An intertidal oyster reef body analysis method and system fusing thermal imaging aerial photography

Abstract

The application discloses a kind of intertidal zone oyster reef body analysis method and system of fusion thermal imaging aerial photography, comprising: obtaining multi-source aerial photography data set, extracting the thermal infrared image data of intertidal zone region, and constructing regional temperature thermal map;Obtain multispectral substrate classification chart, segment the intertidal zone oyster reef body of target region, and obtain oyster reef body distribution chart;Construct the dense three-dimensional point cloud of the intertidal zone of target region, combined with regional temperature thermal map to execute heat-light data fusion operation, generate fusion point cloud data set;Oyster reef body distribution chart is used as spatial mask to be cut and construct intertidal zone reef body three-dimensional model, crack detection and oyster reef body feature measurement are carried out through intertidal zone three-dimensional model, generate multi-modal reef body feature vector;Intertidal zone oyster reef body health in target region is analyzed, and combined with regional ecological data to determine the priority of regional repair, generate early warning information, improve the convenience and accuracy of reef structure parameter measurement.

Description

Technical Field

[0001] This invention relates to the field of intertidal oyster reef analysis technology, and in particular to a method and system for analyzing intertidal oyster reefs by integrating thermal imaging aerial photography. Background Technology

[0002] Oyster reefs are an important component of coastal ecosystems, and monitoring their structure is crucial for coastal ecological restoration. Traditional monitoring methods mainly rely on manual field surveys, requiring personnel to enter the reef area during the low tide window of spring tides to measure reef dimensions using tools such as measuring tapes and sounding rods, and manually record the data. This method is difficult to use for large-scale regional surveys, and the complex topography of the intertidal zone leads to significant measurement errors. In adverse weather or at night, traditional methods face serious safety risks and data loss issues. While visible light aerial photography technology has emerged in recent years, which can improve the operational range, its imaging quality is significantly affected by water reflection, suspended sediment, and lighting conditions, especially in turbid water environments, where the accuracy of reef edge identification is insufficient.

[0003] Therefore, a method and system for analyzing intertidal oyster reefs by integrating thermal imaging aerial photography is proposed to address the above problems, aiming to improve the convenience and accuracy of measuring reef structural parameters. Summary of the Invention

[0004] This invention overcomes the shortcomings of existing technologies and provides a method and system for analyzing intertidal oyster reef structures by integrating thermal imaging aerial photography.

[0005] To achieve the above objectives, the first aspect of this invention provides a method for analyzing intertidal oyster reef structures using fusion thermal imaging aerial photography, comprising:

[0006] A multi-source aerial photography dataset is acquired by patrolling the target area using aerial photography drones. Thermal infrared image data of the intertidal zone is extracted to generate a radiation intensity distribution map. A regional temperature heat map is then constructed using the radiation intensity distribution map.

[0007] Multispectral data was extracted from the multi-source aerial photography dataset to generate a multispectral substrate classification map. The intertidal oyster reef body in the target area was segmented by combining the regional temperature heat map and the reef body edge was optimized to obtain the oyster reef body distribution map.

[0008] Based on the multi-source aerial photography dataset, a dense three-dimensional point cloud of the intertidal zone of the target area is constructed. Combined with the regional temperature heat map, a thermal-optical data fusion operation is performed to generate a fused point cloud dataset.

[0009] The oyster reef distribution map is used as a spatial mask to crop the fused point cloud dataset and construct a three-dimensional model of the intertidal reef. Crack detection and oyster reef feature measurement are performed using the three-dimensional model of the intertidal zone to generate a multimodal reef feature vector.

[0010] Based on the analysis of the multimodal reef feature vector, the health of the intertidal oyster reef in the target area is determined, and regional ecological data is combined to determine the priority of regional restoration and generate early warning information for notification.

[0011] In this solution, the step of acquiring multi-source aerial photography datasets by patrolling the target area using a drone, extracting thermal infrared image data of the intertidal zone to generate a radiation intensity distribution map, and constructing a regional temperature heat map using the radiation intensity distribution map specifically includes:

[0012] A drone equipped with multi-source monitoring equipment is used to cruise the target area. The multi-source monitoring equipment collects thermal infrared image data, visible light image data and multispectral data of the intertidal zone of the target area to generate a multi-source aerial photography dataset.

[0013] Based on the multi-source aerial photography dataset, thermal infrared image data of the intertidal zone is extracted. The thermal infrared image data is then radiometrically calibrated and converted into absolute radiation intensity values ​​based on the temperature-radiation value calibration curve of the sensor's built-in blackbody reference source, generating a radiation intensity distribution map.

[0014] Acquire atmospheric data of the target area during the cruise, input it into the MODTRAN radiation model to calculate atmospheric transmittance and uplink radiation path brightness, retrieve the true surface radiation intensity value through the radiative transfer equation, and perform atmospheric path correction on the radiation intensity distribution map to obtain the corrected radiation intensity distribution map.

[0015] Local binary mode texture calculation is performed on the corrected radiation intensity distribution map, the gray-level variation coefficient of each pixel neighborhood is extracted as a uniformity index, and the texture features are divided by K-means clustering algorithm to generate a divided radiation intensity distribution map.

[0016] By applying the Stefan-Boltzmann law to the radiation intensity distribution map, temperature conversion is performed, and the final output is a regional temperature thermogram.

[0017] In this scheme, the step of extracting multispectral images from the multi-source aerial photography dataset to generate a multispectral seabed classification map, combining this with a regional temperature heatmap to segment the intertidal oyster reef bodies in the target area, and optimizing the reef body edges to obtain an oyster reef body distribution map, specifically includes:

[0018] The multispectral features of oyster reefs are obtained by searching through big data networks to generate a reef spectral feature set, and the reef spectral feature set is mapped to a high-dimensional space using a Gaussian kernel function.

[0019] The minimum hypersphere boundary and hypersphere radius containing the multispectral feature samples of all oyster reef bodies are obtained by solving in the mapped high-dimensional space. A multispectral geological classification model is then established using the minimum hypersphere boundary and hypersphere radius.

[0020] A multi-source aerial photography dataset is acquired, multispectral data is extracted from the multi-source aerial photography dataset and input into the multispectral geological classification model, and the Mahalanobis distance of the input multispectral data in the kernel space is calculated.

[0021] The calculated Mahalanobis distance is compared with the preset hypersphere radius. If it is less than the preset hypersphere radius, the corresponding multispectral data source location is marked as the oyster reef body area, and finally a multispectral geological classification map is generated.

[0022] A regional temperature heat map is obtained, which is then divided into several local blocks. The pixel grayscale information entropy of each block is calculated, and a segmentation threshold is dynamically set based on the calculated local entropy value.

[0023] Based on the multispectral geological classification map, regional features of non-oyster reef bodies are extracted to generate a mask. Combined with the segmentation threshold, preliminary segmentation of oyster reef bodies is performed to obtain a preliminary oyster reef body distribution map.

[0024] A morphological watershed algorithm is introduced to extract temperature gradient features from the temperature heatmap of the region. The initially segmented reef areas are used as marker points for edge optimization to obtain the oyster reef distribution map.

[0025] In this solution, the construction of a dense 3D point cloud of the intertidal zone of the target area based on the multi-source aerial photography dataset, combined with the regional temperature heat map, to perform a thermal-optical data fusion operation and generate a fused point cloud dataset specifically includes:

[0026] A multi-source aerial photography dataset is acquired, visible light image data is extracted from the multi-source aerial photography dataset, the camera pose is calculated using the structure of motion recovery algorithm to generate an initial sparse point cloud, and the point cloud is registered using a multi-view stereo matching algorithm to obtain a dense 3D point cloud.

[0027] A regional temperature heat map is acquired, and the regional temperature heat map is spatially reprojected. The two-dimensional image coordinates of each temperature pixel are converted into three-dimensional spatial rays through back projection calculation using the extrinsic parameter transformation matrix of the monitoring camera pre-calibrated by the aerial drone.

[0028] For each 3D spatial ray, a spherical search domain with a preset spacing is established to search for a set of neighboring points in a dense 3D point cloud, and the spatial distance between the 3D spatial ray and the neighboring points in the dense 3D point cloud is calculated.

[0029] Select the point with the closest spatial distance as the target point, and crop a heat map pixel window with the target point as the center based on a preset window size. Calculate the planar distance between the target point and the center of each pixel in the cropped heat map pixel window.

[0030] The weighting factor is calculated based on the inverse square ratio of the planar distance. The temperature of each pixel within the window is then weighted and fused. The weighted fusion result is used to assign a temperature attribute to the point cloud data corresponding to the target point, and finally, a fused point cloud dataset is generated.

[0031] In this scheme, the step of using the oyster reef distribution map as a spatial mask to crop the fused point cloud dataset and construct a three-dimensional model of the intertidal reef is carried out. The three-dimensional model of the intertidal zone is then used for crack detection and oyster reef feature measurement to generate a multimodal reef feature vector. Specifically, this includes:

[0032] Obtain an oyster reef distribution map and a fused point cloud dataset. Based on the oyster reef distribution map, extract spatial coordinate features of the oyster reef areas to locate them. Generate an initial spatial mask based on the extracted coordinate features.

[0033] A topological buffer is constructed using the initial spatial mask. The boundary lines of the initial spatial mask are extracted and Delaunay triangulation is performed to generate a constrained triangular mesh. Buffer point columns are generated by interpolating the edges of the constrained triangular mesh according to the average spacing of the point cloud, forming a closed buffer zone.

[0034] The closed buffer band is fused with the initial spatial mask to generate a final spatial mask. The final spatial mask is then mapped onto the fused point cloud dataset for point cloud cropping, resulting in a cropped point cloud dataset.

[0035] The cropped point cloud dataset is input into the Poisson reconstruction engine to generate a 3D model of the intertidal reef. Then, all triangular faces connected to the vertices of the 3D model of the intertidal reef are extracted. The square of the area of ​​each face is used as a weighting factor to calculate the weighted average of the unit normal vectors of each face to obtain the mean vector, which is used as the new normal vector of the corresponding vertex to generate a normal vector optimized 3D model.

[0036] In the normal vector optimized 3D model, the shared edges of each triangular facet of the model are scanned one by one, and the normal angle between adjacent facets is calculated. When it is less than a preset threshold, the corresponding shared edge is marked as a potential crack edge segment.

[0037] The temperature gradient vector of each vertex is calculated using the central difference method to generate a gradient magnitude heatmap. The gradient magnitude heatmap is then used to apply temperature constraints to the edge segments of the potential fractures to assess the authenticity of the fractures, and finally, a fracture network contour map is generated.

[0038] Measurement points are randomly set up in the three-dimensional model of the intertidal reef. Multi-level spheres with a preset radius are generated with the measurement points as the center. Morphological analysis algorithms are used to measure the morphological characteristics of the oyster reef and generate reef morphological characteristic information.

[0039] Thermodynamic parameters of the reef are obtained by using a three-dimensional model of the intertidal reef, and the fracture features of the reef are obtained by using the fracture network contour map. A multimodal reef feature vector is constructed by combining the morphological features of the reef.

[0040] In this solution, the analysis of the intertidal oyster reef health in the target area based on the multimodal reef feature vector, combined with regional ecological data to determine the priority of regional restoration, and the generation of early warning information for notification, specifically includes:

[0041] A reef health analysis model is built using a gradient boosting decision tree framework. Historical data is retrieved to obtain historical reef health assessment instances with different health conditions and the reef features corresponding to each instance are extracted. A training set is set to pre-train the reef health analysis model.

[0042] Obtain multimodal reef feature vectors, input the multimodal reef feature vectors into a pre-trained reef health analysis model, analyze the health status of intertidal oyster reefs in the target area through a weighted voting mechanism, and output the reef health score of each oyster reef.

[0043] Oyster reefs with health scores below a preset threshold are screened out, and their actual locations are marked as potential danger points. A ranking table is then generated based on the health scores of each potential danger point.

[0044] Acquire regional ecological and environmental data, extract regional hydrological parameters of potential hazardous points from the regional ecological and environmental data, and perform weighted calculations on each potential hazardous point in combination with the sorting table to generate a remediation priority index for each potential hazardous point.

[0045] The repair priority index of each potential danger point is determined by comparing it with the preset level classification threshold. Based on the level classification results, a regional repair suggestion heat map is generated through visualization processing. Early warning prompts are then issued through the regional repair suggestion heat map.

[0046] A second aspect of the present invention provides a system for analyzing intertidal oyster reef structures based on fusion of thermal imaging aerial photography. The system includes a memory and a processor. The memory contains a program for analyzing intertidal oyster reef structures based on fusion of thermal imaging aerial photography. When the processor executes the program for analyzing intertidal oyster reef structures based on fusion of thermal imaging aerial photography, it performs the following steps:

[0047] A multi-source aerial photography dataset is acquired by patrolling the target area using aerial photography drones. Thermal infrared image data of the intertidal zone is extracted to generate a radiation intensity distribution map. A regional temperature heat map is then constructed using the radiation intensity distribution map.

[0048] Multispectral data was extracted from the multi-source aerial photography dataset to generate a multispectral substrate classification map. The intertidal oyster reef body in the target area was segmented by combining the regional temperature heat map and the reef body edge was optimized to obtain the oyster reef body distribution map.

[0049] Based on the multi-source aerial photography dataset, a dense three-dimensional point cloud of the intertidal zone of the target area is constructed. Combined with the regional temperature heat map, a thermal-optical data fusion operation is performed to generate a fused point cloud dataset.

[0050] The oyster reef distribution map is used as a spatial mask to crop the fused point cloud dataset and construct a three-dimensional model of the intertidal reef. Crack detection and oyster reef feature measurement are performed using the three-dimensional model of the intertidal zone to generate a multimodal reef feature vector.

[0051] Based on the analysis of the multimodal reef feature vector, the health of the intertidal oyster reef in the target area is determined, and regional ecological data is combined to determine the priority of regional restoration and generate early warning information for notification.

[0052] A third aspect of the present invention provides a computer-readable storage medium comprising a method program for analyzing intertidal oyster reefs using fused thermal imaging aerial photography. When executed by a processor, the method program for analyzing intertidal oyster reefs using fused thermal imaging aerial photography implements the steps of the method for analyzing intertidal oyster reefs using fused thermal imaging aerial photography as described in any of the preceding claims. Attached Figure Description

[0053] To more clearly illustrate the technical solutions in the embodiments or examples of the present invention, the drawings used in the embodiments or examples will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained according to these drawings without creative effort.

[0054] Figure 1 A flowchart of a method for analyzing intertidal oyster reef structures using fusion thermal imaging aerial photography, provided in an embodiment of the present invention;

[0055] Figure 2 A flowchart of an auxiliary method for intertidal oyster reef restoration based on fusion thermal imaging aerial photography, provided in an embodiment of the present invention;

[0056] Figure 3A block diagram of an intertidal oyster reef analysis system based on fusion thermal imaging aerial photography, provided in an embodiment of the present invention;

[0057] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0058] To better understand the above-mentioned objectives, features, and advantages of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in these embodiments can be combined with each other.

[0059] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and therefore the scope of protection of the invention is not limited to the specific embodiments disclosed below.

[0060] Figure 1 A flowchart of a method for analyzing intertidal oyster reef structures using fusion thermal imaging aerial photography, provided in an embodiment of the present invention;

[0061] like Figure 1 As shown, this invention provides a flowchart of a method for analyzing intertidal oyster reef structures using fusion thermal imaging aerial photography, including:

[0062] S102, A multi-source aerial photography dataset is acquired by a drone patrolling the target area, thermal infrared image data of the intertidal zone is extracted to generate a radiation intensity distribution map, and a regional temperature heat map is constructed using the radiation intensity distribution map.

[0063] S104. Multispectral data is extracted from the multi-source aerial photography dataset to generate a multispectral substrate classification map. The intertidal oyster reef body in the target area is segmented by combining the regional temperature heat map and the reef body edge is optimized to obtain the oyster reef body distribution map.

[0064] S106, Construct a dense three-dimensional point cloud of the intertidal zone of the target area based on the multi-source aerial photography dataset, and perform a thermal-optical data fusion operation in combination with the regional temperature heat map to generate a fused point cloud dataset;

[0065] S108, the oyster reef distribution map is used as a spatial mask to crop the fused point cloud dataset and construct a three-dimensional model of the intertidal reef. Crack detection and oyster reef feature measurement are performed through the three-dimensional model of the intertidal zone to generate a multimodal reef feature vector.

[0066] S110, based on the multimodal reef feature vector analysis, analyze the health of the intertidal oyster reef in the target area, combine regional ecological data to determine the priority of regional restoration, and generate early warning information for notification.

[0067] Furthermore, in a preferred embodiment of the present invention, the step of acquiring a multi-source aerial photography dataset by patrolling the target area using an aerial photography drone, extracting thermal infrared image data of the intertidal zone to generate a radiation intensity distribution map, and constructing a regional temperature heat map using the radiation intensity distribution map specifically includes:

[0068] A drone equipped with multi-source monitoring equipment is used to cruise the target area. The multi-source monitoring equipment collects thermal infrared image data, visible light image data and multispectral data of the intertidal zone of the target area to generate a multi-source aerial photography dataset.

[0069] Based on the multi-source aerial photography dataset, thermal infrared image data of the intertidal zone is extracted. The thermal infrared image data is then radiometrically calibrated and converted into absolute radiation intensity values ​​based on the temperature-radiation value calibration curve of the sensor's built-in blackbody reference source, generating a radiation intensity distribution map.

[0070] Acquire atmospheric data of the target area during the cruise, input it into the MODTRAN radiation model to calculate atmospheric transmittance and uplink radiation path brightness, retrieve the true surface radiation intensity value through the radiative transfer equation, and perform atmospheric path correction on the radiation intensity distribution map to obtain the corrected radiation intensity distribution map.

[0071] Local binary mode texture calculation is performed on the corrected radiation intensity distribution map, the gray-level variation coefficient of each pixel neighborhood is extracted as a uniformity index, and the texture features are divided by K-means clustering algorithm to generate a divided radiation intensity distribution map.

[0072] By applying the Stefan-Boltzmann law to the radiation intensity distribution map, temperature conversion is performed, and the final output is a regional temperature thermogram.

[0073] It should be noted that in the monitoring operation of intertidal oyster reefs, synchronized data acquisition was performed using a multi-source sensor system carried by an unmanned aerial vehicle (UAV). This system integrates a thermal infrared camera, a high-resolution visible light camera, and a multispectral imager, performing a gridded cruise scan along a preset flight path to capture real-time three-dimensional spatial information and physical field parameters of the target area. During flight, the thermal infrared sensor records the surface thermal radiation distribution, the visible light device acquires texture details, and the multispectral device collects reflectance data in specific bands. These three sensors are fused into a spatiotemporally unified multi-source aerial dataset under a time synchronization and spatial registration mechanism. Subsequently, thermal infrared images are extracted from this dataset, and radiometric calibration transformation is performed: based on the calibration curve of the sensor's built-in blackbody reference source, the original digital values ​​(DN) are converted into absolute radiative intensity values ​​using a piecewise linear interpolation algorithm, generating a radiative intensity distribution map that eliminates the detector's nonlinear response. Simultaneously acquired atmospheric profile data is input into the MODTRAN radiative transfer model to calculate atmospheric transmittance and upward radiative path brightness. The radiative transfer equation is then applied to invert the true surface radiative intensity, and atmospheric path correction is performed on the radiative intensity distribution map, outputting a corrected radiative intensity distribution map that eliminates environmental interference. For the corrected radiance data, a texture feature analysis process is initiated: the Local Binary Pattern (LBP) algorithm is used to calculate the gray-level variation coefficient within the 8-neighborhood of each pixel, quantifying the uniformity index of the spatial distribution of radiance. Using the K-means clustering algorithm with k=3 clusters, the texture features are initially divided into three regions, generating a radiance distribution map. Finally, based on this map, corresponding emissivity values ​​are assigned, and the Stefan-Boltzmann law is applied to convert the radiance intensity into physical temperature, outputting a regional temperature heatmap, providing an interference-free thermal field input for accurate reef segmentation.

[0074] Furthermore, in a preferred embodiment of the present invention, the step of extracting multispectral images from the multi-source aerial photography dataset to generate a multispectral substrate classification map, segmenting the intertidal oyster reef body in the target area using a regional temperature heatmap, and optimizing the reef body edges to obtain an oyster reef body distribution map specifically includes:

[0075] The multispectral features of oyster reefs are obtained by searching through big data networks to generate a reef spectral feature set, and the reef spectral feature set is mapped to a high-dimensional space using a Gaussian kernel function.

[0076] The minimum hypersphere boundary and hypersphere radius containing the multispectral feature samples of all oyster reef bodies are obtained by solving in the mapped high-dimensional space. A multispectral geological classification model is then established using the minimum hypersphere boundary and hypersphere radius.

[0077] A multi-source aerial photography dataset is acquired, multispectral data is extracted from the multi-source aerial photography dataset and input into the multispectral geological classification model, and the Mahalanobis distance of the input multispectral data in the kernel space is calculated.

[0078] The calculated Mahalanobis distance is compared with the preset hypersphere radius. If it is less than the preset hypersphere radius, the corresponding multispectral data source location is marked as the oyster reef body area, and finally a multispectral geological classification map is generated.

[0079] A regional temperature heat map is obtained, which is then divided into several local blocks. The pixel grayscale information entropy of each block is calculated, and a segmentation threshold is dynamically set based on the calculated local entropy value.

[0080] Based on the multispectral geological classification map, regional features of non-oyster reef bodies are extracted to generate a mask. Combined with the segmentation threshold, preliminary segmentation of oyster reef bodies is performed to obtain a preliminary oyster reef body distribution map.

[0081] A morphological watershed algorithm is introduced to extract temperature gradient features from the temperature heatmap of the region. The initially segmented reef areas are used as marker points for edge optimization to obtain the oyster reef distribution map.

[0082] It should be noted that, firstly, historical and measured multispectral reflectance data of oyster reefs are retrieved through big data networks to construct a reef spectral feature set. This feature set is then mapped to a high-dimensional space using a Gaussian kernel function. By solving the Support Vector Data Description (SVDD) optimization problem, the minimum hypersphere boundary and its radius, encompassing all samples, are obtained, thus establishing the core discrimination criterion for the multispectral geological classification model. In the practical application phase, multi-source datasets from UAV aerial photography are acquired, and the multispectral reflectance sequence of each pixel is extracted and input into a pre-trained classification model. The model calculates the Mahalanobis distance (representing the deviation of the current pixel's spectral features from the typical spectral distribution of oyster reefs) in the high-dimensional space and compares it in real-time with the hypersphere radius: if the Mahalanobis distance is less than the radius, the pixel is determined to belong to the oyster reef region, ultimately generating a multispectral geological classification map pixel by pixel. Subsequently, thermal infrared data is introduced to assist in optimization: based on a regional temperature heatmap, it is divided into regular grid blocks, and the information entropy of the pixel grayscale values ​​within each block is calculated (the entropy value reflects the complexity of the temperature distribution). The local adaptive segmentation threshold is dynamically set based on entropy values. Higher thresholds are used in high-entropy areas to capture details of broken reefs, while lower thresholds are used in low-entropy areas to ensure the complete detection of the main reef body. Simultaneously, a multispectral geological classification map is loaded, and regions classified as "non-reef" (such as vegetation and sediment) are extracted to generate exclusion masks. The adaptive threshold segmentation results are overlaid with the masks, and interference areas covered by the masks are removed to obtain a preliminary binary map of oyster reef body distribution. Finally, to address edge breakage issues in the preliminary segmentation, a morphological watershed algorithm is used for optimization: the temperature gradient amplitude of the regional temperature heatmap is used as the watershed topographic surface (gradient values ​​characterize boundary strength), and the initially segmented connected reef areas are used as marker points. By simulating hydrological inundation processes, the segmentation boundaries naturally extend along the temperature gradient ridge, effectively bridging the small gaps between reef fragments and smoothing irregular contours, ultimately outputting a high-precision oyster reef body distribution map.

[0083] Furthermore, in a preferred embodiment of the present invention, the step of constructing a dense three-dimensional point cloud of the intertidal zone of the target area based on the multi-source aerial photography dataset, and performing a thermal-optical data fusion operation in conjunction with the regional temperature heat map to generate a fused point cloud dataset specifically includes:

[0084] A multi-source aerial photography dataset is acquired, visible light image data is extracted from the multi-source aerial photography dataset, the camera pose is calculated using the structure of motion recovery algorithm to generate an initial sparse point cloud, and the point cloud is registered using a multi-view stereo matching algorithm to obtain a dense 3D point cloud.

[0085] A regional temperature heat map is acquired, and the regional temperature heat map is spatially reprojected. The two-dimensional image coordinates of each temperature pixel are converted into three-dimensional spatial rays through back projection calculation using the extrinsic parameter transformation matrix of the monitoring camera pre-calibrated by the aerial drone.

[0086] For each 3D spatial ray, a spherical search domain with a preset spacing is established to search for a set of neighboring points in a dense 3D point cloud, and the spatial distance between the 3D spatial ray and the neighboring points in the dense 3D point cloud is calculated.

[0087] Select the point with the closest spatial distance as the target point, and crop a heat map pixel window with the target point as the center based on a preset window size. Calculate the planar distance between the target point and the center of each pixel in the cropped heat map pixel window.

[0088] The weighting factor is calculated based on the inverse square ratio of the planar distance. The temperature of each pixel within the window is then weighted and fused. The weighted fusion result is used to assign a temperature attribute to the point cloud data corresponding to the target point, and finally, a fused point cloud dataset is generated.

[0089] It should be noted that, firstly, visible light image sequences and regional temperature heatmaps synchronously acquired by the UAV are obtained. The Structure for Motion Restoration (SfM) algorithm is used to calculate the precise camera pose of each frame of the visible light image, and an initial sparse point cloud framework is constructed through feature point matching. Then, the Multi-View Stereo Matching (MVS) algorithm is used to optimize the depth map and densify the point cloud, generating a dense 3D point cloud. At this point, the point cloud only contains geometric coordinates and RGB color attributes. Simultaneously, the regional temperature heatmap is loaded and a spatial reprojection operation is performed. Based on the pre-calibrated extrinsic transformation matrix (including rotation and spatial offset) between the thermal infrared camera and the visible light camera on the aerial drone platform, coordinate transformation is performed on each temperature pixel in the heatmap. Specifically, the 2D image coordinates of the pixels are back-projected into 3D space through the camera perspective model, forming a directional ray emitted from the optical center of the thermal imager. This ray is then transformed to a coordinate system unified with the visible light point cloud using the extrinsic matrix. For each transformed 3D spatial ray, a spherical search domain is set at preset intervals along its propagation path. Within each spherical domain, neighboring point sets are rapidly retrieved in the dense 3D point cloud based on the KD-Tree spatial index, and the Euclidean spatial distance from the current ray point to each neighboring point is calculated. The nearest visible light point cloud point is selected as the target mapping point, and a pre-defined window of the local temperature field is extracted from the original heatmap centered on this point. Within the extracted window, the 2D planar distance between the projected position of the target point and the center of each pixel within the window is calculated. The weighting factor of each pixel is calculated based on the reciprocal of the square of the distance (the closer the distance, the greater the weight), and the temperature values ​​of all pixels within the window are weighted and fused. Finally, the fused temperature value is assigned to the target point cloud point, thereby generating a 3D point cloud dataset with fused spatial coordinates, color, and temperature, laying the data foundation for subsequent 3D reef reconstruction and thermodynamic analysis.

[0090] Furthermore, in a preferred embodiment of the present invention, the step of using the oyster reef distribution map as a spatial mask to crop the fused point cloud dataset and construct a three-dimensional model of the intertidal reef, and then using the three-dimensional model of the intertidal zone to perform crack detection and oyster reef feature measurement to generate a multimodal reef feature vector, specifically includes:

[0091] Obtain an oyster reef distribution map and a fused point cloud dataset. Based on the oyster reef distribution map, extract spatial coordinate features of the oyster reef areas to locate them. Generate an initial spatial mask based on the extracted coordinate features.

[0092] A topological buffer is constructed using the initial spatial mask. The boundary lines of the initial spatial mask are extracted and Delaunay triangulation is performed to generate a constrained triangular mesh. Buffer point columns are generated by interpolating the edges of the constrained triangular mesh according to the average spacing of the point cloud, forming a closed buffer zone.

[0093] The closed buffer band is fused with the initial spatial mask to generate a final spatial mask. The final spatial mask is then mapped onto the fused point cloud dataset for point cloud cropping, resulting in a cropped point cloud dataset.

[0094] The cropped point cloud dataset is input into the Poisson reconstruction engine to generate a 3D model of the intertidal reef. Then, all triangular faces connected to the vertices of the 3D model of the intertidal reef are extracted. The square of the area of ​​each face is used as a weighting factor to calculate the weighted average of the unit normal vectors of each face to obtain the mean vector, which is used as the new normal vector of the corresponding vertex to generate a normal vector optimized 3D model.

[0095] In the normal vector optimized 3D model, the shared edges of each triangular facet of the model are scanned one by one, and the normal angle between adjacent facets is calculated. When it is less than a preset threshold, the corresponding shared edge is marked as a potential crack edge segment.

[0096] The temperature gradient vector of each vertex is calculated using the central difference method to generate a gradient magnitude heatmap. The gradient magnitude heatmap is then used to apply temperature constraints to the edge segments of the potential fractures to assess the authenticity of the fractures, and finally, a fracture network contour map is generated.

[0097] Measurement points are randomly set up in the three-dimensional model of the intertidal reef. Multi-level spheres with a preset radius are generated with the measurement points as the center. Morphological analysis algorithms are used to measure the morphological characteristics of the oyster reef and generate reef morphological characteristic information.

[0098] Thermodynamic parameters of the reef are obtained by using a three-dimensional model of the intertidal reef, and the fracture features of the reef are obtained by using the fracture network contour map. A multimodal reef feature vector is constructed by combining the morphological features of the reef.

[0099] It should be noted that, firstly, the boundary coordinates of the reef region are extracted based on the oyster reef distribution map to generate an initial binary spatial mask. To eliminate the zigzag effect of the boundary and accommodate point cloud registration errors, the mask undergoes topology optimization: the mask outline is extracted and Delaunay triangulation is performed to form a constrained triangular mesh. A dense boundary point array is generated by interpolating along the edge of the triangular mesh according to the average density of the point cloud. A closed buffer zone is constructed by connecting the inner and outer boundaries. The buffer zone is then fused with the initial mask to form the final spatial mask, which preserves the main reef morphology while accommodating uncertain regions at the edges. This mask is mapped to a fused point cloud dataset (including temperature attributes), and the target point cloud of the reef is cropped to remove external interference data. The cropped point cloud is then input into a Poisson surface reconstruction engine: an implicit field function is constructed through octree spatial discretization, and the Poisson equation is solved to generate an initial triangular mesh model. To address potential noise in the normal vectors of the model surface, each vertex is traversed, aggregating all connected triangular faces. A weighted average of the unit normal vectors is calculated using the square of each face's area as the weight, and the resulting mean vector is used as the vertex's corrected normal, outputting a reef 3D model with optimized normal vectors. Subsequently, based on the optimized model, crack detection and morphological analysis are performed simultaneously: all shared edges of the triangular faces are scanned, the angle between the normals of adjacent faces is calculated, and edges with values ​​less than a preset threshold are marked as potential crack segments; point cloud temperature attributes are extracted to construct a vertex-level gradient magnitude heatmap, and temperature verification is applied to the geometrically marked crack segments, retaining only segments with gradient values ​​exceeding a preset gradient threshold, ultimately generating a crack network contour map. Then, measurement points are randomly placed on the model surface, generating multi-level analysis spheres within a preset radius range. The proportion of mesh area contained within each sphere is statistically analyzed, and the surface fractal dimension is calculated using scaling laws; simultaneously, basic morphological parameters such as model volume and convex hull volume are measured. Finally, three types of features are integrated: morphological parameters (fractal dimension, convex hull defect rate) extracted from the geometric model, thermodynamic parameters (surface temperature variation coefficient, interquartile range) derived from the temperature field, and structural parameters (fracture area ratio, average width) extracted from the fracture network, to jointly construct a multimodal reef feature vector.

[0100] Figure 2 A flowchart of an auxiliary method for intertidal oyster reef restoration based on fusion thermal imaging aerial photography, provided in an embodiment of the present invention;

[0101] like Figure 2 As shown, this invention provides a flowchart of an auxiliary method for intertidal oyster reef restoration based on thermal imaging aerial photography, comprising:

[0102] S202, a reef health analysis model is built using a gradient boosting decision tree framework. Historical data is retrieved to obtain historical reef health assessment instances with different health conditions and the reef features corresponding to each instance are extracted. A training set is set to pre-train the reef health analysis model.

[0103] S204, obtain multimodal reef feature vectors, input the multimodal reef feature vectors into a pre-trained reef health analysis model, analyze the health status of intertidal oyster reefs in the target area through a weighted voting mechanism, and output the reef health score of each oyster reef.

[0104] S206, oyster reefs with health scores below a preset health score threshold are screened out, the actual locations of the corresponding oyster reefs are marked as potential danger points, and a ranking table is generated based on the health scores of each potential danger point.

[0105] S208, Obtain regional ecological environment data, extract regional hydrological parameters of potential hazardous points through the regional ecological environment data, and perform weighted calculation on each potential hazardous point in combination with the sorting table to generate a remediation priority index for each potential hazardous point;

[0106] S210, the repair priority index of each potential danger point is judged against the preset level classification threshold, and a regional repair suggestion heat map is generated based on the level classification result through visualization processing. The regional repair suggestion heat map is used to provide early warning prompts.

[0107] It should be noted that a reef health analysis model is constructed based on a gradient boosting decision tree framework. Historical reef samples in different health states are obtained through historical data retrieval, and their multimodal feature vectors (including morphology, fissures, and thermodynamic parameters) are extracted as the training set. Pre-training is completed through hyperparameter tuning. Subsequently, the multimodal reef feature vectors are input into the pre-trained reef health analysis model. A weighted voting mechanism integrating 150 decision trees (each tree's weight is dynamically allocated based on its validation accuracy) is used for comprehensive analysis, outputting a quantitative reef health score that objectively reflects the ecological vitality status of each reef. Oyster reef sample instances in different health states are obtained by retrieving data from a historical ecological monitoring database, and multidimensional feature vectors such as structural stability index, thermal uniformity parameter, and fissure density of each sample are extracted as the training set. A forward stepwise algorithm is used to iteratively generate decision trees, the loss function is optimized through the negative gradient direction, and a learning rate decay strategy is introduced to prevent overfitting. This completes the model's pre-training and cross-validation, outputting the final model. Subsequently, the multimodal reef feature vectors collected on-site were input into the model, and a weighted voting mechanism was used to synthesize the prediction results of each decision tree: voting weights were dynamically allocated based on the F1 score of each tree on the validation set, and the final output was a quantitative health score of the reef. Reefs with health scores below a preset threshold were marked as potential danger points, and a priority ranking table was generated based on the score values, with lower scores ranking higher. Simultaneously, the regional marine environmental database was accessed to extract the regional hydrological parameters (including salinity vertical gradient, current velocity vector magnitude, diurnal temperature variation, etc.) corresponding to the danger points, and a multi-factor weighted calculation was performed based on the ranking table's sequence information: a 40% weight for health score, a 25% weight for salinity fluctuation coefficient, a 20% weight for extreme current values, and a 15% weight for bottom sedimentation risk. A restoration priority index was generated using a linear weighted formula. Finally, an early warning report was generated based on the priority index to guide relevant personnel in understanding and managing the regional situation.

[0108] Furthermore, the method for analyzing intertidal oyster reef structures using fusion thermal imaging aerial photography provided by this invention also includes the following steps:

[0109] Oyster reefs with health scores exceeding a preset health score threshold are screened, potential breeding sites are located, and regional hydrological parameters and bottom sediment type distribution data of potential breeding sites are extracted from the regional marine environment database. These data are then correlated with the multimodal reef feature vectors of potential breeding sites to generate potential breeding site feature information.

[0110] Based on the expected proliferating larvae type, the suitable conditions for proliferating larvae are obtained, and cosine similarity is calculated with the feature information of the potential proliferating sites. The calculated cosine similarity value is used as the proliferating fitness degree to generate a regional proliferating fitness site distribution map.

[0111] The distribution map of the regional proliferation and adaptation points is divided into grids and initial qualified units are set. A dynamic programming algorithm is introduced to calculate the adaptation continuity factor with adjacent units grid by grid starting from the initial qualified units, and finally a regional proliferation and adaptation degree map is generated.

[0112] Based on the regional proliferation adaptation map, a juvenile proliferation and release assistance report is generated to provide recommendations for proliferation areas.

[0113] It should be noted that the condition of regional oyster reefs can reflect the corresponding region's eco-hydrological conditions to a certain extent. Healthier oyster reef areas correspond to areas with better ecosystems, allowing for the utilization of these waters, such as for stock enhancement or designation as aquaculture zones. High-vitality reefs exceeding a preset threshold are selected based on health scores and designated as potential stock enhancement sites. Simultaneously, the regional marine environmental database is accessed to extract regional hydrological parameters and bottom sediment type distribution data for these sites. These data are then spatially correlated with the feature vectors of the potential sites, forming a database of stock enhancement potential characteristics that integrates the reef's structural features with surrounding environmental factors. Next, for the target stock enhancement larvae type, their suitability condition vectors (including optimal temperature range, preferred bottom hardness, and water flow tolerance threshold) are obtained. This vector is then compared with the feature information of each potential site using multidimensional cosine similarity calculations. The resulting similarity value (0-1 interval) is defined as the stock enhancement fit, generating a spatially discrete fit distribution map. To optimize the contiguousness of the stock enhancement area, the fit distribution map is divided into several grid cells, and the cell with the highest fit is selected as the initial qualified point. A dynamic programming algorithm is introduced: starting from the initial unit, the fitness continuity factor (weighted by the fitness difference between units, boundary conformity, and hydrological connectivity) is calculated grid by grid with adjacent units. The comprehensive fitness evaluation value of each unit is iteratively updated, ultimately generating a spatially continuous regional propagation fitness heatmap. Finally, based on the fitness heatmap, highly continuous core areas are identified, and a juvenile propagation and release auxiliary report is generated, marking priority release areas, secondary expansion areas, and ecologically sensitive zones to be avoided, thereby providing actionable decision support for ecological engineering.

[0114] Figure 3 An embodiment of the present invention provides a system 3 for analyzing intertidal oyster reefs using fused thermal imaging aerial photography. The system includes a memory 31 and a processor 32. The memory 31 contains a program for analyzing intertidal oyster reefs using fused thermal imaging aerial photography. When the processor 32 executes the program for analyzing intertidal oyster reefs using fused thermal imaging aerial photography, it performs the following steps:

[0115] A multi-source aerial photography dataset is acquired by patrolling the target area using aerial photography drones. Thermal infrared image data of the intertidal zone is extracted to generate a radiation intensity distribution map. A regional temperature heat map is then constructed using the radiation intensity distribution map.

[0116] Multispectral data was extracted from the multi-source aerial photography dataset to generate a multispectral substrate classification map. The intertidal oyster reef body in the target area was segmented by combining the regional temperature heat map and the reef body edge was optimized to obtain the oyster reef body distribution map.

[0117] Based on the multi-source aerial photography dataset, a dense three-dimensional point cloud of the intertidal zone of the target area is constructed. Combined with the regional temperature heat map, a thermal-optical data fusion operation is performed to generate a fused point cloud dataset.

[0118] The oyster reef distribution map is used as a spatial mask to crop the fused point cloud dataset and construct a three-dimensional model of the intertidal reef. Crack detection and oyster reef feature measurement are performed using the three-dimensional model of the intertidal zone to generate a multimodal reef feature vector.

[0119] Based on the analysis of the multimodal reef feature vector, the health of the intertidal oyster reef in the target area is determined, and regional ecological data is combined to determine the priority of regional restoration and generate early warning information for notification.

[0120] In another aspect, the present invention provides a computer-readable storage medium comprising a method program for analyzing intertidal oyster reefs using fused thermal imaging aerial photography. When executed by a processor, the method program for analyzing intertidal oyster reefs using fused thermal imaging aerial photography implements the steps of the method for analyzing intertidal oyster reefs using fused thermal imaging aerial photography as described in any of the preceding claims.

[0121] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods, such as: multiple units or components can be combined, or integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the various components shown or discussed can be through some interfaces, and the indirect coupling or communication connection between devices or units can be electrical, mechanical, or other forms.

[0122] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units. They may be located in one place or distributed across multiple network units. Some or all of the units may be selected to achieve the purpose of this embodiment according to actual needs.

[0123] In addition, in the various embodiments of the present invention, each functional unit can be integrated into one processing unit, or each unit can be a separate unit, or two or more units can be integrated into one unit; the integrated unit can be implemented in hardware or in the form of hardware plus software functional units.

[0124] Those skilled in the art will understand that all or part of the steps of the above method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it performs the steps of the above method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0125] Alternatively, if the integrated units of this invention are implemented as software functional modules and sold or used as independent products, they can also be stored in a computer-readable storage medium. Based on this understanding, the technical solutions of the embodiments of this invention, or the parts that contribute to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, ROM, RAM, magnetic disks, or optical disks.

[0126] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for analyzing intertidal oyster reef structures using thermal imaging aerial photography, characterized in that, include: A multi-source aerial photography dataset is acquired by patrolling the target area using aerial photography drones. Thermal infrared image data of the intertidal zone is extracted to generate a radiation intensity distribution map. A regional temperature heat map is then constructed using the radiation intensity distribution map. Multispectral data was extracted from the multi-source aerial photography dataset to generate a multispectral substrate classification map. The intertidal oyster reef body in the target area was segmented by combining the regional temperature heat map and the reef body edge was optimized to obtain the oyster reef body distribution map. Based on the multi-source aerial photography dataset, a dense three-dimensional point cloud of the intertidal zone of the target area is constructed. Combined with the regional temperature heat map, a thermal-optical data fusion operation is performed to generate a fused point cloud dataset. The oyster reef distribution map is used as a spatial mask to crop the fused point cloud dataset and construct a three-dimensional model of the intertidal reef. Crack detection and oyster reef feature measurement are performed using the three-dimensional model of the intertidal zone to generate a multimodal reef feature vector. Based on the analysis of the multimodal reef feature vector, the health of the intertidal oyster reef in the target area is determined, and regional ecological data is combined to determine the priority of regional restoration and generate early warning information for notification.

2. The method for analyzing intertidal oyster reef structures using fusion thermal imaging aerial photography as described in claim 1, characterized in that, The process of acquiring multi-source aerial photography datasets by patrolling the target area using a drone, extracting thermal infrared image data of the intertidal zone to generate a radiation intensity distribution map, and constructing a regional temperature heat map using the radiation intensity distribution map specifically includes: A drone equipped with multi-source monitoring equipment is used to cruise the target area. The multi-source monitoring equipment collects thermal infrared image data, visible light image data and multispectral data of the intertidal zone of the target area to generate a multi-source aerial photography dataset. Based on the multi-source aerial photography dataset, thermal infrared image data of the intertidal zone is extracted. The thermal infrared image data is then radiometrically calibrated and converted into absolute radiation intensity values ​​based on the temperature-radiation value calibration curve of the sensor's built-in blackbody reference source, generating a radiation intensity distribution map. Acquire atmospheric data of the target area during the cruise, input it into the MODTRAN radiation model to calculate atmospheric transmittance and uplink radiation path brightness, retrieve the true surface radiation intensity value through the radiative transfer equation, and perform atmospheric path correction on the radiation intensity distribution map to obtain the corrected radiation intensity distribution map. Local binary mode texture calculation is performed on the corrected radiation intensity distribution map, the gray-level variation coefficient of each pixel neighborhood is extracted as a uniformity index, and the texture features are divided by K-means clustering algorithm to generate a divided radiation intensity distribution map. By applying the Stefan-Boltzmann law to the radiation intensity distribution map, temperature conversion is performed, and the final output is a regional temperature thermogram.

3. The method for analyzing intertidal oyster reef structures using fusion thermal imaging aerial photography as described in claim 1, characterized in that, The process involves extracting multispectral images from the multi-source aerial photography dataset to generate a multispectral substrate classification map, segmenting the intertidal oyster reef bodies in the target area using a regional temperature heatmap, and optimizing the reef edges to obtain an oyster reef distribution map. Specifically, this includes: The multispectral features of oyster reefs are obtained by searching through big data networks to generate a reef spectral feature set, and the reef spectral feature set is mapped to a high-dimensional space using a Gaussian kernel function. The minimum hypersphere boundary and hypersphere radius containing the multispectral feature samples of all oyster reef bodies are obtained by solving in the mapped high-dimensional space. A multispectral geological classification model is then established using the minimum hypersphere boundary and hypersphere radius. A multi-source aerial photography dataset is acquired, multispectral data is extracted from the multi-source aerial photography dataset and input into the multispectral geological classification model, and the Mahalanobis distance of the input multispectral data in the kernel space is calculated. The calculated Mahalanobis distance is compared with the preset hypersphere radius. If it is less than the preset hypersphere radius, the corresponding multispectral data source location is marked as the oyster reef body area, and finally a multispectral geological classification map is generated. A regional temperature heat map is obtained, which is then divided into several local blocks. The pixel grayscale information entropy of each block is calculated, and a segmentation threshold is dynamically set based on the calculated local entropy value. Based on the multispectral geological classification map, regional features of non-oyster reef bodies are extracted to generate a mask. Combined with the segmentation threshold, preliminary segmentation of oyster reef bodies is performed to obtain a preliminary oyster reef body distribution map. A morphological watershed algorithm is introduced to extract temperature gradient features from the temperature heatmap of the region. The initially segmented reef areas are used as marker points for edge optimization to obtain the oyster reef distribution map.

4. The method for analyzing intertidal oyster reef structures using fusion thermal imaging aerial photography as described in claim 1, characterized in that, The process of constructing a dense 3D point cloud of the intertidal zone of the target area based on the multi-source aerial photography dataset, and performing a thermal-optical data fusion operation in conjunction with the regional temperature heat map to generate a fused point cloud dataset specifically includes: A multi-source aerial photography dataset is acquired, visible light image data is extracted from the multi-source aerial photography dataset, the camera pose is calculated using the structure of motion recovery algorithm to generate an initial sparse point cloud, and the point cloud is registered using a multi-view stereo matching algorithm to obtain a dense 3D point cloud. A regional temperature heat map is acquired, and the regional temperature heat map is spatially reprojected. The two-dimensional image coordinates of each temperature pixel are converted into three-dimensional spatial rays through back projection calculation using the extrinsic parameter transformation matrix of the monitoring camera pre-calibrated by the aerial drone. For each 3D spatial ray, a spherical search domain with a preset spacing is established to search for a set of neighboring points in a dense 3D point cloud, and the spatial distance between the 3D spatial ray and the neighboring points in the dense 3D point cloud is calculated. Select the point with the closest spatial distance as the target point, and crop a heat map pixel window with the target point as the center based on a preset window size. Calculate the planar distance between the target point and the center of each pixel in the cropped heat map pixel window. The weighting factor is calculated based on the inverse square ratio of the planar distance. The temperature of each pixel within the window is then weighted and fused. The weighted fusion result is used to assign a temperature attribute to the point cloud data corresponding to the target point, and finally, a fused point cloud dataset is generated.

5. The method for analyzing intertidal oyster reef structures using fusion thermal imaging aerial photography as described in claim 1, characterized in that, The process of using the oyster reef distribution map as a spatial mask to crop the fused point cloud dataset and construct a 3D model of the intertidal reef, and then using the 3D model to perform crack detection and oyster reef feature measurement to generate a multimodal reef feature vector, specifically includes: Obtain an oyster reef distribution map and a fused point cloud dataset. Based on the oyster reef distribution map, extract spatial coordinate features of the oyster reef areas to locate them. Generate an initial spatial mask based on the extracted coordinate features. A topological buffer is constructed using the initial spatial mask. The boundary lines of the initial spatial mask are extracted and Delaunay triangulation is performed to generate a constrained triangular mesh. Buffer point columns are generated by interpolating the edges of the constrained triangular mesh according to the average spacing of the point cloud, forming a closed buffer zone. The closed buffer band is fused with the initial spatial mask to generate a final spatial mask. The final spatial mask is then mapped onto the fused point cloud dataset for point cloud cropping, resulting in a cropped point cloud dataset. The cropped point cloud dataset is input into the Poisson reconstruction engine to generate a 3D model of the intertidal reef. Then, all triangular faces connected to the vertices of the 3D model of the intertidal reef are extracted. The square of the area of ​​each face is used as a weighting factor to calculate the weighted average of the unit normal vectors of each face to obtain the mean vector, which is used as the new normal vector of the corresponding vertex to generate a normal vector optimized 3D model. In the normal vector optimized 3D model, the shared edges of each triangular facet of the model are scanned one by one, and the normal angle between adjacent facets is calculated. When it is less than a preset threshold, the corresponding shared edge is marked as a potential crack edge segment. The temperature gradient vector of each vertex is calculated using the central difference method to generate a gradient magnitude heatmap. The gradient magnitude heatmap is then used to apply temperature constraints to the edge segments of the potential fractures to assess the authenticity of the fractures, and finally, a fracture network contour map is generated. Measurement points are randomly set up in the three-dimensional model of the intertidal reef. Multi-level spheres with a preset radius are generated with the measurement points as the center. Morphological analysis algorithms are used to measure the morphological characteristics of the oyster reef and generate reef morphological characteristic information. Thermodynamic parameters of the reef are obtained by using a three-dimensional model of the intertidal reef, and the fracture features of the reef are obtained by using the fracture network contour map. A multimodal reef feature vector is constructed by combining the morphological features of the reef.

6. The method for analyzing intertidal oyster reef structures using fusion thermal imaging aerial photography according to claim 1, characterized in that, The analysis of the intertidal oyster reef health in the target area based on the multimodal reef feature vector, combined with regional ecological data to determine the priority of regional restoration, and generating early warning information for notification, specifically includes: A reef health analysis model is built using a gradient boosting decision tree framework. Historical data is retrieved to obtain historical reef health assessment instances with different health conditions and the reef features corresponding to each instance are extracted. A training set is set to pre-train the reef health analysis model. Obtain multimodal reef feature vectors, input the multimodal reef feature vectors into a pre-trained reef health analysis model, analyze the health status of intertidal oyster reefs in the target area through a weighted voting mechanism, and output the reef health score of each oyster reef. Oyster reefs with health scores below a preset threshold are screened out, and their actual locations are marked as potential danger points. A ranking table is then generated based on the health scores of each potential danger point. Acquire regional ecological and environmental data, extract regional hydrological parameters of potential hazardous points from the regional ecological and environmental data, and perform weighted calculations on each potential hazardous point in combination with the sorting table to generate a remediation priority index for each potential hazardous point. The repair priority index of each potential danger point is determined by comparing it with the preset level classification threshold. Based on the level classification results, a regional repair suggestion heat map is generated through visualization processing. Early warning prompts are then issued through the regional repair suggestion heat map.

7. A system for analyzing intertidal oyster reef structures using thermal imaging aerial photography, characterized in that, The system includes a memory and a processor. The memory contains a program for analyzing intertidal oyster reef structures based on fused thermal imaging aerial photographs. When the processor executes the program for analyzing intertidal oyster reef structures based on fused thermal imaging aerial photographs, it performs the following steps: A multi-source aerial photography dataset is acquired by patrolling the target area using aerial photography drones. Thermal infrared image data of the intertidal zone is extracted to generate a radiation intensity distribution map. A regional temperature heat map is then constructed using the radiation intensity distribution map. Multispectral data was extracted from the multi-source aerial photography dataset to generate a multispectral substrate classification map. The intertidal oyster reef body in the target area was segmented by combining the regional temperature heat map and the reef body edge was optimized to obtain the oyster reef body distribution map. Based on the multi-source aerial photography dataset, a dense three-dimensional point cloud of the intertidal zone of the target area is constructed. Combined with the regional temperature heat map, a thermal-optical data fusion operation is performed to generate a fused point cloud dataset. The oyster reef distribution map is used as a spatial mask to crop the fused point cloud dataset and construct a three-dimensional model of the intertidal reef. Crack detection and oyster reef feature measurement are performed using the three-dimensional model of the intertidal zone to generate a multimodal reef feature vector. Based on the analysis of the multimodal reef feature vector, the health of the intertidal oyster reef in the target area is determined, and regional ecological data is combined to determine the priority of regional restoration and generate early warning information for notification.

8. The intertidal oyster reef analysis system based on thermal imaging aerial photography according to claim 7, characterized in that, The process of acquiring multi-source aerial photography datasets by patrolling the target area using a drone, extracting thermal infrared image data of the intertidal zone to generate a radiation intensity distribution map, and constructing a regional temperature heat map using the radiation intensity distribution map specifically includes: A drone equipped with multi-source monitoring equipment is used to cruise the target area. The multi-source monitoring equipment collects thermal infrared image data, visible light image data and multispectral data of the intertidal zone of the target area to generate a multi-source aerial photography dataset. Based on the multi-source aerial photography dataset, thermal infrared image data of the intertidal zone is extracted. The thermal infrared image data is then radiometrically calibrated and converted into absolute radiation intensity values ​​based on the temperature-radiation value calibration curve of the sensor's built-in blackbody reference source, generating a radiation intensity distribution map. Acquire atmospheric data of the target area during the cruise, input it into the MODTRAN radiation model to calculate atmospheric transmittance and uplink radiation path brightness, retrieve the true surface radiation intensity value through the radiative transfer equation, and perform atmospheric path correction on the radiation intensity distribution map to obtain the corrected radiation intensity distribution map. Local binary mode texture calculation is performed on the corrected radiation intensity distribution map, the gray-level variation coefficient of each pixel neighborhood is extracted as a uniformity index, and the texture features are divided by K-means clustering algorithm to generate a divided radiation intensity distribution map. By applying the Stefan-Boltzmann law to the radiation intensity distribution map, temperature conversion is performed, and the final output is a regional temperature thermogram.

9. The intertidal oyster reef analysis system based on thermal imaging aerial photography according to claim 7, characterized in that, The process involves extracting multispectral images from the multi-source aerial photography dataset to generate a multispectral substrate classification map, segmenting the intertidal oyster reef bodies in the target area using a regional temperature heatmap, and optimizing the reef edges to obtain an oyster reef distribution map. Specifically, this includes: The multispectral features of oyster reefs are obtained by searching through big data networks to generate a reef spectral feature set, and the reef spectral feature set is mapped to a high-dimensional space using a Gaussian kernel function. The minimum hypersphere boundary and hypersphere radius containing the multispectral feature samples of all oyster reef bodies are obtained by solving in the mapped high-dimensional space. A multispectral geological classification model is then established using the minimum hypersphere boundary and hypersphere radius. A multi-source aerial photography dataset is acquired, multispectral data is extracted from the multi-source aerial photography dataset and input into the multispectral geological classification model, and the Mahalanobis distance of the input multispectral data in the kernel space is calculated. The calculated Mahalanobis distance is compared with the preset hypersphere radius. If it is less than the preset hypersphere radius, the corresponding multispectral data source location is marked as the oyster reef body area, and finally a multispectral geological classification map is generated. A regional temperature heat map is obtained, which is then divided into several local blocks. The pixel grayscale information entropy of each block is calculated, and a segmentation threshold is dynamically set based on the calculated local entropy value. Based on the multispectral geological classification map, regional features of non-oyster reef bodies are extracted to generate a mask. Combined with the segmentation threshold, preliminary segmentation of oyster reef bodies is performed to obtain a preliminary oyster reef body distribution map. A morphological watershed algorithm is introduced to extract temperature gradient features from the temperature heatmap of the region. The initially segmented reef areas are used as marker points for edge optimization to obtain the oyster reef distribution map.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a method program for analyzing intertidal oyster reefs using fused thermal imaging aerial photography. When the method program is executed by a processor, it implements the steps of the method for analyzing intertidal oyster reefs using fused thermal imaging aerial photography as described in any one of claims 1 to 6.

Images