A detection system for the aging degree of cableway cables based on image recognition

Abstract

The present invention belongs to the technical field of aging detection of cableway cables, and specifically relates to a cableway cable aging degree detection system based on image recognition, including a multi-dimensional image acquisition module, a background segmentation module, a dirt separation module, a multi-modal damage recognition module, and an aging analysis feedback module. By using a drone to collect spectral images, three-dimensional surface point cloud data, and thermal imaging data of a continuous section of the cableway cable, after background segmentation and dirt separation processing, cracks are identified using spectral images, broken wires are identified using three-dimensional surface point cloud data, and hot spots are identified by combining thermal imaging data with three-dimensional surface point cloud data. On the basis of effectively excluding external interference factors, accurate detection of multi-dimensional aging signs of the cable is achieved, ensuring a high degree of reliability of the detection results. At the same time, comprehensive aging assessment is carried out using the detection results, and hierarchical maintenance is thus implemented, which can optimize resource allocation and reduce unnecessary maintenance costs.

Description

Technical Field

[0001] The present invention belongs to the technical field of aging detection of cableway cables, and particularly relates to a system for detecting the aging degree of cableway cables based on image recognition. Background Art

[0002] Cableway cables play a crucial role in modern transportation and material transportation, and are widely used in tourist attractions, ski resorts, mines, urban public transportation systems and other fields. With the influence of service time and external factors (such as climatic conditions, abrasion, corrosion, etc.), the cables will gradually age, resulting in a decline in their structural strength and safety. Therefore, the aging detection of cableway cables is crucial for ensuring the safe operation of the cableway system.

[0003] In recent years, with the progress of computer vision technology, non-contact methods for detecting the aging of cableway cables have been widely used. For example, a non-contact health detection system for mine cables proposed in Chinese Patent Publication No. CN118408943A uses a high-resolution high-speed camera to capture the appearance image of the cable, and identifies defects such as cracks and abrasions on the cable surface through image processing technology. This method significantly improves the efficiency of cable detection.

[0004] However, on the one hand, the above image-based detection method does not consider the problem of dirt interference. It is inevitable for cableway cables to accumulate dirt in the natural environment. If the dirt is not separated from the image, the dirt will cover the true defect features, making it difficult for the algorithm to accurately identify the normal area and the damaged area, and prone to false judgment or missed detection of defects. On the other hand, different types of defects (such as cracks and abrasions) have different geometric shapes and optical characteristics. Relying only on the appearance images collected by the camera to extract different types of defects may lead to lack of pertinence in detection. Specifically, due to the lack of in-depth analysis of specific defect features, the algorithm may not be able to effectively distinguish different defect types, thereby affecting the reliability of defect recognition. Summary of the Invention

[0005] The purpose of the present invention is to improve the deficiencies existing in the prior art, and provide a system for detecting the aging degree of cableway cables based on image recognition. By collecting multi-dimensional images of spectral images, three-dimensional point cloud data and thermal imaging of the cableway cables, and performing targeted recognition of different defect damages of the cables after dirt separation processing.

[0006] The purpose of the present invention can be achieved by the following technical solutions: A system for detecting the aging degree of cableway cables based on image recognition, including the following modules: a multi-dimensional image acquisition module, which collects spectral images, three-dimensional point cloud data and thermal imaging data of a continuous section of the cableway cable surface through a multi-spectral imaging device, a microscopic imaging device and a thermal imager carried by a drone.

[0007] The background segmentation module performs background segmentation on the spectral images of each cable segment to retain the cable area.

[0008] The dirt separation module divides the dirt probability area for the cable area in the spectral image of each cable segment by capturing the contact area between the cable and the mechanical components and the bent area of the cable trajectory. Then, based on the mapping relationship between the constructed dirt probability area and the polarization light scanning resolution, it distinguishes the dirt-covered area and the cable body by dynamically adjusting the polarization light resolution using the difference in polarization light reflectivity.

[0009] The multi-modal damage identification module fuses the spectral images, three-dimensional point cloud data, and thermal imaging data of the cable body corresponding to each cable segment for damage identification, specifically crack, broken wire, and hot spot identification, and quantifies the damage geometric parameters.

[0010] The aging analysis feedback module generates hierarchical maintenance suggestions based on the damage geometric parameters.

[0011] Combining all the above technical solutions, the positive effects of the present invention are as follows: 1. By using drones to collect the spectral images, three-dimensional point cloud data, and thermal imaging data of continuous segments of the cableway cable, after background segmentation and dirt separation processing, the present invention uses spectral images to identify cracks, three-dimensional point cloud data to identify broken wires, and combines thermal imaging data and three-dimensional point cloud data to identify hot spots, achieving precise detection of multi-dimensional aging signs of the cable on the basis of effectively eliminating external interference factors and ensuring the high reliability of the detection results.

[0012] 2. After the present invention precisely detects the multi-dimensional aging signs of the cable using multi-dimensional images, it conducts a comprehensive assessment of the cable aging, and based on the assessment results and preset warning thresholds, it executes hierarchical maintenance. It can trigger a shutdown and replacement instruction to ensure safety when the aging risk is high, and shorten the inspection cycle and implement load restrictions when the risk is low, thereby optimizing resource allocation, reducing unnecessary maintenance costs, and extending the service life of the cable. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] The present invention is further described with reference to the accompanying drawings, but the embodiments in the drawings do not constitute any limitation to the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained according to the following drawings.

[0014] Figure 1 It is a schematic diagram of the data flow sequence of the system modules of the present invention.

[0015] Figure 2 It is a schematic diagram of the composition of the dirt separation module in the present invention.

[0016] Figure 3 It is an operation diagram of aging analysis using multi-dimensional images of the cableway cable in the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0017] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without making creative efforts belong to the scope of protection of the present invention.

[0018] The present invention proposes a detection system for the aging degree of cableway cables based on image recognition, including a multi-dimensional image acquisition module, a background segmentation module, a dirt separation module, a multi-modal damage recognition module, and an aging analysis and feedback module.

[0019] Please refer to Figure 1 As shown, the data flow order of each module above is that data starts from the multi-dimensional image acquisition module and is sequentially transmitted to the background segmentation module, the dirt separation module, and the multi-modal damage recognition module, and finally converges into the aging analysis and feedback module.

[0020] The multi-dimensional image acquisition module collects spectral images, three-dimensional point cloud data, and thermal imaging data of a continuous section of the cableway cable surface through a multi-spectral camera device, a microscopic imaging device, and a thermal imager carried by a drone.

[0021] It should be understood that the multi-spectral camera device mentioned above is a sensor that can capture images of multiple bands simultaneously or sequentially, and can use the multi-spectral camera device to collect the cable spectral image. Since the reflection characteristics of the cracks and normal regions of the cable in the spectral image are different in different bands, the spectral image helps to distinguish the normal region from the crack region.

[0022] The microscopic imaging device refers to a high-resolution optical microscope or electron microscope system, which can capture the fine structure and detailed features of the object surface at the microscopic scale. The three-dimensional surface image of the cable surface can be collected by using the microscopic imaging device. Since the three-dimensional point cloud in the three-dimensional surface image contains rich geometric information and can accurately reflect the subtle morphological changes of the cable surface, the position, shape, and depth of the broken wires can be clearly reflected in the three-dimensional point cloud.

[0023] A thermal imager is a device used to detect the infrared radiation on the surface of an object and convert it into a visible image, generating a thermal image by measuring the temperature distribution on the surface of the object. The thermal imager can identify temperature abnormal regions caused by rust and wear. When metal rusts, an oxide layer will form on its surface. The thermal conductivity of this oxide layer is usually lower than that of the original metal material, resulting in heat accumulation in these regions instead of rapid diffusion, so that the rust region shows a higher surface temperature than the surrounding healthy regions.

[0024] In a preferred implementation of the above solution, the process of collecting spectral images, three-dimensional surface images, and thermal images of a continuous section of the cableway cable is as follows: Determine the initial flight altitude of the drone based on the technical specifications of the multispectral imaging device, microscopic imaging device, and thermal imager carried on the drone.

[0025] The above technical specifications refer to resolution, focal length, field of view angle, etc.

[0026] In the implementation of the above solution, the initial flight altitude is determined as follows: According to the requirements of the detection task, such as details like cracks and broken wires, set the pixel resolution requirements per centimeter for spectral images, three-dimensional point cloud data, and thermal images.

[0027] When the pixel resolution requirements per centimeter for spectral images, three-dimensional surface images, and thermal images are not very different, for example, the spectral image requires 5 pixels / cm; the three-dimensional surface image requires 8 pixels / cm; the thermal image requires 4 pixels / cm.

[0028] Under the above pixel resolution requirements, usually select the maximum value among the three resolutions as the unified optimal pixel number.

[0029] Based on the unified optimal pixel number and combined with the technical specifications of the multispectral imaging device, microscopic imaging device, and thermal imager, calculate the minimum flight altitude that meets this resolution requirement. The specific calculation has been detailed in the existing technology and will not be elaborated here.

[0030] If the pixel resolution requirements per centimeter for spectral images, three-dimensional surface images, and thermal images are quite different, for example, the spectral image requires 5 pixels / cm, while the three-dimensional surface image requires 20 pixels / cm, and the thermal image requires 9 pixels / cm.

[0031] In the above situation, it is difficult to find a unified optimal pixel number. At this time, calculate the minimum flight altitude for each resolution requirement respectively, and then select the maximum value as the final minimum flight altitude. This is because if a lower flight altitude is adopted, it may cause some images, especially those with high-resolution requirements, to not reach the required resolution, thus affecting the accuracy of the detection results. For example, the broken wire feature in the three-dimensional surface image may be missed or misjudged due to insufficient resolution. Although a higher flight altitude may result in the resolution of some image types being slightly higher than their minimum requirements, this strategy can ensure the data quality of all image types, reduce the repeated flights or additional inspection work caused by insufficient resolution, and thus improve the overall detection efficiency.

[0032] Divide the cable length into multiple segments, and use the geographic information system to locate the collection position of each cable segment.

[0033] In the ways that the above solution can be implemented, the cable segments can be divided at fixed length intervals or according to the topographical features of the environment where the cable is located, such as slope changes and the positions of obstacles to be crossed.

[0034] The acquisition positions of each cable segment are the shooting positions of the multispectral camera device, the microscopic imaging device, and the thermal imager. Exemplarily, the acquisition position is usually set as its geometric center point, that is, the middle position of the cable segment. This layout ensures that the images obtained by the shooting device at this position can cover the entire cable segment, thereby providing comprehensive data.

[0035] Generate a flight route through path planning based on the position information of the cable segment and the topographical features.

[0036] The drone flies along the planned route at the initial flight altitude and continuously detects the wind speed in real time through the on-board anemometer. When the wind speed reaches or exceeds the set wind speed threshold, it gradually increases the flight altitude at a fixed vertical altitude increment and synchronously locates the flight position.

[0037] Specifically, the fixed vertical altitude increment can be 5 meters.

[0038] It should be noted that when the wind speed is high, adjusting the flight altitude at a fixed increment is mainly to optimize the flight conditions. By increasing the flight altitude, the drone can enter a relatively stable air flow layer and reduce the influence of complex air flows near the ground, especially the interference of crosswinds and turbulence on the flight path.

[0039] When the real-time flight position of the drone approaches the acquisition position of a certain cable segment and the real-time wind speed is lower than the wind speed threshold, it gradually descends to the initial flight altitude and hovers after arriving at the specified cable segment acquisition position to collect spectral images, three-dimensional point cloud data, and thermal imaging data, and assigns unique identifiers to the collected spectral images, three-dimensional point cloud data, and thermal imaging data in the order of the cable segments for numbering.

[0040] It should be noted that when the real-time flight position approaches the acquisition position of the cable segment and once the wind speed is lower than the wind speed threshold, the drone gradually descends to the initial flight altitude to be closer to the cable surface for high-precision data acquisition.

[0041] It should be added that when the real-time flight position of the drone approaches the acquisition position of the cable segment, if it is detected that the wind speed is still high, the drone can be controlled to temporarily stay at the current altitude instead of continuing to fly. After the wind speed drops, it gradually descends the flight altitude for data acquisition to prevent the drone from continuing to fly under high wind speed conditions and getting away from the acquisition position, resulting in the shooting device being unable to accurately cover the target area, thus affecting the accuracy and integrity of data acquisition. Moreover, the data collected under high wind speed conditions often has poor quality and may require re-collecting data in the same area, increasing the waste of time and resources.

[0042] When the present invention performs multi-dimensional image acquisition on the cableway cable, the cable is segmented for image acquisition by using a drone. Compared with using a fixed image acquisition device for full-section acquisition, it can obtain and analyze detailed images of each section of the cable, ensure high-quality data for each part, and avoid missing details caused by overall scanning.

[0043] The background segmentation module performs background segmentation on the spectral images of each cable segment to retain the cable area.

[0044] The specific implementation process of the above solution is as follows: The spectral images of each cable segment are denoised and image-enhanced to improve the contrast between the cable and the background.

[0045] According to the spectral characteristics of the cable material, specific bands that can effectively distinguish the cable from the background are selected, and the typical gray value range of the cable material corresponding to the gray image in the selected band is extracted from the reference information library.

[0046] It should be understood that due to the differences in material properties between the cable and the background, they show significantly different spectral characteristics in the spectral response. Among them, the material of the cable is usually metal or plastic, and the background is usually vegetation. The different manifestations of specific spectral characteristics are reflected in the different reflectivities or absorptivities of each band: Metal materials usually show higher reflectivities in the visible and near-infrared bands.

[0047] Plastic materials may show strong absorption characteristics in certain infrared bands. This characteristic makes plastics appear darker in images of specific bands because they absorb more incident energy.

[0048] Vegetation shows a high reflectivity in the visible band because chlorophyll reflects green light strongly.

[0049] In the above, the gray image is selected as the carrier for background segmentation because the gray image is a single-channel image, and each pixel point is represented by a numerical value, which makes the segmentation more efficient, and the differences in reflectivity or absorptivity of different materials in specific bands can be directly mapped to the brightness values in the gray image. By generating a gray image in a specific band, the spectral differences between materials can be observed and analyzed more intuitively.

[0050] It should be added that the typical gray value range of the cable of different materials corresponding to the gray image in the selected band stored in the above-mentioned reference information library comes from systematic laboratory tests on cable samples of various known materials.

[0051] The preprocessed spectral image is converted into a gray image, and the gray value of each pixel point in the selected band is calculated.

[0052] Compare the gray value of each pixel with the typical gray value range, thereby identifying the set of pixels belonging to the cable.

[0053] Use edge detection to identify the regions with significant brightness changes in the image after identifying the cable pixels, and extract the cable contour boundary from them.

[0054] It should be explained that there are significant differences in the reflectivity or absorptivity of the cable and the background in a specific wavelength band. This difference in material properties is directly mapped into the gray image, manifested as different brightnesses. In the selected wavelength band, the pixels belonging to the cable usually fall within its typical gray value range, while the background pixels may show different gray value distributions. Therefore, in the gray image, an obvious brightness boundary will be formed between the cable and the background.

[0055] In the specific implementation process of the above solution, the core of identifying the regions with significant brightness changes lies in calculating the gradient value of each pixel in the image. The gradient is defined as the gray value change rate between a certain pixel in the image and its neighboring pixels. A high gradient value indicates that there is a large brightness change at that position, usually corresponding to the edge or boundary region of an object. Thus, the gradient value can be compared with the configured threshold to screen out the regions with significant brightness changes.

[0056] According to the determined cable contour boundary, use connected component analysis to mark all regions outside the boundary as background regions, and remove them from the original image, retaining the cable region.

[0057] The present invention separates the background from the collected spectral image of the cable segment, retains the cable region, and then separates the dirt, ensuring that subsequent processing is only for the cable itself. And background separation reduces the amount of data to be processed, making subsequent dirt separation and other processing steps more efficient. It reduces unnecessary interference factors and improves the accuracy of cable status assessment.

[0058] The dirt separation module divides the dirt probability distribution regions for the cable regions in each cable segment spectral image by capturing the contact regions between the cable and the mechanical components and the bent regions of the cable trajectory. Then, based on the mapping relationship between the constructed dirt probability distribution regions and the polarization light scanning resolution, by dynamically adjusting the polarization light resolution, it uses the difference in polarization light reflectivity to distinguish the dirt-covered regions from the cable body.

[0059] In the improved implementation of the above module, as shown in Figure 2 The dirt separation module includes a dirt distribution feature capture unit, a resolution-probability mapping unit, and a dirt-body separation unit.

[0060] Specifically, the content of the dirt distribution feature capture unit is as follows: extract the structural features of each mechanical component based on a predefined information library of cableway components, and thus extract and identify the structural features of the mechanical components from the cable areas of each cable segment.

[0061] It should be noted that the cableway cable system not only includes the cable body, but also covers a variety of mechanical components, such as pulleys, support wheels, and fixing clips.

[0062] If the structural features of the mechanical components are recognized, mark the position area of the components in the cable main body area, and calculate the contact area between the cable and the components by analyzing the contour information of this area.

[0063] Extract the cable trajectory from the spectral images of each cable segment to identify whether it is bent. If there is a bend, mark the corresponding bent area in the spectral image and measure the bending angle using geometric analysis.

[0064] It should be understood that the positioning and identification of the contact area between the cable and the components and the bent area of the cable body in the spectral images of each cable segment are mainly because these areas are prone to accumulating dirt. Specifically, at the connection between the cable and the components, due to frequent friction and mechanical contact, lubricating grease, metal debris, or other impurities may accumulate. At the bent part of the cable body, due to frequent deformation, the surface material becomes fatigued, making it easier to adsorb dust, impurities in rainwater, or pollutants in the air.

[0065] The content of the resolution-probability mapping unit is as follows: according to the contact area corresponding to the marked contact area and the bending angle corresponding to the bent area in the spectral images of each cable segment, divide the dirt probability distribution area according to the following rules: High probability area: a) The contact area and the bent area coexist, and either the proportion of the contact area or the bending angle exceeds the allowable threshold.

[0066] It should be added that the proportion of the contact area mentioned above refers to the ratio of the contact area to the cable area.

[0067] b) The contact area and the bent area do not coexist, but the corresponding index exceeds the allowable threshold.

[0068] This grading method emphasizes the risk superposition effect and the significance of a single index exceeding the standard.

[0069] Medium probability area: a) The contact area and the bent area coexist, but both the proportion of the contact area and the bending angle do not exceed the allowable threshold.

[0070] b) The contact area and the bent area do not coexist, but the corresponding index does not exceed the allowable threshold.

[0071] Low probability area: Other areas except the contact area and the bent area.

[0072] These areas generally do not have obvious mechanical stress concentration or friction characteristics, and the possibility of dirt accumulation is the lowest.

[0073] It should be noted that the contact area and the bending area are the main risk points for dirt accumulation on the cable surface. By introducing the contact area ratio and the bending angle as quantitative indicators, the dirt accumulation probability of these areas can be objectively evaluated. The introduction of the allowable threshold makes the division of the dirt probability area have a clear standard, avoiding the influence of subjective judgment. The allowable threshold can be adjusted according to historical data or experimental results to ensure the scientific nature and applicability of the division.

[0074] The present invention divides the dirt probability area of the cable area in the corresponding spectral image of the cable segment by analyzing the contact area between the cable and the mechanical component and the bending angle of the cable. This method provides a targeted direction for subsequent identification of dirt using polarized light scanning according to different probability areas, which helps to improve the regional accuracy of dirt scanning.

[0075] Based on the light sheet parameters of the polarization filter, a scanning resolution range covering from the lowest to the highest is determined, and the upper limit value, the middle value, and the lower limit value of the scanning resolution interval are taken as the high resolution, the medium resolution, and the low resolution.

[0076] It should be added that the polarization filter is an optical element that can selectively filter light of a specific polarization direction. When identifying dirt, the polarization filter is used as an identification tool because the dirt-covered area and the cable body area show different light reflection characteristics due to differences in material and surface state. This difference enables effective differentiation between the two by adjusting the polarization angle. Specifically, at a high polarization angle: specular reflection is significantly reduced, allowing more scattered light to enter the imaging system. This not only enhances the contrast between the dirt and the background but also improves the resolution of the imaging system, helping to capture the detailed features of the dirt.

[0077] At a low polarization angle: more specular reflection light is retained, reducing the proportion of scattered light, resulting in a decrease in the resolution of the imaging system. However, this method is suitable for quickly obtaining the overall contour information of the cable.

[0078] The determination of the scanning resolution range of the polarization filter mentioned above is based on the light sheet parameters through actual experimental measurements. Image samples are obtained at different polarization angles, and the resolution performance of these images is analyzed. Then, based on the experimental data, the resolution range of the imaging system at different polarization angles is statistically calculated, and the upper and lower limit values and the middle value are determined.

[0079] In a specific operation example, the resolution range of the imaging system at different polarization angles is determined as follows: The initial polarization filter angle is set to 0°, and the baseline resolution is obtained.

[0080] Adjust the angle of the polarization filter at a fixed step, such as rotating 5° each time, and record the image resolution after each adjustment.

[0081] Compare the image clarity and detail performance at different angles to determine the polarization angles corresponding to the lowest, middle, and highest resolutions.

[0082] Establish the mapping relationship between the dirt probability distribution area and the polarization light scanning resolution as follows: High probability area: Use high-resolution scanning.

[0083] Medium probability area: Use medium-resolution scanning.

[0084] Low probability area: Use low-resolution scanning.

[0085] It should be understood that for the high probability area, high-resolution scanning is used to ensure that the most subtle dirt features can be captured and the most detailed data support can be provided. For the medium probability area, medium-resolution scanning not only ensures sufficient detail information but also takes into account the data processing efficiency. For the low probability area, low-resolution scanning is used to reduce unnecessary consumption of computing resources and improve the overall detection efficiency.

[0086] The present invention uses the light reflection difference between dirt and the cable body identified by polarized light as the dirt coverage detection principle, and dynamically allocates corresponding scanning resolutions according to different dirt probability distribution areas, which can concentrate computing resources in high-priority areas, avoiding the waste of computing time and resources caused by using high-resolution scanning without zoning, thereby improving the overall recognition efficiency.

[0087] The content of the dirt-body separation unit is as follows: Rotate the polarization filter according to the mapped polarization light scanning resolution for the dirt probability regions divided on the spectral image of each cable segment to obtain the reflected light intensity sequence at the corresponding resolution, and calculate the degree of polarization therefrom.

[0088] The above calculation of the degree of polarization using the reflected light intensity sequence is based on the mature polarized optical theory, and the specific calculation has been introduced in detail in the prior art and will not be elaborated here.

[0089] Compare the calculated degree of polarization with the configured boundary threshold, which is determined based on previous experiments and represents the key boundary for distinguishing the cable body from the dirt-covered area. If the degree of polarization of a certain dirt probability region is lower than the boundary threshold, it is determined that the region is a dirt-covered area; otherwise, it is determined that the region is a cable body area.

[0090] In the above comparison using the degree of polarization and the threshold, the distribution area with the degree of polarization lower than the threshold is determined as the dirt-covered area because dirt usually includes impurities such as dust, grease, and rust. The surfaces of these substances are relatively rough and irregular, resulting in more scattered light rather than specular reflection. Scattered light is usually unpolarized or weakly polarized, so its degree of polarization is low. In contrast, the cable body area has a relatively smooth surface, and the reflected light is mainly specular reflection light with a high degree of polarization.

[0091] During the innovative implementation of the above solution, once the dirt-covered area is determined, the system will automatically generate a corresponding decontamination mask. This mask is used to accurately mark the position of the dirt in the original spectral image, thus providing precise positioning for subsequent cleaning work. In addition, this mask can also be used to enhance or restore the cable body features blocked by dirt, further improving the detection accuracy.

[0092] The multi-modal damage recognition module is used to fuse the spectral images, three-dimensional point cloud data, and thermal imaging data of the cable body corresponding to each cable segment for damage recognition, specifically crack, broken wire, and hot spot recognition, and quantify the damage geometric parameters.

[0093] In an alternative implementation of the above solution, based on the cable body area obtained after dirt separation in the spectral image, by registering and fusing the data of the same area in the three-dimensional surface image and the thermal image, the cable body area can be accurately extracted in the three-dimensional surface image and the thermal image.

[0094] The process of crack recognition in the multi-modal damage recognition module is as follows: Select specific bands that can effectively distinguish split cracks from normal areas according to the spectral characteristics of the cable material.

[0095] The above steps involve analyzing the differences in reflectivity or absorptivity between the cable surface cracks and normal areas in different bands to determine the most suitable band for crack detection. These bands usually have a high contrast, which can clearly highlight the presence of cracks.

[0096] Convert the spectral images of the cable body corresponding to each cable segment in the selected band into grayscale images.

[0097] Calculate the gradient field of the image on the converted grayscale image, compare the gradient of each pixel in the gradient field with a preset limit value, screen out the pixel positions with gradient values higher than the limit value, and mark them as crack areas.

[0098] It should be noted that the gradient is the rate of change of the gray value between a certain pixel point in the image and its neighboring pixel points. Cracks usually appear as local gray mutation regions in the image. Therefore, by calculating the gradient field, these changes can be effectively detected, and then the cracks can be located. The selection of the limit value is usually based on a large amount of experimental data and statistical analysis. By testing known crack samples, a reasonable threshold range can be determined so that the threshold can effectively distinguish between crack and non-crack regions.

[0099] Extract geometric features for the located crack region. Specifically, the geometric features can be length, width, etc. Among them, the length and width can be obtained by measuring the maximum extension distance of the crack as the length and measuring the distance at the widest part as the width.

[0100] Furthermore, the process of broken wire recognition in the multi-modal damage recognition module is as follows: Remove noise and smooth the three-dimensional point cloud data corresponding to each cable segment of the cable body.

[0101] The above steps are mainly to remove isolated noise points to reduce the errors introduced in subsequent operations.

[0102] Calculate the local normal vector of each point for the processed three-dimensional point cloud data.

[0103] The above operation aims to analyze the point distribution in the local neighborhood to determine the direction information of each point.

[0104] Taking each point as the center, set a fixed radius range to define the neighborhood area of this point. All adjacent points around the center point are included in this neighborhood area.

[0105] Compare the normal vector of each point with the average value of the normal vectors of all points in its neighborhood. If the difference between the normal vector of a certain point and the average value of the normal vectors in the neighborhood area is higher than the critical value, then mark this point as a broken wire candidate point.

[0106] Apply a clustering algorithm to cluster the broken wire candidate points together to form a broken wire region.

[0107] In the above, the difference between each point and the average value of the normal vectors of all points in its neighborhood is used as the basis for identifying broken wire candidate points. The reason is that the normal vector distribution on the surface of a normal cable is relatively consistent due to its continuity and uniformity characteristics. In contrast, due to the geometric mutation in the broken wire region, there will be a significant deviation in the local normal vector.

[0108] The geometric parameters of the broken wire region are quantified as follows: Fit a curve along the cable direction through the broken wire region and calculate the length of this curve as the broken wire length.

[0109] Perpendicular to the cable direction, find the farthest point pair on the boundary within the broken wire region, and take the distance between the farthest point pair as the maximum width of the broken wire.

[0110] The depth of the broken wire is obtained by comparing the height difference between the broken wire area and the adjacent area. The specific method is to calculate the average height of the points in the broken wire area and compare it with the average height of the surrounding normal area, and the difference between the two is the depth of the broken wire.

[0111] The process of hot spot identification in the multi-modal damage identification module is as follows: The three-dimensional point cloud data of each cable segment corresponding to the cable body is corrected and denoised to ensure the accuracy of the temperature value.

[0112] The registration algorithm is used to align the three-dimensional point cloud data with the thermal imaging data, which ensures the spatial consistency between the two data sets, enabling the temperature information of each point to be accurately mapped to the corresponding three-dimensional coordinates.

[0113] The corrected and registered thermal imaging data is mapped onto the three-dimensional surface point cloud, and the corresponding temperature value is assigned to each point.

[0114] According to the set temperature difference threshold, the points with temperatures significantly higher than the adjacent area are identified as hot spots.

[0115] The clustering algorithm is applied to cluster adjacent hot spots together to form a continuous hot spot area.

[0116] For the identified hot spot area, edge detection is used to delimit its boundary, and geometric parameters are extracted therefrom. The geometric parameters include length, width, depth, etc. Among them, the length measurement is to fit a curve along the cable direction through the hot spot area and calculate the length of this curve as the length of the hot spot area.

[0117] The width measurement is perpendicular to the cable direction. The farthest point pair on the boundary is found within the hot spot area, and the distance between them is calculated as the maximum width of the hot spot.

[0118] The height measurement can estimate the height difference between the hot spot area and the surrounding normal area in combination with the height information of the three-dimensional point cloud, and this is used as the standard for depth estimation.

[0119] The aging analysis feedback module generates hierarchical maintenance suggestions based on the damage geometric parameters. The specific content is as follows: The aging degree of the cable is comprehensively evaluated based on the damage geometric parameters corresponding to each cable segment.

[0120] As an embodiment of the above solution, the assessment of the aging degree of the cable can be carried out by customizing or assigning weights to various types of damages such as cracks, broken wires, and hot spots based on historical data. After converting the geometric parameters of each cable segment corresponding to various types of damages into damage area or damage volume and performing normalization processing, the specific aging degree of each cable segment is calculated by weighted average using the normalized damage area or damage volume and the weights of the corresponding type of damage. Finally, the maximum aging degree is taken as the final aging degree. This method ensures that the most severely damaged area is fully emphasized, thus more accurately reflecting the overall health status of the cable.

[0121] Define the high warning threshold and low warning threshold for the aging degree. The high warning threshold indicates that the cable is approaching the limit of its service life and there is a relatively high safety risk. The low warning threshold indicates that although the cable does not reach the replacement standard, the monitoring frequency needs to be increased and preventive measures need to be taken to delay further deterioration.

[0122] Compare the aging degree of the cable with the high warning threshold. If the aging degree of the cable reaches the high warning threshold, a cable replacement warning is triggered and a shutdown instruction is generated. This warning signal prompts the operator to arrange a cable replacement plan as soon as possible to avoid potential safety accidents. If the aging degree of the cable does not reach the high level but reaches the low warning threshold, the cable inspection cycle is shortened and the load limit is implemented in association with the ropeway load. This ensures that even under normal conditions, the cable can receive sufficient attention and maintenance, thus maintaining long-term stable operation.

[0123] If the aging degree of the cable is lower than the low warning threshold, the original cable inspection cycle is maintained to ensure that it is continuously monitored under normal conditions.

[0124] An exemplary operation for implementing the load limit in association with the ropeway load in the above is: reducing the maximum number of passengers or the weight of goods per run according to a set ratio to reduce the working load of the cable and extend its service life.

[0125] For the aging analysis operation of the ropeway cable using multi-dimensional images, please refer to Figure 3 as shown.

[0126] After the present invention accurately detects the multi-dimensional aging signs of the cable using multi-dimensional images, it conducts a comprehensive assessment of the cable aging, and performs hierarchical maintenance based on the assessment results and preset warning thresholds. It can trigger a shutdown and replacement instruction by the system to ensure safety when the aging risk is high, and shorten the inspection cycle and implement load limit when the risk is low, thereby optimizing resource allocation, reducing unnecessary maintenance costs, and extending the service life of the cable.

[0127] The above embodiments can be implemented in whole or in part by software, hardware, firmware, or any other combination. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product.

[0128] Those of ordinary skill in the art can realize that the modules and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of this application.

[0129] In addition, the functional modules in each embodiment of this application can be integrated into a processing module, or each module can exist physically alone, or two or more modules can be integrated into one module.

[0130] As mentioned above, this is only the specific implementation manner of this application, but the protection scope of this application is not limited thereto. Any person skilled in the art within the technical scope disclosed in this application can easily think of changes or substitutions, which should all be covered by the protection scope of this application. Therefore, the protection scope of this application should be subject to the protection scope of the claims.

[0131] Finally, the above are only the preferred embodiments of the present invention and are not used to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention shall be included in the protection scope of the present invention.

Claims

1. A detection system for the aging degree of a cableway cable based on image recognition, characterized in that Including: A multi-dimensional image acquisition module that acquires spectral images, three-dimensional point cloud data, and thermal imaging data of a continuous section of the surface of the cableway cable through a multi-spectral camera device, a microscopic imaging device, and a thermal imager carried by a drone; A background segmentation module that performs background segmentation on the spectral images of each cable section to retain the cable area; A dirt separation module that divides the dirt probability area for the cable area in the spectral image of each cable section by capturing the contact area between the cable and the mechanical components and the bent area of the cable trajectory, and differentiates the dirt-covered area and the cable body based on the polarization light reflectivity difference by dynamically adjusting the polarization light resolution according to the mapping relationship between the constructed dirt probability area and the polarization light scanning resolution; A multi-modal damage identification module that fuses the spectral images, three-dimensional point cloud data, and thermal imaging data corresponding to the cable body of each cable section for damage identification, specifically crack, broken wire, and hot spot identification, and quantifies the damage geometric parameters; The crack identification process is as follows: Select a specific band that can effectively distinguish split cracks from normal areas according to the spectral characteristics of the cable material; Convert the spectral image of the cable body area corresponding to each cable section in the selected band into a grayscale image; Calculate the gradient field of the image on the converted grayscale image, compare the gradient of each pixel in the gradient field with a preset limit value, filter out the pixel positions with gradient values higher than the limit value, and mark them as crack areas; The broken wire identification process is as follows: Perform noise removal and smoothing processing on the three-dimensional point cloud data corresponding to the cable body of each cable section; Calculate the local normal vector of each point for the processed three-dimensional point cloud data; Set a fixed radius range centered on each point to define the neighborhood area of the point; Compare the normal vector of each point with the average value of the normal vectors of all points in its neighborhood. If the difference between the normal vector of a certain point and the average value of the normal vectors in the neighborhood area is higher than the critical value, mark this point as a broken wire candidate point; Apply a clustering algorithm to cluster the broken wire candidate points together to form a broken wire area; The hot spot identification process is as follows: Perform calibration and denoising processing on the three-dimensional point cloud data corresponding to the cable body of each cable section; Use a registration algorithm to align the three-dimensional point cloud data with the thermal imaging data; Map the calibrated and registered thermal imaging data onto the three-dimensional surface point cloud, and assign a corresponding temperature value to each point; Identify the points with temperatures significantly higher than the neighborhood area as hot spots according to the set temperature difference threshold; Apply a clustering algorithm to cluster adjacent hot spots together to form a hot spot area; An aging analysis feedback module that generates hierarchical maintenance suggestions based on the damage geometric parameters.

2. The cableway cable aging degree detection system based on image recognition according to claim 1, wherein: The multi-dimensional image acquisition module includes the following: Determine the initial flight altitude of the drone according to the technical specifications of the multi-spectral camera device, the microscopic imaging device, and the thermal imager; Divide the cable length into multiple segments, and use a geographic information system to locate the acquisition position of each cable segment; Generate a flight route through path planning based on the position information of the cable segment and the terrain features; The drone flies along the planned route at the initial flight altitude, and uses an on-board anemometer to detect the wind speed in real time. When the wind speed reaches or exceeds the set wind speed threshold, it gradually increases the flight altitude according to a fixed vertical altitude increment, and synchronously locates the flight position; When the real-time flight position of the UAV approaches the collection position of a cable segment and the real-time wind speed is lower than the wind speed threshold, it gradually descends to the initial flight altitude and hovers after arriving at the designated cable segment collection position to collect spectral images, three-dimensional point cloud data and thermal imaging data.

3. The cableway cable aging degree detection system based on image recognition according to claim 1, wherein: The background segmentation module is implemented as follows: The spectral images of each cable segment are subjected to denoising and image enhancement processing; According to the spectral characteristics of the cable material, a specific band that can effectively distinguish the cable from the background is selected, and the typical grayscale value range of the cable material corresponding to the grayscale image in the selected band is extracted from the reference information library; Convert the preprocessed spectral image into a grayscale image and calculate the grayscale value of each pixel in the selected band; Compare the grayscale value of each pixel with the typical grayscale value range, thereby identifying the pixel set belonging to the cable; Use edge detection to identify areas with significant brightness changes in the image after the cable pixel identification is completed, and extract the cable contour boundary from it; According to the determined cable contour boundary, connected domain analysis is used to mark all areas outside the boundary as background areas, which are removed from the original image and the cable area is retained.

4. The cable rope aging degree detection system based on image recognition according to claim 1, wherein: The dirt separation module includes a dirt distribution feature capture unit, a resolution-probability mapping unit and a dirt-body separation unit, wherein the contents of the dirt distribution feature capture unit are as follows: Extracting and identifying structural features of mechanical components from the cable area of each cable segment according to a predefined cableway component information library; After identifying the structural features of the mechanical components, the locations of these components are marked in the cable area, and the contour information of these areas is analyzed to calculate the contact area between the cable and the components; The cable trajectory is extracted from the cable area of each cable segment to identify whether it is bent. If there is a bend, the corresponding bend area is marked, and the bending angle is determined by geometric analysis.

5. The cableway cable aging degree detection system based on image recognition according to claim 4, wherein: The content of the resolution-probability mapping unit is as follows: According to the dirt distribution characteristics, the contact area and contact area, bending area and bending angle of the unit mark are captured, and the dirt probability area is divided according to the following rules: High probability area: a) The contact area and the bending area coexist, and either the contact area ratio or the bending angle exceeds the allowable threshold; b) The contact area and the bending area do not coexist, but the corresponding index exceeds the allowable threshold; Medium probability area: a) The contact area and the bending area coexist, but the contact area ratio and the bending angle do not exceed the allowable threshold; b) The contact area and the bending area do not coexist, but the corresponding index does not exceed the allowable threshold; Low probability area: other areas except contact area and bending area; Determine a scanning resolution range covering from the lowest to the highest based on the optical sheet parameters of the polarizing filter, and take the upper limit value, the middle value and the lower limit value of the scanning resolution range as the high resolution, the medium resolution and the low resolution; The mapping relationship between the dirt probability area and the polarization light scanning resolution is established as follows: High probability areas: use high-resolution scanning; Medium probability area: use medium resolution scanning; Low probability area: scan with low resolution.

6. The cableway cable aging degree detection system based on image recognition according to claim 5, characterized in that: The contents of the dirt-body separation unit are as follows: By rotating a polarization filter according to the dirt probability regions divided on the cable regions of each cable segment, a sequence of reflected light intensities at the corresponding resolution is obtained, and the degree of polarization is calculated therefrom. The calculated degree of polarization is compared with the configured boundary threshold. If the degree of polarization of a certain dirt probability region is lower than the boundary threshold, it is determined that the region is a dirt-covered region; otherwise, it is determined that the region is the cable body region.

7. The cableway cable aging degree detection system based on image recognition according to claim 1, characterized in that: The generation of graded maintenance suggestions according to the damage geometric parameters is as follows: Comprehensively evaluate the aging degree of the cable based on the damage geometric parameters corresponding to each cable segment; Define the high-level warning threshold and the low-level warning threshold for the aging degree; Compare the cable aging degree with the high-level warning threshold. If the cable aging degree reaches the high-level warning threshold, trigger a cable replacement warning and generate a shutdown instruction. If the cable aging degree does not reach the high level but reaches the low-level warning threshold, shorten the cable inspection cycle and implement load restriction in association with the cableway load. If the cable aging degree is lower than the low-level warning threshold, maintain the original cable inspection cycle.

Images