridge rubber bearing appearance defect detection system proposed in this invention is as follows:

[0043] A bridge rubber bearing appearance defect detection system includes a processor and a memory. The memory stores computer program instructions. When the computer program instructions are executed by the processor, a bridge rubber bearing appearance defect detection method is implemented according to any of the above technical solutions.

[0044] The beneficial effects of this invention are as follows: By constructing an asymmetric distortion index, this invention can sensitively capture trapezoidal and irregular deformations caused by installation misalignment or structural distortion, and strictly distinguish such anomalies from regular parallelogram deformations caused by thermal expansion and contraction. This reduces the false alarm rate caused by traditional visual detection methods that rely solely on angle thresholds, and significantly improves the reliability of the detection results.

[0045] This invention, by constructing a load-shape coupling anomaly degree, can identify abnormal states where, although the support appears to be compressed in geometric height, its surface lacks corresponding stress wrinkles due to voids and lack of stress, or where aging and hardening cause texture features that do not conform to theoretical expectations. This detection logic based on the shape-stress coupling mechanism overcomes the blind spot of existing technologies that treat supports as rigid bodies and ignore the mechanical properties of materials, effectively preventing the missed detection of hidden physical defects.

[0046] This invention logically compares the asymmetric distortion index and load shape coupling anomaly degree with preset safety thresholds. This not only triggers alarms in real time but also automatically determines whether the defect is a structural problem (such as installation misalignment) or a material problem (such as internal voids). This automated diagnostic function provides maintenance personnel with clear decision-making criteria, helps quickly locate the cause of faults, and improves the efficiency and safety of bridge operation and maintenance. Attached Figure Description

[0047] Figure 1 This is a flowchart of the steps of the bridge rubber bearing appearance defect detection method of the present invention;

[0048] Figure 2 This is a scatter plot comparison of data between the present invention and existing technologies in terms of defect identification capabilities;

[0049] Figure 3 This is a schematic diagram of the state analysis of the support in an embodiment of the present invention;

[0050] Figure 4 This is a schematic diagram illustrating the real-time monitoring trend of the asymmetric distortion index of the support in this embodiment of the invention. Detailed Implementation

[0051] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0052] Specific embodiments of the bridge rubber bearing appearance defect detection method proposed in this invention:

[0053] like Figure 1 As shown, the method for detecting visual defects in bridge rubber bearings includes the following steps:

[0054] S1. Acquire a grayscale image of the side of the rubber bearing, extract the region of interest and perform perspective transformation correction on the grayscale image to obtain a frontal projection image.

[0055] Specifically, an industrial high-definition camera installed on the side of the cap beam is used for imaging. To ensure image quality, the camera's resolution is preferably no less than five megapixels. To address the issue that the rubber surface texture is not clearly displayed under natural light, the system uses a side low-angle auxiliary light source. By utilizing the grazing light principle, the shadow contrast of the rubber surface wrinkles is enhanced, thereby acquiring a grayscale image of the rubber support side with rich detail.

[0056] After acquiring the image, the side grayscale image is first processed by Gaussian filtering, for example, using a 5×5 Gaussian kernel for convolution to effectively remove high-frequency noise caused by environmental dust. Subsequently, adaptive thresholding is performed using the Otsu algorithm (maximum inter-class variance method) to automatically calculate the optimal segmentation threshold, accurately separating the black rubber support area from the complex background, and thus extracting the region of interest.

[0057] To address the perspective distortion issue caused by factors such as camera installation location, resulting in near-large and far-small appearances in the bearing, the system uses image processing algorithms to identify the edge lines of the upper and lower steel plates in the rubber bearing area. It then uses Hough transform to detect four edge lines and calculates their intersection points as corner points, constructing a perspective transformation matrix. This matrix is ​​used to correct the trapezoidal image captured from a side view angle into a standard rectangular frontal projection image, ensuring the objectivity and accuracy of subsequent geometric measurements.

[0058] Through the above image acquisition and preprocessing steps, and by using specific light source arrangements and algorithm corrections, the effects of uneven ambient lighting, dust interference, and shooting angle deviations can be effectively eliminated, laying the foundation for subsequent extraction of high-precision geometric and texture features.

[0059] S2. Perform feature extraction on the frontal projection image to obtain the tilt angle between the left and right edges of the support, the left and right heights, the vertical compression ratio, and the texture entropy reflecting the degree of wrinkling on the rubber surface.

[0060] Specifically, the Canny edge detection operator is first used to extract the left and right vertical edges of the support, as well as the horizontal edges of the upper and lower steel plates in the frontal projection image.

[0061] Least squares line fitting is performed on the pixels of the left and right vertical edges respectively, and the angle between the fitted line and the vertical line is calculated. The angle between the left vertical edge and the vertical line is denoted as the left tilt angle. The angle between the right vertical edge and the plumb line is denoted as the right tilt angle. .

[0062] Calculate the vertical pixel distance between the top and bottom endpoints of the left vertical edge as the left height. Calculate the vertical pixel distance between the upper and lower endpoints of the right vertical edge as the right height. Calculate the average of the left and right heights as the current average pixel height of the support. , .

[0063] To eliminate the influence of shooting distance and zoom on dimensional measurement, the system performs pixel equivalent calibration: extracting the horizontal pixel width of the upper or lower steel plate of the support in the frontal projection image, and simultaneously retrieving the factory physical width of the steel plate corresponding to this model of support from the database, then calculating the image scaling factor. Based on this scaling factor, the actual physical height of the support corresponding to the current average pixel height is calculated. The system reads the initial factory height of this model of support from the database, and uses the ratio of the actual physical height to the pre-stored initial factory height as the vertical compression ratio. The smaller the vertical compression ratio, the greater the degree of compression, indicating that the support is bearing a greater load.

[0064] A crop is taken from the center region of the projected image. A rectangular sub-block is used as the texture analysis area to avoid interference from edge stains. A gray-level co-occurrence matrix with a stride of 1 and an orientation of 0 degrees (horizontal) is constructed. This study analyzes the degree of wrinkling on the rubber surface and calculates the texture entropy of the gray-level co-occurrence matrix as an indicator of wrinkle complexity. The physical meaning of this indicator is: the greater the pressure on the rubber, the more wrinkles bulge out on the sides, the more chaotic the image texture, and the higher the texture entropy value; conversely, when the rubber support is not under pressure or is detached, the surface is flat and wrinkle-free, resulting in a lower texture entropy value. To eliminate the influence of light intensity, the texture entropy is normalized to ensure its value falls within a certain range. It should be noted that the calculation method for texture entropy is the same as that for Shannon entropy in existing technologies, and will not be elaborated further here.

[0065] Through the above feature extraction steps, the system obtains the geometric deformation parameters characterizing the outer contour of the rubber bearing and the texture parameters reflecting the surface stress response. It reflects the physical state of the rubber bearing from two aspects: macroscopic geometric features and microscopic texture features, providing an accurate calculation basis for subsequent defect identification.

[0066] S3. Calculate the asymmetric distortion index based on the tilt angle of the left and right edges, the left height and the right height, and calculate the load morphological coupling anomaly based on the texture entropy and the vertical compression ratio.

[0067] The purpose of calculating the asymmetric distortion index in this step is to effectively distinguish between normal functional shear deformation of the branch bearing due to thermal expansion and contraction and installation misalignment defects caused by improper installation. Specifically, the formula for calculating the asymmetric distortion index is:

[0068]

[0069] in, Indicates the asymmetric distortion index; and These represent the left tilt angle and the right tilt angle, respectively. and These represent the left side height and the right side height, respectively. This indicates the pixel width of the bottom edge of the support; This represents the geometric sensitivity coefficient, which is preferably set to a value between 2.0 and 5.0 in practical applications to adjust the algorithm's sensitivity to height differences.

[0070] This formula, by calculating the tangent difference between the left and right tilt angles and the exponential function of the left and right height difference, can distinguish normal deformation of a parallelogram from abnormal deformation of a trapezoid or irregular shape.

[0071] To further illustrate the calculation process of the asymmetric distortion index and its beneficial effects in distinguishing branching states, a detailed explanation is provided below through specific calculation examples.

[0072] The first scenario involves normal functional shear deformation of the support, in which case the support exhibits a regular parallelogram shape. The measured tilt angles on the left and right sides are approximately equal, both around 10 degrees; simultaneously, the heights on the left and right sides are essentially the same. That is... , Correspondingly, , Therefore, the final asymmetric distortion index .

[0073] The second scenario involves installation misalignment of the support, in which case the support will be trapezoidal or irregular in shape. The measured tilt angle on the left side is... right tilt angle Simultaneously, the absolute value of the difference between the height on the left and right sides was measured to be 20 pixels, and the pixel width of the bottom edge of the support was [missing information]. Pixel, Geometric Sensitivity Substituting into the above calculation formula, we get:

[0074] .

[0075] By comparing the calculation results of the two scenarios, it can be seen that when there are abnormal conditions such as installation misalignment, the asymmetric distortion index is significantly higher than that under normal shear deformation conditions, thus verifying the sensitivity and effectiveness of this index in distinguishing different support states.

[0076] This step calculates the load-mode coupling anomaly degree, aiming to solve the challenge of detecting hidden defects such as internal voids or pseudo-compression in rubber bearings. Specifically, the formula for calculating the load-mode coupling anomaly degree is:

[0077]

[0078] In the formula, Indicates the degree of load-mode coupling anomaly; This represents the normalized texture entropy; This represents the zero-prevention safety constant, which is introduced into the formula to prevent the denominator from being zero and to ensure calculation stability. Its value can be set to 0.01. Indicates the distortion weighting coefficient; is the asymmetric distortion index.

[0079] In the formula This represents the theoretical reference texture entropy calculated based on the current vertical compression ratio, and its calculation formula is as follows: .

[0080] In this formula, This represents the vertical strain of the support. This is the material texture coefficient, which is related to the texture generation characteristics of the rubber material surface. In specific implementations, it can be set to 0.8. This is a nonlinear response factor used to reflect the nonlinear mechanical characteristics of rubber materials under pressure. This nonlinear response factor can be set to 1.5. This is the distortion weighting coefficient, used to adjust the influence of the asymmetric distortion index on the final anomaly result.

[0081] To verify the detection capability of this method for hidden defects, a third scenario is assumed: the support has an internal void defect. In this scenario, although the support appears compressed in terms of geometric height, the measured vertical compression ratio... The corresponding vertical strain is 0.2. Theoretically, this compression state should produce obvious surface wrinkles. At this time, the calculated theoretical reference texture entropy is... However, due to the void inside the support, the rubber did not actually bulge under force, and the surface remained smooth, resulting in an inaccurate measured texture entropy. .

[0082] Based on this, the load-mode coupling anomaly degree is calculated. An asymmetric distortion index is assumed. Zero safety constant Substituting into the formula for calculating the load-mode coupling anomaly, we get: .

[0083] In contrast, if the support is under normal stress, its measured texture entropy is... It should be close to the theoretical value of 0.071, at which point the numerator approaches 0, and the final calculated load-mode coupling anomaly degree is obtained. It will also approach 0.

[0084] As can be seen from the above calculation process, the algorithm of the present invention can transform complex physical state assessment into intuitive numerical indicators, effectively identify abnormal states where geometric compression features and mechanical texture features do not match, thereby achieving accurate detection of hidden defects such as voids.

[0085] S4. Compare the asymmetric distortion index with a preset distortion judgment threshold, and compare the load morphology coupling anomaly degree with a preset anomaly degree safety threshold; determine whether the rubber bearing has defects and the type of defects based on the comparison results.

[0086] Specifically, the system first sets an anomaly safety threshold. If the calculated load shape coupling anomaly degree Greater than the anomaly safety threshold or asymmetric distortion index If the distortion exceeds the distortion threshold, the rubber support is determined to be defective.

[0087] Based on this, the system further outputs the specific defect type according to the data characteristics. For example, the asymmetric distortion index... If the distortion exceeds the distortion threshold (for example, if the preset value is set to 0.2), the defect type is determined to be installation misalignment or structural distortion. This determination means that the support may have been improperly installed or that uneven foundation settlement has occurred, and the system will prompt maintenance personnel to adjust the installation position.

[0088] If the asymmetric distortion index Less than or equal to the above distortion judgment threshold, but load morphological coupling anomaly degree Greater than the anomaly safety threshold If the defect type is determined to be internal voiding or material aging, this means that although the support may appear acceptable in appearance, its internal structure is damaged and it has lost its load-bearing capacity. The system will recommend immediate replacement of the support.

[0089] It should be noted that both the distortion judgment threshold and the anomaly safety threshold are system preset values. In practical applications, the anomaly safety threshold is determined as follows: At least 50 bridge rubber bearings in normal working condition, with clean surfaces and no aging cracks, are selected as benchmark samples. These samples cover different normal compression states. The load-mode coupling anomaly degree of each benchmark sample is calculated. Since the samples are all normal bearings, their measured surface texture entropy should be highly close to the theoretical reference texture entropy. Therefore, the calculated load-mode coupling anomaly degree values ​​will be distributed in a low value range close to 0. The mean and standard deviation of this set of benchmark anomaly degree values ​​are calculated according to... The principle is to set a safety threshold for the degree of anomaly. Alternatively, the maximum load morphological coupling anomaly value calculated from the benchmark sample can be directly selected, and a margin of 10% to 20% can be added to it as the safety threshold for the degree of anomaly.

[0090] The method for determining the distortion threshold is as follows: By collecting a sample dataset of bridge rubber bearings containing both normal shear deformation and installation misalignment defect states, the geometric sensitivity coefficient is set to a constant, and the asymmetric distortion index of each sample is calculated. Statistical analysis is used to obtain the distribution boundaries of the index values ​​for the two types of samples. The value located between the maximum statistical value of the normal sample and the minimum statistical value of the abnormal sample is selected as the distortion threshold to effectively distinguish installation misalignment defects.

[0091] Through the above threshold comparison and logical classification, the system can not only issue alarms, but also provide specific fault diagnosis suggestions, which makes it easier for maintenance personnel to quickly locate the cause of the problem, thus realizing an intelligent and automated detection process.

[0092] The following combination Figures 2 to 4 The effects of the present invention will be further explained below.

[0093] Figure 2 The figure shows a scatter plot comparison of the defect identification capabilities of the present invention and existing technologies. The horizontal axis represents the side geometric tilt angle commonly used in existing technologies, and the vertical axis represents the asymmetric distortion index proposed in this invention. Figure 2 As shown, observing the distribution of data points reveals that within the horizontal axis's tilt angle range of 10 to 20 degrees, data points representing normal and abnormal samples are mixed and highly overlapping in the horizontal direction. This indicates that existing technology cannot effectively distinguish between normal shear deformation and installation misalignment based solely on geometric tilt angle. In contrast, by introducing an asymmetric distortion index onto the vertical axis, abnormal sample points representing installation misalignment or structural distortion are significantly elevated to the upper region with higher vertical axis values, while sample points representing normal shear deformation remain consistently within the safe region at the bottom of the vertical axis. This achieves clear stratification and accurate identification of the two types of samples in the data distribution.

[0094] Figure 3 This reflects the correspondence between vertical compression ratio and texture entropy. The solid line in the figure represents the theoretical baseline curve. Data points clustered around the theoretical baseline curve represent normal measured data, indicating that the deformation and texture characteristics of the support conform to physical laws. Abnormal data points distributed in the region far below the theoretical baseline curve represent internal voids or material aging defects. Although these data points show a large vertical compression ratio on the horizontal axis, their texture entropy values ​​on the vertical axis are extremely low, verifying the effectiveness of the load-shape coupling anomaly formula in capturing hidden defects.

[0095] Figure 4 The graph shows a real-time monitoring trend of the bearing asymmetric distortion index over time. The first half of the curve fluctuates smoothly in the low range, representing the bearing in normal working condition. The second half of the curve suddenly jumps and continuously exceeds the set danger alarm threshold, corresponding to the period when the system automatically triggers an alarm. This trend graph demonstrates that the present invention can provide real-time and sensitive early warning for sudden changes in the bearing's condition, such as impact or loosening of fixation.

[0096] Specific embodiments of the bridge rubber bearing appearance defect detection system proposed in this invention:

[0097] The bridge rubber bearing appearance defect detection system includes a processor and a memory. The memory stores computer program instructions. When the computer program instructions are run by the processor, the system executes the bridge rubber bearing appearance defect detection methods in the above embodiments.

[0098] The bridge rubber bearing appearance defect detection system also includes other components well known to those skilled in the art, such as communication buses and communication interfaces. Their settings and functions are known in the art and will not be described in detail here.

[0099] While various embodiments of the invention have been shown and described in this specification, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Many modifications, alterations, and alternatives will occur to those skilled in the art without departing from the spirit and essence of the invention.

Claims

1. A method for detecting appearance defects in bridge rubber bearings, characterized in that, Includes the following steps: S1. Acquire a grayscale image of the side of the rubber bearing, extract the region of interest and perform perspective transformation correction on the grayscale image to obtain a frontal projection image; S2. Perform feature extraction on the frontal projection image to obtain the tilt angle between the left and right edges of the support, the left and right heights, the vertical compression ratio, and the texture entropy reflecting the degree of wrinkling on the rubber surface. S3. Calculate the asymmetric distortion index based on the tilt angle of the left and right edges, the left height and the right height, and calculate the load morphological coupling anomaly based on the texture entropy and the vertical compression ratio. S4. Compare the asymmetric distortion index with a preset distortion judgment threshold, and compare the load morphology coupling anomaly degree with a preset anomaly degree safety threshold; determine whether the rubber bearing has defects and the type of defects based on the comparison results.

2. The method for detecting appearance defects in bridge rubber bearings according to claim 1, characterized in that, Step S1 involves extracting the region of interest and performing perspective transformation correction on the side grayscale image, including: Gaussian filtering was applied to the acquired side grayscale image, and adaptive threshold segmentation was performed using the maximum inter-class variance method to separate the black rubber support area from the background. Identify the edge lines of the upper and lower steel plates in the rubber bearing area, construct a perspective transformation matrix, and correct the side grayscale image into a frontal projection image.

3. The method for detecting appearance defects in bridge rubber bearings according to claim 2, characterized in that, Step S2 specifically includes: The left and right vertical edges of the support, as well as the horizontal edges of the upper and lower steel plates, are extracted from the frontal projection image using an edge detection operator. The angles between the left and right vertical edges and the vertical line are calculated using the least squares line fitting method, and are used as the left tilt angle and the right tilt angle, respectively. The vertical pixel distance between the upper and lower endpoints of the left vertical edge is calculated as the left height, the vertical pixel distance between the upper and lower endpoints of the right vertical edge is calculated as the right height, the average of the left height and the right height is calculated as the average pixel height, and the ratio of the actual physical height corresponding to the average pixel height to the pre-stored initial factory height is used as the vertical compression ratio. A rectangular sub-block is extracted from the central region of the front-view projected image to construct a gray-level co-occurrence matrix. The entropy value of the gray-level co-occurrence matrix is ​​calculated and normalized to obtain the texture entropy.

4. The method for detecting appearance defects in bridge rubber bearings according to claim 3, characterized in that, The formula for calculating the asymmetric distortion index is as follows: In the formula, Indicates the asymmetric distortion index; and These represent the left tilt angle and the right tilt angle, respectively. and These represent the left and right heights, respectively. This indicates the pixel width of the bottom edge of the support; This represents the geometric sensitivity coefficient.

5. The method for detecting appearance defects in bridge rubber bearings according to claim 4, characterized in that, The formula for calculating the load morphological coupling anomaly degree is as follows: In the formula, Indicates the degree of load-mode coupling anomaly; This represents the normalized texture entropy; This represents the theoretical reference texture entropy calculated based on the current vertical compression ratio; Represents the zero safety constant; Indicates the distortion weighting coefficient; is the asymmetric distortion index.

6. The method for detecting appearance defects in bridge rubber bearings according to claim 5, characterized in that, The theoretical reference texture entropy The calculation formula is as follows: In the formula, This indicates the vertical compression ratio; Indicates the material texture coefficient; This represents the nonlinear response factor.

7. The method for detecting appearance defects in bridge rubber bearings according to claim 1, characterized in that, The acquired grayscale image of the side of the rubber bearing includes: The rubber bearing was photographed using a camera installed on the side of the cap beam, along with a low-angle auxiliary light source. The side grayscale image is obtained by enhancing the shadow contrast of the rubber surface wrinkles using the side low-angle auxiliary light source.

8. The method for detecting appearance defects in bridge rubber bearings according to claim 3, characterized in that, The construction of the gray-level co-occurrence matrix includes: In the rectangular sub-block, the direction of the grayscale co-occurrence matrix is ​​set to 0 degrees to capture the texture features of the rubber bulging and wrinkling on the side after being compressed.

9. The method for detecting appearance defects in bridge rubber bearings according to claim 5, characterized in that, The step of determining whether the rubber bearing has defects and the type of defects based on the comparison results includes: If the load-mode coupling anomaly degree is greater than a preset anomaly degree safety threshold, or the asymmetric distortion index is greater than a distortion determination threshold, then the support is determined to have a defect; based on this, the defect type is determined: If the asymmetric distortion index is greater than the distortion determination threshold, the defect type is determined to be installation misalignment or structural distortion. If the asymmetric distortion index is less than or equal to the distortion determination threshold and the load morphology coupling anomaly degree is greater than the preset anomaly degree safety threshold, the defect type is determined to be internal void or material aging.

10. A system for detecting visual defects in bridge rubber bearings, characterized in that, include: The system includes a processor and a memory, the memory storing computer program instructions that, when executed by the processor, implement a method for detecting appearance defects in bridge rubber bearings as described in any one of claims 1-9.

Images