LU504675B1 - A METHOD FOR OBTAINING INFORMATION ABOUT LANDSLIDE HAZARD BASED ON UAV IMAGE DATA IN HIGH MOUNTAIN VALLEY AREAS - Google Patents

Abstract

The present invention discloses a method for obtaining landslide hazard information based on UAV image data in high mountain valley areas, which relates to the technical field of extracting landslide disaster information in high mountain valley areas, and includes the following steps: Step S10, UAV data acquisition , including imagery and LiDAR data; Step S20, image data inspection and quality assessment; Step S30, data product and 3D scene generation; Step S40, extraction of landslide features and identification of disaster information, classification of landslide type and susceptibility, monitoring of landslide changes; Step S50, analyzing the accuracy of landslide information extraction; a quantitative method is used to calculate the extraction accuracy of landslide information. The method of obtaining landslide hazard information based on UAV image data in high mountain valley areas can effectively extract types of landslide hazards and their area changes, etc. At the same time, the spatial and temporal distribution of landslide hazard types can effectively monitor the general occurrence and development trend of landslides, and provide technical support for decision-making in landslide risk assessment and implementation of 20 prevention projects.

Description

A method for obtaining information about the landslide hazard on the LU504675

Basis of UAV image data in high mountain valley areas

Technical part

The invention relates to the technical field of obtaining information about the risk of landslides in mountain valley areas, in particular to a method for obtaining information about the risk of landslides in mountain valley areas on the basis of

UAV image data.

Technology in the background

A landslide is a slope with a certain gradient where natural or from

Man-made destructive forces act, causing part of the slope to slide in a certain direction. Monitoring the risk of landslides in

High valley areas refers to the application of various techniques and methods to provide a comprehensive three-dimensional approach to all types of geological

Hazard activities and the dynamics of various triggering factors at different spatial scales and in a hierarchical manner according to the phase pattern of the

development over time.

Currently, to identify and monitor landslides, earthfills and

Debris flows use the following traditional mapping techniques: on-site surveys, satellite-based remote sensing (optical remote sensing, radar remote sensing) and aerial photographic remote sensing (airborne radar remote sensing, unmanned aerial photographic remote sensing). Given the current state of science and...

Despite increasingly rigorous monitoring, information about the extent, timing and location of landslides still cannot be accurately predicted, and rapid and accurate location of landslide-prone areas is key

Emergency rescue measures after a disaster, but landslides usually occur

Mountainous regions with complex terrain, which makes rescue work significantly more difficult.

Landslide identification and information gathering was first proposed and is still widely used today as a field survey and site investigation technique that provides accurate and precise landslide information, but due to the unstable geology of new landslides, it can easily occur

Secondary slides come that threaten the lives of the investigators. In addition, it often damages and blocks roads, causing inconvenience to traffic and making on-site surveys lengthy, costly and inefficient, resulting in subsequent

Rescue and emergency response after the disaster was significantly affected.

The triggering factors for landslides in high mountain valleys are complex, especially damage caused by human excavations, heavy rains and earthquakes, which often trigger large landslides in a short period of time. In such cases, the conventional

Obtaining information on site is difficult to meet the needs of emergencies.

As a result, remote sensing technology has proven to be an effective technical support for emergency response to large-scale disasters. After a

Disaster, methods such as interpreting and analyzing stereo images from aerial photography can be used to detect and identify landslides. With the gradual increase in the resolution of satellite sensors, the technical research on landslide mining using remote sensing has received more and more attention

Researchers receive and the main data sources for obtaining

Landslide information is SPOT, QUICKBIRD and IKONOS. However, these high spatial resolution remote sensing images-U504675 are difficult and expensive to obtain and are influenced by climate and cloud cover. The advantages of remote sensing are low cost, flexible landing and takeoff, high resolution and variable

shooting angles, etc. In conjunction with the technology for processing

Remote sensing data offers significant advantages for use in emergency rescue, disaster investigation and monitoring, and geological assessment

Disasters. Therefore, the use of UAV remote sensing data in the

Landslide research to obtain information about landslides is gradually increasing

Mainstream.

In summary, in high mountain valleys where landslides occur,

Since sinkholes and debris flows are common geological hazards, efficient collection of high-precision remote sensing data from disaster areas is a priority.

Traditional monitoring methods are not suitable for high mountain valleys where the terrain is heavily hilly and is often hindered by non-human conditions such as

Influence of signals, difficulties in setting up machines and treacherous

Roads compromised, making monitoring difficult. The use of optical

Remote sensing satellites and radar remote sensing is smaller for monitoring

Areas insufficient, and there are other problems such as insufficient image resolution and difficulties in deciphering and identifying miscalculations that the study of landslides, landslides and mudslides as geological

Dangers do not do justice. At the current stage, there are mainly two types of UAV images: three-band visible and four-band visible + near-infrared images, of which the visible images are the most widely used. For this reason, this invention is based on the technology of extracting landslide information from optical

UAV remote sensing data to provide a deeper application of UAV remote sensing in the field of landslide extraction and provide a methodological reference and technical

Support for disaster prevention and mitigation, post-emergency rescue

Catastrophe and other work to offer.

Content of the invention (1) Technical problem solved

The present invention provides a method for obtaining information about the

Landslide hazard in high mountain valleys based on UAV image data is available to address the disadvantages of existing methods for obtaining information about the

To overcome the danger of landslides, such as high costs, long period of time, strong influence of

Clouds and terrain shadows and poor security. (2) Technical solutions

In order to achieve the above purpose, the present invention is realized through the following technical solution: a method for obtaining information about

Landslide disasters in high alpine valley areas based on UAV image data, the present invention provides a method for obtaining information about

Landslide disasters based on UAV imagery, which includes the following steps:

Step 1, UAV data collection, including field investigation,

Ground control point setup and measurement, route setup, parameter setting,

Flight operations, obtaining UAV aerial image data; LU504675

Step 2, image data inspection and quality assessment, including

Data quality inspection, aerial photography, pre-processing.

Step 3, data product and 3D scene generation, data including DRG (Digital

Raster Map), DLG (Digital Line Delineation Map), DEM (Digital Elevation Model) and DOM (Digital Orthophoto Map);

The technological process of 3D scene construction mainly includes the

Inspection and pre-processing of image data, the comparison of features from several

Views, the triangulation of aerial images, the creation of a dense point cloud, the

Construction of a 3D TIN grid, texture mapping and creation of a digital one

product;

Step 4, extraction of landslide features and identification of

Disaster information, landslide type, classification of landslide susceptibility,

Monitoring landslide changes based on analysis of image spectra,

Shape features, terrain features, texture features, multi-scale segmentation of

remote sensing images, classification of remote sensing images, visual interpretation,

Landslide extraction

Step 5, analysis of the accuracy of landslide information extraction.

The present invention provides a method for the extraction of

Landslide disaster information based on UAV image data, which includes the following steps: Step S10, UAV data acquisition, including image data and

LiDAR data; Step S20, image data checking and quality evaluation, including

Data quality check, pre-processing of aerial photographs; Step S30, generation of

Data products and 3D scenes, including inspection and pre-processing of image data,

Matching features from multiple views, aerial triangulation, generation of

Dense matching point clouds, 3D TIN grid construction, texture mapping and

Generation of digital product results; Step S40, extraction of landslide features and identification of hazard information, classification of landslide types and susceptibilities, monitoring of landslide changes; Step S50, analyzing the accuracy of landslide information extraction; Use of quantitative methods for

Calculating the accuracy of landslide information extraction, using two

Accuracy measures of the error T.E and the Kappa coefficient for the quantitative

Assessment of landslide extraction accuracy, cross matrix of extraction results.

Furthermore, in S10, including the following: S11, recording an image card of the

study area, essentially understanding the surrounding terrain and

Terrain and traffic from the picture, and based on the information received, the

Conducting a survey of the field and selecting a suitable location as

UAV takeoff and landing site; S12, Laying Ground Control Points and Acquiring Three-Dimensional Coordinate Information: The actual laying is carried out according to the extent of the test area and the height differences of the ground based on the

Field survey.

Furthermore, in S12 there will be a total of 15 ground control points at the corners and in the

Laid out in the middle of the test area with the ground control points in an L-shaped pattern, with the middle control point L-shaped white and the others L-shaped with a blue tape pattern.

Furthermore, after S12, there is also: S13, collecting three-dimensional frY904675

Coordinate information, including longitude, dimension and altitude, of the circular

Ground control point position using a dynamic

real-time measuring instrument; S14, route relocation, parameter setting and flight operations, obtaining aerial photo data.

Further, in S14, including the following: S141, UAV battery assembly,

remote control battery; S142, after selecting a suitable location, according to the specific terrain, combined with the actual situation for laying the route; S143, visually estimate the length of the subject, and choose an appropriate flight direction; S144, Depending on the complexity of the topic, the study area is divided and the route on the

Based on the surface condition of the test area and the range of the drone, determining an overlap of 90% of the course and 80% of the sides.

Furthermore, in S14 the flight altitude of the UAV is set to 200 m

Flight speed is Sm/s, the lens is shot vertically down to the ground, the flight takes place along the planned route, and the UAV image data and

LiDAR data is collected partition by partition.

Further included in S20: S21, data quality check: the acquired image data is checked, the check includes: image exposure, image blur, image integrity and continuity, if there is a problem in the check content, make or reshoot; S22,

Pre-processing of the aerial image: the continuity of the image data, the integrity of the information is checked, the image with large distortion is rejected, and the image is with

Noise removal, enhancement and color matching pre-processed.

Furthermore, step S30: S31, after completion of the UAV aerial survey, includes

performing the 3D scene construction and the image composition processing;, S32, using the 3D ground measurement coordinate information of the ground control points set in step S10, performing the geometric correction and the

Conversion of the spatial projection of the aerial survey images to orthophotos

centimeter resolution and create digital surface models; where the selected

Ground control points must appear on at least five aerial survey images in order to achieve the

To spatially align UAV image data twice in the test area.

Furthermore, step S40 includes: S41, acquiring landslide information on the

Basis of image spectral features, terrain features, shape features,

Texture features, multi-scale segmentation of remote sensing images, classification of

remote sensing images and visual interpretation; S42, Extracting landslide features using object-oriented methods based on orthophotos taken from

UAV image data was generated.

Furthermore, S42: S421 includes multiscale segmentation of the image data obtained

creation of image objects; S422, construction of a feature space and analysis of various

Characteristics of the image objects, including shape, texture, vegetation, spectrum and others

Characteristics; S423, Implementing Landslide Extraction After an Earthquake on the

Basis of the German eCognition software platform. (3) Beneficial effects

The present invention provides a method for extracting information about

Landslide disasters in high mountain valleys based on UAV imagery. It has the following beneficial effects:

The method of obtaining information about landslide hazards on dec 504675

Based on UAV image data can effectively extract information about landslide hazard types and their area changes etc. At the same time, the spatial and temporal

Distribution of landslide hazard types effectively monitor the frequency occurrence and development trend of 5 landslides, and provide technical support for decision-making in landslide risk assessment and the deployment of prevention measures.

Description of the attached drawings

Figure 1 shows a flowchart of extracting landslide hazard information in the high mountain valley from the UAV image data in the present invention;

Figure 2 shows a schematic diagram of a digital elevation model of the

study area for the UAV image data in the present invention;

Figure 3 shows a schematic diagram of the orthophoto of the UAV in the present

Invention;

Figure 4 shows a schematic diagram of the surface landscape of the UAV image data within the scope of the present invention;

Figure 5 shows a schematic diagram of the external field survey site of the present invention.

Detailed description

The technical solutions in the embodiments of the present invention will be discussed below in conjunction with the accompanying drawings in Figs

Embodiments of the present invention are clearly and completely described, and it is to be understood that the described embodiments represent only a part of the embodiments of the present invention, and not all of them. Starting from the

Embodiments of the present invention, all other embodiments which can be achieved by one skilled in the art without any creative work are included within the scope of the present invention, which are described hereinafter in connection with certain

Embodiments will be described in more detail.

Embodiment 1

Referring to Figures 1-5, the present invention provides a method

Obtaining information about landslide risk based on UAV imagery in Alpine gorge areas, the method comprising the following steps:

Step 1, UAV data acquisition, based on field survey, ground control point laying and measurement, route laying, parameter setting and flight operation, to

Obtain aerial photography data.

The field survey includes the following elements:

laying ground control points and collecting 3D coordinate information;

Laying ground control points using circular patterns as control points, the diameter of the circle being at least 10 times the image resolution;

Collect 3D geographical coordinate information corresponding to the circle center of the circular pattern, or longitude, latitude and altitude to obtain 3D coordinate information.

The UAV image data is collected by setting the route, setting the parameters and flying the operation; when collecting the UAV image data, the UAV follows the planned route, setting 80% course overlap and 70% side overlap, and a flight altitude of 50-120m, and collects the drone image data several times, each time with the same aerial survey parameters, where the collection time is in

Freezing and thawing periods are divided into to obtain the drone image data.

Step 2, review of the image data and quality assessment, including review of the

Data quality, aerial photography, pre-processing, texture analysis, model analysis, etc,

Distortion correction, | Zero-triple encryption, geometric correction and

image enhancement;

Assembling the UAV image data and performing geometric correction and spatial alignment; Combining ground control points and their measured 3D geographic coordinate information, merging and geometrically correcting the

UAV imagery, where the ground control points appear on at least 5 individual UAV imagery and multiple spatial alignments of the UAV imagery are performed using ground control points of the same name.

Step 3, data product and 3D scene generation: The technological process of 3D scene generation mainly includes the inspection and pre-processing of image data, the matching of image features from multiple views, the aerial triangulation, the

Generation of a dense point cloud, the construction of a 3D TIN grid, the

Texture mapping and the creation of a digital product.

Step 4, extraction of landslide features and identification of

Hazard information, landslide type, classification of landslide susceptibility and

Monitoring landslide changes.

On the basis of spectral analysis of images, analysis of shape and

Terrain features, the analysis of texture features, the multi-level segmentation of

remote sensing images, the classification of remote sensing images and the visual

Interpretation, landslide information is extracted.

Based on the orthophoto generated from the UAV image data, an object-oriented approach is used to extract landslide features.

The basic idea is as follows: First, a multi-scale segmentation of the captured image data is carried out to generate image objects; then a

Feature space is constructed and various features of the image objects are analyzed, including shape, texture, vegetation, spectrum and other features; eventually it will

Post-earthquake landslide extraction performed based on the German eCognition software platform.

Step 5, evaluate the analysis of the accuracy of landslide extraction on the

Basis of UAV image data.

The present embodiment further illustrates the invention in conjunction with the accompanying drawings.

This application is based on the mudslide valley study area of the

Dabai Nai River in the southwest of Dongchuan County, Kunming City, Yunnan Province, which includes the Dabai Nai River and the Xiaobai Nai River, which are both tributaries of the

Xiaojiang River Basin, extending from southwest to northeast, with a geographical location of 103° 9' 50" E~103° 13' 13" E, 25° 55' 58" N~26° 01' 13"

N

The valley has a long and narrow gorge, and the terrain in the depression is steep, with high slopes, steep mountains, and sparse vegetation on both sides of the gorge. The

The surface of the gully is riddled with constant leaching and accumulation of sediment/504675, with the water flowing from the center of the gully through the entire valley, the ground in the middle being soft and prone to collapse, while on both sides of the gully there are huge

Sediment mounds form and the soil is harder. The complex terrain is excellent

Study area for researching the use of UAV remote sensing technology for

Monitoring of mudslides.

The procedure for obtaining information about the danger of permafrost on the

UAV image data basis includes the following steps:

Step 1, UAV data collection, including image data and LiDAR data.

The process of UAV remote sensing data collection includes field survey, the

Setting and measuring ground control points, route setting, parameter setting, flight operations and the acquisition of aerial image data.

Field survey: Mudslides are mostly in mountainous areas with complex

Find terrain.

Before deployment, a field inspection of the study area was required to determine the basic conditions of the study area, the presence of

Understand no-fly zones in the area and the landing and takeoff conditions of the UAV.

First, the image map of the study area is created using the

Aowei mapping software created to gain a basic understanding of the surrounding terrain and traffic from the image. Based on the won

information, a field inspection will be carried out and a suitable location will be chosen as a starting point

Landing point selected for the UAV.

Laying ground control points and capturing three-dimensional

Coordinate information: Depending on the extent of the test area and the height differences

The actual laying takes place on the basis of the terrain survey.

A total of 15 were found at the four corners and in the center of the test area

Ground control points laid out in an L-shaped pattern, with the center control point marked with white spray paint and the others with blue tape.

Subsequently, with the help of real-time kinematic measuring devices (Real-TimeKinematic,

RTK) 3D coordinate information, including longitude, dimension and elevation, the

Ground control points captured in the center of the circle. The 3D coordinates come from the

CORS (ContinuouslyOperatingReferenceStation) network of Yunnan receiving stations

Provincial Bureau of Surveying, Mapping and Geoinformation with an accuracy of

centimeters.

The model of the RTK meter in this embodiment is: Southern Thousand Seek

SRmini, but not limited to.

In this variant, the 3D coordinates of the ground control points come from, among other things, the network of continuously operating receiving stations for

Yunnan Provincial Bureau of Surveying Satellite Positioning Services (YNCORS),

Mapping and geoinformation, but not limited to.

The route is determined, the parameters are set and the flight operations are carried out to obtain aerial photography data: The external aerial photography is carried out when the weather is clear, there are no low altitude clouds, the

Wind speed is less than 8m/s, visibility is more than Skm and the

The angle of the sun is greater than 45°. LU504675

First, the battery of the drone and the battery of the remote control are assembled.

After selecting a suitable location, the route is determined according to the specific terrain and in conjunction with the actual situation. Since the

If the drone is at the bottom of the valley, try to determine the altitude depending on the location of the object, fly to the highest point where the object is located, make sure

Make sure there are fewer obstacles nearby, the signal is stable and safe, and select the current altitude as the flight altitude; in turn, visually estimate the length of the object and choose a suitable direction of flight.

Finally, the study area is divided into zones depending on the complexity of the object, and the route is planned based on the surface condition of the test area and the range of the drone, setting a 90 percent overlap in the direction and an 80 percent overlap in the side.

Due to the hilly terrain of the test area, the UAV was set to fly at an altitude of 200m and at a speed of 5m/s with the camera pointing vertically downward at the ground. The UAV is operated to fly along the planned route and collect the UAV image data and LiDAR data partition by partition.

The UAV aerial survey platform in this example is the DJI M300 UAV with the

Camera RainpooM6P2020081S; The LiDAR data is used, among other things, with the HASCO

Navigation AA450 LiDAR measurement system captured, but not limited to.

Step 2, image data review and quality assessment, including review of the

Data quality and preprocessing of aerial photographs.

Data quality verification: After the UAV completes flight operations, the captured image data needs to be verified. The test includes: image exposure,

Image blur, image integrity and continuity. If there are problems with the content of the inspection, the recordings must be repeated or retaken to obtain the best data.

Pre-processing of aerial images: The continuity of the image data and the integrity of the information are checked. At the same time, images with strong distortions and

Underexposures caused by weather and lighting factors during the shooting process are eliminated, and pre-processing steps such as

Noise removal, enhancement and color matching are performed on the images to ensure that the quality of the simultaneous image data is as consistent as possible. After loading the data, you can view the waypoint information in the 3DViewer viewport for initial review.

Step 3, data product and 3D scene generation: The technological process of 3D scene generation mainly includes inspection and pre-processing of image data, feature matching from multiple views, aerial triangulation, the

Generation of a dense point cloud, the creation of a 3D TIN grid, the

Texture mapping and the creation of a digital product.

After completing the UAV aerial survey, the 3D scene construction is done with the

ContextCaptureCentert software from Bently. After completion of the

UAV aerial photo survey, Pix4DMapper software is used for image stitching processing, and geometric correction and geospatial projection conversion de}4504675

Aerial survey images are created using the 3D coordinate information of the in

Step one established ground control points. The final orthophoto with a

Resolution of one centimeter and the digital surface model (DigitalSurfaceModel,

DSM) are created.

In this step, it is important to note that the selected ground control points should be present on at least 5 aerial photographs. The spatial orientation of the

UAV image data for the test area is generated twice using the same name

Ground control points.

The image data obtained through UAV photogrammetry is rich in image information and, in addition to RGB color information, also contains POS data (longitude, latitude and...

Elevation data) that provide auxiliary data for triangulation from the air. 3D models, digital orthophotos (DigitalOrthophotoMap, DOM), digital surface models (Digital SurfaceModel, DSM), digital elevation models (DigitalElevationModel, DEM) and closely coordinated point cloud data can be obtained. The data sources are intuitive, rich in 3D detailed texture information, and can be used as key data for extraction and

Analysis of monetary characteristics can be used while the point clouds

Color information included, making the data rich in results, but also the

Data processing difficult.

Step 4, extraction of landslide features and identification of

Hazard information, landslide type, classification of landslide susceptibility and

Monitoring landslide changes.

On the basis of spectral analysis of images, terrain feature analysis, shape feature analysis, texture feature analysis, multi-stage

Segmentation of remote sensing images, classification of remote sensing images and visual interpretation are used to extract landslides.

Based on the orthophoto generated from the UAV image data, an object-oriented approach is used to extract landslide features.

The basic idea: First, a multi-stage segmentation of the captured image data is performed to generate image objects; second, a feature space is constructed and various features of the image objects are analyzed, including shape, texture, vegetation, spectral and other features; Finally, the extraction of landslides is carried out after

Earthquake conducted based on the German eCognition software platform.

The topographic and geomorphological features of the landslide are relatively obvious, and coupled with the fact that the landslide is also rich

features (such as color tone, geometry, texture, etc.) on the remote sensing image, landslide extraction can be achieved by comparing and analyzing the above features.

When interpreting landslide features, the following aspects are combined: (1) Color tone features, landslides appear in light color tones such as white, gray, light gray and off-white on optical remote sensing images. In areas where landslides occur, there is little or no sparse vegetation on the ground surface, and large accumulations of rocks, debris, and others occur at the base of the landslide

Rubble as the trees in the area are destroyed.

(2) Image features, The landslide is a combination of back wall and side wall irr/504675

Shape of a closed chair, and the landslide exhibits a variety of morphological

Features, usually in the form of a long tongue, oval, horseshoe, etc. (3) Vegetation features, as part of the slope along the corresponding surface for

Shear movement and the formation of landslides, so that the occurrence of landslides

Will cause damage to surrounding vegetation and surface once the

Landslide occurs, the vegetation on the landslide body is also influenced by the vegetation in the landslide

Environment may be different, you can observe that the vegetation around the landslide tilted and directly covered the landslide body, so the vegetation is an important

feature to identify the landslide. (4) Textural features, the sliding effect of the landslide causes the landslide to have different spectral reflectance characteristics in different places, i.e. the

Texture of landslide has gross and fine difference.

Step 5, landslide information extraction accuracy analysis, the invention uses the UAV ultra-low-altitude remote sensing data, so that the quantitative method

Calculating the extraction accuracy of the landslide information, with the error T.E and

Kappa coefficient two accuracy measures for the quantitative assessment of landslide

Extraction accuracy, extraction results of the cross matrix.

Combined with the above embodiments, it can be illustrated that the landslide hazard information extraction method based on UAV image data in the present invention can effectively extract landslide hazard types and their area

Changes, etc. At the same time, the spatial and temporal distribution of

Landslide hazard species can effectively monitor the co-occurrence and the

Development trend of landslides, and provide technical support for the

Decision making of landslide risk risk assessment and prevention project deployment.

Although embodiments of the present invention have been shown and described, it will be apparent to those skilled in the art that various variations, modifications,

Substitutions and variations to these embodiments may be made without departing from the principles and spirit of the present invention, the scope of which is limited by the appended claims and their equivalents.

Claims (10)

Claims LU504675 1. A method for obtaining landslide hazard information based on UAV image data in high mountain valley areas, characterized by: comprising the following steps: Step S10, UAV data acquisition, including image data and LIDAR data; Step S20, inspection and quality assessment of the image data, including data quality inspection and preprocessing of aerial image data; Step S30, data product and 3D scene generation, including image data inspection and preprocessing, multi-view image feature matching, aerial triangulation, dense matching point cloud generation, 3D TIN mesh construction, texture mapping, and digital product result generation; Step S40, extraction of landslide features and identification of hazard information, classification of landslide types and susceptibility, monitoring of landslide changes; Step S50, analyzing the accuracy of landslide information extraction; Using a quantitative method to calculate the extraction accuracy of landslide information, using two accuracy measures of error T.E and Kappa coefficient for quantitative assessment of landslide extraction accuracy, cross matrix of extraction results. 2. The method for obtaining landslide hazard information based on UAV image data in high mountain valley areas according to claim 1, characterized in that: in S10 the following steps are included: S11, taking an image map of the study area, obtaining a basic understanding of the surrounding terrain and traffic from the image, conducting a survey of the area based on the information obtained and selecting a suitable location as a UAV take-off and landing site; S12, laying ground control points and collecting three-dimensional coordinate information: according to the extent of the test area and ground survey waves, based on the field survey, the actual laying. 3. The method for obtaining landslide hazard information based on UAV image data in high mountain valley areas according to claim 2, characterized in that: in S12, a total of 15 ground control points are laid at the four corners and in the middle of the test area, and the ground control points are in are arranged in an L-shaped pattern, of which the control points in the center are arranged in an L-shaped white spray paint and the others in an L-shaped blue band pattern. 4. The method for obtaining landslide hazard information based on UAV image data in high mountain valley areas according to claim 3, characterized in that: after S12, further comprising the steps: S13, collecting three-dimensional coordinate information including longitude, dimension and altitude , the circular center position of the ground control point using a real-time dynamic measuring instrument; S14, setting the route, setting parameters, performing flight operations and obtaining aerial photography data. 5. The procedure for obtaining information about the danger of landslides on the UAV image data base in high mountain valley areas according to claim 4, characterized in that: in S14, the steps include: S141, assembling the UAV battery, the remote control battery; S142, after selecting a suitable location, laying out a route according to the specific terrain, taking into account the actual situation; S143, visual estimation of the length of the object and selection of an appropriate flight direction; S144, depending on the complexity of the object and based on the surface condition of the test area and the range of the drone, the study area is divided, the route is planned and a 90% overlap of the flight direction and an 80% overlap of the sides are set. 6. The method for obtaining information about the risk of landslides based on UAV image data in high mountain valley areas according to claim 5, characterized in that in S14 the UAV flight altitude is set to 200m, the flight speed is 5m / s, the lens is vertical is aimed at the ground below, the flight takes place along the planned route and the UAV image data and LiDAR data are collected part by part. 7. The method for obtaining information about landslide risk based on UAV image data in high mountain valley areas according to claim 1, characterized in that: in S20 comprises the following steps: S21, data quality check: the acquired image data is checked, and the Inspection includes: image exposure, image blur, image integrity and continuity, and if there are problems in the inspection content, a touch-up or reshoot is made; S22, Aerial image pre-processing: The continuity of image data and the integrity of information are checked, the image with large distortions is rejected, the image is pre-processed by noise removal, enhancement and color adjustment. 8. The method for obtaining landslide hazard information based on UAV image data in high mountain valley areas according to claim 7, characterized in that: in step S30, the following steps are carried out: S31, after completion of the UAV aerial survey, the construction of a 3D -Scene and performing image composition processing; S32, using the 3D ground measurement coordinate information of the ground control points set in step S10, performing geometric correction and geospatial projection conversion of the aerial survey images to generate centimeter resolution orthophotos and digital surface models; The selected ground control points must appear on at least five aerial survey images to spatially align the two UAV imagery of the test area. 9. The method for obtaining landslide hazard information based on UAV image data in high mountain valley areas according to claim 1, characterized in that: in S40, the following steps include: S41, obtaining landslide information based on image spectral feature analysis, terrain feature analysis, shape feature analysis , Texture feature analysis, multi-scale segmentation of remote sensing images, classification of remote sensing images and visual interpretation; S42, Extraction of landslide features based on the orthophoto generated from the UAV imagers rU504675 using an object-oriented approach. 10. The method for obtaining landslide hazard information based on UAV image data in high mountain valley areas according to claim 9, characterized in that: S42 comprises the following steps: S421, performing multi-scale segmentation on the obtained image data to generate image objects; S422, constructing a feature space and analyzing various features of the image object, including shape, texture, vegetation, spectral and other features; S423, Performing post-earthquake landslide extraction based on the German eCognition software platform.