Geological disaster hidden danger sample set construction method and construction system

Abstract

The invention discloses a geological disaster hidden danger sample set construction method and system, and belongs to the technical field of environment monitoring, and the method comprises the steps: collecting the InSAR data and image of a target region; identifying a first area according to the image, and establishing a label for the first area; obtaining a second area with the gradient smaller than a first threshold value, and establishing a label for the second area; obtaining a mountainous area of the target area according to the image and vision; segmenting the mountainous area according to the boundary of the mountainous area to obtain segmented bodies; based on the InSAR data, the deformation quantity of the segmentation body is obtained; and marking the segmented body according to the deformation quantity to obtain sample data and a sample set. The sample data considers the four-season change, the gradient, the deformation quantity and the mountain segmentation body, so that the labeling of the sample set is more accurate; and the accuracy of geological disaster hidden danger identification is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of environmental monitoring, and in particular to a method and system for constructing a sample set of geological disaster hidden dangers. Background Art

[0002] Geological disaster deformation monitoring technology based on remote sensing data has been widely used in recent years. Remote sensing technologies such as synthetic aperture radar SAR, synthetic aperture radar interferometry InSAR, optical imaging and laser radar Li DAR can obtain surface deformation information over a large area and in a non-contact manner, generate digital elevation models, and calculate the deformation amount and deformation rate of the surface. Remote sensing technology has shown significant advantages in the monitoring of geological disasters such as landslides, surface subsidence, and debris flows, especially in inaccessible mountainous areas and dangerous areas.

[0003] Based on machine learning methods, landslide prediction has become an important research method. The construction of sample sets is an important part of machine learning, which affects the accuracy and reliability of the prediction model. However, in the current measurement and prediction of landslides, only the deformation rate is considered, and the terrain factors are not considered, which leads to deviations in the measurement and prediction results and low accuracy. Summary of the invention

[0004] In view of the above technical problems existing in the prior art, the present invention provides a method and system for constructing a sample set of geological disaster hazards to improve the accuracy and reliability of data annotation.

[0005] The present invention discloses a method for constructing a sample set of geological disaster hazards, comprising the following steps: collecting InSAR data and images of a target area; identifying a first area according to the image, and establishing a label for the first area, the first area including a farmland irrigation area and an ice and snow covered area; obtaining the slope of the target area; obtaining a second area with a slope less than a first threshold, and establishing a label for the second area; obtaining a mountainous area in the target area according to the image and vision; segmenting the mountainous area according to the boundary of the mountainous area to obtain a segmented body; obtaining a shape variable of the segmented body based on the InSAR data; and labeling the segmented body according to the shape variable to obtain sample data and a sample set.

[0006] Preferably, the method for training the data set includes:

[0007] Based on a machine learning method, the data set is trained to obtain a first prediction model.

[0008] Preferably, the specific method for training the data set is:

[0009] Filter the sample data of the segmented body from the data set, and construct a training set through the sample data of the segmented body;

[0010] Based on a machine learning method, the training set is trained to obtain a second prediction model.

[0011] Preferably, the method for obtaining landslide hazards based on the knowledge base or knowledge graph also includes:

[0012] Build a knowledge base or knowledge graph through data sets;

[0013] Processing the image of the area to be predicted and the InSAR data to obtain the area of ​​interest or segmentation;

[0014] Match the region or segment with the knowledge base or knowledge graph to obtain the sample data with the greatest similarity and its annotations;

[0015] According to the matched sample data and its annotations, the landslide hazard of the area or segment is obtained.

[0016] Preferably, the first area is identified based on images and vision.

[0017] Preferably, a digital elevation model is obtained based on InSAR data;

[0018] The slope of the target area is obtained according to the digital elevation model.

[0019] Preferably, the method for obtaining the mountain boundary comprises:

[0020] Based on the image and its visual features, the boundary lines of the mountain area are obtained, wherein the boundary lines include valley lines, ridge lines, road surface lines and mountain bottom boundary lines;

[0021] Based on the dividing lines and connected domains, the segmented volume is obtained.

[0022] Preferably, the method for marking the segmented body is selected from the following methods:

[0023] According to labeling rules; according to expert experience; according to historical data of landslides; based on machine learning methods; based on labeling model labeling; the features of the segmented bodies are graded and labeled according to the classification.

[0024] The present invention also provides a construction system for implementing the above-mentioned sample set construction method, comprising a collection module, a first annotation module, a second annotation module, a segmentation module and a third annotation module.

[0025] The acquisition module is used to collect InSAR data and images of the target area;

[0026] The first annotation module is used to identify a first area according to the image and create an annotation for the first area;

[0027] The second annotation module is used to obtain the slope of the target area; obtain a second area with a slope less than the first threshold, and create an annotation for the second area;

[0028] The segmentation module is used to obtain the mountainous area of ​​the target area according to the image and vision; segment the mountainous area according to the boundary of the mountainous area to obtain a segmented body;

[0029] The third annotation module is used to obtain the shape variable of the segmented body based on the InSAR data; and to annotate the segmented body according to the shape variable to obtain sample data and a sample set.

[0030] Preferably, the construction system further comprises a prediction module,

[0031] The prediction module is used to train the data set based on a machine learning method to obtain a first prediction model; the InSAR data and images of the area to be predicted are analyzed according to the first prediction module to obtain the geological disaster hazards in the area to be predicted;

[0032] Alternatively, the prediction module is used to process the image and InSAR data of the area to be predicted to obtain the area or segment of interest; match the area or segment with the data set to obtain sample data with the greatest similarity and its annotation; and obtain the landslide hazard of the area or segment based on the matched sample data and its annotation.

[0033] Compared with the prior art, the beneficial effects of the present invention are as follows: the sample data takes into account the seasonal changes, slope, deformation and mountain segmentation, making the annotation of the sample set more accurate; and improving the accuracy of identifying geological disaster hazards. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 It is a flow chart of the method for constructing a sample set of geological disaster hidden dangers of the present invention;

[0035] Figure 2 It is a schematic diagram of a mountainous dissection;

[0036] Figure 3 This is another schematic diagram of a mountainous divide;

[0037] Figure 4 It is the logic block diagram of the prediction system;

[0038] Figure 5 It is a framework diagram of the computing device provided by the present invention. DETAILED DESCRIPTION

[0039] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0040] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present disclosure can be combined with each other. The present invention is further described in detail below in conjunction with the accompanying drawings:

[0041] Overview: Synthetic Aperture Radar Interferometry (InSAR) is a radar technology used in surveying and remote sensing. It uses synthetic aperture radar to perform coherent processing on two complex value image data observed in the same area to obtain surface elevation information. Using synthetic aperture radar interferometry (InSAR) technology to generate DEM is an efficient and high-precision method. Based on synthetic aperture radar interferometry technology, it is an important means to measure terrain shape variables.

[0042] Digital Elevation Model (DEM) has a wide range of applications in geographic information systems, environmental science, civil engineering, etc. It can provide basic data support for terrain analysis, watershed division, flood simulation, soil erosion prediction, geological disaster assessment, engineering design and infrastructure planning.

[0043] like Figure 1 The first aspect of the present invention provides a sample set construction method, comprising the following steps:

[0044] Step 101: Collect InSAR data and images of the target area.

[0045] Step 102: Identify a first area according to the image and create a label for the first area.

[0046] The first region includes areas where deformation varies with the seasons, such as farmland irrigation areas and ice and snow covered areas. The marking of these areas can be set with a fixed reference value, such as 0, so that the impact of the deformation in the first region on the geological disaster risk is not considered.

[0047] Step 103: Obtain the slope of the target area according to the digital elevation model of the target area.

[0048] Step 104: Obtain a second area with a slope less than a first threshold, and create a label for the second area.

[0049] If the slope is less than 5-8°, it is considered a plain area and its label can be set to 0. The deformation variable in the second area is mainly ground subsidence.

[0050] Step 105: Obtain the mountainous area of ​​the target area according to the image.

[0051] Step 106: Based on the image and visual features, the boundary of the mountain area is obtained.

[0052] Step 107: Segment the mountainous area according to the boundary to obtain a segmented volume.

[0053] The boundaries of a mountain are usually divided by valley lines, ridge lines, road surfaces and the bottom boundary of the mountain. Figure 2 and Figure 3 The wireframe of shows the recognition result of the segmentation body. The segmentation body can be obtained based on the boundary line and the connected domain; or the segmentation body can be obtained based on the boundary line and its constraint.

[0054] Step 108: Based on the InSAR data, the deformation of the mountain area and its segmented bodies is obtained.

[0055] A mapping between the segmented body and the deformation amount is established to obtain the deformation amount of the segmented body.

[0056] Step 109: label the segmented volume according to the deformation amount to obtain sample data and a sample set.

[0057] Step 110: Identify geological disaster risks in the target area based on the annotations.

[0058] The labeling rules can be selected from the following methods: setting based on expert experience; labeling based on historical landslide data; labeling based on machine learning; labeling based on labeling models; grading the characteristics of the segmented body and labeling according to the grading. More specifically, the characteristics of the segmented body are graded and classified and respectively assigned scores, and the larger the score, the greater the risk of geological hazards.

[0059] Specifically, the characteristics of the segmented body are selected from: average slope, soil looseness coefficient, average deformation rate, rainfall, soil moisture, etc. The specific marking rules are shown in Table 1.

[0060] feature Average deformation rate Average slope (°) Soil looseness coefficient Hazard classification Points 1 >5m / s 10 0.5 high 10 2 3m / min-5m / s 20 0.5 high 9 3 1.8m / h-3m / min 25 0.7 middle 5 4 13m / month-1.8m / h 38 0.5 middle 6 5 1.6m / year-13m / month 10 0.6 Low 3 6 1.6mm / year-1.6m / year 35 0.7 Low 2

[0061] The sample data of the present invention takes into account the seasonal changes, slope, deformation and mountain segmentation, making the annotation of the sample set more accurate and improving the accuracy of identifying geological disaster hazards.

[0062] In a specific embodiment, the prediction model can be trained by the data set to make the prediction result of the prediction model more accurate. Specifically, the prediction model can be trained by machine learning methods such as LSTM (Long Short-Term Memory Network) and Convolutional Neural Networks (Region-based Convolutional Neural Networks, R-CNN) to train the data set and obtain the prediction model.

[0063] More specifically, the sample data of the segmented body can be screened from the data set, and a training set can be constructed through the sample data of the segmented body; the training set can be trained based on a machine learning method to obtain a second prediction model. The second prediction model is used to predict the geological disaster risks in mountainous areas and their segmented bodies.

[0064] In another specific embodiment, a knowledge base or knowledge graph can be constructed through a data set; the image and InSAR data of the area to be predicted are processed to obtain an area or segment of interest; the area or segment is matched with the data set to obtain sample data with the greatest similarity and its annotation, thereby obtaining the landslide hazard of the area or segment.

[0065] Knowledge Graph is a structured tool for representing and displaying knowledge. It consists of nodes (entities) and edges (relationships). Nodes represent entities, and edges represent the relationships between entities. It uses visualization technology to present knowledge in a graphical way to help users better understand and use knowledge. Knowledge graph aims to describe concepts, entities, events and the relationships between them in the objective world. It is an advanced technical means that can be used to store and manage various knowledge and to intuitively display various knowledge.

[0066] When identifying geological hazards, the deformation variables of InSAR data, optical data, DEM, etc. are combined to make a judgment. Slopes with InSAR deformation have greater landslide hazards, so the above two characteristics are fully considered when constructing the sample set.

[0067] The second aspect of the present invention provides a construction system for implementing the above-mentioned sample set construction method, such as Figure 4 As shown, it includes a collection module 1, a first labeling module 2, a second labeling module 3, a segmentation module 4 and a third labeling module 5.

[0068] The acquisition module 1 is used to acquire InSAR data and images of the target area.

[0069] The first labeling module 2 is used to identify a first area according to the image and to create a label for the first area.

[0070] The second labeling module 3 is used to obtain the slope of the target area; obtain a second area with a slope less than a first threshold, and create a label for the second area.

[0071] The segmentation module 4 is used to obtain the mountainous area of ​​the target area according to the image and vision; and to segment the mountainous area according to the boundary of the mountainous area to obtain a segmented body.

[0072] The third labeling module 5 is used to obtain the shape amount of the segmented body based on the InSAR data; and label the segmented body according to the shape amount to obtain sample data and a sample set.

[0073] The construction system further comprises a prediction module 6,

[0074] The prediction module 6 is used to train the data set based on a machine learning method to obtain a first prediction model; the InSAR data and images of the area to be predicted are analyzed according to the first prediction module to obtain the geological disaster hazards of the area to be predicted;

[0075] Or it is used to process the image and InSAR data of the area to be predicted to obtain the area or segment of interest; match the area or segment with the data set to obtain the sample data with the greatest similarity and its annotation; and obtain the landslide hazard of the area or segment based on the matched sample data and its annotation.

[0076] It should be noted that the search result analysis device provided in the above embodiment is only illustrated by the division of the above functional modules. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above.

[0077] A third aspect of the present invention provides a computing device, such as Figure 5 The computing device 11 includes a memory 12 and a processor 13. The processor may be a multi-core processor or may include multiple processors. In some embodiments, the processor may include a general-purpose main processor and one or more special coprocessors, such as a graphics processing unit (GPU), a digital signal processor (DSP), etc. In some embodiments, the processor is implemented using a customized circuit, such as an application-specific integrated circuit (ASIC) or a field programmable gate array (FPGA).

[0078] The memory may include various types of storage units, such as system memory, read-only memory (ROM), and permanent storage. Among them, ROM can store static data or instructions required by the processor or other modules of the computer. The permanent storage device may be a readable and writable storage device. The permanent storage device may be a non-volatile storage device that does not lose the stored instructions and data even after the computer is powered off. In some embodiments, the permanent storage device uses a large-capacity storage device (such as a magnetic or optical disk, flash memory) as a permanent storage device. In some other embodiments, the permanent storage device may be a removable storage device (such as a floppy disk, optical drive). The system memory may be a readable and writable storage device or a volatile readable and writable storage device, such as a dynamic random access memory. The system memory may store some or all instructions and data required by the processor at run time. In addition, the memory may include any combination of computer-readable storage media, including various types of semiconductor memory chips (DRAM, SRAM, SDRAM, flash memory, programmable read-only memory), and disks and / or optical disks may also be used. In some embodiments, the memory may include a removable storage device that can be read and / or written, such as a laser disc (CD), a read-only digital versatile disc (e.g., DVD-ROM, double-layer DVD-ROM), a read-only Blu-ray disc, an ultra-density optical disc, a flash memory card (e.g., SD card, min SD card, Micro-SD card, etc.), a magnetic floppy disk, etc. The computer-readable storage medium does not contain a carrier wave and an instantaneous electronic signal transmitted wirelessly or by wire. The memory stores executable code, and when the executable code is processed by the processor, the processor can execute the sample set construction method described above.

[0079] In some of the processes described in the specification and claims of this application and the above-mentioned figures, multiple operations that appear in a specific order are included, but it should be clearly understood that these operations may not be executed in the order in which they appear in this article or executed in parallel. The serial numbers of the operations, such as 101, 102, etc., are only used to distinguish different operations, and the serial numbers themselves do not represent any execution order. In addition, these processes may include more or fewer operations, and these operations may be executed in sequence or in parallel. It should be noted that the descriptions of "first", "second", etc. in this article are used to distinguish different messages, devices, modules, etc., and do not represent the order of precedence, nor do they limit "first" and "second" to be different types.

[0080] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for constructing a sample set of geological hazards, characterized in that: The following steps are involved: Collect InSAR data and images of the target area; According to the image, a first area is identified and a label is established for the first area, wherein the first area includes a farmland irrigation area and an ice and snow covered area; Get the slope of the target area; Obtain a second area whose slope is less than a first threshold, and create a label for the second area; Obtaining a mountainous area of ​​a target area based on the image and vision; Segmenting the mountain area according to the boundary of the mountain area to obtain a segmented body; Based on InSAR data, the shape of the segmented body is obtained; The segmented body is labeled according to the deformation amount to obtain sample data and a sample set.

2. The sample set construction method according to claim 1, characterized in that: It also includes methods for training datasets: Based on a machine learning method, the data set is trained to obtain a first prediction model.

3. The sample set construction method according to claim 2, characterized in that: Specific method for training the data set: Filter the sample data of the segmented body from the data set, and construct a training set through the sample data of the segmented body; Based on a machine learning method, the training set is trained to obtain a second prediction model.

4. The sample set construction method according to claim 1, characterized in that: It also includes methods for obtaining landslide hazards based on knowledge base or knowledge graph: Build a knowledge base or knowledge graph through data sets; Process the image and InSAR data of the area to be predicted to obtain the area of ​​interest or segmentation; Match the region or segment with the knowledge base or knowledge graph to obtain the sample data with the greatest similarity and its annotations; According to the matched sample data and its annotations, the landslide hazard of the area or segment is obtained.

5. The sample set construction method according to claim 1, characterized in that: Based on the image and vision, a first region is identified.

6. The sample set construction method according to claim 1, characterized in that: Obtain digital elevation model based on InSAR data; The slope of the target area is obtained according to the digital elevation model.

7. The sample set construction method according to claim 1, characterized in that: Methods for obtaining mountain boundaries include: Based on the image and its visual features, the boundary lines of the mountain area are obtained, wherein the boundary lines include valley lines, ridge lines, road lines and mountain bottom boundary lines; Based on the dividing lines and connected domains, the segmented volume is obtained.

8. The sample set construction method according to claim 1, characterized in that: The method for marking the segmented body is selected from the following methods: According to labeling rules; according to expert experience; according to historical data of landslides; based on machine learning methods; based on labeling model labeling; the features of the segmented bodies are graded and labeled according to the classification.

9. A construction system, characterized in that: The method for constructing a sample set according to any one of claims 1 to 8 comprises a collection module, a first labeling module, a second labeling module, a segmentation module and a third labeling module. The acquisition module is used to collect InSAR data and images of the target area; The first annotation module is used to identify a first area according to the image and create an annotation for the first area; The second annotation module is used to obtain the slope of the target area; Obtain a second area whose slope is less than a first threshold, and create a label for the second area; The segmentation module is used to obtain the mountainous area of ​​the target area according to the image and vision; segment the mountainous area according to the boundary of the mountainous area to obtain a segmented body; The third annotation module is used to obtain the shape amount of the segmented body based on the InSAR data; and to annotate the segmented body according to the shape amount to obtain sample data and a sample set.

10. The construction system according to claim 9, characterized in that: It also includes a prediction module, The prediction module is used to train the data set based on a machine learning method to obtain a first prediction model; the InSAR data and images of the area to be predicted are analyzed according to the first prediction module to obtain the geological disaster hazards of the area to be predicted; Or it is used to process the image and InSAR data of the area to be predicted to obtain the area or segment of interest; match the area or segment with the data set to obtain the sample data with the greatest similarity and its annotation; and obtain the landslide hazard of the area or segment based on the matched sample data and its annotation.

Images