Earthquake landslide detection method and device combining coherence enhancement and vegetation damage

Abstract

The present invention discloses a method and device for detecting earthquake landslides by combining coherence increase and vegetation damage, wherein the method comprises: determining coherence maps when an earthquake occurs and after an earthquake according to earthquake coherence data; comparing coherence maps when an earthquake occurs and after an earthquake to generate a post-earthquake coherence increase map; quantifying vegetation damage in a target area to be detected before and after an earthquake based on a dual-polarization radar vegetation index to generate a post-earthquake vegetation damage map; fusing the post-earthquake coherence increase map with the post-earthquake vegetation damage map using wavelet transform; segmenting the fused image into a landslide area and a non-landslide area to obtain a segmented fused image, determining a mask file based on satellite optical data, performing mask processing on the segmented fused image using the mask file, and performing earthquake landslide detection on the target area to be detected based on the masked fused image. The present invention can comprehensively and accurately identify landslide areas, and improve the accuracy of earthquake landslide detection.

Description

Technical Field

[0001] The present invention relates to the technical field of landslide disaster detection, and in particular to an earthquake landslide detection method and device combining coherence increase and vegetation damage. Background Art

[0002] This section is intended to provide a background or context to the embodiments of the invention recited in the claims. No admission is made that the description herein is prior art by inclusion in this section.

[0003] Earthquake landslide detection is a technical means to identify and locate landslide areas by analyzing the surface changes and landslide characteristics caused by earthquakes. Accurate earthquake landslide detection can quickly obtain spatial distribution information of landslides, providing key data support for disaster emergency response and subsequent disaster prevention and mitigation work. Therefore, the study of earthquake landslide detection technology is of great significance for assessing earthquake secondary disasters, reducing disaster losses, and protecting people's lives and property.

[0004] At present, the existing landslide detection methods mainly rely on coherence changes to identify landslide areas. However, this method only uses single coherence change information, which may be interfered by natural environmental changes (such as precipitation, vegetation growth, etc.) or noise, resulting in false detection or missed detection, thus affecting the accuracy and reliability of landslide detection results. Summary of the invention

[0005] The embodiment of the present invention provides a method for detecting earthquake landslides by combining coherence increase and vegetation damage, which is used for detecting earthquake landslides by combining post-earthquake coherence increase and polarization vegetation damage to improve the accuracy of earthquake landslide detection. The method includes:

[0006] Acquire seismic coherence data covering the target area to be detected;

[0007] According to the earthquake coherence data, the coherence diagram when the earthquake occurs and the coherence diagram after the earthquake are determined;

[0008] Compare the coherence map when the earthquake occurs with the coherence map after the earthquake to determine the difference in coherence changes and generate a post-earthquake coherence increase map;

[0009] Based on the dual-polarization radar vegetation index, the vegetation damage in the target area to be detected before and after the earthquake is quantified, the vegetation damage is detected, and the post-earthquake vegetation damage map is generated;

[0010] Wavelet transform is used to fuse the post-earthquake coherence increase map with the post-earthquake vegetation damage map to obtain a fused image.

[0011] The fused image is segmented into landslide area and non-landslide area to obtain a segmented fused image. A mask file is determined based on satellite optical data. The mask file is used to perform mask processing on the segmented fused image. Earthquake landslide detection is performed on the target area to be detected based on the masked fused image.

[0012] The embodiment of the present invention further provides an earthquake landslide detection device combining coherence increase and vegetation damage, which is used for earthquake landslide detection combining post-earthquake coherence increase and polarization vegetation damage to improve the accuracy of earthquake landslide detection. The device includes:

[0013] An acquisition unit, used for acquiring seismic coherence data covering the target area to be detected;

[0014] A coherence map generating unit, used for determining a coherence map when an earthquake occurs and a coherence map after the earthquake according to the earthquake coherence data;

[0015] A coherence increase map generating unit is used to compare the coherence map when the earthquake occurs with the coherence map after the earthquake, determine the difference in coherence changes, and generate a post-earthquake coherence increase map;

[0016] A post-earthquake vegetation damage map generation unit is used to quantify the vegetation damage in the target area to be detected before and after the earthquake based on the dual-polarization radar vegetation index, perform change detection on the vegetation damage, and generate a post-earthquake vegetation damage map;

[0017] A wavelet transform fusion unit is used to fuse the post-earthquake coherence increase map with the post-earthquake vegetation damage map using wavelet transform to obtain a fused image;

[0018] The earthquake landslide detection unit is used to segment the fused image into a landslide area and a non-landslide area to obtain a segmented fused image, determine a mask file based on satellite optical data, perform mask processing on the segmented fused image using the mask file, and perform earthquake landslide detection on the target area to be detected based on the masked fused image.

[0019] An embodiment of the present invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above-mentioned earthquake landslide detection method combining coherence increase and vegetation damage when executing the computer program.

[0020] An embodiment of the present invention further provides a computer-readable storage medium storing a computer program, and when the computer program is executed by a processor, the above-mentioned earthquake landslide detection method combining coherence increase and vegetation damage is implemented.

[0021] An embodiment of the present invention further provides a computer program product, which includes a computer program. When the computer program is executed by a processor, the above-mentioned earthquake landslide detection method combining coherence increase and vegetation damage is implemented.

[0022] Compared with the prior art which mainly relies on single coherence change information to detect landslide areas, which may be disturbed by natural environment changes or noise, resulting in false detection or missed detection, thereby affecting the accuracy and reliability of landslide detection results, the earthquake landslide detection scheme of combining coherence increase and vegetation damage provided by the embodiment of the present invention uses the generated post-earthquake coherence increase map and post-earthquake vegetation damage map as data sources, adopts the wavelet transform fusion method to realize the fusion of the post-earthquake coherence increase map and the post-earthquake vegetation damage map, divides the fused image into landslide areas and non-landslide areas, obtains the fused image after segmentation processing, determines the mask file based on the satellite optical data, uses the mask file to perform mask processing on the fused image after segmentation processing, and performs earthquake landslide detection on the target area to be detected based on the fused image after mask processing. The landslide detection result finally generated can clearly and intuitively present the spatial distribution of the landslide area. The scheme can more comprehensively and accurately identify the landslide area, improve the accuracy of earthquake landslide detection, and provide reliable data support for post-earthquake landslide disaster assessment and emergency response. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. In the drawings:

[0024] Figure 1 It is a flow chart of a method for detecting earthquake landslides by combining coherence increase and vegetation damage in an embodiment of the present invention;

[0025] Figure 2 It is a flow chart of an earthquake landslide detection method combining post-earthquake coherence increase and polarization vegetation damage in another embodiment of the present invention;

[0026] Figure 3 A schematic diagram of the post-earthquake coherence change provided by the present invention;

[0027] Figure 4 A typical coherence area map of post-earthquake landslides in an embodiment of the present invention is added;

[0028] Figure 5 A typical vegetation damage area map of post-earthquake landslide in an embodiment of the present invention;

[0029] Figure 6 The earthquake landslide detection diagram of the track raising and the earthquake landslide detection diagram of the track lowering in the embodiment of the present invention;

[0030] Figure 7 A comparison diagram of earthquake landslide detection and field verification of the combined track in an embodiment of the present invention;

[0031] Figure 8 This is a diagram for evaluating the accuracy of earthquake landslide detection in an embodiment of the present invention;

[0032] Fig. 9 It is a schematic diagram of the structure of an earthquake landslide detection device combining coherence increase and vegetation damage in an embodiment of the present invention. DETAILED DESCRIPTION

[0033] To make the purpose, technical solution and advantages of the embodiments of the present invention more clear, the embodiments of the present invention are further described in detail below in conjunction with the accompanying drawings. Here, the exemplary embodiments of the present invention and their descriptions are used to explain the present invention, but are not intended to limit the present invention.

[0034] The acquisition, storage, use, and processing of data in the technical solution of this application comply with the relevant provisions of laws and regulations.

[0035] Considering the technical problems existing in the existing landslide detection, the inventors found that post-earthquake vegetation damage is one of the important characteristics of landslides. Landslides are often accompanied by large-scale destruction of vegetation, exposing bare soil and rock areas. This change can provide supplementary and verification information for landslide detection based on coherence changes. Therefore, the embodiment of the present invention proposes an earthquake landslide detection scheme that combines coherence increase and vegetation damage. In the embodiment of the present invention, the quantification of vegetation damage information adopts the dual-polarization radar vegetation index (DpRVI). The scheme uses the all-weather characteristics of radar data to overcome the limitation of difficulty in obtaining optical vegetation index due to poor weather after the earthquake, thereby ensuring the timeliness and reliability of landslide detection. Therefore, in earthquake landslide detection, combining the post-earthquake coherence increase information with the vegetation damage information based on DpRVI not only improves the accuracy of detection, but also reflects the characteristics of the landslide area more comprehensively. This scheme overcomes the limitations of traditional single detection methods and provides a more effective solution for landslide disaster detection under complex terrain and severe meteorological conditions. The earthquake landslide detection scheme combining coherence enhancement and vegetation damage is introduced in detail below.

[0036] Figure 1 FIG. 4 is a flow chart of a method for detecting earthquake landslides by combining coherence enhancement with vegetation damage in an embodiment of the present invention. Figure 1 As shown, the method comprises the following steps:

[0037] Step 101: Acquire seismic coherence data covering the target area to be detected;

[0038] Step 102: determining a coherence graph when an earthquake occurs and a coherence graph after the earthquake according to the earthquake coherence data;

[0039] Step 103: Compare the coherence map when the earthquake occurs with the coherence map after the earthquake, determine the difference in coherence change, and generate a post-earthquake coherence increase map;

[0040] Step 104: quantifying vegetation damage in the target area to be detected before and after the earthquake based on the dual-polarization radar vegetation index, performing change detection on the vegetation damage, and generating a post-earthquake vegetation damage map;

[0041] Step 105: using wavelet transform to fuse the post-earthquake coherence increase map with the post-earthquake vegetation damage map to obtain a fused image;

[0042] Step 106: Segment the fused image into landslide areas and non-landslide areas to obtain a segmented fused image, determine a mask file based on satellite optical data, perform mask processing on the segmented fused image using the mask file, and perform earthquake landslide detection on the target area to be detected based on the masked fused image.

[0043] The earthquake landslide detection method combining coherence increase and vegetation damage provided by the embodiment of the present invention, when working, uses the generated post-earthquake coherence increase map and post-earthquake vegetation damage map as data sources, adopts the wavelet transform fusion method to realize the fusion of the post-earthquake coherence increase map and the post-earthquake vegetation damage map, divides the fused image into landslide area and non-landslide area, obtains the fused image after segmentation processing, determines the mask file based on the satellite optical data, uses the mask file to perform mask processing on the fused image after segmentation processing, and performs earthquake landslide detection on the target area to be detected based on the fused image after mask processing.

[0044] Compared with the prior art that mainly relies on single coherence change information to detect landslide areas, which may be affected by natural environment changes or noise, resulting in false detection or missed detection, thereby affecting the accuracy and reliability of landslide detection results, the earthquake landslide detection method of the joint coherence increase and vegetation damage provided by the embodiment of the present invention can finally generate a landslide detection result that can clearly and intuitively present the spatial distribution of the landslide area. This method can more comprehensively and accurately identify landslide areas, improve the accuracy of earthquake landslide detection, and provide reliable data support for post-earthquake landslide disaster assessment and emergency response. The following is a detailed introduction to the earthquake landslide detection method of the joint coherence increase and vegetation damage.

[0045] In view of the problems currently existing in earthquake landslide detection using coherence changes, the purpose of the embodiments of the present invention is to provide an earthquake landslide detection method that combines post-earthquake coherence increase and polarization vegetation damage (denoted as CCI-PVD method) to solve the problems of inaccurate earthquake landslide detection relying on a single coherence change and difficulty in obtaining the optical vegetation index due to bad weather after an earthquake.

[0046] To achieve the above purpose, the embodiment of the present invention adopts the following implementation scheme:

[0047] Figure 2 The flowchart of the earthquake landslide detection method combining post-earthquake coherence increase and polarization vegetation damage in another embodiment of the present invention comprises the following steps:

[0048] Step S1: Acquire the Sentinel-1 radar elevation and descending orbit data and Sentinel-2 optical data covering the study area and pre-process them respectively, i.e., the pre-preparation step before the above step 101.

[0049] 1.1: The Sentinel-1 data (satellite radar ascending and descending orbit data) is strip-separated and the polarization mode is determined to be VV polarization. After precise orbit determination, the burst is spliced. Specifically, after strips containing information about the study area are separated from the Sentinel-1 data, precise orbit data is used to improve the accuracy of its geographic positioning, helping to achieve sub-pixel-level precise registration of SAR images at different time points.

[0050] 1.2: Perform radiometric calibration, atmospheric correction and geometric correction on Sentinel-2 data (satellite optical data). Specifically, according to the radiometric calibration coefficient in Sentinel-2 data, the DN value is converted to the surface radiometric value, the influence of aerosol, water vapor and other factors in the atmosphere is removed, and the radiometric value is converted to the surface reflectivity. At the same time, the influence of terrain, height and sensor geometric errors is eliminated, so that the image has accurate geographic positioning.

[0051] Step S2: Analyze the preprocessed Sentinel-1 data and optimize the coherence data, that is, use the coherence decomposition theory as a constraint to optimize the key steps of coherence map generation, and then calculate the coherence maps before and after the earthquake, that is, the above steps 101 to 102.

[0052] 2.1: The specific method of optimizing the key steps of coherence map generation is as follows: Optimizing the key steps of coherence data, including calculating the spatial baseline to select the optimal image pair, enhancing registration and checking the registration quality, and optimizing the coherence estimation window size to ensure the generation of high-quality coherence maps. It mainly includes the following three points: First, by calculating the spatial baseline, the optimal image pair for generating the earthquake and post-earthquake coherence map is selected to ensure the reliability of coherence; second, for the problem of incoherence that may be caused by data processing, the data is enhanced and the registration quality is strictly checked; third, for the incoherence caused by other factors, the most appropriate coherence estimation window size is selected to further improve the quality and accuracy of the coherence map.

[0053] It can be seen from the above that, in one embodiment, the above-mentioned earthquake landslide detection method combining coherence increase and vegetation damage may further include performing the following optimization processing operation on the earthquake coherence data to obtain the optimized earthquake coherence data:

[0054] By calculating the spatial baseline, the optimal image pairs for generating the coherence maps at the time of earthquake and after the earthquake are selected from the seismic coherence data to ensure the reliability of seismic coherence;

[0055] In view of the incoherence caused by the seismic coherence data processing, the seismic coherence data enhancement registration processing is carried out, and the data enhancement registration processing results are verified;

[0056] In view of the incoherence caused by other factors, the most appropriate coherence estimation window size is selected to improve the quality and accuracy of the coherence map generated based on seismic coherence data.

[0057] In specific implementation, the above-mentioned implementation method of optimizing the seismic coherence data can further improve the accuracy of subsequent incoherence map generation, thereby further improving the accuracy of earthquake landslide detection.

[0058] 2.2: Based on the optimized coherence data, the coherence diagrams before and after the earthquake are calculated. The principle of coherence increase at the landslide after the earthquake is as follows: Figure 3 As shown, Figure 3 A schematic diagram of the post-earthquake coherence change provided by the present invention, wherein Figure 3 Figure a is a schematic diagram before the earthquake. Figure 3 Middle b is a schematic diagram after the earthquake. Before the earthquake, the landslide area in the mountain valley was usually covered with dense vegetation and had low coherence (corresponding to Figure 3 After the earthquake triggered the landslide, the vegetation coverage area was destroyed, and the surface structure of the landslide area became more uniform and stable, resulting in a significant increase in the coherence of the landslide area after the earthquake (corresponding to Figure 3 b), which provides an important basis for landslide detection.

[0059] In step S2, the pre-earthquake image pair is precisely registered by enhanced multi-spectral (ESD) to generate the earthquake coherence map (denoted as γ Co-seismic ) and the coherence graph after the earthquake (denoted as γ Post-earthquake ).

[0060] As can be seen from the above, in one embodiment, determining the coherence graph when an earthquake occurs and the coherence graph after the earthquake according to the earthquake coherence data may include:

[0061] Performing the above optimization processing operation on the seismic coherence data to obtain optimized seismic coherence data;

[0062] Performing enhanced multi-spectral precise registration on the pre-earthquake image pairs in the optimized seismic coherence data to obtain enhanced multi-spectral precise registration pre-earthquake image pairs, and obtaining seismic coherence data after secondary optimization;

[0063] Based on the seismic coherence data after secondary optimization processing, the optimized coherence diagrams during and after the earthquake are generated.

[0064] In specific implementation, the above-mentioned implementation method of determining the coherence map when an earthquake occurs and the coherence map after the earthquake can further improve the accuracy of coherence map generation, thereby further improving the accuracy of earthquake landslide detection.

[0065] In specific implementation, in the subsequent step 103, the coherence map when the earthquake occurs is compared with the coherence map after the earthquake to determine the difference in coherence changes. When generating the post-earthquake coherence increase map, the optimized coherence map when the earthquake occurs and after the earthquake can be used to further improve the accuracy of generating the post-earthquake coherence increase map, thereby improving the accuracy of earthquake landslide detection.

[0066] Step S3: Compare the coherence maps before and after the earthquake, calculate the difference in coherence changes, and generate a post-earthquake coherence increase map, i.e., the above step 103, to identify areas where landslides may occur. That is, based on the optimized coherence data, calculate the difference between the post-earthquake coherence map and the coherence map during the earthquake, obtain the post-earthquake coherence increase map, and adjust the coherence increase threshold according to the coherence histogram distribution. The optical images before and after the earthquake and the coherence change images are shown in Figure 1. Figure 4 As shown, Figure 4 is a typical area of ​​increased coherence of post-earthquake landslide in the embodiment of the present invention, wherein: Figure 4 a and c are before the earthquake. Figure 4 b and d in the middle are after the earthquake. Figure 4 The e in the figure corresponds to the coherence diagram when an earthquake occurs. Figure 4 f in the figure corresponds to the coherence diagram after the earthquake. Figure 4 In the middle, g is the earthquake time. Figure 4 In step S3, h is the time after the earthquake. γ Post-earthquake minus γ Co-seismic , the coherence increase map after the earthquake is obtained and de-dimensionalized and contrast-enhanced (denoted as Δ N(Coherence) ).

[0067] From the above, it can be seen that, in one embodiment, the above earthquake landslide detection method combining coherence increase and vegetation damage may also include the following operation of obtaining the optimal post-earthquake coherence increase map:

[0068] According to the coherence histogram distribution, the coherence increase threshold of the post-earthquake coherence increase map is adjusted;

[0069] Optimizing the post-earthquake coherence increase map according to the coherence increase threshold value to obtain an optimized post-earthquake coherence increase map;

[0070] The optimized post-earthquake coherence increase map is de-dimensionalized and contrast-enhanced to obtain the optimal post-earthquake coherence increase map.

[0071] In specific implementation, the above-mentioned implementation method of obtaining the optimal post-earthquake coherence increase map can further improve the accuracy of generating the post-earthquake coherence increase map, thereby further improving the accuracy of earthquake landslide detection.

[0072] In the subsequent step 105, the post-earthquake coherence increase map and the post-earthquake vegetation damage map are fused by wavelet transform. When the fused image is obtained, the optimal post-earthquake coherence increase map can be used to further improve the fusion accuracy, thereby improving the accuracy of earthquake landslide detection.

[0073] Step S4: Based on the dual-polarization radar vegetation index, quantify the vegetation damage in the study area before and after the earthquake, perform change detection, and generate a post-earthquake vegetation damage map, that is, the above step 104, that is, using the pre-earthquake radar vegetation index as a benchmark, using the pre-earthquake index map minus the post-earthquake index map to obtain the post-earthquake vegetation damage map Δ DpRVI :Δ DpRVI = Post DpRVI - Pre DpRVI , and adjust the vegetation damage threshold according to the image histogram. The changes of optical images and dual-polarization radar vegetation index before and after the earthquake are shown in the figure. Figure 5 As shown, Figure 5 This is a typical vegetation damage area of ​​post-earthquake landslide in the embodiment of the present invention, where: Figure 5 a and c in the middle are before the earthquake. Figure 5 b and d in the middle are after the earthquake. Figure 5The middle e corresponds to the dual-polarization vegetation index map before the earthquake. Figure 5 The figure f in the middle corresponds to the dual-polarization vegetation index map after the earthquake. Figure 5 The middle g is before the earthquake, Figure 5 The h in the middle is after the earthquake.

[0074] In step S4, the C2 polarization matrix is ​​generated using Sentinel-1 data, and the polarization information is decomposed. The dual-polarization radar vegetation index (denoted as DpRVI) is derived from this polarization information. The DpRVI before the earthquake in the study area is calculated as Pre DpRVI , calculate the DpRVI of the study area after the earthquake and record it as Post DpRVI .calculate Post DpRVI and Pre DpRVI The difference between the values ​​of Δ and Δ is used to obtain the vegetation damage map after the earthquake, which is recorded as Δ DpRVI .

[0075] As can be seen from the above, in one embodiment, based on the dual-polarization radar vegetation index, quantifying the vegetation damage in the target area to be detected before and after the earthquake, performing change detection on the vegetation damage, and generating a post-earthquake vegetation damage map may include:

[0076] The polarization matrix is ​​generated using Sentinel-1 data (the first set of satellite data, i.e. satellite radar ascending and descending orbit data);

[0077] Decomposing the polarization matrix to obtain polarization information;

[0078] Derived according to the polarization information, a dual-polarization radar vegetation index before the earthquake and a dual-polarization radar vegetation index after the earthquake in the target area to be detected;

[0079] Based on the difference between the dual-polarization radar vegetation index before the earthquake and the dual-polarization radar vegetation index after the earthquake in the target area to be detected, the post-earthquake vegetation damage map is obtained.

[0080] In specific implementation, the above-mentioned method of generating a post-earthquake vegetation damage map can further improve the accuracy of generating a post-earthquake vegetation damage map, solve the problem of difficulty in obtaining the optical vegetation index due to bad weather after an earthquake, and further improve the accuracy of earthquake landslide detection.

[0081] Step S5: Fuse the post-earthquake coherence increase map with the vegetation damage map, i.e., the above step 105. Specifically, in order to fully consider the impact of coherence increase and vegetation damage on landslide detection, wavelet transform (WT) is used to fuse the results of post-earthquake coherence increase and vegetation damage. The specific calculation formula is as follows:

[0082] CCI-PVD=IWT[ w x WT(Δ N(Coherence) )+ w y WT(Δ DpRVI )]

[0083] The final fused image result can effectively combine Δ N(Coherence) and Δ DpRVI The advantages of the two types of change information further improve the reliability of landslide detection in complex terrain and vegetation-covered areas.

[0084] Specifically, in step S5, the data fusion operation of the post-earthquake coherence increase map and the vegetation damage map is as follows:

[0085] The Δ obtained in step S3 N(Coherence) and the Δ obtained in step S4 DpRVI Perform wavelet decomposition to decompose the image into multi-scale and multi-directional sub-bands, denoted as WT(Δ N(Coherence) ) and WT(Δ DpRVI ):

[0086] WT(Δ N(Coherence) )={ L C ,H Cx ,H Cy ,H Cd}

[0087] In the formula, WT is wavelet transform, L is the low frequency subband, L C represents the large-scale coherence change in the earthquake-landslide region, L D Indicates the overall vegetation damage in the landslide area.

[0088] WT(Δ DpRVI )={ L D ,H Dx ,H Dy ,H Dd}

[0089] H is a high-frequency subband, representing detail information (such as edges, textures, etc.). x , y , d Denotes horizontal, vertical, and diagonal directions. Different weights are then assigned to the two images:

[0090] w x WT(Δ N(Coherence) )+ w y WT(Δ DpRVI )

[0091] w x To measure Δ N(Coherence) Contributions to landslide detection, e.g. for highlighting surface stability changes of earthquake-induced landslides. w y To measure Δ DpRVI Contributions to landslide detection, such as enhanced reflection of vegetation cover changes.

[0092] In one embodiment, the weight can be adjusted according to the importance or quality of the actual data using the following formula:

[0093]

[0094] in, σ C and σ D Respectively represent Δ N(Coherence) and Δ DpRVI The signal strength (such as contrast or standard deviation) of w x Adding weight to the graph for post-seismic coherence, w y is the weight of the vegetation damage map after the earthquake, σ C represents the signal strength of the post-earthquake coherence increase map, σ D They respectively represent the signal strength of the vegetation damage map after the earthquake.

[0095] In specific implementation, the above implementation method of adjusting the weight according to the importance or quality of actual data can further improve the accuracy of image fusion, thereby further improving the accuracy of earthquake landslide detection.

[0096] The fused result is expressed in the wavelet domain as:

[0097] WT fusion = w x WT(Δ N(Coherence) )+ w y WT(Δ DpRVI )

[0098] Through the inverse wavelet transform (IWT), the fused sub-bands are reconstructed into the final fused image:

[0099] CCI-PVD=IWT(WT fusion )

[0100] The image reconstructed by IWT will retain the advantages of both images. The low-frequency sub-band provides the trend distribution of the macroscopic landslide area, and the high-frequency sub-band highlights the edge of the landslide area and the detailed features of vegetation damage.

[0101] From the above, it can be seen that in one embodiment, using wavelet transform to fuse the post-earthquake coherence increase map with the post-earthquake vegetation damage map to obtain a fused image may include:

[0102] The post-earthquake coherence increase map is decomposed by wavelet transform to obtain multi-scale and multi-directional sub-bands corresponding to the post-earthquake coherence increase map; the multi-scale and multi-directional sub-bands corresponding to the post-earthquake coherence increase map include low-frequency sub-bands representing the coherence change of the earthquake landslide area, and high-frequency sub-bands corresponding to the post-earthquake coherence increase map representing the detailed information in the horizontal, vertical and diagonal directions;

[0103] The post-earthquake vegetation damage map is decomposed by wavelet transform to obtain multi-scale and multi-directional sub-bands corresponding to the post-earthquake vegetation damage map; the multi-scale and multi-directional sub-bands corresponding to the post-earthquake vegetation damage map include low-frequency sub-bands representing the overall vegetation damage in the earthquake landslide area, and high-frequency sub-bands corresponding to the post-earthquake vegetation damage map representing detailed information in the horizontal, vertical and diagonal directions;

[0104] Assign weights to the post-earthquake coherence increase map and the post-earthquake vegetation damage map. The weight of the post-earthquake coherence increase map is used to measure the contribution of the post-earthquake coherence increase map to landslide detection, and the weight of the post-earthquake vegetation damage map is used to measure the contribution of the post-earthquake vegetation damage map to landslide detection.

[0105] According to the multi-scale and multi-directional sub-bands corresponding to the post-earthquake coherence increase map, the multi-scale and multi-directional sub-bands corresponding to the post-earthquake vegetation damage map, and the weights assigned to the post-earthquake coherence increase map and the post-earthquake vegetation damage map, the fused sub-bands in the wavelet domain are obtained;

[0106] The fused sub-bands are reconstructed into the final fused image through inverse wavelet transform; in the final fused image, the low-frequency sub-bands provide the trend distribution of the macroscopic landslide area, and the high-frequency sub-bands highlight the edge of the landslide area and the detailed features of vegetation damage.

[0107] In specific implementation, the above-mentioned implementation method of using wavelet transform to fuse the post-earthquake coherence increase map with the post-earthquake vegetation damage map, the final generated landslide detection result can clearly and intuitively present the spatial distribution of the landslide area, can further improve the accuracy of image fusion, and further improve the accuracy of earthquake landslide detection.

[0108] Step S6: post-process the fused image, and perform mask processing on the fused image based on the mask file calculated based on the Sentinel-2 optical data, and finally carry out detection of the earthquake landslide area based on the fused image, that is, the above-mentioned step 106.

[0109] In step S6, the specific operations of post-processing and landslide detection on the CCI-PVD fusion result (fused image) obtained in step S5 may be:

[0110] 6.1: Post-processing of the fused image: Calculate the histogram of the image within the pixel value range of the fused CCI-PVD image, and traverse each possible grayscale value as the segmentation threshold. Then, the image is divided into foreground and background according to the segmentation threshold. Calculate the weights, average values, and inter-class variance of the foreground and background, and find the segmentation threshold that maximizes the inter-class variance. Based on this segmentation threshold, the CCI-PVD image is segmented into landslide areas and non-landslide areas. Subsequently, the segmented fused image is subjected to image morphological processing to eliminate small islands and holes.

[0111] In specific implementation, the above specific implementation of the operation of post-processing the fused image can further improve the accuracy of the post-processing operation on the fused image, thereby improving the accuracy of subsequent earthquake landslide detection.

[0112] Specifically, when calculating the segmentation threshold, the CCI-PVD fusion image is segmented into landslide areas and non-landslide areas according to the following formula:

[0113]

[0114] argmax represents the MATLAB function that takes the maximum value. is the inter-class variance of the foreground and background of the CCI-PVD image (fused image), t The maximum grayscale threshold that can achieve the best segmentation is used as the segmentation threshold, that is, the threshold that maximizes the inter-class variance is found. t , we can get the best segmentation threshold θ After the segmentation threshold is obtained, the landslide area is separated from the non-landslide area using the segmentation threshold:

[0115]

[0116] Among them, CCI-PVD (x,y) is the pixel position in the CCI-PVD fusion image ( x,y ), which is the optimal segmentation threshold θ Binarization of CCI-PVD (x,y)), a gray value (optimal segmentation threshold) of 1 represents a landslide area, and a gray value of 0 represents a non-landslide area. That is, in one embodiment, based on the optimal segmentation threshold that maximizes the inter-class variance, the fused image is segmented into a landslide area and a non-landslide area to obtain a segmented fused image, including: an optimal segmentation threshold of 1 represents a landslide area, and an optimal segmentation threshold of 0 represents a non-landslide area.

[0117] In specific implementation, the specific implementation method of segmenting the fused image into landslide areas and non-landslide areas using the above formula can further improve the accuracy of post-processing operations on the fused image, thereby improving the accuracy of subsequent earthquake landslide detection.

[0118] From the above, it can be seen that in one embodiment, the above-mentioned earthquake landslide detection method combining coherence increase and vegetation damage can also include: performing image morphological processing on the segmented image (for example, performing morphological opening and closing operations to effectively remove small interference points) to make the landslide edge more complete, thereby obtaining a segmented image after morphological processing.

[0119] In specific implementation, the above implementation method of performing image morphology processing on the segmented images can improve the accuracy of post-processing operations on the fused images, thereby improving the accuracy of subsequent earthquake landslide detection.

[0120] When the mask file is used to perform mask processing on the fused image in the subsequent process, the mask file can be used to perform mask processing on the segmented image after the morphological processing, which can improve the accuracy of subsequent earthquake landslide detection.

[0121] 6.2: Calculate the water mask of the study area based on sentinel-2 data With building mask , using water mask and building mask to Binarization (CCI-PVD (x,y) ) (the segmented image obtained according to the fused image) is subjected to mask processing operation to eliminate the interference of non-landslide factors, and finally a raster image is obtained as the result of the mask processing operation.

[0122] As can be seen from the above, in one embodiment, determining a mask file based on satellite optical data and performing mask processing on the fused image using the mask file may include:

[0123] Determine the mask file based on satellite optical data to determine the water mask and building mask of the target area to be detected;

[0124] The water body mask and the building mask are used to perform mask processing on the fused image after segmentation processing, and finally a raster image is obtained as the fused image after mask processing.

[0125] 6.3: Finally, earthquake landslide detection is performed based on the fused image after the mask processing. The above raster image is converted into a vector file and the coordinate system is unified to obtain a vector file with unified coordinates. The vector file with unified coordinate system is intersected with the vector file of the field survey result, and then fused and deduplicated, etc., to obtain the correctly detected earthquake landslide result. The result of the joint track is as follows Figure 6 As shown, the results of orbit raising and orbit lowering are as follows Figure 7 As shown, Figure 6 ( Figure 6 a) and the down-track earthquake landslide detection map ( Figure 6 b), Figure 7 This is an earthquake landslide detection diagram of a combined track (upward track + downward track) in an embodiment of the present invention ( Figure 7 (a) and the field verification comparison chart ( Figure 7 (b)

[0126] Based on the above, in one embodiment, performing earthquake landslide detection on the target area to be detected based on the fused image after mask processing may include:

[0127] Convert raster images to vector files;

[0128] Unify the coordinate system of the vector file to obtain the vector file with the unified coordinate system;

[0129] The vector file after the unified coordinate system is intersected with the vector file of the field survey results of the target area to be detected to obtain the earthquake landslide detection results.

[0130] Step S7: Verify the detection results in combination with geological survey data to evaluate the accuracy and reliability of the method. Specifically, in order to evaluate the accuracy of the CCI-PVD method (the earthquake landslide detection method combining coherence increase and vegetation damage in the embodiment of the present invention) in detecting landslides, field survey data is used as reference data, and the correct detection rate, false detection rate and missed detection rate are used as accuracy evaluation indicators to verify the accuracy of the detected results, so as to assess the reliability of the method. In the attribute table, the total number of landslides detected by the method and the correctly detected landslides are counted, and the above three indicators are calculated to obtain the specific detection accuracy as follows: Figure 8 As shown, Figure 8 This is a diagram for evaluating the accuracy of earthquake landslide detection in an embodiment of the present invention.

[0131] From the above, it can be seen that in one embodiment, the above-mentioned earthquake landslide detection method of combining coherence increase and vegetation damage can also include: using field survey data as reference data, using correct detection rate, false detection rate and missed detection rate as accuracy evaluation indicators, and verifying the accuracy of earthquake landslide detection results, so as to evaluate the reliability of the method.

[0132] In summary, the beneficial effects of the embodiments of the present invention are as follows: the embodiments of the present invention use the post-earthquake coherence increase map and the post-earthquake vegetation damage map as data sources, and adopt the wavelet transform fusion method to achieve multi-source fusion of post-earthquake landslide detection information. This method makes full use of the advantages of the post-earthquake coherence increase map in reflecting the changes in surface stability and the ability of the vegetation damage map to characterize the characteristics of ecological damage, effectively combines the advantageous information of the two types of data, and suppresses the noise and shortcomings of a single data source. Through multi-scale feature extraction and weighted fusion, this method can more comprehensively and accurately identify landslide areas, which not only improves the accuracy of landslide detection, but also has good adaptability under complex terrain and vegetation coverage conditions. The landslide detection results finally generated can clearly and intuitively present the spatial distribution of the landslide area, provide reliable data support for post-earthquake landslide disaster assessment and emergency response, and have important application value for geological disaster monitoring and disaster prevention and mitigation.

[0133] The present invention also provides a combined coherence increase and vegetation damage earthquake landslide detection device, as described in the following embodiments. Since the principle of the device to solve the problem is similar to the combined coherence increase and vegetation damage earthquake landslide detection method, the implementation of the device can refer to the implementation of the combined coherence increase and vegetation damage earthquake landslide detection method, and the repeated parts will not be repeated.

[0134] Fig. 9 FIG. 1 is a schematic diagram of the structure of an earthquake landslide detection device combining coherence increase and vegetation damage according to an embodiment of the present invention. Fig. 9 As shown, the device comprises:

[0135] An acquisition unit 01 is used to acquire seismic coherence data covering the target area to be detected;

[0136] A coherence map generating unit 02 is used to determine the coherence map when the earthquake occurs and the coherence map after the earthquake according to the earthquake coherence data;

[0137] A coherence increase map generating unit 03 is used to compare the coherence map when the earthquake occurs with the coherence map after the earthquake, determine the difference in coherence changes, and generate a post-earthquake coherence increase map;

[0138] The post-earthquake vegetation damage map generating unit 04 is used to quantify the vegetation damage in the target area to be detected before and after the earthquake based on the dual-polarization radar vegetation index, perform change detection on the vegetation damage, and generate a post-earthquake vegetation damage map;

[0139] The wavelet transform fusion unit 05 is used to fuse the post-earthquake coherence increase map with the post-earthquake vegetation damage map by using wavelet transform to obtain a fused image;

[0140] The earthquake landslide detection unit 06 is used to segment the fused image into a landslide area and a non-landslide area to obtain a segmented fused image, determine a mask file based on satellite optical data, perform mask processing on the segmented fused image using the mask file, and perform earthquake landslide detection on the target area to be detected based on the masked fused image.

[0141] In one embodiment, the above-mentioned coherence graph generation unit is specifically used for:

[0142] Performing the above optimization processing operation on the seismic coherence data to obtain optimized seismic coherence data;

[0143] Performing enhanced multi-spectral precise registration on the pre-earthquake image pairs in the optimized seismic coherence data to obtain enhanced multi-spectral precise registration pre-earthquake image pairs, and obtaining seismic coherence data after secondary optimization;

[0144] Based on the seismic coherence data after secondary optimization processing, the optimized coherence diagrams during and after the earthquake are generated.

[0145] In one embodiment, the above-mentioned earthquake landslide detection device combining coherence increase and vegetation damage may further include an optimal post-earthquake coherence increase map determination unit, which is used to perform the following operations to obtain the optimal post-earthquake coherence increase map:

[0146] According to the coherence histogram distribution, the coherence increase threshold of the post-earthquake coherence increase map is adjusted;

[0147] Optimizing the post-earthquake coherence increase map according to the coherence increase threshold value to obtain an optimized post-earthquake coherence increase map;

[0148] The optimized post-earthquake coherence increase map is de-dimensionalized and contrast-enhanced to obtain the optimal post-earthquake coherence increase map.

[0149] In one embodiment, the post-earthquake vegetation damage map generating unit is specifically used to:

[0150] Generate polarization matrix using satellite radar ascending and descending orbit data;

[0151] Decomposing the polarization matrix to obtain polarization information;

[0152] Derived according to the polarization information, a dual-polarization radar vegetation index before the earthquake and a dual-polarization radar vegetation index after the earthquake in the target area to be detected;

[0153] Based on the difference between the dual-polarization radar vegetation index before the earthquake and the dual-polarization radar vegetation index after the earthquake in the target area to be detected, the post-earthquake vegetation damage map is obtained.

[0154] In one embodiment, the above-mentioned earthquake landslide detection device combining coherence increase and vegetation damage may further include an image morphology processing unit for performing image morphology processing on the segmented image to make the landslide edge more complete, thereby obtaining a segmented image after morphological processing.

[0155] In one embodiment, earthquake landslide detection is performed on a target area to be detected based on the fused image after mask processing, including:

[0156] Convert raster images to vector files;

[0157] Unify the coordinate system of the vector file to obtain the vector file with the unified coordinate system;

[0158] The vector file after the unified coordinate system is intersected with the vector file of the field survey results of the target area to be detected to obtain the earthquake landslide detection results.

[0159] In one embodiment, the above-mentioned earthquake landslide detection device combining coherence increase and vegetation damage may also include a verification unit, which is used to: use field survey data as reference data, adopt correct detection rate, false detection rate and missed detection rate as accuracy evaluation indicators, verify the accuracy of earthquake landslide detection results, so as to evaluate the reliability of the method.

[0160] In one embodiment, the wavelet transform fusion unit is specifically used for:

[0161] The post-earthquake coherence increase map is decomposed by wavelet transform to obtain the multi-scale and multi-directional sub-bands corresponding to the post-earthquake coherence increase map; the multi-scale and multi-directional sub-bands corresponding to the post-earthquake coherence increase map include: low-frequency sub-bands representing the coherence change of the earthquake landslide area, and high-frequency sub-bands corresponding to the post-earthquake coherence increase map representing the detailed information in the horizontal, vertical and diagonal directions;

[0162] The post-earthquake vegetation damage map is decomposed by wavelet transform to obtain the multi-scale and multi-directional sub-bands corresponding to the post-earthquake vegetation damage map; the multi-scale and multi-directional sub-bands corresponding to the post-earthquake vegetation damage map include: low-frequency sub-bands representing the overall vegetation damage in the earthquake landslide area, and high-frequency sub-bands corresponding to the post-earthquake vegetation damage map representing the detailed information in the horizontal, vertical and diagonal directions;

[0163] Assign weights to the post-earthquake coherence increase map and the post-earthquake vegetation damage map; the weight of the post-earthquake coherence increase map is used to measure the contribution of the post-earthquake coherence increase map to landslide detection, and the weight of the post-earthquake vegetation damage map is used to measure the contribution of the post-earthquake vegetation damage map to landslide detection;

[0164] According to the multi-scale and multi-directional sub-bands corresponding to the post-earthquake coherence increase map, the multi-scale and multi-directional sub-bands corresponding to the post-earthquake vegetation damage map, and the weights assigned to the post-earthquake coherence increase map and the post-earthquake vegetation damage map, the fused sub-bands in the wavelet domain are obtained;

[0165] The fused sub-bands are reconstructed into the final fused image through inverse wavelet transform; in the final fused image, the low-frequency sub-bands provide the trend distribution of the macroscopic landslide area, and the high-frequency sub-bands highlight the edge of the landslide area and the detailed features of vegetation damage.

[0166] In one embodiment, the above-mentioned earthquake landslide detection device combining coherence increase and vegetation damage may further include a weight adjustment unit, which is used to adjust the weight according to the importance or quality of the actual data according to the following formula:

[0167]

[0168] in, w x Adding weight to the graph for post-seismic coherence, w y is the weight of the vegetation damage map after the earthquake, σ C It represents the signal strength corresponding to the post-earthquake coherence increase map, σ D They respectively represent the signal strength corresponding to the vegetation damage map after the earthquake.

[0169] In one embodiment, the earthquake landslide detection unit is specifically used for:

[0170] In the pixel value range of the fused image, the histogram of the fused image is determined, and each gray value of the histogram is traversed as a threshold;

[0171] According to the threshold, the fused image is divided into a foreground and a background;

[0172] Determine the weights, average values, and inter-class variances of the foreground and background, and find the optimal segmentation threshold that maximizes the inter-class variance based on the weights, average values, and inter-class variances;

[0173] Based on the optimal segmentation threshold that maximizes the inter-class variance, the fused image is segmented into landslide area and non-landslide area to obtain a segmented fused image.

[0174] In one embodiment, the earthquake landslide detection unit is specifically used for:

[0175] Determine the water mask and building mask of the target area to be detected based on satellite optical data;

[0176] The water body mask and the building mask are used to perform mask processing on the fused image after segmentation processing, and finally a raster image is obtained as the fused image after mask processing.

[0177] In one embodiment, the above-mentioned earthquake landslide detection device combining coherence increase and vegetation damage may further include an optimization unit, which is used to perform the following optimization processing operations on the earthquake coherence data to obtain the optimized earthquake coherence data:

[0178] By calculating the spatial baseline, the optimal image pairs for generating the coherence maps at the time of earthquake and after the earthquake are selected from the seismic coherence data to ensure the reliability of seismic coherence;

[0179] In view of the loss of coherence caused by the processing of seismic coherence data, enhanced registration processing of seismic coherence data is performed, and the results of data enhanced registration processing are verified;

[0180] In view of the incoherence caused by other factors, the most appropriate coherence estimation window size is selected to improve the quality and accuracy of the coherence map generated based on seismic coherence data.

[0181] In one embodiment, segmenting the fused image into landslide areas and non-landslide areas to obtain a segmented fused image includes: segmenting the fused image into landslide areas and non-landslide areas according to the following formula:

[0182]

[0183] In the formula, argmax represents the MATLAB function that takes the maximum value, is the inter-class variance of the foreground and background of the CCI-PVD image (fused image), t The maximum grayscale threshold that can achieve the best segmentation is used as the segmentation threshold, that is, the threshold that maximizes the inter-class variance is found. t , we can get the best segmentation threshold θ , after finding the segmentation threshold, the landslide area is separated from the non-landslide area using the following formula and the segmentation threshold:

[0184]

[0185] Among them, CCI-PVD (x,y) is the pixel position in the CCI-PVD fusion image ( x,y ), which is the optimal segmentation threshold θ Binarization of CCI-PVD (x,y) ), the gray value (optimal segmentation threshold) of 1 represents the landslide area, and the gray value of 0 represents the non-landslide area.

[0186] An embodiment of the present invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above-mentioned earthquake landslide detection method combining coherence increase and vegetation damage when executing the computer program.

[0187] An embodiment of the present invention further provides a computer-readable storage medium storing a computer program, and when the computer program is executed by a processor, the above-mentioned earthquake landslide detection method combining coherence increase and vegetation damage is implemented.

[0188] An embodiment of the present invention further provides a computer program product, which includes a computer program. When the computer program is executed by a processor, the above-mentioned earthquake landslide detection method combining coherence increase and vegetation damage is implemented.

[0189] Compared with the prior art which mainly relies on single coherence change information to detect landslide areas, which may be disturbed by natural environment changes or noise, resulting in false detection or missed detection, thereby affecting the accuracy and reliability of landslide detection results, the earthquake landslide detection scheme of combining coherence increase and vegetation damage provided by the embodiment of the present invention uses the generated post-earthquake coherence increase map and post-earthquake vegetation damage map as data sources, adopts the wavelet transform fusion method to realize the fusion of the post-earthquake coherence increase map and the post-earthquake vegetation damage map, divides the fused image into landslide areas and non-landslide areas, obtains the fused image after segmentation processing, determines the mask file based on the satellite optical data, uses the mask file to perform mask processing on the fused image after segmentation processing, and performs earthquake landslide detection on the target area to be detected based on the fused image after mask processing. The landslide detection result finally generated can clearly and intuitively present the spatial distribution of the landslide area. The scheme can more comprehensively and accurately identify the landslide area, improve the accuracy of earthquake landslide detection, and provide reliable data support for post-earthquake landslide disaster assessment and emergency response.

[0190] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0191] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0192] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0193] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable device. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0194] The specific embodiments described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for detecting earthquake landslides by combining coherence increase and vegetation destruction, characterized in that: include: Acquire seismic coherence data covering the target area to be detected; According to the earthquake coherence data, the coherence diagram when the earthquake occurs and the coherence diagram after the earthquake are determined; Compare the coherence map when the earthquake occurs with the coherence map after the earthquake to determine the difference in coherence changes and generate a post-earthquake coherence increase map; Based on the dual-polarization radar vegetation index, the vegetation damage in the target area to be detected before and after the earthquake is quantified, the vegetation damage is detected, and the post-earthquake vegetation damage map is generated; Wavelet transform is used to fuse the post-earthquake coherence increase map with the post-earthquake vegetation damage map to obtain a fused image. The fused image is segmented into landslide area and non-landslide area to obtain a segmented fused image. A mask file is determined based on satellite optical data. The mask file is used to perform mask processing on the segmented fused image. Earthquake landslide detection is performed on the target area to be detected based on the masked fused image.

2. The method according to claim 1, characterized in that Wavelet transform is used to fuse the post-earthquake coherence increase map with the post-earthquake vegetation damage map to obtain a fused image, including: The post-earthquake coherence increase map is decomposed by wavelet transform to obtain the multi-scale and multi-directional sub-bands corresponding to the post-earthquake coherence increase map; the multi-scale and multi-directional sub-bands corresponding to the post-earthquake coherence increase map include: low-frequency sub-bands representing the coherence change of the earthquake landslide area, and high-frequency sub-bands corresponding to the post-earthquake coherence increase map representing the detailed information in the horizontal, vertical and diagonal directions; The post-earthquake vegetation damage map is decomposed by wavelet transform to obtain the multi-scale and multi-directional sub-bands corresponding to the post-earthquake vegetation damage map; the multi-scale and multi-directional sub-bands corresponding to the post-earthquake vegetation damage map include: low-frequency sub-bands representing the overall vegetation damage in the earthquake landslide area, and high-frequency sub-bands corresponding to the post-earthquake vegetation damage map representing the detailed information in the horizontal, vertical and diagonal directions; Assign weights to the post-earthquake coherence increase map and the post-earthquake vegetation damage map; the weight of the post-earthquake coherence increase map is used to measure the contribution of the post-earthquake coherence increase map to landslide detection, and the weight of the post-earthquake vegetation damage map is used to measure the contribution of the post-earthquake vegetation damage map to landslide detection; According to the multi-scale and multi-directional sub-bands corresponding to the post-earthquake coherence increase map, the multi-scale and multi-directional sub-bands corresponding to the post-earthquake vegetation damage map, and the weights assigned to the post-earthquake coherence increase map and the post-earthquake vegetation damage map, the fused sub-bands in the wavelet domain are obtained; The fused sub-bands are reconstructed into the final fused image through inverse wavelet transform; in the final fused image, the low-frequency sub-bands provide the trend distribution of the macroscopic landslide area, and the high-frequency sub-bands highlight the edge of the landslide area and the detailed features of vegetation damage.

3. The method according to claim 2, characterized in that It also includes: adjusting the weight according to the importance or quality of the actual data according to the following formula: in, w x Adding weight to the graph for post-seismic coherence, w y is the weight of the vegetation damage map after the earthquake, σ C It represents the signal strength corresponding to the post-earthquake coherence increase map, σ D They respectively represent the signal strength corresponding to the post-earthquake vegetation damage map.

4. The method according to claim 1, characterized in that The fused image is segmented into landslide area and non-landslide area to obtain a segmented fused image, including: In the pixel value range of the fused image, the histogram of the fused image is determined, and each gray value of the histogram is traversed as a threshold; According to the threshold, the fused image is divided into a foreground and a background; Determine the weights, average values, and inter-class variances of the foreground and background, and find the optimal segmentation threshold that maximizes the inter-class variance based on the weights, average values, and inter-class variances; Based on the optimal segmentation threshold that maximizes the inter-class variance, the fused image is segmented into landslide area and non-landslide area to obtain a segmented fused image.

5. The method according to claim 1, characterized in that The mask file is determined based on the satellite optical data, and the mask file is used to perform mask processing on the fused image after the segmentation processing, including: Determine the water mask and building mask of the target area to be detected based on satellite optical data; The water body mask and the building mask are used to perform mask processing on the fused image after segmentation processing, and finally a raster image is obtained as the fused image after mask processing.

6. The method according to claim 1, characterized in that It also includes the following optimization processing operations on the seismic coherence data to obtain the seismic coherence data after optimization processing: By calculating the spatial baseline, the optimal image pairs for generating the coherence maps at the time of earthquake and after the earthquake are selected from the seismic coherence data to ensure the reliability of seismic coherence; In view of the incoherence caused by the seismic coherence data processing, the seismic coherence data enhancement registration processing is carried out, and the data enhancement registration processing results are verified; In view of the incoherence caused by other factors, the most appropriate coherence estimation window size is selected to improve the quality and accuracy of the coherence map generated based on seismic coherence data.

7. An earthquake landslide detection device combining coherence increase and vegetation damage, characterized in that: The device is used to execute the earthquake landslide detection method combining coherence increase and vegetation damage as described in claim 1, and the device comprises: An acquisition unit, used for acquiring seismic coherence data covering the target area to be detected; A coherence map generating unit, used for determining a coherence map when an earthquake occurs and a coherence map after the earthquake according to the earthquake coherence data; A coherence increase map generating unit is used to compare the coherence map when the earthquake occurs with the coherence map after the earthquake, determine the difference in coherence changes, and generate a post-earthquake coherence increase map; A post-earthquake vegetation damage map generation unit is used to quantify the vegetation damage in the target area to be detected before and after the earthquake based on the dual-polarization radar vegetation index, perform change detection on the vegetation damage, and generate a post-earthquake vegetation damage map; A wavelet transform fusion unit is used to fuse the post-earthquake coherence increase map with the post-earthquake vegetation damage map using wavelet transform to obtain a fused image; The earthquake landslide detection unit is used for segmenting the fused image into landslide area and non-landslide area to obtain the segmented fused image, determining the mask file based on the satellite optical data, performing mask processing on the segmented fused image using the mask file, and performing earthquake landslide detection on the target area to be detected based on the masked fused image.

8. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the method according to any one of claims 1 to 6 is implemented.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

10. A computer program product, characterized in that The computer program product comprises a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

Images