Coherence change earthquake landslide detection method and device based on intensity difference enhancement

Abstract

The invention discloses a coherence change earthquake landslide detection method and device based on intensity difference enhancement. The method comprises the following steps: acquiring satellite data covering a to-be-detected area range; coherence information is extracted from the satellite data; determining a post-earthquake coherence map and an earthquake coherence map according to the coherence information; generating a post-earthquake coherence increase graph according to a difference value between the post-earthquake coherence graph and the earthquake coherence graph; generating a backscattering intensity difference chart according to the difference value of the backscattering coefficient chart after the earthquake and the backscattering coefficient chart before the earthquake; fusing the post-earthquake coherence enhancement image and the backscattering intensity difference image to obtain a fusion image of backscattering intensity difference enhancement and coherence change information; and performing seismic landslide detection on the to-be-detected area according to the fused image. According to the method, landslide detection is carried out by analyzing the backscattering intensity difference graph and the coherence change graph before and after the earthquake and integrating the advantages of the two graph characteristics in space structure and energy change, so that the precision and stability of earthquake landslide detection are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of earthquake-induced landslide disaster detection, and in particular to a method and device for detecting earthquake landslides based on coherence change with enhanced intensity difference. Background Art

[0002] This section is intended to provide a background or context to the embodiments of the invention that are recited in the claims. No statement herein is admitted to be prior art by virtue of its inclusion in this section.

[0003] Earthquake landslide detection utilizes advanced technologies such as remote sensing image analysis and ground-based monitoring equipment to rapidly identify and assess earthquake-triggered landslide hazards. By acquiring information on the location, extent, and extent of landslide damage, it provides crucial support for post-disaster rescue, risk warning, and reconstruction. Efficient earthquake landslide detection not only shortens emergency response times but also helps decision-makers develop effective disaster prevention strategies, minimizing casualties and economic losses. This has far-reaching implications for enhancing regional disaster prevention and control capabilities.

[0004] Currently, most landslide detection methods based on InSAR (Interferometric Synthetic Aperture Radar) technology rely on coherence changes as their primary criterion, using the changing coherence characteristics of radar images before and after an earthquake to locate landslide locations. However, relying solely on coherence changes has certain limitations: coherence changes can not only be caused by landslides but are also easily affected by factors such as surface moisture, terrain undulation, and vegetation disturbance. This can lead to false positives or false negatives in landslide detection, compromising the overall accuracy and stability of earthquake landslide detection. Summary of the Invention

[0005] The present invention provides a method for detecting earthquake landslides based on coherence change based on intensity difference enhancement. The method analyzes backscatter intensity difference maps and coherence change maps before and after an earthquake, integrating the advantages of these two types of map features in terms of spatial structure and energy change to detect landslides. The method includes:

[0006] Obtain satellite data covering the area to be detected;

[0007] Extracting coherence information from satellite data covering the area to be detected;

[0008] Determine the post-earthquake coherence map and the coherence map during the earthquake based on the coherence information; generate the post-earthquake coherence increase map based on the difference between the post-earthquake coherence map and the coherence map during the earthquake;

[0009] Generate a backscatter intensity difference map based on the difference between the backscatter coefficient map after the earthquake and the backscatter coefficient map before the earthquake;

[0010] The post-earthquake coherence enhancement map and the backscatter intensity difference map are fused at the feature level to obtain a fusion image of the backscatter intensity difference enhancement and coherence change information;

[0011] Earthquake landslide detection is performed in the area to be detected based on the fusion image of backscatter intensity difference enhancement and coherence change information.

[0012] The present invention also provides an earthquake landslide detection device based on coherence change with intensity difference enhancement. The device is used to detect landslides by analyzing the backscatter intensity difference map and the coherence change map before and after the earthquake, integrating the advantages of the two types of map features in spatial structure and energy change. The device includes:

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

[0014] An extraction unit, configured to extract coherence information from satellite data covering an area to be detected;

[0015] A post-earthquake coherence increase map generating unit is used to determine a post-earthquake coherence map and a coherence map at the time of the earthquake based on the coherence information; and to generate a post-earthquake coherence increase map based on a difference between the post-earthquake coherence map and the coherence map at the time of the earthquake;

[0016] A backscatter intensity difference map generating unit, configured to generate a backscatter intensity difference map based on a difference between a backscatter coefficient map after an earthquake and a backscatter coefficient map before an earthquake;

[0017] A fusion unit is used to fuse the post-earthquake coherence enhancement map with the backscatter intensity difference map at the feature level to obtain a fused image of the backscatter intensity difference enhancement and coherence change information;

[0018] The earthquake landslide detection unit is used to perform earthquake landslide detection on the detection area based on the fusion image of backscatter intensity difference enhancement and coherence change information.

[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. When the processor executes the computer program, the above-mentioned method for detecting earthquake landslides based on coherence changes enhanced by intensity differences is implemented.

[0020] An embodiment of the present invention further provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, the method for detecting earthquake landslides based on coherence changes with enhanced intensity differences 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, it implements the above-mentioned earthquake landslide detection method based on coherence change of intensity difference enhancement.

[0022] Compared with the technical solutions in the prior art that rely on a single coherence change to detect earthquake landslides, which are inaccurate and have low precision, the beneficial technical effect of the coherence change earthquake landslide detection solution based on intensity difference enhancement provided by the embodiment of the present invention is: it realizes landslide detection by analyzing the backscattering intensity difference map and the coherence change map before and after the earthquake, integrating the advantages of the two types of map features in spatial structure and energy change, thereby improving the accuracy and stability of earthquake landslide detection. It is particularly suitable for areas with complex terrain or harsh post-earthquake conditions, and has good adaptability and practical value. 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 the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative work. In the drawings:

[0024] Figure 1 Schematic diagram of the flow of a method for detecting earthquake landslides based on coherence change with enhanced intensity difference according to an embodiment of the present invention;

[0025] Figure 2 A and Figure 2 b is a schematic diagram of radar coherence change in an embodiment of the present invention;

[0026] Figure 3 Figure a is an earthquake landslide detection diagram of the track raising embodiment of the present invention. Figure 3 Middle b is a diagram of the detection of the lowering track earthquake landslide in an embodiment of the present invention;

[0027] Figure 4 Figure a is an earthquake landslide detection diagram of the combined lifting rail in an embodiment of the present invention. Figure 4 b is a comparison diagram of field verification in an embodiment of the present invention;

[0028] Figure 5 This is a graph showing the accuracy assessment of the landslide detection method in an embodiment of the present invention;

[0029] Figure 6 Schematic diagram of the structure of the earthquake landslide detection device based on coherence change of intensity difference enhancement in an embodiment of the present invention. DETAILED DESCRIPTION

[0030] To make the purpose, technical solutions and advantages of the embodiments of the present invention more clear, the embodiments of the present invention are further described in detail below with reference to 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.

[0031] The acquisition, storage, use, and processing of data in the technical solution of this application comply with relevant laws and regulations.

[0032] The inventors discovered a technical problem with the existing technology: earthquake landslides are often accompanied by drastic changes in the surface structure, such as rock and soil collapse, vegetation destruction, or the expansion of exposed areas. These changes are usually manifested in SAR (Synthetic Aperture Radar) images as significant changes in radar backscatter intensity. As an important indicator reflecting the radar scattering characteristics of ground objects, backscatter intensity can characterize the changing characteristics of energy reflection in the landslide area and is an important supplementary information for coherence changes. Introducing intensity change information into landslide detection can help enhance the sensitivity of earthquake landslide detection to post-earthquake landform changes and improve the accuracy of landslide area detection.

[0033] Therefore, based on the above-mentioned existing technical problems, the embodiment of the present invention proposes a coherence change detection scheme based on intensity difference enhancement (Intensity Difference-enhanced Coherence Change Detection, ID-CCD). By analyzing the backscatter intensity difference map and the coherence change map before and after the earthquake, and using multi-scale wavelet fusion, the advantages of the two types of map features in spatial structure and energy change are integrated. This method effectively overcomes the limitations of traditional detection based on a single coherence change, improves the accuracy and robustness of landslide detection, and is particularly suitable for areas with complex terrain or harsh post-earthquake conditions. It has good adaptability and practical value. The following is a detailed introduction to the earthquake landslide detection scheme based on coherence change enhancement based on intensity difference enhancement.

[0034] Figure 1 FIG. 1 is a flow chart of a method for detecting earthquake landslides based on coherence change with enhanced intensity difference according to an embodiment of the present invention. Figure 1 As shown, the method includes the following steps:

[0035] Step 101: Acquire satellite data covering the area to be detected;

[0036] Step 102: extracting coherence information from satellite data covering the area to be detected;

[0037] Step 103: determining a post-earthquake coherence map and a coherence map at the time of the earthquake based on the coherence information; generating a post-earthquake coherence increase map based on the difference between the post-earthquake coherence map and the coherence map at the time of the earthquake;

[0038] Step 104: generating a backscatter intensity difference map based on the difference between the backscatter coefficient map after the earthquake and the backscatter coefficient map before the earthquake;

[0039] Step 105: Fusing the post-earthquake coherence enhancement map and the backscatter intensity difference map at the feature level to obtain a fused image of the backscatter intensity difference enhancement and coherence change information;

[0040] Step 106: Perform earthquake landslide detection on the area to be detected based on the fused image of the backscatter intensity difference enhancement and the coherence change information.

[0041] The embodiment of the present invention provides a method for detecting earthquake landslides based on coherence change based on intensity difference enhancement. During operation, the method comprises the following steps: acquiring satellite data covering an area to be detected; extracting coherence information from the satellite data covering the area to be detected; determining a post-earthquake coherence map and a coherence map at the time of the earthquake based on the coherence information; generating a post-earthquake coherence increase map based on the difference between the post-earthquake coherence map and the coherence map at the time of the earthquake; generating a backscattering intensity difference map based on the difference between the post-earthquake backscattering coefficient map and the pre-earthquake backscattering coefficient map; fusing the post-earthquake coherence enhancement map with the backscattering intensity difference map at a feature level to obtain a fused image of the backscattering intensity difference enhancement and the coherence change information; and performing earthquake landslide detection on the area to be detected based on the fused image of the backscattering intensity difference enhancement and the coherence change information.

[0042] Compared to existing solutions that rely on single coherence changes for earthquake landslide detection, which are inaccurate and lack high precision, the earthquake landslide detection method based on coherence changes based on intensity difference enhancement provided by the present invention can analyze the backscatter intensity difference map before and after the earthquake and the coherence change map, integrating the advantages of these two types of map features in spatial structure and energy changes for landslide detection. This method improves the accuracy and stability of landslide detection and is particularly suitable for areas with complex terrain or harsh post-earthquake conditions, showing good adaptability and practical value. A detailed description is given below.

[0043] In view of the problems existing in the current use of coherence change for earthquake landslide detection, the purpose of the embodiment of the present invention is to provide a coherence change detection method based on intensity difference enhancement (denoted as ID-CCD method) to solve the problem of inaccurate and low precision of earthquake landslide detection based on single coherence change. Figures 2 to 5 Provide a detailed introduction.

[0044] To achieve the above-mentioned purpose, the coherence change detection method based on intensity difference enhancement provided by the embodiment of the present invention may include the following steps:

[0045] 1. Step 101: Obtain satellite data covering the area to be detected.

[0046] In step 101 above, satellite data covering the entire study area is obtained, such as ascending and descending data from the Sentinel-1 radar, and the radar images are preprocessed as follows:

[0047] 1.1: The Sentinel-1 SLC data (Single Look Complex, or SLC) are cropped to retain only the study area (the area to be inspected) to avoid data redundancy. After extracting the study area, the ephemeris orbit is corrected using precise orbit data. This involves fine-tuning the image positioning to improve the data's geolocation accuracy and achieve subpixel alignment between SAR images of different phases.

[0048] 1.2: Radiometrically calibrate cropped satellite data, such as Sentinel-1 data, extracting intensity information from the SLC data and saving the calibrated image in complex format. Based on the complex data, further inversion is performed to obtain radar backscatter coefficient maps for subsequent change detection analysis.

[0049] 2. Step 102: extracting coherence information from satellite data covering the area to be detected.

[0050] Analyze the preprocessed Sentinel-1 data to extract and optimize its coherence information. Guided by coherence decomposition theory, optimize the key steps in generating coherence maps to improve the stability and accuracy of post-earthquake landslide detection:

[0051] The coherence optimization process mainly includes the following three aspects: (1) calculating the spatiotemporal baseline to select the optimal image pair; (2) optimizing the SAR registration algorithm to minimize the loss of coherence caused by data processing; and (3) selecting the optimal coherence estimation window to improve the quality and accuracy of the coherence map. In other words, the specific method of optimizing the information affecting coherence is to estimate the spatiotemporal baseline of the data to select the optimal interferometric image pair, optimize the image registration algorithm, and select the optimal coherence window size to ensure the generation of high-quality interferometric coherence maps.

[0052] 3. Step 103: Determine a post-earthquake coherence map and a coherence map at the time of the earthquake based on the coherence information; and generate a post-earthquake coherence increase map based on the difference between the post-earthquake coherence map and the coherence map at the time of the earthquake.

[0053] Based on the optimized data, coherence images were calculated both during and after the earthquake to construct a multi-temporal coherence dataset. The coherence images before and after the earthquake were then interpolated to extract coherence change characteristics. This post-earthquake coherence increase map was generated to identify potential landslide areas.

[0054] The principle of coherence change in landslide area after earthquake is as follows: Figure 2 Before the earthquake, the potential landslide areas in the high mountain valleys were usually covered by dense vegetation. Due to the complex vegetation structure, the coherence was generally low (e.g. Figure 2 After the earthquake triggered the landslide, the vegetation was destroyed and the surface was exposed, which led to the surface structure of the landslide area becoming uniform and stable, and the post-earthquake coherence level increased significantly (e.g. Figure 2 (b) This change provides an important basis for landslide detection.

[0055] In specific implementation, the difference between the post-earthquake coherence map and the coherence map at the time of the earthquake is calculated to generate a post-earthquake coherence increase map. Simultaneously, the coherence change threshold is dynamically adjusted based on the coherence histogram distribution to optimize the initial identification of landslide areas.

[0056] In specific implementation, based on the coherence generation condition optimized in step 102, a coherence map (denoted as γ Co-event ) and the post-earthquake coherence graph (denoted as γ Post-event ).

[0057] In specific implementation, use γ Post-event Subtract γ Co-event , we get the coherence increase diagram Δ after the earthquake γ , and perform image normalization processing (denoted as Δ Norm(γ) ), that is, Δ Norm(γ) This is the post-earthquake coherence enhancement map after image normalization.

[0058] 4. Step 104: Generate a backscatter intensity difference map based on the difference between the backscatter coefficient map after the earthquake and the backscatter coefficient map before the earthquake.

[0059] Based on the radar backscatter coefficient, the changes in backscatter intensity in the landslide area of ​​the study area before and after the earthquake were quantified for change detection, generating a post-earthquake backscatter intensity difference map. That is, using the post-earthquake backscatter coefficient map as a benchmark, the post-earthquake coefficient map was subtracted from the pre-earthquake coefficient map to obtain the post-earthquake backscatter intensity difference map: And adjust the backscatter intensity difference threshold according to the image histogram.

[0060] In practice, the intensity information in the Sentinel-1 SLC data is used to save the radiometrically calibrated image in a complex format. Subsequently, the intensity information is extracted from the complex format to invert the radar backscatter coefficient map and convert it into dB form (denoted as σ dB ). The σ before the earthquake in the study area is dB Recorded as The σ after the earthquake dB Recorded as calculate and The difference between the backscattering intensity and the backscattering intensity difference is obtained, which is recorded as Δ σ .

[0061] As can be seen from the above, in one embodiment, generating a backscatter intensity difference map based on the difference between the backscatter coefficient map after the earthquake and the backscatter coefficient map before the earthquake includes:

[0062] Using the intensity information in the satellite data, the calibrated image is saved in a complex format to obtain a complex format image;

[0063] Extracting intensity information from the complex format image to invert a backscatter coefficient map of the area to be detected after the earthquake and a backscatter coefficient map before the earthquake, converting the backscatter coefficient map of the area to be detected after the earthquake into a backscatter coefficient map in dB format after the earthquake, and converting the backscatter coefficient map of the area to be detected before the earthquake into a backscatter coefficient map in dB format before the earthquake;

[0064] The difference between the backscatter coefficient map in dB format after the earthquake in the area to be detected and the backscatter coefficient map in dB format before the earthquake in the area to be detected is determined to obtain a backscatter intensity difference map.

[0065] In specific implementation, the above specific implementation method of obtaining the backscattering intensity difference map can improve the quality and accuracy of generating the backscattering intensity difference map.

[0066] 5. Step 105: Fusing the post-earthquake coherence enhancement map and the backscattering intensity difference map at the feature level to obtain a fused image of the backscattering intensity difference enhancement and coherence change information.

[0067] The post-earthquake coherence enhancement map and the backscatter intensity difference map are fused at the feature level. That is, in order to fully consider the joint impact of the post-earthquake coherence increase and the backscatter intensity difference on landslide detection, the wavelet transform (WT) method is used to perform multi-scale decomposition on the post-earthquake coherence enhancement map and the backscatter intensity difference map, and complete feature fusion in the wavelet domain. During the feature fusion process, the spatial structure information of the image in the multi-scale wavelet domain is retained, such as the landslide edge, crack direction and other detailed features. Through the fusion of sub-bands of different scales and directions, the expression ability of "spatial structure" in texture and morphology in the image is reflected, thereby improving the structural integrity of landslide detection. The specific calculation formula is as follows:

[0068] ID-CCD=IDWT[X γ ·DWT(Δ Norm(γ) )+X σ ·DWT(Δ σ )]

[0069] The fused image results can effectively combine Δ Norm(γ) and Δ σ The advantages of the two types of change characteristics further improve the reliability and accuracy of landslide detection in complex terrain and vegetation-covered areas.

[0070] In specific implementation, the data fusion operation of the post-earthquake coherence increase map and the backscatter intensity difference map is as follows:

[0071] The Δ obtained in step 103 N orm(γ) and Δ obtained in step 104 σ Perform wavelet decomposition to obtain sub-bands of different scales and directions, denoted as DWT(Δ Norm(γ) ) and DWT(Δ σ ):

[0072]

[0073] In the formula, DWT(·) represents the wavelet decomposition operation, and the Daubechies-4 (db4) wavelet is used for single-layer two-dimensional discrete wavelet transform. γ is a low-frequency sub-band, reflecting the overall coherence variation trend of the landslide area; H γ , V γ , D γ They are high-frequency sub-bands in the horizontal, vertical and diagonal directions, which mainly describe the texture details such as the edge and cracks of the landslide area.

[0074]

[0075] A σ Indicates the change of large-scale backscattering coefficient in the landslide area; H σ , Vσ , D σ They are high-frequency sub-bands in the horizontal, vertical and diagonal directions, which mainly describe the intensity details of the landslide area.

[0076] Then, in order to balance the contribution of the two images in landslide detection, the fusion weight coefficient X is defined σ With X γ , respectively control Δ σ and Δ N orm (γ) The contribution ratio in the fusion process is as follows:

[0077] X γ ·DWT(Δ Norm(γ) )+X σ ·DWT(Δ σ )

[0078] The weight can be adjusted according to the importance or quality of the actual data. The formula is as follows:

[0079]

[0080] Among them, S σ With S γ Represents the image Δ σ With Δ γ (or Δ N orm(γ)), ​​X γ is the fusion weight coefficient of the post-earthquake coherence enhancement image sub-band, X σ is the fusion weight coefficient of the backscatter intensity difference map subband, that is, Xσ and Xγ represent the normalized weights of the two types of image features in the fusion process, which are used for linear weighted fusion. The sum of these two weights is 1, which represents the proportional relationship of the weight distribution. Finally, the corresponding subbands are linearly weighted fused in the wavelet domain, and the fused wavelet domain expression is shown as follows:

[0081] DWT fusion =X γ ·DWT(Δ Norm(γ) )+X σ ·DWT(Δ σ )

[0082] A fused image is reconstructed by performing an inverse wavelet transform (IDWT) operation. In one embodiment, the post-seismic coherence enhancement map and the backscattering intensity difference map are fused at a feature level to obtain a fused image of backscattering intensity difference enhancement and coherence change information. The fused image can be obtained according to the following formula:

[0083] ID-CCD=IDWT·(DWT fusion );

[0084] The inverse transformed image retains both coherence change and intensity difference information. ID-CCD represents the fused image obtained from the coherence change detection method based on intensity difference enhancement. This method combines both intensity and coherence change information, improving the sensitivity and stability of landslide detection.

[0085] As can be seen from the above, in one embodiment, the post-earthquake coherence enhancement map and the backscatter intensity difference map are fused at the feature level to obtain a fused image of backscatter intensity difference enhancement and coherence change information, including:

[0086] Perform wavelet decomposition on the post-earthquake coherence enhancement map to obtain sub-bands of the post-earthquake coherence enhancement map at different scales and directions; perform wavelet decomposition on the backscattering intensity difference map to obtain sub-bands of the backscattering intensity difference map at different scales and directions;

[0087] Determine the weight of the sub-band of the post-earthquake coherence enhancement map and the weight of the sub-band of the backscattering intensity difference map according to the signal strength of the post-earthquake coherence enhancement map and the signal strength of the backscattering intensity difference map;

[0088] According to the weights of the sub-bands of the post-earthquake coherence enhancement map and the sub-bands of the backscattering intensity difference map, the corresponding sub-bands are linearly weighted fused in the wavelet domain to obtain the fused wavelet domain expression;

[0089] Through the inverse wavelet transform operation, the fused image of backscattering intensity difference enhancement and coherence change information is reconstructed according to the fused wavelet domain expression.

[0090] In specific implementation, the above specific implementation method of obtaining the fused image can improve the sensitivity and stability of landslide detection.

[0091] 6. Step 106: Perform earthquake landslide detection on the area to be detected based on the fused image of the backscatter intensity difference enhancement and the coherence change information.

[0092] The fusion results are post-processed and used as a basis for detecting earthquake landslide areas. Specifically, the pixel grayscale value range is extracted from the fused ID-CCD image and the image histogram is calculated based on this. Each possible grayscale value (a possible grayscale value refers to a pixel grayscale value that actually appears in the fused ID-CCD image. Its characteristic is that the number of corresponding pixels in the image histogram is not zero, that is, there is at least one pixel in the image with this grayscale value. These grayscale values ​​reflect the image response intensity of different regions and indirectly reveal the change characteristics such as the degree of surface disturbance) is used as a candidate threshold. That is, the grayscale level with a non-zero frequency in the grayscale histogram is used as the candidate threshold. Then, based on the Otsu method (Otsu's Method, the Otsu method is a classic image binarization segmentation method, which automatically determines the optimal segmentation threshold of the image by maximizing the inter-class variance, thereby dividing the image into foreground and background), according to the principle of maximizing the inter-class variance, the optimal segmentation threshold is automatically determined. Specifically, by calculating the weights, average values ​​and inter-class variance of the foreground and background, the grayscale threshold that can maximize the inter-class difference is found.

[0093] To facilitate understanding of the present invention's implementation, let's consider an example of determining the optimal segmentation threshold: In a post-earthquake remote sensing image, an interferometric coherence map or vegetation index map is preprocessed to form a grayscale image, where the pixel grayscale values ​​reflect the extent of surface change. The inventors aim to use image segmentation to distinguish areas with potential landslides (foreground) from stable background areas.

[0094] For example, let's assume the grayscale value of the image ranges from 0 to 1, representing the strength of the landslide possibility. The inventors tried to use a certain grayscale threshold (such as 0.35) as the dividing point, and the area less than or equal to 0.35 is considered the background (stable area), and the area greater than 0.35 is considered the foreground (possible landslide area). Next:

[0095] 1. Count the proportion of the number of pixels in the foreground and background in the total image to obtain weights; 2. Calculate the average grayscale value of the foreground and background areas respectively to reflect their typical characteristics; 3. Use the difference between the weights and the mean of these two parts to calculate the between-class variance and evaluate the effectiveness of the segmentation.

[0096] The inventors repeatedly experimented with different thresholds and calculated the corresponding inter-class variance. Ultimately, they selected the grayscale threshold that maximized the inter-class variance as the optimal segmentation threshold for distinguishing landslide from non-landslide areas. This approach effectively improves the accuracy of identifying suspected landslide areas and reduces human intervention.

[0097] Based on the determined threshold, the ID-CCD fusion image is segmented into landslide and non-landslide areas. Subsequently, morphological processing is performed on the initial segmentation results to remove small isolated noise points and void areas, further optimizing the spatial integrity and boundary clarity of landslide detection.

[0098] In the specific implementation, the raster image (fused image) is converted into a vector file and the coordinate system is unified. It is intersected with the vector file of the field survey results, and then fused and deduplicated, and the correctly detected earthquake landslide can be obtained. The results of the ascending and descending orbits are as follows: Figure 3 Middle a, Figure 3 In b, the result of the joint track is as follows Figure 4 As shown in a, the verification results are as follows Figure 4 As shown in b.

[0099] In the specific implementation, the fusion result of the ID-CCD method obtained in step 105 is processed. Specifically, the automatic threshold segmentation method is used to divide the fused image into landslide areas and non-landslide areas. The selection of the threshold θ is based on the principle of maximizing the inter-class variance, which is mathematically expressed as follows:

[0100]

[0101] Where t represents the candidate threshold, It represents the inter-class variance between the foreground and the background when the candidate threshold is t, that is, t represents the candidate image grayscale segmentation threshold, which is used to calculate the inter-class variance between the foreground and the background under different thresholds. This indicator reflects the effect of image segmentation. The greater the inter-class difference, the better the segmentation. The function argmax is used to obtain t corresponding to the optimal segmentation threshold that maximizes the inter-class variance. This process can be implemented by the OTSU function in MATLAB. After determining the optimal segmentation threshold, the fused image is binarized, and the segmentation expression is as follows, that is, in one embodiment, based on the fused image of the backscattering intensity difference enhancement and the coherence change information, earthquake landslide detection is performed on the area to be detected, including determining the landslide area and the non-landslide area according to the following formula:

[0102]

[0103] Among them, ID-CCD (x,y) Indicates the gray value of the pixel (x, y) in the fused image, and the valued result B(ID-CCD (x,y) ), where the segmentation result is 1, it indicates a potential landslide area, while 0 indicates a non-landslide area. θ represents the optimal segmentation threshold, which is the value of t that maximizes the inter-class variance. In short, t is the ergodic variable, and the entire formula is the threshold that achieves the optimal segmentation effect. That is, θ is the optimal grayscale threshold selected for automatic image segmentation, which is used to binarize the fused image into "landslide area" and "non-landslide area."

[0104] In specific implementation, using the above segmentation expression to perform earthquake landslide detection can improve the efficiency and accuracy of earthquake landslide detection.

[0105] As can be seen from the above, in one embodiment, earthquake landslide detection is performed on a to-be-detected area based on a fusion image of backscatter intensity difference enhancement and coherence change information, including:

[0106] Extracting the pixel gray value range from the fused image of backscatter intensity difference enhancement and coherence change information;

[0107] Determine the grayscale histogram of the fused image according to the pixel grayscale value range;

[0108] The gray levels with non-zero frequency in the gray histogram are used as candidate thresholds;

[0109] By calculating the inter-class variance of the histogram foreground and background, the grayscale threshold that can maximize the inter-class difference is found from the candidate thresholds as the optimal segmentation threshold;

[0110] Based on the optimal segmentation threshold, the fused image is divided into landslide area and non-landslide area.

[0111] In specific implementation, using the above segmentation expression to perform earthquake landslide detection can improve the efficiency and accuracy of earthquake landslide detection.

[0112] In specific implementation, the earthquake landslide detection method based on coherence change enhanced by intensity difference provided by the embodiment of the present invention may also include: processing the segmented image using morphological opening and closing operations to effectively remove isolated noise points and small interference areas, thereby making the landslide boundary more continuous and complete, further improving the accuracy of earthquake landslide detection.

[0113] In a further preferred embodiment of the present invention, the following steps may be included: comparing and verifying the recognition results with the field survey data, and comprehensively evaluating the performance of the proposed method in terms of accuracy and stability. That is, in order to evaluate the accuracy of the ID-CCD method for detecting landslides, the landslide 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 evaluate the accuracy of the method. The results are as follows: Figure 5 As shown, Figure 5 This is an accuracy evaluation diagram of the landslide detection method in an embodiment of the present invention.

[0114] During the specific implementation, the accuracy of the earthquake landslide detection results of ID-CCD is verified, including:

[0115] The accuracy of the detected results was verified using three indicators: landslide correct detection rate, false detection rate, and missed detection rate, in order to assess the reliability of the method.

[0116] The beneficial effects of the embodiments of the present invention are as follows: using post-earthquake coherence increase maps and intensity difference maps as data sources, the present embodiment employs a wavelet fusion method to achieve effective complementarity between coherence information and backscattering intensity information in earthquake landslide detection. By leveraging the sensitivity of coherence changes to structural disturbances in the landslide region and the ability of intensity differences to characterize sudden changes in surface scattering properties, the present embodiment integrates the advantageous information of these two types of features, improving the spatial classification of landslide targets while mitigating the shortcomings of single feature sources, which are susceptible to interference from factors such as terrain obscuration and vegetation disturbance. Through multi-scale wavelet decomposition and weighted fusion, this method can more comprehensively explore the local edge features and overall distribution characteristics of the landslide region, maintaining good adaptability and robustness under complex terrain conditions and strong radar scattering backgrounds. The final output of the landslide detection result has clear boundaries and accurate positioning, and can intuitively reflect the spatial distribution of the landslide on the post-earthquake surface. This provides reliable data support for the rapid assessment and precise response to earthquake-induced landslide disasters, and has important practical value and promotional significance.

[0117] The present invention also provides an apparatus for detecting earthquake landslides based on coherence changes using enhanced intensity differences, as described in the following embodiments. Because the principles underlying the apparatus are similar to those of the method for detecting earthquake landslides based on coherence changes using enhanced intensity differences, the implementation of the apparatus can be referenced to the implementation of the method for detecting earthquake landslides based on coherence changes using enhanced intensity differences, and any repetitions will not be repeated.

[0118] Figure 6 FIG. 1 is a structural diagram of a coherence change earthquake landslide detection device based on intensity difference enhancement in an embodiment of the present invention. Figure 6 As shown, the device includes:

[0119] Acquisition unit 01, used to acquire satellite data covering the area to be detected;

[0120] Extraction unit 02, used to extract coherence information from satellite data covering the area to be detected;

[0121] The post-earthquake coherence increase map generating unit 03 is used to determine the post-earthquake coherence map and the earthquake coherence map according to the coherence information; and generate the post-earthquake coherence increase map according to the difference between the post-earthquake coherence map and the earthquake coherence map;

[0122] A backscatter intensity difference map generating unit 04 is configured to generate a backscatter intensity difference map based on the difference between the backscatter coefficient map after the earthquake and the backscatter coefficient map before the earthquake;

[0123] Fusion unit 05 is used to fuse the post-earthquake coherence enhancement map with the backscatter intensity difference map at the feature level to obtain a fused image of the backscatter intensity difference enhancement and coherence change information;

[0124] The earthquake landslide detection unit 06 is used to perform earthquake landslide detection on the area to be detected based on the fusion image of the backscatter intensity difference enhancement and the coherence change information.

[0125] In one embodiment, the fusion unit is specifically configured to:

[0126] Perform wavelet decomposition on the post-earthquake coherence enhancement map to obtain sub-bands of the post-earthquake coherence enhancement map at different scales and directions; perform wavelet decomposition on the backscattering intensity difference map to obtain sub-bands of the backscattering intensity difference map at different scales and directions;

[0127] Determine the weight of the sub-band of the post-earthquake coherence enhancement map and the weight of the sub-band of the backscattering intensity difference map according to the signal strength of the post-earthquake coherence enhancement map and the signal strength of the backscattering intensity difference map;

[0128] According to the weights of the sub-bands of the post-earthquake coherence enhancement map and the sub-bands of the backscattering intensity difference map, the corresponding sub-bands are linearly weighted fused in the wavelet domain to obtain the fused wavelet domain expression;

[0129] Through the inverse wavelet transform operation, the fused image of backscattering intensity difference enhancement and coherence change information is reconstructed according to the fused wavelet domain expression.

[0130] In one embodiment, the fusion unit is specifically configured to obtain a fused image according to the following formula:

[0131] ID-CCD=IDWT·(DWT fusion );

[0132] DWT fusion =X γ ·DWT(Δ Norm(γ) )+X σ ·DWT(Δ σ )

[0133]

[0134] Among them, ID-CCD represents the fusion image of backscatter intensity difference enhancement and coherence change information, IDWT represents the inverse wavelet transform operation, and DWT fusion is the wavelet domain expression after fusion; Δ N orm(γ) is the post-earthquake coherence enhancement map, Δ σ is the backscatter intensity difference map, X γ is the fusion weight coefficient of the post-earthquake coherence enhancement image sub-band, Xσ is the fusion weight coefficient of the backscatter intensity difference map subband, DWT(Δ Norm(γ) ) is the post-earthquake coherence enhancement subband, DWT(Δ σ ) is the backscatter intensity difference map subband; S γ Representative post-earthquake coherence enhancement map Δ N The signal intensity of orm(γ), S σ Represents the backscatter intensity difference map Δ σ signal strength.

[0135] In one embodiment, the backscatter intensity difference map generating unit is specifically configured to:

[0136] Using the intensity information in the satellite data, the calibrated image is saved in a complex format to obtain a complex format image;

[0137] Extracting intensity information from the complex format image to invert a backscatter coefficient map of the area to be detected after the earthquake and a backscatter coefficient map before the earthquake, converting the backscatter coefficient map of the area to be detected after the earthquake into a backscatter coefficient map in dB format after the earthquake, and converting the backscatter coefficient map of the area to be detected before the earthquake into a backscatter coefficient map in dB format before the earthquake;

[0138] The difference between the backscatter coefficient map in dB format after the earthquake in the area to be detected and the backscatter coefficient map in dB format before the earthquake in the area to be detected is determined to obtain a backscatter intensity difference map.

[0139] In one embodiment, the earthquake landslide detection unit is specifically used to:

[0140] Extracting the pixel grayscale value range from the fused image of backscatter intensity difference enhancement and coherence change information;

[0141] Determine the grayscale histogram of the fused image according to the pixel grayscale value range;

[0142] The gray levels with non-zero frequency in the gray histogram are used as candidate thresholds;

[0143] By calculating the inter-class variance of the histogram foreground and background, the grayscale threshold that can maximize the inter-class difference is found from the candidate thresholds as the optimal segmentation threshold;

[0144] Based on the optimal segmentation threshold, the fused image is divided into landslide area and non-landslide area.

[0145] In one embodiment, the earthquake landslide detection unit is specifically configured to determine the landslide area and the non-landslide area according to the following formula:

[0146]

[0147] Where t represents the candidate threshold, It represents the inter-class variance between the foreground and the background when the candidate threshold is t. Argmax is used to obtain t that maximizes the inter-class variance as the optimal segmentation threshold. θ is the optimal segmentation threshold. ID-CCD (x,y) Indicates the gray value of the pixel (x, y) in the fused image, B(ID-CCD (x,y) ) represents the valued segmentation result. The area with a segmentation result of 1 represents the potential landslide area, while 0 represents the non-landslide area.

[0148] 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. When the processor executes the computer program, the above-mentioned method for detecting earthquake landslides based on coherence changes enhanced by intensity differences is implemented.

[0149] An embodiment of the present invention further provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, the method for detecting earthquake landslides based on coherence changes with enhanced intensity differences is implemented.

[0150] 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, it implements the above-mentioned earthquake landslide detection method based on coherence change of intensity difference enhancement.

[0151] Compared with the existing technical solutions that rely on a single coherence change to detect earthquake landslides, which are inaccurate and have low precision, the beneficial technical effect of the earthquake landslide detection solution based on coherence change enhanced by intensity difference provided by the embodiment of the present invention is: by analyzing the backscatter intensity difference map and the coherence change map before and after the earthquake, the advantages of the two types of map features in spatial structure and energy change are integrated to perform landslide detection, thereby improving the accuracy and stability of landslide detection. It is particularly suitable for areas with complex terrain or harsh post-earthquake conditions, and has good adaptability and practical value.

[0152] It will be understood by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely 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 magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0153] 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 flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, 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 produce 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 flowcharts and / or block diagrams. 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.

[0154] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work 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 The function specified in one or more boxes.

[0155] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0156] 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 based on coherence change based on intensity difference enhancement, characterized in that: include: Obtain satellite data covering the area to be detected; Extracting coherence information from satellite data covering the area to be detected; Determine the post-earthquake coherence map and the coherence map during the earthquake based on the coherence information; generate the post-earthquake coherence increase map based on the difference between the post-earthquake coherence map and the coherence map during the earthquake; Generate a backscatter intensity difference map based on the difference between the backscatter coefficient map after the earthquake and the backscatter coefficient map before the earthquake; The post-earthquake coherence enhancement map and the backscatter intensity difference map are fused at the feature level to obtain a fusion image of the backscatter intensity difference enhancement and coherence change information; Earthquake landslide detection is performed in the area to be detected based on the fusion image of backscatter intensity difference enhancement and coherence change information.

2. The method according to claim 1, wherein The post-earthquake coherence enhancement map and the backscatter intensity difference map are fused at the feature level to obtain a fused image of backscatter intensity difference enhancement and coherence change information, including: Perform wavelet decomposition on the post-earthquake coherence enhancement map to obtain sub-bands of the post-earthquake coherence enhancement map at different scales and directions; perform wavelet decomposition on the backscattering intensity difference map to obtain sub-bands of the backscattering intensity difference map at different scales and directions; Determine the weight of the sub-band of the post-earthquake coherence enhancement map and the weight of the sub-band of the backscattering intensity difference map according to the signal strength of the post-earthquake coherence enhancement map and the signal strength of the backscattering intensity difference map; According to the weights of the sub-bands of the post-earthquake coherence enhancement map and the sub-bands of the backscattering intensity difference map, the corresponding sub-bands are linearly weighted fused in the wavelet domain to obtain the fused wavelet domain expression; Through the inverse wavelet transform operation, the fused image of backscattering intensity difference enhancement and coherence change information is reconstructed according to the fused wavelet domain expression.

3. The method according to claim 1, wherein The post-earthquake coherence enhancement map and the backscatter intensity difference map are fused at the feature level to obtain a fused image of the backscatter intensity difference enhancement and coherence change information, including obtaining the fused image according to the following formula: ID-CCD=IDWT·(DWT fusion ); DWT fusion =X γ ·DWT(D Norm(γ) )+X σ ·DWT(D σ ) Among them, ID-CCD represents the fusion image of backscatter intensity difference enhancement and coherence change information, IDWT represents the inverse wavelet transform operation, and DWT fusion is the wavelet domain expression after fusion; Δ Norm(γ) is the post-earthquake coherence enhancement diagram, Δ σ is the backscatter intensity difference map, X γ is the fusion weight coefficient of the post-earthquake coherence enhancement image sub-band, X σ is the fusion weight coefficient of the backscatter intensity difference map subband, DWT(Δ Norm(γ) ) is the post-earthquake coherence enhancement subband, DWT(Δ σ ) is the backscatter intensity difference map subband; S γ Representative post-earthquake coherence enhancement map Δ Norm(γ) The signal strength, S σ Represents the backscatter intensity difference map Δ σ signal strength.

4. The method according to claim 1, wherein Based on the difference between the backscatter coefficient map after the earthquake and the backscatter coefficient map before the earthquake, a backscatter intensity difference map is generated, including: Using the intensity information in the satellite data, the calibrated image is saved in a complex format to obtain a complex format image; Extracting intensity information from the complex format image to invert a backscatter coefficient map of the area to be detected after the earthquake and a backscatter coefficient map before the earthquake, converting the backscatter coefficient map of the area to be detected after the earthquake into a backscatter coefficient map in dB format after the earthquake, and converting the backscatter coefficient map of the area to be detected before the earthquake into a backscatter coefficient map in dB format before the earthquake; The difference between the backscatter coefficient map in dB format after the earthquake in the area to be detected and the backscatter coefficient map in dB format before the earthquake in the area to be detected is determined to obtain a backscatter intensity difference map.

5. The method according to claim 1, wherein Based on the fusion image of backscatter intensity difference enhancement and coherence change information, earthquake landslide detection is performed on the detection area, including: Extracting the pixel grayscale value range from the fused image of backscatter intensity difference enhancement and coherence change information; Determine the grayscale histogram of the fused image according to the pixel grayscale value range; The gray levels with non-zero frequency in the gray histogram are used as candidate thresholds; By calculating the inter-class variance of the histogram foreground and background, the grayscale threshold that can maximize the inter-class difference is found from the candidate thresholds as the optimal segmentation threshold; Based on the optimal segmentation threshold, the fused image is divided into landslide area and non-landslide area.

6. The method according to claim 1, wherein Based on the fusion image of backscatter intensity difference enhancement and coherence change information, earthquake landslide detection is performed in the detection area, including determining the landslide area and non-landslide area according to the following formula: Where t represents the candidate threshold, It represents the inter-class variance between the foreground and the background when the candidate threshold is t. Argmax is used to obtain t that maximizes the inter-class variance as the optimal segmentation threshold. θ is the optimal segmentation threshold. ID-CCD (x,y) Indicates the gray value of the pixel (x, y) in the fused image, B(ID-CCD (x,y) ) represents the valued segmentation result. The area with a segmentation result of 1 represents the potential landslide area, while 0 represents the non-landslide area.

7. A coherence change earthquake landslide detection device based on intensity difference enhancement, characterized in that: include: An acquisition unit, used for acquiring satellite data covering the area to be detected; An extraction unit, configured to extract coherence information from satellite data covering an area to be detected; A post-earthquake coherence increase map generating unit is used to determine a post-earthquake coherence map and a coherence map at the time of the earthquake based on the coherence information; and to generate a post-earthquake coherence increase map based on a difference between the post-earthquake coherence map and the coherence map at the time of the earthquake; A backscatter intensity difference map generating unit, configured to generate a backscatter intensity difference map based on a difference between a backscatter coefficient map after an earthquake and a backscatter coefficient map before an earthquake; A fusion unit is used to fuse the post-earthquake coherence enhancement map with the backscatter intensity difference map at the feature level to obtain a fused image of the backscatter intensity difference enhancement and coherence change information; The earthquake landslide detection unit is used to perform earthquake landslide detection on the detection area based on the fusion image of backscatter intensity difference enhancement and coherence change information.

8. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: 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