Forest disturbance detection method, system, equipment and medium based on full polarization SAR image

Abstract

The present invention discloses a forest disturbance detection method based on full-polarization SAR images, including: obtaining full-polarization SAR images in a target area, and dividing the images into reference phase and monitoring phase; preprocessing the divided full-polarization SAR images to obtain false color synthetic SAR images; including: SAR image import, SAR image polarization matrix conversion, Refined Lee filtering and false color synthesis; inputting the false color synthetic SAR images into an optimized surface change detection model to extract surface change spots; and combining with radar vegetation index RVI images to obtain initial forest disturbance spots; superimposing and verifying the initial forest disturbance spots to obtain forest disturbance spots in the target area, and generating visualization results. It can observe the ground all day and all weather, can penetrate clouds and fog, can be widely used in forest resource monitoring, environmental protection and other fields, and can better serve ecological protection and restoration supervision.

Description

Technical Field

[0001] The present invention relates to the field of remote sensing technology, and more particularly to a forest disturbance detection method, system, equipment and medium based on full-polarization SAR images. Background Art

[0002] In recent years, my country has attached great importance to ecological environment protection and ecological civilization construction, implemented a series of effective ecological environment protection measures, and the quality of the ecological environment has been significantly improved.

[0003] Forests have functions such as soil and water conservation, water conservation, wind and sand fixation, and biodiversity protection. They are an important part of the earth's ecosystem and ecological balance. Especially for areas with important ecological functions and fragile ecological environment, overexploitation of forests is an important factor leading to the decline of regional ecological service functions. In recent years, illegal destruction of forest resources has been rampant, seriously threatening the country's ecological security. Therefore, it is urgent to strengthen the monitoring of forest development, which is crucial to improving the diversity, stability and sustainability of the ecosystem.

[0004] Remote sensing technology is widely used in the field of surface cover change monitoring. At present, surface change detection technology based on high-resolution optical remote sensing images has become the main means of forest land change and other businesses. However, optical images are greatly affected by weather. In areas with high forest coverage, cloudy and rainy areas, there are often problems such as difficulty in obtaining optical data or poor image quality, which to some extent affects the accuracy and timeliness of remote sensing monitoring data results.

[0005] Compared with optical satellites, the number of SAR satellites is relatively small, and the data source has limited its wide application to a certain extent. At the same time, SAR images are difficult to monitor human activities due to factors such as coherent speckle noise, geometric distortion, overlap and shadow, and there are few related application studies. In recent years, SAR satellite technology has developed rapidly, and many high-resolution satellites in different bands have been launched. The polarization mode has also developed from single polarization to dual polarization, multi-polarization and full polarization, which has better expanded the application fields and directions of SAR satellite remote sensing.

[0006] Therefore, how to design a forest disturbance detection method based on fully polarized SAR images to make up for the insufficient coverage of high-resolution optical images in cloudy and rainy areas and optimize the monitoring of forest development activities is an urgent problem that technicians in this field need to solve. Summary of the invention

[0007] In view of this, the present invention provides a forest disturbance detection method based on fully polarized SAR images, which can be applied to the monitoring of forest disturbance in important ecological spaces. It can timely discover forest disturbance activities, serve the supervision of ecological protection and restoration, and thus improve the diversity, stability and sustainability of the ecosystem.

[0008] In order to achieve the above object, the present invention adopts the following technical solution:

[0009] In a first aspect, the present invention provides a forest disturbance detection method based on full-polarization SAR images, comprising:

[0010] S1. Acquire a fully polarized SAR image in a target area, and divide the fully polarized SAR image into a reference phase and a monitoring phase image;

[0011] S2, preprocessing the divided full-polarization SAR image to obtain a false color synthetic SAR image; including: SAR image import, SAR image polarization matrix conversion, Refined Lee filtering and false color synthesis;

[0012] S3, inputting the false color synthetic SAR image into the optimized surface change detection model to extract the surface change spots; and combining with the radar vegetation index RVI image to obtain the initial forest disturbance spots;

[0013] S4. Overlay and verify the initial forest disturbance map to obtain forest disturbance maps in the target area and generate visualization results.

[0014] The step S2, preprocessing the divided full-polarization SAR image to obtain a false color synthetic SAR image, specifically includes:

[0015] S21, SAR image import: convert the fully polarized SAR images of the reference phase and the monitoring phase into a standard data format, and perform radiation correction processing to obtain fully polarized single-view complex data X:

[0016] X={X i :i=1,2,…,M}

[0017] Among them, X i ={xpq i =Ipq i +jQ i :p,q∈{H,V}}

[0018] M is the total number of pixels, p is the antenna transmission mode, q is the antenna reception mode, i is the pixel index, j is the imaginary unit, xpq i represents the complex scattering coefficient of the i-th pixel under pq polarization mode, Ipq i For xpq i The real part, Qpq i For xpq i The imaginary part of , H represents horizontal polarization, V represents vertical polarization;

[0019] S22. SAR image polarization matrix conversion: construct a polarization scattering matrix S based on the fully polarized single-view complex data X:

[0020] S={S i :i=1,2,…,M}

[0021]

[0022] The target scattering vector set K is defined using the polarization scattering matrix S:

[0023] K={K i :i=1,2,…,M}

[0024]

[0025] The target scattering vector set K is converted into a polarization coherence matrix T:

[0026] T={T i :i=1,2,…,M}

[0027]

[0028] Among them, * represents conjugation operation, <> represents multi-view processing;

[0029] S23, Refined Lee filtering: performing Refined Lee filtering on the polarization coherence matrix T to obtain a filtered polarization coherence matrix T1, including:

[0030]

[0031] Where N represents the number of views, and m represents the pixel index of the image after multi-view processing;

[0032] The azimuth resolution is reduced to 1 / N of the original, while the range resolution remains unchanged, and we obtain:

[0033]

[0034] in, represents the mean value of the same boundary alignment window in the local window where the m-th pixel is located, and w is the filtering weight value;

[0035] S24, false color synthesis: Use the Pauli synthesis method to perform full polarization false color synthesis on the filtered polarization conversion matrix T1 to obtain the false color synthetic SAR image S1 and the false color synthetic SAR image S2.

[0036] Furthermore, in S3, the surface change detection model optimization includes:

[0037] S31. Creating a labeled image dataset based on surface disturbance interpretation signs;

[0038] S32, training a surface change detection model using the labeled image dataset;

[0039] S33, classifying the false color synthetic SAR image through the trained surface change detection model to obtain a classification result;

[0040] S34, performing vector conversion on the classification result, calculating the cross entropy loss value using the cross entropy loss function, and performing model evaluation;

[0041] S35. If the cross entropy loss value is less than the cross entropy loss threshold, output the classification result; otherwise, return to S31 and use the classification result to supplement the labeled image data set.

[0042] Furthermore, in S31, constructing the surface disturbance interpretation mark includes:

[0043] S311, acquiring the first two late high-resolution optical remote sensing images that are consistent with the fully polarized SAR image;

[0044] S312, inputting the first two late high-resolution optical remote sensing images into a pre-trained neural network model to extract newly added surface disturbance patches;

[0045] S313, constructing a surface disturbance interpretation mark in combination with the false color synthetic SAR image.

[0046] Furthermore, in S3, the Freeman method is used to calculate the radar vegetation index RVI:

[0047] RVI=Fv / (Fv+Fd+Fs)

[0048] Among them, Fv, Fd and Fs represent the odd scattering component, even scattering component and volume scattering component obtained by decomposing the scattering matrix after the polarization matrix conversion of the SAR image, respectively.

[0049] Furthermore, in S4, the initial forest disturbance map is overlaid and verified, including:

[0050] S411, superimposing and verifying the initial forest land disturbance map and the forestry land data to obtain the forestry land disturbance map;

[0051] S412: spatially superimpose and verify the forest land disturbance map and slope data to obtain the forest land disturbance map within a preset slope range.

[0052] Furthermore, in S4, generating a visualization result includes:

[0053] S421, exporting the forest disturbance patches in the target area to a vector data layer;

[0054] S422, labeling the vector data layer in combination with the original attribute information;

[0055] S423. Generate various types of visualization results based on the annotated vector data layer, including: forest property classification map, forest disturbance hotspot map and forest disturbance trend map.

[0056] In a second aspect, the present invention provides a forest disturbance detection system based on full-polarization SAR images, comprising:

[0057] Data acquisition module: used to acquire full-polarization SAR images in the target area, and divide the full-polarization SAR images into reference phase and monitoring phase images;

[0058] Data preprocessing module: used to preprocess the divided full-polarization SAR images to obtain false color synthetic SAR images; including: SAR image import, SAR image polarization matrix conversion, Refined Lee filtering and false color synthesis;

[0059] Feature extraction module: used to input the false color synthetic SAR image into the optimized surface change detection model to extract the surface change spots; and combine the radar vegetation index RVI image to obtain the initial forest disturbance spots;

[0060] Overlay verification module: used to perform overlay verification on the initial forest disturbance map, obtain the forest disturbance map in the target area, and generate visualization results.

[0061] In a third aspect, the present invention provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above-mentioned forest disturbance detection method when executing the computer program.

[0062] In a fourth aspect, the present invention provides a computer-readable storage medium, wherein the storage medium stores a computer program, and the computer program implements the above-mentioned forest disturbance detection method when executed by a processor.

[0063] The descriptions of the second to fourth aspects of the present invention can refer to the detailed description of the first aspect; and the beneficial effects of the descriptions of the second to fourth aspects can refer to the analysis of the beneficial effects of the first aspect, which will not be repeated here.

[0064] It can be seen from the above technical solutions that, compared with the prior art, the forest disturbance detection method based on full-polarization SAR images provided by the present invention has the following beneficial effects:

[0065] 1. This method uses full-polarization SAR images, combined with the division of baseline phase and monitoring phase images, to improve the accuracy of detection and analysis of forest disturbance in the target area. Fully polarization SAR images are weather-independent and have remote sensing penetration, and can obtain high-quality image data under various weather conditions.

[0066] 2. By preprocessing the fully polarized SAR images, including SAR image import, SAR image polarization matrix conversion, Refined Lee filtering and false color synthesis, high-quality false color synthetic SAR images are generated. It can provide more intuitive and clear surface information, which is conducive to subsequent surface change detection and spot extraction.

[0067] 3. Based on the optimized surface change detection model, the surface change patches can be extracted more accurately. Combined with the radar vegetation index (RVI) image, the disturbance of forest land can be better distinguished, thus improving the accuracy and reliability of the detection results.

[0068] 4. Finally, by superimposing and verifying the initial forest disturbance map, the detection accuracy of forest disturbance in the target area can be further improved, reducing the false detection rate and missed detection rate, generating more reliable and accurate visualization results, and providing important reference information for forest protection and management. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0070] Figure 1 A flow chart of a forest disturbance detection method based on full-polarization SAR images provided by an embodiment of the present invention;

[0071] Figure 2 A flow chart of a method for preprocessing the divided fully polarized SAR images provided in an embodiment of the present invention;

[0072] Figure 3 A schematic diagram of a fully polarized single-view complex data image after the SAR image is imported according to an embodiment of the present invention;

[0073] Figure 4 A schematic diagram of an image after polarization matrix conversion provided by an embodiment of the present invention;

[0074] Figure 5 A schematic diagram of an image after Refined Lee filtering provided by an embodiment of the present invention;

[0075] Figure 6 The false color synthesized SAR image provided by the embodiment of the present invention;

[0076] Figure 7 A flow chart of a method for optimizing a surface change detection model provided by an embodiment of the present invention;

[0077] Figure 8 A flow chart of a method for constructing a surface disturbance interpretation mark provided by an embodiment of the present invention;

[0078] Fig. 9 A schematic diagram of a newly added surface disturbance pattern provided in an embodiment of the present invention;

[0079] Fig.10 A schematic diagram of a surface disturbance interpretation mark provided by an embodiment of the present invention;

[0080] Fig.11 A radar vegetation index image provided by an embodiment of the present invention;

[0081] Fig.12 A schematic diagram of an initial forest disturbance pattern provided by an embodiment of the present invention;

[0082] Fig.13 A flow chart of a method for superimposing and checking initial forest disturbance patches provided in an embodiment of the present invention;

[0083] Fig.14 A flow chart of a method for generating visualization results provided by an embodiment of the present invention;

[0084] Fig.15 A schematic diagram of the structure of a forest disturbance detection system based on full-polarization SAR images provided by an embodiment of the present invention;

[0085] Fig.16 A schematic diagram of the structure of an electronic device is provided for an embodiment of the present invention. DETAILED DESCRIPTION

[0086] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0087] The detection method provided in the embodiment of the present application can be applied to a detection server. The above-mentioned detection server can be hardware or software. When the detection server is hardware, it can be implemented as a distributed server cluster that provides detection services, or it can be implemented as a single server. When the detection server is software, it can be installed in the servers listed above. It can be implemented as multiple software or software modules, or it can be implemented as a single software or software module, which is not specifically limited here.

[0088] Embodiment 1:

[0089] like Figure 1 As shown, this embodiment takes an area within a certain range of a city as a target area and provides a forest disturbance detection method based on full polarization SAR images, including:

[0090] S1. Acquire a fully polarized SAR image in a target area, and divide the fully polarized SAR image into a reference phase and a monitoring phase image;

[0091] The domestically produced GF-3 satellite has multiple polarization modes such as single polarization, dual polarization and full polarization. The GF-3 full polarization data covering a certain range of a city is used, with a spatial resolution of 8 meters. An interpretation mark library is established to improve the recognition accuracy, and the GF-1 optical data is used as auxiliary data with a spatial resolution of 2 meters.

[0092] S2, preprocessing the divided full-polarization SAR image to obtain a false color synthetic SAR image; including: SAR image import, SAR image polarization matrix conversion, Refined Lee filtering and false color synthesis;

[0093] S3, inputting the false color synthetic SAR image into the optimized surface change detection model to extract the surface change spots; and combining with the radar vegetation index RVI image to obtain the initial forest disturbance spots;

[0094] S4. Overlay and verify the initial forest disturbance map to obtain forest disturbance maps in the target area and generate visualization results.

[0095] This method has efficient, accurate and reliable detection capabilities, and can observe the ground through clouds and fog around the clock and in all weather conditions. It can be widely used in forest resource monitoring, environmental protection and other fields, especially in remote sensing monitoring of forest development activities in cloudy and rainy areas where optical images lack effective coverage. It can better serve ecological protection and restoration supervision, thereby improving the diversity, stability and sustainability of ecosystems.

[0096] The above steps are further explained in detail below:

[0097] like Figure 2As shown, in this embodiment S2, the divided full-polarization SAR image is preprocessed to obtain a false color synthetic SAR image, specifically including:

[0098] S21, SAR image import: convert the fully polarized SAR images of the reference phase and the monitoring phase into a standard data format, and perform radiation correction processing to obtain fully polarized single-view complex data X:

[0099] X={X i :i=1,2,…,M}

[0100] Among them, X i ={xpq i =Ipq i +jQ i :p,q∈{H,V}}

[0101] M is the total number of pixels, p is the antenna transmission mode, q is the antenna reception mode, i is the pixel index, j is the imaginary unit, xpq i represents the complex scattering coefficient of the i-th pixel under pq polarization mode, Ipq i For xpq i The real part, Qpq i For xpq i The imaginary part of , H represents horizontal polarization, V represents vertical polarization;

[0102] like Figure 3 As shown, the single-view complex data without subsequent processing has a lot of noise and poor image readability, and cannot be directly used for human activity pattern interpretation.

[0103] S22. SAR image polarization matrix conversion: construct a polarization scattering matrix S based on the fully polarized single-view complex data X:

[0104] S={S i :i=1,2,…,M}

[0105]

[0106] The target scattering vector set K is defined using the polarization scattering matrix S:

[0107] K={K i :i=1,2,…,M}

[0108]

[0109] The target scattering vector set K is converted into a polarization coherence matrix T:

[0110] T={T i :i=1,2,…,M}

[0111]

[0112] Among them, * represents conjugation operation, <> represents multi-view processing;

[0113] like Figure 4 As shown, the converted polarization coherence matrix includes 9 images.

[0114] S23, Refined Lee filtering: performing Refined Lee filtering on the polarization coherence matrix T, using a window size of 7*7 Refined Lee filtering to obtain a filtered polarization coherence matrix T1, including:

[0115]

[0116] Where N represents the number of views, and m represents the pixel index of the image after multi-view processing;

[0117] The azimuth resolution is reduced to 1 / N of the original, while the range resolution remains unchanged, and we obtain:

[0118]

[0119] in, represents the mean value of the same boundary alignment window in the 7*7 local window where the m-th pixel is located, and w is the filtering weight value;

[0120] like Figure 5 As shown in the figure, the image noise after Refined Lee filtering with a window size of 7*7 is moderate, while the image detail information is retained to the maximum extent.

[0121] S24, false color synthesis: Use the Pauli synthesis method to perform full polarization false color synthesis on the filtered polarization conversion matrix T1 to obtain the false color synthetic SAR image S1 and the false color synthetic SAR image S2.

[0122] like Figure 6 As shown in the figure, the false color synthesized SAR image has rich color information and the visibility of different ground objects is significantly improved, which is conducive to the interpretation of human activities.

[0123] like Figure 7 As shown, in this embodiment S3, the surface change detection model optimization includes:

[0124] S31. Creating a labeled image dataset based on surface disturbance interpretation signs;

[0125] S32, training a surface change detection model using the labeled image dataset;

[0126] S33, classifying the false color synthetic SAR image through the trained surface change detection model to obtain a classification result;

[0127] S34, performing vector conversion on the classification result, calculating the cross entropy loss value using the cross entropy loss function, and performing model evaluation;

[0128] S35. If the cross entropy loss value is less than the cross entropy loss threshold, output the classification result; otherwise, return to S31 and use the classification result to supplement the labeled image data set.

[0129] like Figure 8 As shown, the construction of surface disturbance interpretation signs includes:

[0130] S311, acquiring the first two late high-resolution optical remote sensing images that are consistent with the fully polarized SAR image;

[0131] S312, such as Fig. 9 As shown, the first two late high-resolution optical remote sensing images are input into a pre-trained neural network model to extract newly added surface disturbance patches;

[0132] S313, such as Fig.10 As shown, the false color synthetic SAR image is combined to construct the surface disturbance interpretation mark.

[0133] like Fig.11 As shown, the Freeman method is used to calculate the radar vegetation index RVI:

[0134] RVI=Fv / (Fv+Fd+Fs)

[0135] Among them, Fv, Fd and Fs represent the odd scattering component, even scattering component and volume scattering component obtained by decomposing the scattering matrix after the polarization matrix conversion of the SAR image, respectively.

[0136] like Fig.12 As shown in the figure, the initial forest disturbance map is obtained by combining the radar vegetation index RVI.

[0137] like Fig.13 As shown, in this embodiment S4, the initial forest disturbance pattern is overlaid and checked, including:

[0138] S411, superimposing and verifying the initial forest land disturbance map and the forestry land data to obtain the forestry land disturbance map;

[0139] S412: spatially superimpose and verify the forest land disturbance map and slope data to obtain the forest land disturbance map within a preset slope range.

[0140] The following is a detailed description of the superposition and verification of the initial forest disturbance map:

[0141] Overlay and check the initial forest disturbance map. In the GIS (Geographic Information System) software, overlay and check the initial forest disturbance map, especially with the forestry land data and slope data for spatial overlay and verification:

[0142] Data preparation: This includes: Forest land data: Make sure you have a vector data layer that contains the boundaries and related information of the forest land. Slope data: You also need a vector or raster data layer that contains slope information. This can be slope data calculated based on a DEM (digital elevation model). Initial forest disturbance map: This is the layer you need to check, which may be a vector layer that contains the location and extent of forest disturbance.

[0143] Import data: Open ArcGIS software and create a new map project. Use the "Add Data" function to import forest land data, slope data, and initial forest disturbance patches into the map project.

[0144] Spatial overlay analysis:

[0145] Overlay with forestry land data: In the ArcGIS toolbox, find the "Overlay Analysis" tool. Select the "Intersect" or "Identify" tool, depending on the result you want. If you just want to find the disturbed patches that overlap with forestry land, you can use the "Intersect" tool; if you also want to keep other information about the initial forest disturbance patches, you can use the "Identify" tool. In the dialog box that pops up, set the input features (initial forest disturbance patches) and overlay features (forestry land data).

[0146] Set the location and name of the output layer, then click OK to run the analysis. The result will be a new layer containing the disturbed forest land patches that overlap with the forest land. Overlay with slope data (assuming the slope data is raster data): If the slope data is raster data, you may need to use the "Raster to Vector" tool to convert it to vector data to facilitate overlay analysis. Once the conversion is complete, use the "Spatial Selection" or "Select by Location" tools to select the vector slope data with slopes within a preset range. Next, use the same overlay analysis tools as the forest land data (such as "Intersect" or "Identity") to overlay the selected vector slope data with the disturbed forest land patches. Set the parameters and run the analysis. The result will be a new layer containing the disturbed forest land patches that are within the preset slope range.

[0147] Result processing:

[0148] Perform necessary processing and cleaning on the results of the overlay analysis, such as deleting duplicates, merging adjacent patches, etc. If necessary, you can modify the style and symbols of the layer to better display the results. You can use the export function of ArcGIS to export the results to the required format (such as Shapefile, GeoJSON, KMZ, etc.) for sharing with other software or platforms.

[0149] Through the above steps, the initial forest land disturbance map is superimposed and checked with the forestry land data and slope data based on ArcGIS software, and the forestry land disturbance map within the required preset slope range is generated.

[0150] like Fig.14 As shown, the visualization results are generated, including:

[0151] S421, exporting the forest disturbance patches in the target area to a vector data layer;

[0152] S422, labeling the vector data layer in combination with the original attribute information;

[0153] S423. Generate various types of visualization results based on the annotated vector data layer, including: forest property classification map, forest disturbance hotspot map and forest disturbance trend map.

[0154] Forest property classification map: A visual chart that classifies forest disturbance spots according to their properties and represents them with different colors or symbols. Through this chart, you can clearly understand the different types of forest disturbances in the target area, such as deforestation, fire, natural disasters, etc.

[0155] Woodland Disturbance Heat Map: This can show woodland disturbance hotspots within a target area. Hotspots are usually represented as high-density points, heat maps, or density maps to indicate the concentration and distribution of woodland disturbance. This helps identify areas that require focus and action.

[0156] Forest disturbance trend graph: Using time series data, a forest disturbance trend graph can be generated. This visualization can show the changing trend of forest disturbance over time, helping analysts understand the evolution and influencing factors of forest disturbance.

[0157] The technical solution in this embodiment realizes accurate detection and visualization of forest disturbance by combining fully polarized SAR images and radar vegetation index RVI, as well as optimization of surface change detection model and construction of surface disturbance interpretation mark. It plays an important role in forest resource management, environmental monitoring and natural disaster assessment, and provides more effective data support and decision-making basis for related fields.

[0158] Embodiment 2:

[0159] like Fig.15 As shown, this embodiment provides a forest disturbance detection system based on full-polarization SAR images, including:

[0160] Data acquisition module: used to acquire full-polarization SAR images in the target area, and divide the full-polarization SAR images into reference phase and monitoring phase images;

[0161] Data preprocessing module: used to preprocess the divided full-polarization SAR images to obtain false color synthetic SAR images; including: SAR image import, SAR image polarization matrix conversion, Refined Lee filtering and false color synthesis;

[0162] Feature extraction module: used to input the false color synthetic SAR image into the optimized surface change detection model to extract the surface change spots; and combine the radar vegetation index RVI image to obtain the initial forest disturbance spots;

[0163] Overlay verification module: used to perform overlay verification on the initial forest disturbance map, obtain the forest disturbance map in the target area, and generate visualization results.

[0164] Embodiment 3:

[0165] like Fig.16 As shown, this embodiment provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, and when the processor executes the computer program, the forest disturbance detection method in the above embodiment is implemented.

[0166] Embodiment 4:

[0167] This embodiment provides a computer-readable storage medium, wherein the storage medium stores a computer program, and when the computer program is executed by a processor, the forest disturbance detection method in the above embodiment is implemented.

[0168] In the technical solution of the above embodiment, the fully polarized SAR image can collect more detailed surface information through different polarization states. Compared with traditional non-polarized SAR images, fully polarized SAR images can provide additional information about the surface structure and material, which is especially important for forest disturbance detection. Because disturbances in forest areas often lead to changes in surface structure, fully polarized SAR images can help identify and monitor forest disturbance events by capturing these subtle structural changes.

[0169] The preprocessing of SAR images includes image import, polarization matrix conversion and Refined Lee filtering, which can remove or reduce the impact of noise and improve image quality. False color synthesis can highlight key features while suppressing irrelevant background information, making the detection of forest disturbance more accurate.

[0170] After optimization, the surface change detection model can process pre-processed SAR images more efficiently and accurately extract surface change patches. The stacking verification process can determine the specific scope and extent of forest disturbance by comparing SAR images from different time periods. It can reduce the error in a single measurement and ensure the accuracy of the final disturbance patch. Converting complex SAR image data into easy-to-understand visualization results provides decision makers and relevant personnel with fast and accurate information, helping them to respond in a timely manner and take appropriate protection measures.

[0171] Through the effective use of full-polarimetric SAR images, the application of preprocessing and false color synthesis, the optimization of the surface change detection model, the stacking verification and the generation of visualization results, this method has high accuracy and reliability in practical applications and can effectively support forest management and protection.

[0172] In the above embodiments provided in the present application, it should be understood that the disclosed methods, systems, devices and media can be implemented in other ways. The above described methods, systems, devices and media embodiments are only schematic. For example, the division of modules or units is only a logical function division, and there may be other division methods in actual implementation. Each functional unit can be integrated into a processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0173] The units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0174] The computer program includes computer program code, which may be in source code form, object code form, executable file or some intermediate form, etc. Computer readable media may include: any entity or device capable of carrying computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory, random access memory, electric carrier signal, telecommunication signal and software distribution medium, etc.

[0175] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A forest disturbance detection method based on full polarization SAR images, characterized in that: include: S1. Acquire a fully polarized SAR image in a target area, and divide the fully polarized SAR image into a reference phase and a monitoring phase image; S2, preprocessing the divided full-polarization SAR image to obtain a false color synthetic SAR image; including: SAR image import, SAR image polarization matrix conversion, Refined Lee filtering and false color synthesis; S3, inputting the false color synthetic SAR image into the optimized surface change detection model to extract the surface change spots; and combining with the radar vegetation index RVI image to obtain the initial forest disturbance spots; S4. Overlaying and verifying the initial forest disturbance spots to obtain forest disturbance spots in the target area and generate visualization results; wherein overlaying and verifying the initial forest disturbance spots includes: S411, superimposing and verifying the initial forest land disturbance map and the forestry land data to obtain the forestry land disturbance map; S412: spatially superimpose and verify the forest land disturbance map and slope data to obtain the forest land disturbance map within a preset slope range.

2. The method for detecting forest disturbance based on full polarization SAR images according to claim 1, characterized in that: The S2, preprocessing the divided full-polarization SAR image to obtain a false color synthetic SAR image, specifically includes: S21, SAR image import: convert the fully polarized SAR images of the reference phase and the monitoring phase into a standard data format, and perform radiation correction processing to obtain fully polarized single-view complex data X: X={X i :i=1,2,…,M} Among them, X i ={xpq i =Ipq i +jQ i :p,q∈{H,V}} M is the total number of pixels, p is the antenna transmission mode, q is the antenna reception mode, i is the pixel index, j is the imaginary unit, xpq i represents the complex scattering coefficient of the i-th pixel under pq polarization mode, Ipq i For xpq i The real part, Qpq i For xpq i The imaginary part of , H represents horizontal polarization, V represents vertical polarization; S22. SAR image polarization matrix conversion: construct a polarization scattering matrix S based on the fully polarized single-view complex data X: S={S i :i=1,2,…,M} The target scattering vector set K is defined using the polarization scattering matrix S: K={K i :i=1,2,…,M} The target scattering vector set K is converted into a polarization coherence matrix T: T={T i :i=1,2,…,M} Among them, * represents conjugation operation, <> represents multi-view processing; S23, Refined Lee filtering: performing Refined Lee filtering on the polarization coherence matrix T to obtain a filtered polarization coherence matrix T1, including: Where N represents the number of views, and m represents the pixel index of the image after multi-view processing; The azimuth resolution is reduced to 1 / N of the original, while the range resolution remains unchanged, and we obtain: in, represents the mean value of the same boundary alignment window in the local window where the m-th pixel is located, and w is the filtering weight value; S24, false color synthesis: Use the Pauli synthesis method to perform full polarization false color synthesis on the filtered polarization conversion matrix T1 to obtain the false color synthetic SAR image S1 and the false color synthetic SAR image S2.

3. The method for detecting forest disturbance based on full polarization SAR images according to claim 1, characterized in that: In S3, the surface change detection model optimization includes: S31. Creating a labeled image dataset based on surface disturbance interpretation signs; S32, training a surface change detection model using the labeled image dataset; S33, classifying the false color synthetic SAR image through the trained surface change detection model to obtain a classification result; S34, performing vector conversion on the classification result, calculating the cross entropy loss value using the cross entropy loss function, and performing model evaluation; S35. If the cross entropy loss value is less than the cross entropy loss threshold, output the classification result; otherwise, return to S31 and use the classification result to supplement the labeled image data set.

4. The method for detecting forest disturbance based on full polarization SAR images according to claim 3 is characterized in that: In S31, the construction of the surface disturbance interpretation mark includes: S311, acquiring the first two late high-resolution optical remote sensing images that are consistent with the fully polarized SAR image; S312, inputting the first two late high-resolution optical remote sensing images into a pre-trained neural network model to extract newly added surface disturbance patches; S313, constructing a surface disturbance interpretation mark in combination with the false color synthetic SAR image.

5. The method for detecting forest disturbance based on full polarization SAR images according to claim 1, characterized in that: In S3, the radar vegetation index RVI is calculated using the Freeman method: RVI=Fv / (Fv+Fd+Fs) Among them, Fv, Fd and Fs represent the odd scattering component, even scattering component and volume scattering component obtained by decomposing the scattering matrix after the polarization matrix conversion of the SAR image, respectively.

6. The method for detecting forest disturbance based on full polarization SAR images according to claim 1, characterized in that: In S4, a visualization result is generated, including: S421, exporting the forest disturbance patches in the target area to a vector data layer; S422, labeling the vector data layer in combination with the original attribute information; S423. Generate various types of visualization results based on the annotated vector data layer, including: forest property classification map, forest disturbance hotspot map and forest disturbance trend map.

7. A forest disturbance detection system based on full polarization SAR images, characterized in that: include: Data acquisition module: used to acquire full-polarization SAR images in the target area, and divide the full-polarization SAR images into reference phase and monitoring phase images; Data preprocessing module: used to preprocess the divided full-polarization SAR images to obtain false color synthetic SAR images; including: SAR image import, SAR image polarization matrix conversion, Refined Lee filtering and false color synthesis; Feature extraction module: used to input the false color synthetic SAR image into the optimized surface change detection model to extract the surface change spots; and combine the radar vegetation index RVI image to obtain the initial forest disturbance spots; The superposition verification module is used to perform superposition verification on the initial forest disturbance map, obtain the forest disturbance map in the target area, and generate a visualization result; wherein, the superposition verification on the initial forest disturbance map includes: Superimposing and verifying the initial forest land disturbance map and forestry land data to obtain forestry land disturbance map; The forestry land disturbance map and slope data are spatially superimposed and verified to obtain the forestry land disturbance map within a preset slope range.

8. An electronic 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 forest disturbance detection method according to any one of claims 1 to 6 is implemented.

9. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the forest disturbance detection method according to any one of claims 1 to 6 is implemented.

Images