CN117611661A - Slope displacement monitoring method and system - Google Patents
Slope displacement monitoring method and system Download PDFInfo
- Publication number
- CN117611661A CN117611661A CN202311353505.1A CN202311353505A CN117611661A CN 117611661 A CN117611661 A CN 117611661A CN 202311353505 A CN202311353505 A CN 202311353505A CN 117611661 A CN117611661 A CN 117611661A
- Authority
- CN
- China
- Prior art keywords
- monitoring
- position information
- convolution layer
- identification
- points
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01B—MEASURING LENGTH, THICKNESS OR SIMILAR LINEAR DIMENSIONS; MEASURING ANGLES; MEASURING AREAS; MEASURING IRREGULARITIES OF SURFACES OR CONTOURS
- G01B11/00—Measuring arrangements characterised by the use of optical techniques
- G01B11/02—Measuring arrangements characterised by the use of optical techniques for measuring length, width or thickness
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/50—Depth or shape recovery
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10028—Range image; Depth image; 3D point clouds
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A10/00—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE at coastal zones; at river basins
- Y02A10/23—Dune restoration or creation; Cliff stabilisation
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Computational Linguistics (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a slope displacement monitoring method and a system, 1) a marking point is arranged in a slope monitoring range, a control point is arranged, and the marking point and the initial position information of the control point are obtained; 2) Collecting depth images of identification points and control points, and performing format conversion on the depth images to generate point cloud data; 3) Detecting point cloud data through a deep learning model, and processing a depth image according to a detection result; 4) Dividing the processed depth image into identification points and control points through a division model, and generating monitoring relative position information of the identification points and the control points according to a division result; generating a slope displacement detection result according to the initial position information and the monitoring relative position information of the identification points and the control points; the identification of the marker is more accurate, so that the positioning is more accurate, meanwhile, the damage displayed by the slope form is monitored in the slope monitoring process, and related personnel are timely warned, so that the method has good practicability.
Description
Technical Field
The invention relates to the technical field of slope displacement monitoring, in particular to a slope displacement monitoring method and system.
Background
The slope is a slope formed naturally or artificially, is one of the most basic geological environments in human engineering activities, and is the most common engineering form in engineering construction. The geological disaster of the side slope can seriously endanger the safety and smoothness of road transportation. Slope displacement monitoring is one of very important means for guaranteeing slope safety. In modern buildings, slope displacement monitoring may be used in a variety of settings, such as highways, railways, reservoir dams, subway tunnels, and the like. The deformation signs can be found in time and early warning can be carried out by monitoring the slope displacement, so that the danger of the slope is avoided. The slope displacement monitoring generally adopts the following method: laser displacement monitoring based on an electronic ranging technology, satellite displacement monitoring based on a high-precision GPS technology, microseismic monitoring based on a magnetometer and an accelerometer and monitoring based on a photographing technology, wherein the photographing monitoring technology is based on photographing equipment installed on a side slope, a photo of the side slope is photographed at regular time, and then an image processing technology is utilized to perform feature extraction and analysis on the photo so as to acquire displacement data of the side slope, but in monitoring of the photographing technology, the influence of environmental factors of the side slope is usually received, so that the feature extraction and analysis of the photo are inaccurate, and further the measurement of the follow-up displacement data is inaccurate.
Disclosure of Invention
The invention provides a slope displacement monitoring method and a system for solving the problems in the background technology.
The scheme of the invention is as follows:
a slope displacement monitoring method comprises the following steps:
1) Setting an identification point in a side slope monitoring range, setting a control point, and acquiring initial position information of the identification point and the control point;
2) Collecting depth images of identification points and control points, and performing format conversion on the depth images to generate point cloud data;
3) Detecting point cloud data through a deep learning model, and processing a depth image according to a detection result;
4) Dividing the processed depth image into identification points and control points through a division model, and generating monitoring relative position information of the identification points and the control points according to a division result; and generating a slope displacement detection result according to the initial position information and the monitoring relative position information of the identification points and the control points.
As a preferable technical scheme, the identification points and the control points adopt a regular pattern, and the central point position of the regular pattern is used as initial position information.
As an optimal technical scheme, the deep learning model adopts an image information processing model, the image information processing model is structurally an input layer and a rotation alignment network, a first convolution layer and a first graph convolution layer are connected with the rotation network, the output results of the first convolution layer and the first graph convolution layer are weighted and fused and are connected to a second graph convolution layer, the first convolution layer is also connected with the second convolution layer, the weighting and fusion results of the second convolution layer and the second graph convolution layer are input to a third graph convolution layer, and the output end of the third graph convolution layer is sequentially connected with a full connection layer and an output layer.
As a preferable technical scheme, the segmentation model adopts a full convolution neural network.
As an optimal technical scheme, the generation process of the slope displacement detection result is as follows:
in the segmentation result, taking the central pixel of the identification point and the control point as a monitoring pixel, calculating relative position information of the identification point and the control point according to the three-dimensional position information of the monitoring pixel in a camera coordinate system according to the three-dimensional position information of the monitoring pixel, converting the relative position information in the camera coordinate system into monitoring relative position information in the world coordinate system according to the conversion relation of the camera coordinate system and the world coordinate system, determining the monitoring position information of the identification point in the monitoring process according to the monitoring relative position information in the world coordinate system and the initial position information of the control point, and generating a slope displacement detection result according to the monitoring position information and the initial position information of the identification point.
The invention also discloses a corresponding monitoring system based on the slope displacement monitoring method, which comprises a positioning module, an acquisition module, a detection module and a displacement monitoring module;
the positioning module is used for acquiring initial position information of the identification point and the control point;
the acquisition processing module is used for acquiring depth images of the identification points and the control points, performing format conversion on the depth images and generating point cloud data;
the detection module is used for detecting the point cloud data through the deep learning model and processing the deep image according to the detection result;
the displacement monitoring module is used for dividing the identification points and the control points of the processed depth image through the division model, and generating monitoring relative position information of the identification points and the control points according to the division result; and generating a slope displacement detection result according to the initial position information and the monitoring relative position information of the identification points and the control points.
In the preferred technical scheme, in the detection module, the deep learning model adopts an image information processing model, wherein the structure is an input layer and a rotation alignment network, a first convolution layer and a first picture convolution layer which are connected with the rotation network are connected to the first picture convolution layer, the output results of the first convolution layer and the first picture convolution layer are weighted and fused to be connected to a second picture convolution layer, the first convolution layer is also connected with the second convolution layer, the weighted and fused results of the second convolution layer and the second picture convolution layer are input to a third picture convolution layer, and the output end of the third picture convolution layer is sequentially connected with a full connection layer and an output layer.
In the preferred technical scheme, in the displacement monitoring module, the segmentation model adopts a full convolution neural network.
In the preferred technical scheme, in the displacement monitoring module, the generation process of the slope displacement detection result is that,
in the segmentation result, taking the central pixel of the identification point and the control point as a monitoring pixel, calculating relative position information of the identification point and the control point according to the three-dimensional position information of the monitoring pixel in a camera coordinate system according to the three-dimensional position information of the monitoring pixel, converting the relative position information in the camera coordinate system into monitoring relative position information in the world coordinate system according to the conversion relation of the camera coordinate system and the world coordinate system, determining the monitoring position information of the identification point in the monitoring process according to the monitoring relative position information in the world coordinate system and the initial position information of the control point, and generating a slope displacement detection junction according to the monitoring position information and the initial position information of the identification point.
By adopting the technical scheme, the slope displacement monitoring method and the slope displacement monitoring system 1) set the identification points in the slope monitoring range and set the control points to obtain the initial position information of the identification points and the control points; 2) Collecting depth images of identification points and control points, and performing format conversion on the depth images to generate point cloud data; 3) Detecting point cloud data through a deep learning model, and processing a depth image according to a detection result;
4) Dividing the processed depth image into identification points and control points through a division model, and generating monitoring relative position information of the identification points and the control points according to a division result; and generating a slope displacement detection result according to the initial position information and the monitoring relative position information of the identification points and the control points.
The invention has the advantages that:
the invention can remove the influence factors of the surrounding environment, monitor whether damage occurs in real time, determine the position information of the marker in the monitoring process by a method of automatically identifying the marker and resolving the position after removing the influence factors, and calculate the displacement of the side slope according to the initial position information of the marker and the position information of the current marker.
After the environmental influence is removed, the identification of the marker is more accurate, the positioning is more accurate, meanwhile, in the slope monitoring process, the damage displayed by the slope form is monitored, related personnel are timely warned, and the method has good practicability.
Drawings
Fig. 1 is a schematic flow chart of an embodiment of the present invention.
Detailed Description
The invention is further described in connection with the following embodiments in order to make the technical means, the creation features, the achievement of the purpose and the effect of the invention easy to understand.
Example 1
As shown in fig. 1, the present invention provides a slope displacement monitoring method, which is as follows:
corresponding identification points are arranged on the slope to be monitored, the identification points can be selected to be identical or different in shape for identification, corresponding control points are arranged in a stable area, and the relative distance between the identification points and the control points in the monitoring process is determined through shooting and monitoring related information in advance, so that the displacement of the slope is monitored. In the process, related factors are removed through a related information detection method, marker information and control point information in an image are effectively identified subsequently, and slope displacement information is determined through displacement information of corresponding marker points.
The method comprises the steps that corresponding identification points are arranged on an unstable side slope, the identification points adopt identification points which have obvious contrast with the environmental color of the side slope and can be effectively displayed in an acquired image, in order to more highlight corresponding identification point information, patterns such as regular patterns, circles, squares or cross shapes can be selected, marks of corresponding identifiers are marked in the patterns so as to distinguish different identifiers, the central positions of the identification points are used as positioning points of the different identifiers, meanwhile, in the setting process, random distribution or uniform distribution of a certain range can be carried out, a certain distance is arranged between the identifiers, and the current three-dimensional position of the identifiers is ensured to be determined through an RTK measurement technology or a total station or other high-precision positioning devices, so that the initial positions of the identifiers are used as follow-up side slope displacement monitoring;
the control points need to be arranged in the areas near the side slopes, such as highways or bedrock, where displacement is not easy to occur, obvious regular patterns can be adopted, different specific symbols are marked in the center to be displayed to determine different control points, the control points and the identification points can be distinguished through different colors or marking information, the position of the center point of the control points is determined through RTK measurement technology or total station or other high-precision positioning devices, and the position of the control points is recorded to be used as verification information of subsequent identifiers.
The identification points and the control points are shot by the depth camera, the depth camera can shoot manually, the site shoots or the unmanned aerial vehicle is adopted to shoot correspondingly, in the shooting process, the three-dimensional position of the depth camera is required to be determined by a relevant high-precision positioning device, the absolute position of the pixels in the image is calculated by the absolute position of the depth camera, the corresponding shooting image is acquired through depth, the image contains the identification points and the control points which are not less than a threshold value, only the image of the slope range is shot at the same time, and the area outside the slope range is not shot, so that the accuracy of subsequent monitoring is ensured. Wherein the threshold is manually input, for example, the threshold of the identification point is 2, and the control point is 3. And (3) carrying out position calculation of different pixels in the corresponding image through a positioning principle of the depth camera, determining the positions of different pixel points in the image, converting the positions of the identification points into absolute positions through a control point, and taking the position of the central pixel of the marker as the position of the marker in the corresponding image in the position determination related content of the marker.
In the monitoring process, removing the area which possibly affects the identification of the identification point in the image by acquiring numerical value information and position information of pixels of the image in the monitoring process in advance;
the invention provides a deep learning model which adopts an image information processing model, wherein the image information processing model is recorded according to three-dimensional position information of different pixels provided by a depth camera, the processed three-dimensional position information can select position information corresponding to a coordinate system by using the depth camera, denoising and image enhancement preprocessing is carried out on an image acquired by the depth camera, pixels in the preprocessed image are converted into gray images from RGB images, wherein different weights are set for different values of RGB channels, the weights of R and B channel values are set to 0.2, the numerical weight of G channel is set to 0.6, the weights of different passing values are set to be weighted and calculated, corresponding gray images are finally generated, gray values of pixels in the gray images are recorded, corresponding point cloud data can be generated according to the gray value information of the pixels and the three-dimensional position information corresponding to the pixels, and point cloud data are processed through the image information processing model.
The image information processing model performs related extraction by fusing the characteristics of pixels in the image and different structural characteristics among different pixels, the characteristics of the specific pixels can be extracted by selecting a corresponding three-dimensional convolution layer, the three-dimensional structural characteristics of the pixels are extracted by a graph convolution neural network (GCN), the structural characteristics are characteristic information of a neighborhood formed by each point and surrounding points, and for convenience of later description, the graph convolution neural network (GCN) is subsequently described as a graph convolution layer; the convolution layer and the graph convolution layer are fused in a weighted sum mode, a specific image information processing model is generated, the image information processing model mainly removes influences of plants, other obstacles and cracks in the environment, the other obstacles comprise waste objects, the accuracy of the identification of the follow-up markers is improved, and meanwhile whether deformation damage occurs to the side slope or not is monitored.
The method comprises the steps of constructing a correlation model by using Pytorch software, wherein the specific structure of an image information processing model is an input layer and a rotation alignment network, a first convolution layer and a first graph convolution layer which are connected with the rotation network are connected, the output results of the first convolution layer and the first graph convolution layer are weighted and fused, the first convolution layer is connected to a second graph convolution layer, the second convolution layer is also connected, the weighted and fused results of the second convolution layer and the second graph convolution layer are input to a third graph convolution layer, and the output end of the third graph convolution layer is sequentially connected with a full connection layer and an output layer. The loss function of the network model is set as an NLLloss function, and specific parameter adjustment of the relevant network and the convolution layer can be obtained according to routine adjustment of those skilled in the art, and details are not described herein.
Specific implementation procedures of the image information processing model are as follows: firstly, processing the data through a rotation alignment network, rotating the network comprising a normalized T-net network and a matrix multiplication matrix, processing image pixels based on depth camera acquisition information through the T-net network, namely, input data, generating a corresponding affine transformation matrix, multiplying the reflection transformation matrix with the input data through the matrix multiplier, outputting a final operation rotation alignment operation result, extracting structural features of a data three-dimensional local structure from the rotation alignment result through a first convolution layer, extracting pixel features of the data pixel points per se through the first convolution layer, weighting and fusing the structural features and the pixel features, wherein weights can be preset according to manual experience, the content corresponding to the fusion result comprises local three-dimensional structural information features and pixel features per se, the fusion result is input into a next convolution layer, namely, a second convolution layer, the pixel features are input into the next convolution layer, namely, the second convolution layer, extracting the structural features of the data pixel points per se through the first convolution layer, weighting and fusing the pixel features, namely, the weighting and fusing the pixel features are output into the third convolution layer, and the third convolution layer is connected with the input into the third convolution layer, and the total-depth matrix is set, and the total-depth matrix is input, and the total-depth matrix is output, and the total-depth matrix is not input, and the three-obstacle matrix is input, and the three-level is input. And removing pixels corresponding to the point clouds identified as plants and other obstacles in the original image, reserving pixels corresponding to the rest point clouds, and setting the pixels which do not need to be reserved in the image as white values. Meanwhile, when the crack occurs and the occupied range is large, the crack should be immediately alarmed to relevant personnel for processing, so as to prevent the side slope from being further damaged.
In the process of identifying the images processed by the method, corresponding manual labeling can be carried out on the shot images in the monitoring process of the images by related personnel, as obvious differences exist between the identifiers and control points in the images and the slope environment, or the identifiers can be subjected to simple image semantic segmentation by a simple full-convolution neural network, in the process of segmentation, related preprocessing, denoising or image enhancement and other modes are carried out on the processed images in advance, then a full-convolution neural network for semantic segmentation is built by Pytorch software, the full-convolution neural network is composed of an input layer, a convolution structure and a deconvolution structure, the convolution structure comprises a plurality of convolution layers and a maximum pooling layer, the deconvolution structure comprises a plurality of deconvolution layers, the final result is output by a softmax function, the characteristics of the identifiers of the corresponding images are extracted by the convolution layers, the corresponding images output by the convolution layers are restored to the original sizes by the deconvolution layers, whether the pixels in the images are the output categories of the identifiers or not are labeled by the softmax function, the pixels in the images are output categories of the identifiers, the positions of the corresponding pixels in the images are automatically identified, and the positions of the images are identified. The number of the convolution layers and the number of the deconvolution layers are the same and different parameters correspond to each other, and a person related to the image processing field can construct and set the deconvolution layers according to related prior knowledge, which is not described in detail herein. The full convolutional neural network and the image information processing model belong to a deep learning model, the deep learning model is optimized by acquiring historical data or downloading corresponding data from a related website to form a corresponding training set, test set or verification set, and the training mode of the deep learning model is common knowledge in the art and is not repeated here.
After identification of the marker is completed in the monitoring process, the position of the marker is determined, wherein identification is firstly carried out on marked information in the marker through a neural network, whether the marked information is a marker point type or a control point is identified, the absolute position of the marker is calculated through the relative positions of the control point and the marker point under a camera coordinate system and the absolute position of the control point, wherein three-dimensional positioning information of a center point pixel of the marker and the center point of the control point, which are acquired by a depth camera, is determined, the three-dimensional positioning information of the marker point and the control point is unified to the same coordinate system, for example, the relative positions of the marker point and the control point are calculated under the camera coordinate system, the relative position relation between the marker point and the control point under the world coordinate system is generated through the conversion relation between the camera coordinate system and the absolute position, namely the world coordinate system, the absolute position of the marker point and the control point is calculated according to the absolute position of the control point, the absolute position of the marker point is calculated according to the three-point calibration method, and the actual displacement is calculated finally.
According to the technical scheme, the influence factors of the surrounding environment can be removed, whether damage occurs or not can be monitored in real time, after the influence factors are removed, the position information of the marker in the monitoring process is determined through an automatic identification marker and position resolving method, and the displacement of the side slope is calculated according to the initial position information of the marker and the position information of the current marker. After the environmental influence is removed, the identification of the marker is more accurate, the positioning is more accurate, meanwhile, in the slope monitoring process, the damage displayed by the slope form is monitored, related personnel are timely warned, and the method has good practicability.
Example two
A corresponding monitoring system based on a slope displacement monitoring method comprises a positioning module, an acquisition module, a detection module and a displacement monitoring module; the positioning module is used for acquiring initial position information of the identification point and the control point; the acquisition processing module is used for acquiring depth images of the identification points and the control points, performing format conversion on the depth images and generating point cloud data; the detection module is used for detecting the point cloud data through the deep learning model and processing the deep image according to the detection result; the displacement monitoring module is used for dividing the identification points and the control points of the processed depth image through the division model, and generating monitoring relative position information of the identification points and the control points according to the division result; and generating a slope displacement detection result according to the initial position information and the monitoring relative position information of the identification points and the control points. The system corresponds to the method content and is not described in detail herein.
The foregoing has shown and described the basic principles, main features and advantages of the present invention. It will be understood by those skilled in the art that the present invention is not limited to the embodiments described above, and that the above embodiments and descriptions are merely illustrative of the principles of the present invention, and various changes and modifications may be made without departing from the spirit and scope of the invention, which is defined in the appended claims. The scope of the invention is defined by the appended claims and equivalents thereof.
Claims (9)
1. The slope displacement monitoring method is characterized by comprising the following steps of:
1) Setting an identification point in a side slope monitoring range, setting a control point, and acquiring initial position information of the identification point and the control point;
2) Collecting depth images of identification points and control points, and performing format conversion on the depth images to generate point cloud data;
3) Detecting point cloud data through a deep learning model, and processing a depth image according to a detection result;
4) Dividing the processed depth image into identification points and control points through a division model, and generating monitoring relative position information of the identification points and the control points according to a division result; and generating a slope displacement detection result according to the initial position information and the monitoring relative position information of the identification points and the control points.
2. A slope displacement monitoring method as in claim 1, wherein: the identification points and the control points adopt regular patterns, and the positions of central points of the regular patterns are used as initial position information.
3. A slope displacement monitoring method as in claim 1, wherein: the deep learning model adopts an image information processing model, the image information processing model is structurally an input layer and a rotary alignment network, a first convolution layer and a first graph convolution layer are connected with the rotary alignment network, the output results of the first convolution layer and the first graph convolution layer are weighted and fused, the first convolution layer is further connected with a second convolution layer, the weighting and fusion results of the second convolution layer and the second graph convolution layer are input into a third graph convolution layer, and the output end of the third graph convolution layer is sequentially connected with a full connection layer and an output layer.
4. A slope displacement monitoring method as in claim 1, wherein: the segmentation model adopts a full convolution neural network.
5. The slope displacement monitoring method of claim 1, wherein the slope displacement detection result is generated by:
in the segmentation result, taking the central pixel of the identification point and the control point as a monitoring pixel, calculating relative position information of the identification point and the control point according to the three-dimensional position information of the monitoring pixel in a camera coordinate system according to the three-dimensional position information of the monitoring pixel, converting the relative position information in the camera coordinate system into monitoring relative position information in the world coordinate system according to the conversion relation of the camera coordinate system and the world coordinate system, determining the monitoring position information of the identification point in the monitoring process according to the monitoring relative position information in the world coordinate system and the initial position information of the control point, and generating a slope displacement detection result according to the monitoring position information and the initial position information of the identification point.
6. A corresponding monitoring system based on a slope displacement monitoring method according to any one of claims 1 to 6, characterized in that: the device comprises a positioning module, an acquisition module, a detection module and a displacement monitoring module;
the positioning module is used for acquiring initial position information of the identification point and the control point;
the acquisition processing module is used for acquiring depth images of the identification points and the control points, performing format conversion on the depth images and generating point cloud data;
the detection module is used for detecting the point cloud data through the deep learning model and processing the deep image according to the detection result;
the displacement monitoring module is used for dividing the identification points and the control points of the processed depth image through the division model, and generating monitoring relative position information of the identification points and the control points according to the division result; and generating a slope displacement detection result according to the initial position information and the monitoring relative position information of the identification points and the control points.
7. The system for monitoring slope displacement based on the slope displacement monitoring method of claim 6, wherein: in the detection module, the deep learning model adopts an image information processing model, wherein the structure is an input layer and a rotation alignment network, a first convolution layer and a first graph convolution layer which are connected with the rotation network are connected to the first graph convolution layer, the output results of the first convolution layer and the first graph convolution layer are weighted and fused, the first convolution layer is also connected with a second convolution layer, the weighting and fusion results of the second convolution layer and the second graph convolution layer are input to a third graph convolution layer, and the output end of the third graph convolution layer is sequentially connected with a full connection layer and an output layer.
8. The system for monitoring slope displacement according to claim 6, wherein: in the displacement monitoring module, the segmentation model adopts a full convolution neural network.
9. The system for monitoring slope displacement according to claim 6, wherein: in the displacement monitoring module, the generation process of the slope displacement detection result is that,
in the segmentation result, taking the central pixel of the identification point and the control point as a monitoring pixel, calculating relative position information of the identification point and the control point according to the three-dimensional position information of the monitoring pixel in a camera coordinate system according to the three-dimensional position information of the monitoring pixel, converting the relative position information in the camera coordinate system into monitoring relative position information in the world coordinate system according to the conversion relation of the camera coordinate system and the world coordinate system, determining the monitoring position information of the identification point in the monitoring process according to the monitoring relative position information in the world coordinate system and the initial position information of the control point, and generating a slope displacement detection result according to the monitoring position information and the initial position information of the identification point.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311353505.1A CN117611661A (en) | 2023-10-19 | 2023-10-19 | Slope displacement monitoring method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311353505.1A CN117611661A (en) | 2023-10-19 | 2023-10-19 | Slope displacement monitoring method and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117611661A true CN117611661A (en) | 2024-02-27 |
Family
ID=89943206
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311353505.1A Pending CN117611661A (en) | 2023-10-19 | 2023-10-19 | Slope displacement monitoring method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117611661A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118863142A (en) * | 2024-07-02 | 2024-10-29 | 内蒙古高新科技控股有限责任公司 | Slope displacement prediction method and device |
-
2023
- 2023-10-19 CN CN202311353505.1A patent/CN117611661A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118863142A (en) * | 2024-07-02 | 2024-10-29 | 内蒙古高新科技控股有限责任公司 | Slope displacement prediction method and device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110223341B (en) | Intelligent water level monitoring method based on image recognition | |
| KR102346676B1 (en) | Method for creating damage figure using the deep learning-based damage image classification of facility | |
| CN118692226B (en) | A flood analysis method based on artificial intelligence | |
| CN113034689A (en) | Laser point cloud-based terrain three-dimensional model, terrain map construction method and system, and storage medium | |
| CN112528979B (en) | Transformer substation inspection robot obstacle distinguishing method and system | |
| CN102768022A (en) | Tunnel surrounding rock deformation detection method adopting digital camera technique | |
| US20250378566A1 (en) | Beach erosion detection and quantitative calculation method based on unmanned aerial vehicle mapping | |
| CN117516487B (en) | Medium-small river video flow test method | |
| CN113935096B (en) | Method and system for monitoring deformation of foundation pit in real time | |
| CN118171918B (en) | Air-space-ground landslide monitoring method and system based on multi-source data | |
| CN100580692C (en) | A method for detection of changes in water bodies and settlements based on aerial images | |
| CN118586757A (en) | A bridge construction quality supervision method based on small multi-rotor drone | |
| CN118298326B (en) | Ocean garbage area estimation method based on deep learning and unmanned aerial vehicle remote sensing | |
| CN119984416A (en) | A method, device and product for real-time monitoring of river flow | |
| CN119048932A (en) | Landslide crack identification method based on unmanned aerial vehicle orthographic image and pyramid scene analysis network image segmentation algorithm | |
| CN117611661A (en) | Slope displacement monitoring method and system | |
| CN117911754A (en) | Method, system and storage medium for detecting defect degree of power transmission line tree obstacle | |
| CN115984355A (en) | Method for calculating length of dry beach of tailing pond based on deep learning | |
| CN112613437B (en) | Method for identifying illegal buildings | |
| KR20200012373A (en) | Apparatus and method for calculating water lever based on image processing | |
| CN120014015A (en) | A method for measuring hidden dangers in transmission line channels based on point cloud image fusion | |
| CN118898589A (en) | Surface damage detection method of concrete bridge based on drone | |
| CN118799710A (en) | A visibility detection method, device, electronic device and storage medium | |
| CN117346742A (en) | Hydropower station surveying and mapping system based on airborne lidar and oblique photogrammetry | |
| TWI597405B (en) | System and method for monitoring slope with tree displacement |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |