Landslide early warning method based on ground-based radar high-frequency surface domain data

Abstract

The invention relates to a landslide early warning method based on ground-based radar high-frequency surface domain data. The method comprises the following steps: extracting a landslide main body region through an adaptive region growing method; extracting data about deformation and coherence of the landslide main body area, wherein the data at least comprises deformation standard deviation and average coherence of different deformation pixel points at the same time; the landslide early warning method is established according to the statistical characteristics of the landslide area data, and at least comprises the step of determining that the side slope is in a temporary sliding stage under the condition that the tangent angle of the deformation degree of the comprehensive standardized area at different moments is greater than a threshold value. According to the embodiment of the invention, the early warning method is combined with ground-based radar high-frequency surface domain data and has universality, and powerful guarantee is provided for landslide prevention.

Description

Technical Field

[0001] The present invention relates to the field of geological disaster monitoring and remote sensing technology, and in particular to a landslide early warning method based on high-frequency surface data of ground-based radar. Background Art

[0002] Compared with traditional monitoring methods, a major innovation of ground-based radar monitoring is that it can realize surface monitoring through regional scanning, which can obtain multiple basic data such as the number of deformation pixels, coherence change data and slope displacement in the target monitoring area. However, the current landslide deformation feature extraction of ground-based radar still uses the single-point displacement curve analysis method of traditional slope monitoring technology, which fails to fully utilize the data characteristics of high-frequency surface domain of ground-based radar. Summary of the invention

[0003] The present disclosure intends to provide a landslide early warning method based on ground-based radar high-frequency surface data, which is a general-purpose early warning method combined with ground-based radar high-frequency surface data to provide a strong guarantee for the prevention of landslides.

[0004] According to one of the solutions disclosed herein, a landslide early warning method based on high-frequency surface data of ground-based radar is provided, comprising:

[0005] The main area of ​​the landslide is extracted by adaptive region growing method;

[0006] Extracting deformation and coherence data of the main area of ​​the landslide, the data at least includes: deformation standard deviation and average coherence of different deformation pixels at the same time;

[0007] A landslide early warning method is established based on the statistical characteristics of landslide surface data, which at least includes: when the tangent angle of the comprehensive standardized regional deformation degree at different times is greater than a threshold, it is determined that the slope is in the pre-slip stage.

[0008] In some embodiments, the method of extracting the main area of ​​the landslide by an adaptive region growing method includes:

[0009] Identify deformation areas by mapping ground-based radar pixels to actual slope topography;

[0010] The adaptive region growing method is used to extract the main deformation area identified by ground-based radar.

[0011] In some embodiments, the method of extracting the deformation main body region identified by the ground-based radar using an adaptive region growing method includes:

[0012] Based on the ground-based radar deformation map, the deformation threshold is calculated;

[0013] The obtained threshold is used as the seed point of the region growing method and region growing is performed to extract the main area of ​​the landslide.

[0014] In some embodiments, calculating the deformation threshold based on the ground-based radar deformation map includes:

[0015] Construct a two-dimensional histogram, calculate the gray level and neighborhood gray level of each pixel in the radar deformation map, and increase the count at the corresponding position in the two-dimensional histogram;

[0016] Calculate the between-class variance based on the means of multiple regions and the global mean of the two-dimensional histogram;

[0017] The segmentation threshold that maximizes the inter-class variance is taken as the optimal threshold.

[0018] In some embodiments, the obtained threshold is used as a seed point of a region growing method and region growing is performed, including:

[0019] Starting from the seed point, if the pixels in its area meet the conditions, they are added to the area where the current seed point is located, and the cycle continues until there are no pixels that meet the conditions and can be added.

[0020] In some embodiments, extracting data about deformation and coherence of the main area of ​​the landslide includes:

[0021] Extract deformation data and accumulate them to obtain a deformation curve cluster of each pixel point that changes cumulatively over time;

[0022] Extract the coherence change of each pixel over time;

[0023] The obtained data is used to calculate the deformation standard deviation and average coherence of different deformed pixels in the deformation main area at the same time.

[0024] In some embodiments, a landslide early warning method is established based on the statistical characteristics of landslide surface data, including:

[0025] Using the obtained deformation standard deviation and average coherence, the comprehensive standardized regional deformation degree is calculated and the comprehensive standardized regional deformation degree-time curve is constructed;

[0026] Calculate the tangent angle of the comprehensive standardized regional deformation degree-time curve;

[0027] When the tangent angle is greater than the threshold angle, it indicates that the slope is in the stage of imminent sliding, and an alarm is issued. In some embodiments, the field pixel value of the seed point satisfies:

[0028] SPV-SPV*K threshold ≤DPV≤SPV+SPV*K threshold

[0029] Where: DPV is the domain pixel value of the seed point; SPV is the seed point value; Kthreshold It is the ratio of the difference between the domain value and the growth point value, which is used to control the growth rate of the region and the number of regions after growth. The range is [0,1]. If the value is smaller, the number of regions after regional growth will increase.

[0030] In some embodiments, the coherence formula:

[0031]

[0032] Where: k is the number of pixels in the window, A is the backscattering coefficient of the target, λ is the radar operating wavelength, ΔR is the target radar line of sight deformation, and Δβ is the scattering phase.

[0033] In some embodiments, the formula for calculating the tangent angle of the comprehensive normalized regional deformation degree-time curve is:

[0034]

[0035] Where: α is the tangent angle; CSRDD i With CSRDD i are the standardized regional deformation degrees at the i-th moment and the i-1-th moment respectively; t i and t i-1 are the i-th moment and the i-1-th moment respectively.

[0036] The landslide early warning method based on high-frequency surface data of ground-based radar in various embodiments of the present disclosure extracts the main area of ​​the landslide at least by adaptive regional growing method; extracts the deformation and coherence data of the main area of ​​the landslide, which at least includes: the deformation standard deviation and average coherence of different deformation pixels at the same time; establishes a landslide early warning method according to the statistical characteristics of the landslide surface data, which at least includes: when the tangent angle of the comprehensive standardized regional deformation degree at different times is greater than the threshold, it is determined that the slope is in the pre-slip stage, thereby comprehensively considering the deformation change trend and coherence change trend of all pixels in the deformation area. After multiplying the regional deformation standard deviation by the inverse of the average coherence, the comprehensive standardized regional deformation degree (CSRDD) curve shows a stronger change trend in the landslide stage. By calculating the tangent angle of the CSRDD curve, the division of the landslide stage can be realized, and the landslide early warning criterion can be determined in combination with the specific slope characteristics. The method disclosed in the present disclosure forms a comprehensive quantitative early warning indicator representing the overall characteristics of these indicators by integrating various indicators reflecting deformation trends. In addition, the diversified slope data detected by the ground-based radar was fully utilized to reduce the false alarm rate, improve the warning accuracy, make up for the shortcomings of the single-point displacement curve analysis method of traditional slope monitoring technology, and provide a new direction for the application of ground-based radar in landslide early warning.

[0037] It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the disclosure, as claimed. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In the drawings, which are not necessarily drawn to scale, similar reference numerals in different views may represent similar components. Similar reference numerals with letter suffixes or similar reference numerals with different letter suffixes may represent different instances of similar components. The drawings generally illustrate various embodiments by way of example and not limitation, and together with the description and claims, serve to explain the disclosed embodiments.

[0039] Figure 1 A schematic diagram of the implementation process of a landslide early warning method based on high-frequency surface data of ground-based radar according to an embodiment of the present disclosure is shown. DETAILED DESCRIPTION

[0040] In order to make the purpose, technical solution and advantages of the embodiments of the present disclosure clearer, the technical solution of the embodiments of the present disclosure will be clearly and completely described below in conjunction with the drawings of the embodiments of the present disclosure. Obviously, the described embodiments are part of the embodiments of the present disclosure, not all of the embodiments. Based on the described embodiments of the present disclosure, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present disclosure.

[0041] In order to deal with geological disasters such as landslides caused by extremely diverse terrain and landforms, multi-source stereoscopic monitoring technology integrating sky, air and ground has been used to achieve early identification and monitoring and early warning of geological disaster risks, which will further enhance the proactive prevention capabilities of landslides.

[0042] In the monitoring and early warning of disaster accidents, accurate measurement of the target's micro-deformation is the basis for judging the target state and realizing forecasting and early warning. Traditional geological disaster monitoring methods mainly rely on crack meters, precision leveling and inclinometers to measure the deformation of feature points, and then obtain the overall deformation information of feature points. However, although these methods have high single-point monitoring accuracy, they are costly and cannot be applied to areas that are difficult for personnel to access or large areas. In contrast, ground-based micro-change monitoring radar has great potential in landslide monitoring due to its all-day, all-weather imaging capabilities, as well as its wide observation area, high precision, high resolution and non-contact monitoring advantages, effectively making up for the shortcomings of traditional monitoring methods in terms of spatial range, monitoring point density and dangerous area monitoring.

[0043] Landslide early warning is a systematic project that integrates multiple disciplines such as geology, mechanics, engineering and monitoring technology. Limited by the basic theory of geological body destruction and quantitative analysis methods, it is difficult to integrate the theoretical achievements of various disciplines with research methods. Therefore, landslide early warning prediction and analysis has always been a long-term and arduous challenge faced by the engineering community. At present, the commonly used early warning methods are Saito model, Fukuzono model and Voight model, which are typical representatives of creep theory. In addition, methods such as dynamics, statistics, limit equilibrium, neural networks and machine learning are also widely used in landslide early warning. However, although the existing methods are widely used, they have not been combined with multivariate data obtained by ground-based radar. In the cumulative displacement cloud map obtained by ground-based radar, each deformation pixel point in the deformation area has an independent displacement time curve. Although the selection of a single point can reflect the change trend of the landslide to a certain extent, it is impossible to accurately reflect the change trend of the entire landslide due to the inevitable deviation of the point-to-surface approach.

[0044] In summary, ground-based radar is widely used in landslide monitoring due to its advantages, but the existing early warning methods do not consider the data characteristics of the high-frequency domain of ground-based radar. Therefore, there is an urgent need for an early warning method that combines the high-frequency domain data of ground-based radar and has a certain degree of versatility to provide a strong guarantee for the prevention of landslides.

[0045] In combination with the above-mentioned background technology, the present disclosure records the corresponding solutions in the form of embodiments to solve the defects in the prior art, but it is not intended to limit the scope of protection of the patent right claimed by the present disclosure.

[0046] As one of the solutions, Figure 1 The following is a schematic diagram of the implementation process of a landslide early warning method based on ground-based radar high-frequency surface data according to an embodiment of the present disclosure. The embodiment of the present disclosure provides a landslide early warning method based on ground-based radar high-frequency surface data, including:

[0047] The main area of ​​the landslide is extracted by adaptive region growing method;

[0048] Extracting deformation and coherence data of the main area of ​​the landslide, the data at least includes: deformation standard deviation and average coherence of different deformation pixels at the same time;

[0049] A landslide early warning method is established based on the statistical characteristics of landslide surface data, which at least includes: when the tangent angle of the comprehensive standardized regional deformation degree at different times is greater than a threshold, it is determined that the slope is in the pre-slip stage.

[0050] In view of the above contents, the embodiments of the present disclosure aim to propose a landslide early warning method based on high-frequency surface data of ground-based radar. First, the main area of ​​the landslide is extracted by an adaptive regional growing method. Secondly, the deformation and coherence data of the main area of ​​the landslide are extracted. Finally, a landslide early warning method is established based on the statistical characteristics of the landslide surface data. The landslide early warning method proposed in the embodiment of the present disclosure comprehensively considers the deformation change trend and coherence change trend of all pixel points in the deformation area. After multiplying the standard deviation of the regional deformation by the inverse of the average coherence, the CSRDD curve shows a stronger change trend in the landslide stage, and the landslide early warning criterion is determined in combination with the specific slope characteristics. The method disclosed in the present disclosure forms a comprehensive quantitative early warning indicator representing the overall characteristics of these indicators by integrating various indicators reflecting the deformation trend. In addition, the diversified slope data detected by the ground-based radar is fully utilized to reduce the false alarm rate and improve the early warning accuracy.

[0051] The steps of the embodiments of the present disclosure are exemplary. The following takes steps S1 to S6 as an example to further illustrate the landslide early warning method based on high-frequency surface data of ground-based radar.

[0052] In some specific embodiments, the present disclosure may be to extract the main area of ​​the landslide by an adaptive region growing method, including:

[0053] Identify deformation areas by mapping ground-based radar pixels to actual slope topography;

[0054] The adaptive region growing method is used to extract the main deformation area identified by ground-based radar.

[0055] Step S1: Map the ground-based radar pixels to the actual slope terrain to identify the deformation area.

[0056] Step S2: using the adaptive region growing method to extract the deformation main area identified by the ground-based radar.

[0057] The disclosed embodiment uses an adaptive regional growing method (Otsu-RGM) to extract the deformation main area. The Otsu-RGM method combines the two-dimensional Otsu method (Otsu) with the regional growing method (RGM), and uses the two-dimensional Otsu algorithm to automatically calculate the threshold function and the RGM algorithm to identify the regional connectivity function to extract the landslide deformation main body.

[0058] In some specific implementations, the present disclosure may be to extract the deformation main area identified by the ground-based radar using an adaptive region growing method, including:

[0059] Based on the ground-based radar deformation map, the deformation threshold is calculated;

[0060] The obtained threshold is used as the seed point of the region growing method and region growing is performed to extract the main area of ​​the landslide.

[0061] In some specific embodiments, the present disclosure may be, based on the ground-based radar deformation map, calculating the deformation threshold, including:

[0062] Construct a two-dimensional histogram, calculate the gray level and neighborhood gray level of each pixel in the radar deformation map, and increase the count at the corresponding position in the two-dimensional histogram;

[0063] Calculate the between-class variance based on the means of multiple regions and the global mean of the two-dimensional histogram;

[0064] The segmentation threshold that maximizes the inter-class variance is taken as the optimal threshold.

[0065] In some specific embodiments, the present disclosure may be to use the obtained threshold as a seed point of a region growing method and perform region growing, including:

[0066] Starting from the seed point, if the pixels in its area meet the conditions, they are added to the area where the current seed point is located, and the cycle continues until there are no pixels that meet the conditions and can be added.

[0067] The step S2 of the embodiment of the present disclosure may further include steps S21 to S22.

[0068] Step S21: Based on the ground-based radar deformation map, the deformation threshold is calculated using the two-dimensional Otsu method.

[0069] Step S211: construct a two-dimensional histogram, calculate the gray level and the neighborhood gray level of each pixel in the radar deformation map, and increase the count at the corresponding position in the two-dimensional histogram.

[0070] Step S212: Calculate the inter-class variance, divide the two-dimensional histogram into multiple regions (usually two regions, representing the foreground and the background), and calculate the mean of each region and the global mean. Then, calculate the inter-class variance by traversing all possible segmentation thresholds.

[0071] Step S213: Determine the optimal segmentation threshold. The segmentation threshold that maximizes the inter-class variance is considered to be the optimal threshold.

[0072] Step S22: Using the optimal threshold obtained in step S21 as the seed point of the region growing method and performing region growing.

[0073] Step S221: Starting from the seed point, check whether the pixels in its area meet the following conditions and add them to the area where the current seed point is located:

[0074] SPV-SPV*K threshold ≤DPV≤SPV+SPV*K threshold

[0075] Where: DPV is the domain pixel value of the seed point; SPV is the seed point value; K threshold It is the ratio of the difference between the domain value and the growth point value, which is used to control the growth rate of the region and the number of regions after growth. The range is [0,1]. If the value is smaller, the number of regions after regional growth will increase.

[0076] Step S222: Take these new pixels as new seed points and continue the process of step S221 until no pixels meeting the conditions can be included, thus extracting the main area of ​​the landslide.

[0077] In some specific embodiments, the present disclosure may be to extract data on deformation and coherence of the main area of ​​the landslide, including:

[0078] Extract deformation data and accumulate them to obtain a deformation curve cluster of each pixel point that changes cumulatively over time;

[0079] Extract the coherence change of each pixel over time;

[0080] The obtained data is used to calculate the deformation standard deviation and average coherence of different deformed pixels in the deformation main area at the same time.

[0081] The embodiment of the present disclosure may further include step S3 based on the above step S2.

[0082] Step S3: Extract the deformation and coherence data of each pixel in the deformed main area obtained in step S22.

[0083] Step S31: extracting deformation data and accumulating them to obtain a deformation curve cluster of each pixel point with time. Assuming that there are N pixels in the deformation main area, N deformation curves with time can be obtained. Take a certain pixel point as an example:

[0084]

[0085] Among them: CD k is the cumulative deformation value, k is the time window; n is the total number of radar deformation images; D ij is the deformation value of the j-th pixel in the i-th deformation image, j = 1···2···N.

[0086] Step S32: extract the coherence change of each pixel over time. Assuming that there are N pixels in the deformation main area, N curves of coherence change over time can be obtained. The coherence formula is as follows:

[0087]

[0088] Where: k is the number of pixels in the window, A is the backscatter coefficient of the target, λ is the radar operating wavelength, ΔR is the deformation of the target radar line of sight, and Δβ is the scattering phase. If the target does not change between two echoes, the coherence of the target is equal to 1, indicating that the target is completely coherent. If the target deforms between two echoes, the coherence coefficient will vary between 0 and 1. The closer to 0, the weaker the coherence. When the coherence coefficient reaches 0, the target is completely uncorrelated. As the slope deformation speed gradually increases, the coherence will gradually decrease. When the slope accelerates, its coherence will show a rapid decrease trend. The closer to the landslide, the faster the decrease trend.

[0089] The embodiment of the present disclosure may further include step S4 based on the above step S3.

[0090] Step S4: Using the data obtained in step S3, calculate the deformation standard deviation and average coherence of different deformed pixel points in the deformation main area at the same time:

[0091]

[0092]

[0093] Where: d is the standard deviation of regional deformation; d i is the deformation value of different pixels at the same time; is the average deformation value of different pixels at the same time; is the regional average coherence; γ i is the coherence value of different pixels at the same time; n is the number of different pixels at the same time.

[0094] In some specific embodiments, the present disclosure may be to establish a landslide early warning method based on the statistical characteristics of landslide surface data, including:

[0095] Using the obtained deformation standard deviation and average coherence, the comprehensive standardized regional deformation degree is calculated and the comprehensive standardized regional deformation degree-time curve is constructed;

[0096] Calculate the tangent angle of the comprehensive standardized regional deformation degree-time curve;

[0097] When the tangent angle is greater than the threshold angle, it indicates that the slope is in the stage of imminent sliding and an alarm is issued.

[0098] The embodiment of the present disclosure may further include steps S5 to S7 based on the above step S4.

[0099] Step S5: Calculate the comprehensive standardized regional deformation degree (CSRDD) using the deformation standard deviation and average coherence obtained in step S4 and construct a CSRDD-time curve.

[0100]

[0101] Among them: CSRDD i is the comprehensive standardized regional deformation degree at different times; δ di is the standard deviation of deformation at different times; is the average coherence at different times.

[0102] Step S6: Calculate the tangent angle of the CSRDD-time curve, and the calculation formula is as follows:

[0103]

[0104] Where: α is the tangent angle; CSRDD i With CSRDD i are the standardized regional deformation degrees at the i-th moment and the i-1-th moment respectively; t i and t i-1 are the i-th moment and the i-1-th moment respectively, usually t i With t i-1 The interval can be set to 1 hour or 1 day etc.

[0105] Step S7: When the tangent angle is less than 85 degrees, it indicates that the slope is not in the stage of imminent sliding, and the observation continues; when the tangent angle is greater than 85 degrees, it indicates that the slope is in the stage of imminent sliding, and an alarm is issued.

[0106] In some specific embodiments, the landslide early warning device based on high-frequency surface data of ground-based radar disclosed in the present invention includes one or more processing modules configured to execute the specific implementation methods of steps S1 to S7 in the foregoing text.

[0107] Specifically, one of the inventive concepts of the present disclosure is that the landslide early warning method based on high-frequency surface data of ground-based radar of various embodiments of the present disclosure extracts the main area of ​​the landslide at least by adaptive regional growing method; extracts the deformation and coherence data of the main area of ​​the landslide, and the data at least includes: the deformation standard deviation and average coherence of different deformation pixels at the same time; establishes a landslide early warning method according to the statistical characteristics of the landslide surface data, at least including: when the tangent angle of the comprehensive standardized regional deformation degree at different times is greater than the threshold, it is determined that the slope is in the pre-slip stage, thereby comprehensively considering the deformation change trend and coherence change trend of all pixels in the deformation area. After multiplying the regional deformation standard deviation by the inverse of the average coherence, the comprehensive standardized regional deformation degree (CSRDD) curve shows a stronger change trend in the landslide stage. By calculating the tangent angle of the CSRDD curve, the landslide stage division can be realized, and the landslide early warning criterion can be determined in combination with the specific slope characteristics. The method of the present disclosure forms a comprehensive quantitative early warning index representing the overall characteristics of these indicators by integrating various indicators reflecting deformation trends. In addition, the diversified slope data detected by the ground-based radar was fully utilized to reduce the false alarm rate, improve the warning accuracy, make up for the shortcomings of the single-point displacement curve analysis method of traditional slope monitoring technology, and provide a new direction for the application of ground-based radar in landslide early warning.

[0108] The present disclosure also provides a computer-readable storage medium having computer executable instructions stored thereon. When the computer executable instructions are executed by a processor, the landslide early warning method based on high-frequency surface data of ground-based radar is mainly implemented, which at least includes:

[0109] The main area of ​​the landslide is extracted by adaptive region growing method;

[0110] Extracting deformation and coherence data of the main area of ​​the landslide, the data at least includes: deformation standard deviation and average coherence of different deformation pixels at the same time;

[0111] A landslide early warning method is established based on the statistical characteristics of landslide surface data, which at least includes: when the tangent angle of the comprehensive standardized regional deformation degree at different times is greater than a threshold, it is determined that the slope is in the pre-slip stage.

[0112] The above embodiments are only exemplary embodiments of the present disclosure and are not intended to limit the present disclosure. The protection scope of the present disclosure is defined by the claims. Those skilled in the art may make various modifications or equivalent substitutions to the present disclosure within the essence and protection scope of the present disclosure, and such modifications or equivalent substitutions shall also be deemed to fall within the protection scope of the present disclosure.

Claims

1. Landslide early warning method based on high-frequency surface data of ground-based radar, including: The main area of ​​the landslide is extracted by adaptive region growing method; Extracting deformation and coherence data of the main area of ​​the landslide, the data at least includes: deformation standard deviation and average coherence of different deformation pixels at the same time; A landslide early warning method is established based on the statistical characteristics of landslide surface data, which at least includes: when the tangent angle of the comprehensive standardized regional deformation degree at different times is greater than a threshold, it is determined that the slope is in the pre-slip stage.

2. The method according to claim 1, wherein: The main area of ​​the landslide is extracted by the adaptive region growing method, including: Identify deformation areas by mapping ground-based radar pixels to actual slope topography; The adaptive region growing method is used to extract the main deformation area identified by ground-based radar.

3. The method according to claim 2, wherein: The adaptive region growing method is used to extract the deformation main area identified by the ground-based radar, including: Based on the ground-based radar deformation map, the deformation threshold is calculated; The obtained threshold is used as the seed point of the region growing method and region growing is performed to extract the main area of ​​the landslide.

4. The method according to claim 3, wherein: Based on the ground-based radar deformation map, the deformation threshold is calculated, including: Construct a two-dimensional histogram, calculate the gray level and neighborhood gray level of each pixel in the radar deformation map, and increase the count at the corresponding position in the two-dimensional histogram; Calculate the between-class variance based on the means of multiple regions and the global mean of the two-dimensional histogram; The segmentation threshold that maximizes the inter-class variance is taken as the optimal threshold.

5. The method according to claim 4, wherein: The obtained threshold is used as the seed point of the region growing method and region growing is performed, including: Starting from the seed point, if the pixels in its area meet the conditions, they are added to the area where the current seed point is located, and the cycle continues until there are no pixels that meet the conditions and can be added.

6. The method according to claim 5, wherein: Extract deformation and coherence data about the main area of ​​the landslide, including: Extract deformation data and accumulate them to obtain a deformation curve cluster of each pixel point that changes cumulatively over time; Extract the coherence change of each pixel over time; The obtained data is used to calculate the deformation standard deviation and average coherence of different deformed pixels in the deformation main area at the same time.

7. The method according to claim 6, wherein: The landslide early warning method is established based on the statistical characteristics of landslide surface data, including: Using the obtained deformation standard deviation and average coherence, the comprehensive standardized regional deformation degree is calculated and the comprehensive standardized regional deformation degree-time curve is constructed; Calculate the tangent angle of the comprehensive standardized regional deformation degree-time curve; When the tangent angle is greater than the threshold angle, it indicates that the slope is in the stage of imminent sliding and an alarm is issued.

8. The method according to claim 5, wherein: Satisfy the conditions, including: whether the pixel value of the seed point satisfies: SPV-SPV*K threshold ≤DPV≤SPV+SPV*K threshold Where: DPV is the domain pixel value of the seed point; SPV is the seed point value; K threshold It is the ratio of the difference between the domain value and the growth point value, which is used to control the growth rate of the region and the number of regions after growth. The range is [0,1]. If the value is smaller, the number of regions after regional growth will increase.

9. The method according to claim 6, wherein the coherence formula is: in: k is the number of pixels in the window, A is the backscatter coefficient of the target, λ is the radar operating wavelength, ΔR is the target radar line of sight deformation, and Δβ is the scattering phase.

10. The method according to claim 7, wherein the formula for calculating the tangent angle of the comprehensive standardized regional deformation degree-time curve is: in: α is the tangent angle; CSRDD i With CSRDD i are the standardized regional deformation degrees at the i-th moment and the i-1-th moment respectively; t i and t i-1 are the i-th moment and the i-1-th moment respectively.

Images