Slope stability monitoring method and system based on machine vision

Abstract

The invention discloses a side slope stability monitoring method and system based on machine vision, and the method comprises the steps: obtaining a standard side slope image of a target visual angle and a corresponding current real-time side slope image at a target position; performing edge feature recognition to obtain edge feature points, and constructing a first polygon and a second polygon; deformation information is obtained through calculation, and whether displacement exists in the slope or not is judged based on the deformation information; when the slope has displacement, acquiring a real-time multispectral image of the slope through an unmanned aerial vehicle based on a preset slope scanning path; carrying out slope surface feature recognition on the multispectral image, and determining the displacement direction of the slope; and based on the displacement direction, continuously calculating the displacement rate of the slope in a preset period, and determining the slope stability grade according to the displacement rate. The technical problems that in the prior art, the slope monitoring stability is low, the matching success rate is low, and full-natural landforms cannot be recognized depending on artificial texture features are solved.

Description

A Machine Vision-Based Method and System for Monitoring Slope Stability Technical Field

[0001] This invention relates to the field of machine vision image data processing technology, and in particular to a method and system for monitoring slope stability based on machine vision. Background Technology

[0002] Slope stability is a critical issue in fields such as geological engineering, civil engineering, and mining. Slope instability can lead to natural disasters such as landslides and collapses, causing huge casualties and economic losses. Traditional slope monitoring methods mainly rely on manual inspections and sensors (such as displacement gauges and inclinometers). However, deep learning is now being used to monitor slopes.

[0003] Currently, deep learning-based crack identification and change detection models require extensive labeled data for training. However, "abnormal" samples of slope disasters (such as images of landslide processes) are extremely rare, making the models prone to overfitting and exhibiting weak generalization ability. For slopes with different geological, geomorphological, and vegetation conditions, the models may need to be retrained or adjusted. Furthermore, traditional feature point matching algorithms for slopes rely on rich and unchanging artificial texture features. For slopes with entirely natural terrain, homogeneous rock, and scarce texture, feature points are sparse and unstable, resulting in a low matching success rate. Summary of the Invention

[0004] This invention provides a slope stability monitoring method and system based on machine vision to solve the technical problems of low slope monitoring stability, low matching success rate, and inability to identify natural landforms by relying on artificial texture features in the prior art.

[0005] To address the aforementioned technical problems, embodiments of the present invention provide a slope stability monitoring method based on machine vision, comprising:

[0006] At the target location, acquire a standard slope image from the target perspective and a corresponding real-time slope image; perform edge feature recognition on the standard slope image and the real-time slope image to obtain edge feature points, and construct a first polygon and a second polygon corresponding to the standard slope image and the real-time slope image, respectively, based on the edge feature points; calculate deformation information based on the first polygon and the second polygon, and determine whether the slope has displacement based on the deformation information; if the slope has displacement, acquire real-time multispectral images of the slope using a drone based on a preset slope scanning path; perform slope surface feature recognition on the multispectral images, and determine the displacement direction of the slope based on the slope surface features and the deformation information; continuously calculate the slope displacement rate within a preset period based on the displacement direction, and determine the slope stability level based on the displacement rate to complete slope stability monitoring.

[0007] As a preferred embodiment, acquiring a standard slope image from the target perspective and a corresponding real-time slope image at the target location specifically includes: locating the current position of the UAV and calculating the distance and orientation between the UAV and the target location; controlling the UAV to reach the target location based on the distance and orientation between the UAV and the target location; after the UAV reaches the target location, adjusting the UAV's shooting angle according to the target perspective and acquiring the current real-time slope image; and acquiring the standard slope image corresponding to the target location and the target perspective from a preset database.

[0008] As a preferred embodiment, the step of performing edge feature recognition on the standard slope image and the real-time slope image to obtain edge feature points, and constructing a first polygon and a second polygon corresponding to the standard slope image and the real-time slope image respectively based on the edge feature points, specifically includes: performing edge feature recognition on the standard slope image and the real-time slope image to obtain edge features; identifying abrupt change points on the identified edge features to obtain edge feature points; and connecting the edge feature points on the standard slope image and the real-time slope image to obtain a first polygon and a second polygon corresponding to the standard slope image and the real-time slope image respectively.

[0009] As a preferred embodiment, the step of performing edge feature recognition on the standard slope image and the real-time slope image to obtain edge features specifically includes: performing edge feature recognition on the standard slope image and the real-time slope image, and extracting vegetation cover lines, soil-rock edge lines, and soil-rock crack lines from the standard slope image and the real-time slope image respectively; performing noise filtering on the vegetation cover lines, soil-rock edge lines, and soil-rock crack lines to obtain the final vegetation edge features, soil-rock edge features, and crack edge features; wherein, the edge features include vegetation edge features, soil-rock edge features, and crack edge features.

[0010] As a preferred embodiment, the step of identifying abrupt change points of the identified edge features to obtain edge feature points specifically includes: determining the slope of the change of the edges of the vegetation edge features, rock and soil edge features, and crack edge features respectively, and determining the abrupt change points corresponding to the vegetation edge features, rock and soil edge features, and crack edge features based on the slope of the change; and taking the abrupt change points where the vegetation edge features, rock and soil edge features, and crack edge features overlap in pairs or all three as edge feature points.

[0011] As a preferred embodiment, the step of calculating deformation information based on the first polygon and the second polygon, and determining whether there is displacement of the slope based on the deformation information, specifically includes: calculating the number of edge feature points of the first polygon and the second polygon; when the number of edge feature points of the first polygon is the same as the number of edge feature points of the second polygon, then the first polygon and the second polygon are projected to coincide in a way that the edges coincide, the non-coincident edges are identified, and the deformation information of the edges of the non-coincident first polygon and the non-coincident second polygon is calculated, and it is determined that there is displacement of the slope; when the number of edge feature points of the first polygon is the same as the number of edge feature points of the second polygon, the first polygon and the second polygon are projected to ... deformation information of the edges of the first polygon and the second polygon is calculated, and it is determined that there is displacement of the slope. If the number of edge feature points is different, the first polygon and the second polygon are projected to overlap in the manner that maximizes the number of overlapping edge feature points. Non-overlapping edge feature points are identified, and the vector data between the non-overlapping edge feature points of the first polygon and the non-overlapping edge feature points of the second polygon is calculated as deformation information, and it is determined whether the slope has displacement. When the number of edge feature points of the first polygon and the number of edge feature points of the second polygon are the same, and no non-overlapping edge feature points are identified when the first polygon and the second polygon are overlapped, the deformation data of the first polygon and the second polygon is zero, and it is determined whether the slope has displacement.

[0012] As a preferred embodiment, the step of identifying slope surface features from the multispectral image and determining the displacement direction of the slope based on the slope surface features and the deformation information specifically includes: identifying slope surface features from the multispectral image to obtain the features of the slope surface; wherein, the slope surface features include: vegetation features, rock and soil features, and crack features; and determining the displacement direction of the slope based on the vegetation features, rock and soil features, and crack features, combined with the deformation information.

[0013] This embodiment also provides a slope stability monitoring system based on machine vision, including: an image acquisition module, an edge recognition module, a deformation displacement module, a multispectral imaging module, a displacement direction module, and a stability module; the acquisition module is used to acquire a standard slope image from a target perspective and a corresponding real-time slope image at a target location; the edge recognition module is used to perform edge feature recognition on the standard slope image and the real-time slope image to obtain edge feature points, and based on the edge feature points, construct a first polygon and a second polygon corresponding to the standard slope image and the real-time slope image, respectively; the deformation displacement module is used to, according to the first... The system uses a first polygon and a second polygon to calculate deformation information and determines whether the slope has displacement based on the deformation information. A multispectral imaging module is used to acquire real-time multispectral images of the slope via a drone based on a preset slope scanning path when displacement exists. A displacement direction module is used to identify slope surface features from the multispectral images and determine the displacement direction of the slope based on the slope surface features and the deformation information. A stability module is used to continuously calculate the slope displacement rate within a preset period based on the displacement direction and determine the slope stability level based on the displacement rate, thus completing slope stability monitoring.

[0014] This embodiment also provides a terminal device, which includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements the machine vision-based slope stability monitoring method as described above.

[0015] This embodiment also provides a computer-readable storage medium, which includes a stored computer program, wherein the computer program, when running, controls the device where the computer-readable storage medium is located to execute the slope stability monitoring method based on machine vision as described above.

[0016] Compared with existing technologies, the embodiments of the present invention have the following beneficial effects: The technical solution of the present invention acquires a standard slope image from the target perspective and a corresponding real-time slope image at the target location, performs edge feature recognition, thereby identifying edge feature points, and constructs a first polygon and a second polygon corresponding to the standard slope image and the real-time slope image, respectively, to calculate deformation information and determine whether the slope has displacement. It can directly judge displacement based on the polygon information of edge feature points on different images, avoiding the displacement information shift caused by slopes with different geological, geomorphological, and vegetation conditions, which would cause interference. At the same time, after displacement occurs, real-time multispectral images of the slope are acquired by UAV based on a preset slope scanning path to determine the slope surface features, and the slope displacement direction is determined by combining deformation information. Thus, the slope displacement rate is continuously calculated in a preset period, which efficiently monitors slope stability, avoids being limited to the recognition and monitoring of artificial texture features, and improves the applicability and accuracy of slope recognition. Attached Figure Description

[0017] Figure 1: A flowchart of the steps of a slope stability monitoring method based on machine vision provided in an embodiment of the present invention; Figure 2: A structural diagram of a slope stability monitoring system based on machine vision provided in an embodiment of the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] Example 1 refers to Figure 1, which is a slope stability monitoring method based on machine vision provided by an embodiment of the present invention, including the following steps S101-S106: S101: At the target location, acquire a standard slope image from the target perspective and the corresponding real-time slope image.

[0020] As a preferred embodiment, acquiring a standard slope image from the target perspective and a corresponding real-time slope image at the target location specifically includes: locating the current position of the UAV and calculating the distance and orientation between the UAV and the target location; controlling the UAV to reach the target location based on the distance and orientation between the UAV and the target location; after the UAV reaches the target location, adjusting the UAV's shooting angle according to the target perspective and acquiring the current real-time slope image; and acquiring the standard slope image corresponding to the target location and the target perspective from a preset database.

[0021] In this embodiment, GPS positioning technology is used to accurately obtain the current location coordinates (longitude, latitude, and altitude) of the UAV. Simultaneously, sensors such as an inertial measurement unit (IMU) are combined to improve the accuracy and stability of positioning. Therefore, based on the coordinates of the target location and the coordinates of the UAV's current location, geometric calculations or Geographic Information System (GIS) algorithms are used to calculate the straight-line distance and azimuth between them. The azimuth represents the direction from the UAV's current location to the target location. Then, based on the calculated distance and azimuth, the flight control system adjusts the UAV's flight direction and speed. The UAV automatically navigates to the target location according to the preset flight path and speed parameters. During flight, the UAV's position and attitude are monitored in real time to ensure it flies along the predetermined trajectory.

[0022] In this embodiment, after the drone reaches the target location, its shooting angle is adjusted. Based on the requirements of the target angle (such as pitch and azimuth), the angle of the drone's gimbal or camera is adjusted to ensure its shooting angle matches the target angle. The target angle is typically preset according to monitoring needs to ensure the image covers key areas of the slope. Preferably, the same target position and angle as a standard slope image are used for shooting. The standard slope image is pre-captured and stored in a pre-defined database. The corresponding shooting position and angle of the standard slope image are also stored. Therefore, in subsequent shooting processes, the shooting position and angle of the standard slope image can be directly used as the target position and angle for shooting, enabling more accurate monitoring of slope changes.

[0023] In this embodiment, after adjusting the shooting angle, the camera is activated to acquire real-time slope images. These real-time slope images will serve as the basis for subsequent analysis. Then, from a pre-set database, corresponding standard slope images are retrieved based on the target location and target viewing angle. Standard images are typically acquired under specific conditions using high-precision measuring equipment and are used as a reference benchmark for comparison and analysis with the real-time images. Subsequently, the acquired real-time slope images are compared with the standard slope images to analyze slope changes, such as the appearance of cracks and changes in displacement.

[0024] S102: Perform edge feature recognition on the standard slope image and the real-time slope image to obtain edge feature points, and construct a first polygon and a second polygon corresponding to the standard slope image and the real-time slope image, respectively, based on the edge feature points.

[0025] As a preferred embodiment, the step of performing edge feature recognition on the standard slope image and the real-time slope image to obtain edge feature points, and constructing a first polygon and a second polygon corresponding to the standard slope image and the real-time slope image respectively based on the edge feature points, specifically includes: performing edge feature recognition on the standard slope image and the real-time slope image to obtain edge features; identifying abrupt change points on the identified edge features to obtain edge feature points; and connecting the edge feature points on the standard slope image and the real-time slope image to obtain a first polygon and a second polygon corresponding to the standard slope image and the real-time slope image respectively.

[0026] In this embodiment, standard slope images and real-time slope images are preprocessed, including grayscale conversion, noise reduction, and contrast enhancement, to improve image quality. Then, edge detection algorithms (such as Canny edge detection, Sobel operator, etc.) are used to identify edge features in the image. The edge detection algorithm detects areas with significant brightness changes in the image by calculating the gradient information of the image, thereby extracting the edge detection results. Finally, continuous edge segments are extracted from the edge detection results. The edge segments reflect the contour and structural features of the slope.

[0027] In this embodiment, abrupt change points refer to points where significant changes occur in edge features, such as the initiation of cracks or inflection points of slope contours. These are typically important indicators of slope structural changes. By identifying abrupt change points in the slope, changes in slope shape can be accurately located. Slope displacement usually originates from unstable locations with abrupt edge changes; therefore, locating abrupt change points ensures the accuracy of slope displacement identification. Abrupt change points are identified by calculating the curvature and direction changes of edge features or by using deep learning models (such as convolutional neural networks). For example, points with large curvature or drastic direction changes can be identified as abrupt change points. These identified abrupt change points are marked on standard and real-time slope images; these points will serve as key feature points for subsequent polygon construction.

[0028] In this embodiment, abrupt change points on the standard slope image and the real-time slope image are connected to form polygons. The connection rules can be based on information such as the spatial location and edge direction of the abrupt change points. Preferably, the center point of all abrupt change points can be determined by their spatial location, and a ray is drawn from the center point to the right, then the abrupt change points swept by the ray are connected sequentially in a counter-clockwise direction to form polygons. It should be noted that the first polygon and the second polygon generated on the standard slope image and the real-time slope image, respectively, represent the key contour areas of the slope under standard and current conditions. By comparing the shape, area, and position parameters of the first and second polygons, slope changes, such as crack expansion and slope displacement, can be analyzed.

[0029] As a preferred embodiment, the step of performing edge feature recognition on the standard slope image and the real-time slope image to obtain edge features specifically includes: performing edge feature recognition on the standard slope image and the real-time slope image, and extracting vegetation cover lines, soil-rock edge lines, and soil-rock crack lines from the standard slope image and the real-time slope image respectively; performing noise filtering on the vegetation cover lines, soil-rock edge lines, and soil-rock crack lines to obtain the final vegetation edge features, soil-rock edge features, and crack edge features; wherein, the edge features include vegetation edge features, soil-rock edge features, and crack edge features.

[0030] In this embodiment, vegetation cover lines are used to identify the edges of vegetated areas on the slope, typically achieved through color segmentation (e.g., green areas of vegetation) combined with edge detection. Soil and rock edge lines are used to identify the edges of soil and rock areas on the slope, usually extracted through grayscale variations or texture features. Soil and rock crack lines are used to identify the edges of cracks on the slope; cracks typically appear as narrow areas with a different color or texture from the surrounding soil and rock, and can be detected using high-pass filtering or deep learning methods. By identifying vegetation edge features, soil and rock edge features, and crack edge features, the influence of different types of cover edges on the slope can be obtained, thus enabling the identification of vegetation edges, soil and rock edges, and crack edges. In this embodiment, actual image edge feature detection and identification may be affected by factors such as lighting variations, camera pixel noise, and image resolution, resulting in noise in the extracted edge features. Preferably, edge features can be smoothed using smoothing filters such as Gaussian filtering to reduce noise interference. For cases where vegetation edge features and soil edge features overlap in multiple directions, deep learning models (such as convolutional neural networks, CNN) can be used to filter noise from the edge features, and the model can be trained to learn the difference between noise and the real edge.

[0031] Understandably, by performing edge feature recognition on standard slope images and real-time slope images, and filtering noise from the extracted features, vegetation edge features, rock and soil edge features, and crack edge features can be accurately extracted. This improves the accuracy of feature extraction, enhances the reliability of slope monitoring, increases analysis efficiency, and has wide applicability.

[0032] As a preferred embodiment, the step of identifying abrupt change points of the identified edge features to obtain edge feature points specifically includes: determining the slope of the change of the edges of the vegetation edge features, rock and soil edge features, and crack edge features respectively, and determining the abrupt change points corresponding to the vegetation edge features, rock and soil edge features, and crack edge features based on the slope of the change; and taking the abrupt change points where the vegetation edge features, rock and soil edge features, and crack edge features overlap in pairs or all three as edge feature points.

[0033] In this embodiment, the slope of the edge change needs to be determined first. For each edge feature (vegetation edge feature, soil and rock edge feature, and crack edge feature), its local slope is calculated. The slope can be determined by calculating the angle change or gradient change between adjacent points on the edge feature. For example, for an edge feature line, a slope sequence can be obtained by calculating the slope between every two adjacent points. By analyzing the changes in the slope sequence, points of significant slope change are found as abrupt change points of the edge feature, such as the starting point of a crack or the turning point of a soil and rock edge. Abrupt change points can be identified by setting a threshold (e.g., the slope change exceeds a certain threshold) or by using a difference method (e.g., calculating the second derivative of the slope).

[0034] In this embodiment, abrupt change points refer to points on edge features where the slope changes significantly, reflecting important changes in the slope structure, such as crack expansion and collapse of the soil-rock edge. For each edge feature, abrupt change points are determined based on slope changes. Then, the abrupt change points of vegetation edge features, soil-rock edge features, and crack edge features are compared and analyzed to identify abrupt change points that overlap in pairs or all three. For example, if a point appears simultaneously in the abrupt change points of vegetation edge features and soil-rock edge features, or simultaneously in the abrupt change points of all three edge features, that point is identified as an edge feature point.

[0035] Understandably, identifying edge feature points can capture complex changes in slope structure. For example, the intersection of cracks and soil / rock edges is often a critical area for slope instability. That is, when vegetation edge features, soil / rock edge features, and crack edge features overlap in pairs or all three, this is usually an area where slope displacement is likely to occur. It should be noted that at abrupt changes where vegetation cover and soil / rock edges overlap, soil displacement is easily caused by vegetation erosion and decay. Conversely, at abrupt changes where soil / rock edges and crack edges overlap, displacement is easily caused by the influence of weathering, sunlight, and rainwater on crack stress. Therefore, multi-feature overlap analysis reduces false alarms that may arise from single-feature analysis, improving the reliability of monitoring results and avoiding false alarms that may occur from single-abrupt-point feature analysis.

[0036] S103: Based on the first polygon and the second polygon, calculate the deformation information, and based on the deformation information, determine whether there is displacement of the slope.

[0037] As a preferred embodiment, the step of calculating deformation information based on the first polygon and the second polygon, and determining whether there is displacement of the slope based on the deformation information, specifically includes: calculating the number of edge feature points of the first polygon and the second polygon; when the number of edge feature points of the first polygon is the same as the number of edge feature points of the second polygon, then the first polygon and the second polygon are projected to coincide in a way that the edges coincide, the non-coincident edges are identified, and the deformation information of the edges of the non-coincident first polygon and the non-coincident second polygon is calculated, and it is determined that there is displacement of the slope; when the number of edge feature points of the first polygon is the same as the number of edge feature points of the second polygon, the first polygon and the second polygon are projected to ... deformation information of the edges of the first polygon and the second polygon is calculated, and it is determined that there is displacement of the slope. If the number of edge feature points is different, the first polygon and the second polygon are projected to overlap in the manner that maximizes the number of overlapping edge feature points. Non-overlapping edge feature points are identified, and the vector data between the non-overlapping edge feature points of the first polygon and the non-overlapping edge feature points of the second polygon is calculated as deformation information, and it is determined whether the slope has displacement. When the number of edge feature points of the first polygon and the number of edge feature points of the second polygon are the same, and no non-overlapping edge feature points are identified when the first polygon and the second polygon are overlapped, the deformation data of the first polygon and the second polygon is zero, and it is determined whether the slope has displacement.

[0038] In this embodiment, the number of edge feature points of the first polygon and the second polygon are calculated from the standard slope image (first polygon) and the real-time slope image (second polygon) respectively, so as to perform projection overlap and deformation analysis.

[0039] In this embodiment, when the number of feature points is the same, meaning the key abrupt change points in both the standard slope image and the real-time slope image are identical, it indicates that vegetation and cracks have little impact on the slope and will not affect the accuracy of slope identification. Therefore, the first and second polygons are projected to coincide at their edges. This involves using geometric transformations (such as translation and rotation) to align the edges of the two polygons as closely as possible. During the overlap process, non-overlapping edge portions are identified, and deformation information, such as displacement vectors and angular changes, between the edges of the non-overlapping first and second polygons is calculated. Based on this deformation information, it is determined whether the slope has displacement. If no non-overlapping edge portions exist, it indicates that the slope in the real-time slope image does not exhibit significant displacement compared to the standard slope. Since the number of feature points is the same, there is no significant difference in external influences between the first and second polygons. Therefore, by directly analyzing the projection overlap of the polygon edges, the displacement caused by offset errors can be identified to the greatest extent possible, improving identification accuracy. Preferably, the following feature point overlap method can be used, but it is not as accurate and robust as edge overlap analysis.

[0040] In this embodiment, when the number of feature points is different, the first and second polygons are projected to coincide in a manner that maximizes the overlap of edge feature points. This involves using geometric transformation to match as many edge feature points as possible between the two polygons. During the coincidence process, non-coincident edge feature points are identified, and the vector data between the non-coincident edge feature points of the first and second polygons is calculated as deformation information. Based on this deformation information, it is determined whether the slope has displacement. It is understandable that, due to the different number of feature points, if the analysis is still based on the projection coincidence of polygon edges, it is impossible to determine the areas without displacement and the areas with displacement. Therefore, by coinciding edge feature points, while ensuring that the displacement information at some locations remains unchanged, it is possible to more intuitively and accurately locate the moving feature points and extract the corresponding deformation information.

[0041] In this embodiment, when the number of feature points is the same, the first polygon and the second polygon are overlapped. If no non-overlapping edge feature points are identified, the two polygons are considered to be completely overlapped. At this time, the deformation data of the first polygon and the second polygon is zero. Based on the fact that the deformation data is zero, it is determined that there is no displacement of the slope.

[0042] S104: When the slope is displaced, real-time multispectral images of the slope are acquired by a drone based on a preset slope scanning path.

[0043] In this embodiment, the preset slope scanning path can be set manually. After determining that the slope has displacement, a drone can be used to perform a detailed scan of the non-overlapping edge feature points to obtain real-time multispectral images of the corresponding slope for further monitoring.

[0044] S105: Perform slope surface feature identification on the multispectral image, and determine the displacement direction of the slope based on the slope surface features and the deformation information.

[0045] As a preferred embodiment, the step of identifying slope surface features from the multispectral image and determining the displacement direction of the slope based on the slope surface features and the deformation information specifically includes: identifying slope surface features from the multispectral image to obtain the features of the slope surface; wherein, the slope surface features include: vegetation features, rock and soil features, and crack features; and determining the displacement direction of the slope based on the vegetation features, rock and soil features, and crack features, combined with the deformation information.

[0046] In this embodiment, the multispectral image contains information from multiple bands, such as visible light (red, green, and blue), near-infrared, and shortwave infrared. Different bands provide different ground feature information, helping to distinguish different types of slope surface features. Vegetated areas are identified using the ratio of near-infrared to visible light bands in the multispectral image (e.g., Normalized Difference Vegetation Index, NDVI), where areas with higher NDVI values ​​typically indicate vegetation cover. Soil and rock areas are identified by analyzing the combination of shortwave infrared bands and other bands in the multispectral image. The spectral reflectance characteristics of soil and rock are usually different from those of vegetation and can be distinguished by spectral features. Cracks typically appear as linear or irregular low-reflectance areas. Crack features can be identified by combining edge detection algorithms (e.g., Canny edge detection) with specific bands of the multispectral image (e.g., shortwave infrared).

[0047] It should be noted that the identified vegetation, soil, and crack features are spatially matched with deformation information. For example, crack feature points are associated with deformation vectors to determine the crack displacement direction. The displacement direction of the vegetation area is analyzed to determine whether the vegetation-covered area has undergone overall displacement. The displacement direction of the soil and rock area is analyzed to determine the sliding or deformation direction of the soil and rock. The displacement direction of the cracks is analyzed to determine whether the cracks are expanding or moving, and their expansion direction. Furthermore, by combining the displacement directions of vegetation, soil and rock, and cracks, the overall displacement direction of the slope is comprehensively determined. For example, if the displacement directions of cracks and soil and rock are consistent, and the vegetation area also shows a similar displacement trend, the overall displacement direction of the slope can be determined. Then, by combining the deformation information obtained in the above steps, it is determined whether the displacement directions of vegetation, soil and rock, and cracks are the same as the directions corresponding to the deformation information, thus determining the final displacement direction.

[0048] Preferably, vegetation features, soil and rock features, and crack features can be spatially matched with deformation information to obtain corresponding spatial vectors. These spatial vectors are then added together with preset weights to obtain the overall displacement direction. The preset weights can be manually adjusted and set according to actual needs and measurement accuracy.

[0049] S106: Based on the displacement direction, continuously calculate the slope displacement rate within a preset period, and determine the slope stability level according to the displacement rate to complete the slope stability monitoring.

[0050] In this embodiment, a fixed monitoring period (such as hourly, daily, or weekly) is set. Within each preset period, slope displacement data is acquired using monitoring devices (such as GPS, InSAR, drones, etc.). Based on multispectral imagery or feature recognition, the displacement direction is determined, and the slope displacement changes in that direction are continuously monitored. Furthermore, the displacement amount within each period is calculated, and combined with the period time, the displacement rate can be obtained. Thus, the displacement rate is continuously recorded over multiple preset periods to form time-series data for analyzing the dynamic changes of the slope.

[0051] In this embodiment, statistical analysis is performed on displacement rate data from multiple periods to calculate statistical quantities such as the average and standard deviation, assess the trend of displacement rate changes, analyze the time series of displacement rates, and determine the displacement trend of the slope (e.g., acceleration, deceleration, or stability). The displacement rate and stability assessment criteria can be set according to actual needs. By setting different displacement rate thresholds, slope stability is divided into different levels (e.g., stable, slight deformation, moderate deformation, and severe deformation). The slope stability level is determined by comparing the calculated displacement rate with the preset threshold. Based on continuous monitoring data, the slope stability level is dynamically adjusted to reflect changes in slope stability in a timely manner.

[0052] Implementing the above embodiments yields the following effects: The technical solution of the present invention acquires a standard slope image from the target perspective and a corresponding real-time slope image at the target location, performs edge feature recognition, identifies edge feature points, and constructs a first polygon and a second polygon corresponding to the standard slope image and the real-time slope image, respectively, to calculate deformation information and determine whether the slope has displacement. It can directly determine displacement based on the polygon information of edge feature points on different images, avoiding the displacement information shift caused by slopes with different geological, geomorphological, and vegetation conditions, thus avoiding interference. At the same time, after displacement occurs, real-time multispectral images of the slope are acquired by a UAV based on a preset slope scanning path to determine the slope surface features and combine them with deformation information to determine the slope displacement direction, thereby continuously calculating the slope displacement rate within a preset period, efficiently monitoring slope stability, avoiding limitations on monitoring based on artificial texture features, and improving the applicability and accuracy of slope identification.

[0053] Example 2, please refer to Figure 2, which provides a slope stability monitoring system based on machine vision, including: an image acquisition module 201, an edge recognition module 202, a deformation displacement module 203, a multispectral imaging module 204, a displacement direction module 205, and a stability module 206; the acquisition module 201 is used to acquire a standard slope image from a target perspective and a corresponding real-time slope image at a target location; the edge recognition module 202 is used to perform edge feature recognition on the standard slope image and the real-time slope image to obtain edge feature points, and based on the edge feature points, construct a first polygon and a second polygon corresponding to the standard slope image and the real-time slope image, respectively; deformation... The displacement module 203 is used to calculate deformation information based on the first polygon and the second polygon, and to determine whether there is displacement of the slope based on the deformation information; the multispectral imaging module 204 is used to acquire real-time multispectral images of the slope through a drone based on a preset slope scanning path when the slope has displacement; the displacement direction module 205 is used to identify slope surface features from the multispectral images, and to determine the displacement direction of the slope based on the slope surface features and the deformation information; the stability module 206 is used to continuously calculate the displacement rate of the slope within a preset period based on the displacement direction, and to determine the slope stability level based on the displacement rate, thereby completing slope stability monitoring.

[0054] As a preferred embodiment, acquiring a standard slope image from the target perspective and a corresponding real-time slope image at the target location specifically includes: locating the current position of the UAV and calculating the distance and orientation between the UAV and the target location; controlling the UAV to reach the target location based on the distance and orientation between the UAV and the target location; after the UAV reaches the target location, adjusting the UAV's shooting angle according to the target perspective and acquiring the current real-time slope image; and acquiring the standard slope image corresponding to the target location and the target perspective from a preset database.

[0055] As a preferred embodiment, the step of performing edge feature recognition on the standard slope image and the real-time slope image to obtain edge feature points, and constructing a first polygon and a second polygon corresponding to the standard slope image and the real-time slope image respectively based on the edge feature points, specifically includes: performing edge feature recognition on the standard slope image and the real-time slope image to obtain edge features; identifying abrupt change points on the identified edge features to obtain edge feature points; and connecting the edge feature points on the standard slope image and the real-time slope image to obtain a first polygon and a second polygon corresponding to the standard slope image and the real-time slope image respectively.

[0056] As a preferred embodiment, the step of performing edge feature recognition on the standard slope image and the real-time slope image to obtain edge features specifically includes: performing edge feature recognition on the standard slope image and the real-time slope image, and extracting vegetation cover lines, soil-rock edge lines, and soil-rock crack lines from the standard slope image and the real-time slope image respectively; performing noise filtering on the vegetation cover lines, soil-rock edge lines, and soil-rock crack lines to obtain the final vegetation edge features, soil-rock edge features, and crack edge features; wherein, the edge features include vegetation edge features, soil-rock edge features, and crack edge features.

[0057] As a preferred embodiment, the step of identifying abrupt change points of the identified edge features to obtain edge feature points specifically includes: determining the slope of the change of the edges of the vegetation edge features, rock and soil edge features, and crack edge features respectively, and determining the abrupt change points corresponding to the vegetation edge features, rock and soil edge features, and crack edge features based on the slope of the change; and taking the abrupt change points where the vegetation edge features, rock and soil edge features, and crack edge features overlap in pairs or all three as edge feature points.

[0058] As a preferred embodiment, the step of calculating deformation information based on the first polygon and the second polygon, and determining whether there is displacement of the slope based on the deformation information, specifically includes: calculating the number of edge feature points of the first polygon and the second polygon; when the number of edge feature points of the first polygon is the same as the number of edge feature points of the second polygon, then the first polygon and the second polygon are projected to coincide in a way that the edges coincide, the non-coincident edges are identified, and the deformation information of the edges of the non-coincident first polygon and the non-coincident second polygon is calculated, and it is determined that there is displacement of the slope; when the number of edge feature points of the first polygon is the same as the number of edge feature points of the second polygon, the first polygon and the second polygon are projected to ... deformation information of the edges of the first polygon and the second polygon is calculated, and it is determined that there is displacement of the slope. If the number of edge feature points is different, the first polygon and the second polygon are projected to overlap in the manner that maximizes the number of overlapping edge feature points. Non-overlapping edge feature points are identified, and the vector data between the non-overlapping edge feature points of the first polygon and the non-overlapping edge feature points of the second polygon is calculated as deformation information, and it is determined whether the slope has displacement. When the number of edge feature points of the first polygon and the number of edge feature points of the second polygon are the same, and no non-overlapping edge feature points are identified when the first polygon and the second polygon are overlapped, the deformation data of the first polygon and the second polygon is zero, and it is determined whether the slope has displacement.

[0059] As a preferred embodiment, the step of identifying slope surface features from the multispectral image and determining the displacement direction of the slope based on the slope surface features and the deformation information specifically includes: identifying slope surface features from the multispectral image to obtain the features of the slope surface; wherein, the slope surface features include: vegetation features, rock and soil features, and crack features; and determining the displacement direction of the slope based on the vegetation features, rock and soil features, and crack features, combined with the deformation information.

[0060] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working process of the device described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0061] Implementing the above embodiments yields the following effects: The technical solution of the present invention acquires a standard slope image from the target perspective and a corresponding real-time slope image at the target location, performs edge feature recognition, identifies edge feature points, and constructs a first polygon and a second polygon corresponding to the standard slope image and the real-time slope image, respectively, to calculate deformation information and determine whether the slope has displacement. It can directly determine displacement based on the polygon information of edge feature points on different images, avoiding the displacement information shift caused by slopes with different geological, geomorphological, and vegetation conditions, thus avoiding interference. At the same time, after displacement occurs, real-time multispectral images of the slope are acquired by a UAV based on a preset slope scanning path to determine the slope surface features and combine them with deformation information to determine the slope displacement direction, thereby continuously calculating the slope displacement rate within a preset period, efficiently monitoring slope stability, avoiding limitations on monitoring based on artificial texture features, and improving the applicability and accuracy of slope identification.

[0062] Accordingly, the present invention also provides a terminal device, comprising: a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor executes the computer program to implement the machine vision-based slope stability monitoring method as described in any of the above embodiments.

[0063] The terminal device of this embodiment includes a processor, a memory, and a computer program and computer instructions stored in the memory and executable on the processor. When the processor executes the computer program, it implements the various steps in Embodiment 1 above, such as steps S101 to S16 shown in FIG1. ​​Alternatively, when the processor executes the computer program, it implements the functions of each module / unit in the above device embodiment, such as the edge recognition module 202.

[0064] For example, the computer program can be divided into one or more modules / units, which are stored in the memory and executed by the processor to complete the present invention. The one or more modules / units can be a series of computer program instruction segments capable of performing specific functions, which describe the execution process of the computer program in the terminal device. For example, the edge recognition module 202 is used to perform edge feature recognition on the standard slope image and the real-time slope image to obtain edge feature points, and based on the edge feature points, construct a first polygon and a second polygon corresponding to the standard slope image and the real-time slope image, respectively.

[0065] The terminal device may be a desktop computer, laptop, handheld computer, or a cloud-based pre-set database, etc. The terminal device may include, but is not limited to, a processor and memory. Those skilled in the art will understand that the schematic diagram is merely an example of a terminal device and does not constitute a limitation on the terminal device. It may include more or fewer components than illustrated, or combine certain components, or different components. For example, the terminal device may also include input / output devices, network access devices, buses, etc.

[0066] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor. The processor is the control center of the terminal device, connecting all parts of the terminal device via various interfaces and lines.

[0067] The memory can be used to store the computer programs and / or modules. The processor implements various functions of the terminal device by running or executing the computer programs and / or modules stored in the memory and by calling data stored in the memory. The memory may mainly include a program storage area and a data storage area. The program storage area may store the operating system, applications required for at least one function, etc.; the data storage area may store data created based on the use of the mobile terminal, etc. In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, RAM, plug-in hard disk, smart media card (SMC), secure digital card (SD card), flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0068] Wherein, if the modules / units integrated in the terminal device are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when the computer program is executed by a processor, it can implement the steps of the various method embodiments described above. Wherein, the computer program includes computer program code, which can be in the form of source code, object code, executable file, or some intermediate form, etc. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, portable hard drive, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content contained in the computer-readable medium can be appropriately added or removed according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the computer-readable medium does not include electrical carrier signals and telecommunication signals.

[0069] Accordingly, the present invention also provides a computer-readable storage medium, the computer-readable storage medium including a stored computer program, wherein, when the computer program is executed, it controls the device where the computer-readable storage medium is located to perform the slope stability monitoring method based on machine vision as described in any of the above embodiments.

[0070] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. In particular, it should be noted that any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention for those skilled in the art.

Claims

1. A slope stability monitoring method based on machine vision, characterized in that, include: At the target location, acquire a standard slope image from the target perspective and the corresponding real-time slope image. Edge feature recognition is performed on the standard slope image and the real-time slope image to obtain edge feature points. Based on the edge feature points, a first polygon and a second polygon corresponding to the standard slope image and the real-time slope image are constructed respectively. Deformation information is calculated based on the first polygon and the second polygon, and based on the deformation information, it is determined whether there is displacement of the slope. When the slope is displaced, real-time multispectral images of the slope are acquired by a drone based on a preset slope scanning path; the slope surface features are identified by the multispectral images, and the displacement direction of the slope is determined based on the slope surface features and the deformation information; based on the displacement direction, the displacement rate of the slope is continuously calculated in a preset period, and the slope stability level is determined based on the displacement rate to complete the slope stability monitoring.

2. The slope stability monitoring method based on machine vision as described in claim 1, characterized in that, The step of acquiring a standard slope image from the target perspective and a corresponding real-time slope image at the target location specifically includes: locating the current position of the UAV and calculating the distance and orientation between the UAV and the target location; controlling the UAV to reach the target location based on the distance and orientation between the UAV and the target location; after the UAV reaches the target location, adjusting the UAV's shooting angle according to the target perspective and acquiring the current real-time slope image; and acquiring the standard slope image corresponding to the target location and the target perspective from a preset database.

3. The slope stability monitoring method based on machine vision as described in claim 1, characterized in that, The step of performing edge feature recognition on the standard slope image and the real-time slope image to obtain edge feature points, and constructing a first polygon and a second polygon corresponding to the standard slope image and the real-time slope image respectively based on the edge feature points, specifically includes: performing edge feature recognition on the standard slope image and the real-time slope image to obtain edge features; identifying abrupt change points on the identified edge features to obtain edge feature points; and connecting the edge feature points on the standard slope image and the real-time slope image to obtain a first polygon and a second polygon corresponding to the standard slope image and the real-time slope image respectively.

4. The slope stability monitoring method based on machine vision as described in claim 3, characterized in that, The step of performing edge feature recognition on the standard slope image and the real-time slope image to obtain edge features specifically includes: performing edge feature recognition on the standard slope image and the real-time slope image, extracting vegetation cover lines, soil-rock edge lines, and soil-rock crack lines from the standard slope image and the real-time slope image respectively; filtering noise from the vegetation cover lines, soil-rock edge lines, and soil-rock crack lines to obtain the final vegetation edge features, soil-rock edge features, and crack edge features; wherein, the edge features include vegetation edge features, soil-rock edge features, and crack edge features.

5. The slope stability monitoring method based on machine vision as described in claim 4, characterized in that, The step of identifying abrupt change points in the identified edge features to obtain edge feature points specifically includes: determining the slope of the change of the edges of the vegetation edge features, rock and soil edge features, and crack edge features respectively, and determining the abrupt change points corresponding to the vegetation edge features, rock and soil edge features, and crack edge features based on the slope of the change; and taking the abrupt change points where the vegetation edge features, rock and soil edge features, and crack edge features overlap in pairs or all three as edge feature points.

6. The slope stability monitoring method based on machine vision as described in claim 5, characterized in that, The step of calculating deformation information based on the first polygon and the second polygon, and determining whether there is slope displacement based on the deformation information, specifically includes: calculating the number of edge feature points of the first polygon and the second polygon; when the number of edge feature points of the first polygon is the same as the number of edge feature points of the second polygon, the first polygon and the second polygon are projected to coincide with each other in a way that the edges coincide, identifying non-coincident edges, and calculating the deformation information of the edges of the non-coincident first polygon and the non-coincident second polygon, and determining that there is slope displacement; when the number of edge feature points of the first polygon is different from the number of edge feature points of the second polygon, the first polygon and the second polygon are projected to coincide with each other in a way that the number of edge feature points coincides the most, identifying non-coincident edge feature points, and calculating the vector data between the edge feature points of the non-coincident first polygon and the non-coincident second polygon as deformation information, and determining that there is slope displacement; when the number of edge feature points of the first polygon is the same as the number of edge feature points of the second polygon, and no non-coincident edge feature points are identified when the first polygon and the second polygon are coincided, the deformation data of the first polygon and the second polygon is zero, and determining whether there is slope displacement.

7. The slope stability monitoring method based on machine vision as described in claim 6, characterized in that, The step of identifying slope surface features from the multispectral image and determining the displacement direction of the slope based on the slope surface features and deformation information specifically includes: identifying slope surface features from the multispectral image to obtain slope surface features; wherein, the slope surface features include: vegetation features, rock and soil features, and crack features; and determining the displacement direction of the slope based on the vegetation features, rock and soil features, and crack features, combined with the deformation information.

8. A slope stability monitoring system based on machine vision, characterized in that, include: Image acquisition module, edge recognition module, deformation and displacement module, multispectral imaging module, displacement direction module, and stability module; The acquisition module is used to acquire a standard slope image from the target perspective and the corresponding real-time slope image at the target location. An edge recognition module is used to perform edge feature recognition on the standard slope image and the real-time slope image to obtain edge feature points, and based on the edge feature points, to construct a first polygon and a second polygon corresponding to the standard slope image and the real-time slope image, respectively. The deformation displacement module is used to calculate deformation information based on the first polygon and the second polygon, and to determine whether there is displacement of the slope based on the deformation information. The multispectral imaging module is used to acquire real-time multispectral images of the slope via a drone based on a preset slope scanning path when the slope is displaced; the displacement direction module is used to identify slope surface features from the multispectral images and determine the displacement direction of the slope based on the slope surface features and the deformation information. The stability module is used to continuously calculate the displacement rate of the slope within a preset period based on the displacement direction, and determine the slope stability level according to the displacement rate, thereby completing the slope stability monitoring.

9. A terminal device, characterized in that, The method includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor, when executing the computer program, implements the machine vision-based slope stability monitoring method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored computer program, wherein, when the computer program is executed, it controls the device containing the computer-readable storage medium to perform the slope stability monitoring method based on machine vision as described in any one of claims 1 to 7.

Images