A method and system for intelligent identification of coastal mangroves based on UAV remote sensing

Abstract

This invention relates to the field of image recognition technology, and more particularly to a method and system for intelligent identification of coastal mangroves based on UAV remote sensing. The method includes: acquiring multi-band spectral data using remote sensing equipment, generating a standardized spectral feature dataset, calculating multi-band differences between adjacent pixels, and constructing a difference matrix to extract quantification indicators; marking potential transition points using a threshold segmentation algorithm, and performing temporal fusion with multi-temporal data to adjust dynamic offsets to obtain corrected transition features; grouping the corrected transition features using a clustering algorithm, determining critical position coordinates, and generating a refined boundary map using geographic information system data; and using interpolation methods to repair missing contour segments, ensuring contour continuity and obtaining the final mangrove boundary location result. This invention solves the problem of insufficient accuracy of traditional mangrove boundary extraction methods in dynamic environments and achieves high-precision intelligent identification of mangrove boundaries based on UAV remote sensing.

Description

Technical Field

[0001] This invention relates to the field of image recognition technology, and in particular to a method and system for intelligent identification of coastal mangroves based on UAV remote sensing. Background Technology

[0002] Coastal mangroves are vital ecosystems with significant ecological functions, such as protecting coastlines, regulating climate, and maintaining biodiversity. However, with increasing global climate change and human activities, mangrove ecosystems face severe challenges, particularly the growing impacts of natural factors like coastal erosion, storm surges, and tidal cycles. Therefore, accurate monitoring and protection of mangroves have become crucial topics in ecological and environmental research and management.

[0003] Traditional mangrove monitoring methods primarily rely on ground surveys and remote sensing technology. While ground surveys provide high-precision data, their high cost and limited coverage make them unsuitable for large-scale, long-term monitoring. Remote sensing technology, especially UAV remote sensing, has become a crucial tool for mangrove monitoring due to its high resolution, low cost, and flexibility. However, traditional remote sensing methods still face certain technical challenges in accurately identifying mangrove boundaries, capturing dynamic changes, and achieving high-precision positioning. Particularly in complex coastal environments, natural factors such as tidal cycles and storm surges can significantly alter the spectral characteristics of mangrove boundaries, posing a challenge to accurate boundary identification.

[0004] Existing remote sensing technologies mostly rely on static images or single-temporal data for mangrove boundary extraction, lacking effective adaptation to dynamic environmental factors. This results in low boundary positioning accuracy under dynamic environments such as tidal changes and storm surges. Furthermore, traditional methods suffer from insufficient rapid response to dynamic changes and the inability to repair missing segments, affecting the completeness and accuracy of mangrove boundary extraction results.

[0005] To address the aforementioned issues, this invention proposes an intelligent identification method and system for coastal mangroves based on UAV remote sensing technology, combined with time-series data fusion, dynamic environmental impact analysis, and high-precision boundary positioning methods. This method effectively captures the impact of environmental factors such as tidal cycles and storm surges on mangrove boundaries and provides more accurate mangrove boundary positioning results through high-resolution remote sensing data. It not only overcomes the limitations of existing methods in dynamic environments but also enables real-time monitoring and precise extraction of mangrove boundaries, providing strong support for the ecological protection and management of mangroves. Summary of the Invention

[0006] This invention provides a method and system for intelligent identification of coastal mangroves based on UAV remote sensing, which overcomes the problem of insufficient accuracy of mangrove boundary identification in dynamic environments by traditional methods. In particular, it can effectively improve the accuracy and stability of mangrove boundaries under the influence of natural factors such as tidal cycles and storm surges.

[0007] In a first aspect, the present invention provides a method for intelligent identification of coastal mangroves based on unmanned aerial vehicle (UAV) remote sensing, the method comprising:

[0008] Step S1: Collect multi-band spectral data of the coastal mangrove area using remote sensing equipment to obtain a raw dataset containing spectral feature changes. Standardize the raw dataset to obtain a standardized spectral feature dataset.

[0009] Step S2: Calculate the multi-band difference between adjacent pixels based on the spectral feature dataset, and construct the difference matrix to determine the quantitative index reflecting the change of spectral features;

[0010] Step S3: Use a threshold segmentation algorithm to process the quantization index in the difference matrix, mark potential jump points, and obtain a preliminary jump position set; integrate the preliminary jump position set with multi-temporal spectral data through a time series fusion method, adjust the dynamic offset, and obtain the corrected jump features;

[0011] Step S4: Based on the corrected jump features, apply a clustering algorithm to group the boundary region pixels and determine the critical position coordinates; overlay the geographic information system data with the critical position coordinates to generate a boundary vector layer and obtain a refined boundary map;

[0012] Step S5: Extract the edge contours based on the refined boundary map, fill the missing segments of the edge contours using an interpolation method, and determine the continuity of the contours to obtain the final mangrove boundary positioning result.

[0013] As a preferred embodiment of the present invention, step S1 involves obtaining a standardized spectral feature dataset, including:

[0014] Multi-band spectral data of coastal mangrove areas are acquired using satellite remote sensing equipment. The multi-band spectral data includes at least near-infrared, red, and green bands. Time-series sampling is performed based on the dynamic environment under the influence of tidal cycle disturbances and storm surges to determine the sampling time interval. Multi-temporal spectral data are collected according to the sampling time interval to generate a raw dataset containing spectral feature changes. The raw dataset is preprocessed to remove noisy data, resulting in a standardized spectral feature dataset. The multi-band spectral dataset includes the spectral features and spatial location information of each pixel.

[0015] As a preferred embodiment of the present invention, step S2, constructing a difference matrix to determine a quantitative index reflecting changes in spectral characteristics, includes:

[0016] Based on the multi-band spectral values ​​of each pixel and its neighboring pixels in the spectral feature dataset; calculate the difference between the pixel and its neighboring pixels in each band to generate a multi-band difference set; construct a difference matrix based on the multi-band difference set; perform statistical analysis on the difference matrix to extract quantitative indicators that reflect the spectral jump pattern, wherein the quantitative indicators include at least the mean and variance of the differences.

[0017] As a preferred embodiment of the present invention, step S3, obtaining the preliminary set of transition positions, includes:

[0018] Based on the quantization index in the difference matrix, a preset threshold is set; if the quantization index exceeds the preset threshold, the corresponding pixel is marked as a potential jump point; for the potential jump point, the difference distribution characteristics of its adjacent regions are calculated; the spatial consistency of the difference distribution characteristics is compared by an iterative method; if the difference distribution characteristics meet the preset spatial connectivity conditions, the potential jump point is included in the preliminary jump position set, which includes the spatial coordinates and spectral feature values ​​of the jump point.

[0019] As a preferred embodiment of the present invention, step S3, obtaining the modified jump feature, includes:

[0020] Obtain the coordinates of the transition points in the preliminary transition position set; analyze the impact of tidal periodic disturbances on the transition points based on the multi-temporal spectral data; dynamically offset and adjust the coordinates of the transition points using a time-series fusion method to generate a set of transition points after offset correction; extract corrected transition features based on the set of transition points after offset correction, the corrected transition features including the spectral features and positional stability of the transition points.

[0021] As a preferred embodiment of the present invention, step S4, determining the critical position coordinates, includes:

[0022] Based on the corrected abrupt change features, the spectral feature vectors of the pixels in the boundary region are extracted; the spectral feature vectors are grouped using a clustering algorithm to generate mutation clusters; if the mutation clusters show significant changes in spectral features, they are confirmed as mangrove boundary transition zones; based on the center point of the mutation clusters, the critical position coordinates are calculated, and the critical position coordinates include the spatial location information of the boundary region.

[0023] As a preferred embodiment of the present invention, step S4, obtaining the refined boundary map, includes:

[0024] The spatial location information in the critical location coordinates is obtained; the spatial location information and terrain data are superimposed using geographic information system data; a boundary vector layer is generated based on the superposition result; the positioning accuracy of the boundary vector layer is verified for the spectral jump phenomenon under the influence of storm surge, and a refined boundary map is generated, which includes the spatial distribution of mangrove boundaries.

[0025] As a preferred embodiment of the present invention, step S5, obtaining the final mangrove boundary positioning result, includes:

[0026] Based on the refined boundary map, the edge contour of the mangrove boundary is extracted; for the missing contour segments caused by the dynamic environment, an interpolation method is used to generate continuous contour lines; the spatial connectivity of the continuous contour lines is analyzed to determine the contour continuity; if the contour continuity meets the preset conditions, the final mangrove boundary positioning result is generated, and the final mangrove boundary positioning result includes the spatial coordinates of the boundary contour.

[0027] Secondly, the present invention also provides an intelligent identification system for coastal mangroves based on unmanned aerial vehicle (UAV) remote sensing, for implementing the above-mentioned method, the system comprising:

[0028] The data acquisition unit is used to acquire multi-band spectral data of the coastal mangrove area through remote sensing equipment, obtain a raw dataset containing spectral feature changes, and standardize the raw dataset to obtain a standardized spectral feature dataset.

[0029] The difference calculation unit is used to calculate the multi-band difference between adjacent pixels based on the spectral feature dataset, and construct the difference matrix to determine the quantitative index reflecting the change of spectral features;

[0030] The threshold segmentation unit is used to process the quantization index in the difference matrix through the threshold segmentation algorithm, mark potential jump points, and obtain a preliminary jump position set; the preliminary jump position set is integrated with multi-temporal spectral data through a time series fusion method, and the dynamic offset is adjusted to obtain the corrected jump features;

[0031] The clustering analysis unit is used to apply a clustering algorithm to group boundary region pixels according to the corrected jump features, determine critical position coordinates, and generate a boundary vector layer by overlaying geographic information system data with the critical position coordinates to obtain a refined boundary map.

[0032] The contour extraction unit is used to extract edge contours based on the refined boundary map, fill in the missing segments of the edge contours using an interpolation method, and determine the continuity of the contours to obtain the final mangrove boundary positioning result.

[0033] Thirdly, the present invention also provides a computer-readable storage medium storing instructions that, when executed by a processor, implement the above-described method.

[0034] The beneficial effects of this invention are as follows:

[0035] This invention collects multi-band spectral data of coastal mangrove areas using remote sensing equipment, and standardizes the original dataset to generate a standardized spectral feature dataset, providing a consistent and stable data foundation for subsequent analysis. By calculating the multi-band differences between adjacent pixels, a difference matrix is ​​constructed to extract quantitative indicators reflecting changes in spectral features, providing spectrally sensitive quantitative data for subsequent boundary localization, effectively capturing spectral jumps caused by environmental changes. A threshold segmentation algorithm is used to process the quantification indicators in the difference matrix, marking potential jump points and forming a preliminary set of jump locations. Through temporal fusion with multi-temporal spectral data, the potential jump points are dynamically offset and adjusted to obtain corrected jump features. This not only captures the trend of environmental changes but also effectively addresses changes caused by dynamic environmental factors such as tides and storm surges, ensuring the stability and accuracy of boundary identification. After identifying the jump characteristics, a clustering algorithm is used to group the pixels in the boundary region, generating mutation clusters. Critical position coordinates are determined through spatial location calculations, ultimately generating a boundary vector layer and a refined boundary map. This effectively identifies and groups regions exhibiting significant spectral changes, providing strong support for the precise location of mangrove boundaries. By analyzing the refined boundary map, interpolation methods are used to repair missing edge contour segments caused by dynamic environmental changes, ensuring contour continuity and determining spatial connectivity. Finally, a complete mangrove boundary location result is obtained, ensuring accurate identification and precise location of mangrove boundaries even under the influence of environmental factors such as storm surges. This significantly improves the accuracy and robustness of boundary location. Through the synergy of these technical solutions, the problem of traditional methods being unable to cope with dynamic environmental changes is overcome, providing technical support for the precise monitoring and protection of coastal mangroves. Attached Figure Description

[0036] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0037] Figure 1 This is a flowchart illustrating an intelligent identification method for coastal mangroves based on UAV remote sensing, as shown in the embodiment.

[0038] Figure 2This is a flowchart illustrating the method for obtaining the initial jump position set in the embodiment;

[0039] Figure 3 This is a flowchart of the method for obtaining modified transition features in the embodiment;

[0040] Figure 4 This is a structural diagram of a coastal mangrove intelligent identification system based on UAV remote sensing, as shown in the embodiment. Detailed Implementation

[0041] This invention provides a method and system for intelligent identification of coastal mangroves based on UAV remote sensing. The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" or "having" and any variations thereof are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0042] For ease of understanding, the specific process of the embodiments of the present invention will be described below, such as... Figure 1 As shown in the figure, an intelligent identification method for coastal mangroves based on UAV remote sensing in an embodiment of the present invention includes:

[0043] Step S1: Collect multi-band spectral data of the coastal mangrove region using remote sensing equipment to obtain a raw dataset containing spectral feature variations. Standardize the raw dataset to obtain a standardized spectral feature dataset; specifically including:

[0044] Multi-band spectral data of coastal mangrove areas are acquired using satellite remote sensing equipment. The multi-band spectral data includes at least near-infrared, red, and green bands. Time-series sampling is performed based on the dynamic environment under the influence of tidal cycle disturbances and storm surges to determine the sampling time interval. Multi-temporal spectral data are collected according to the sampling time interval to generate a raw dataset containing spectral feature changes. The raw dataset is preprocessed to remove noisy data, resulting in a standardized spectral feature dataset. The multi-band spectral dataset includes the spectral features and spatial location information of each pixel.

[0045] Specifically, in this embodiment, multi-band spectral data of the coastal mangrove region is acquired using the aforementioned satellite remote sensing equipment. This multi-band spectral data includes at least near-infrared, red, and green bands to ensure comprehensive capture of the vegetation reflectance characteristics of the mangrove region. Furthermore, to address the impact of dynamic environmental factors such as tidal cycles and storm surges on the mangrove boundary, a time-series sampling method is used during the data acquisition process to determine an appropriate sampling time interval. This ensures that changes caused by tidal fluctuations and storm surges are captured. The specific sampling interval is dynamically adjusted based on the influence of tidal cycles and storm surges. A suitable sampling time interval is calculated based on historical tidal data, tidal cycle variation patterns, and storm surge information in weather forecasts. This time interval is determined to be a period of data acquisition at regular intervals to ensure that the acquired data accurately reflects the impact of tides and storm surges on the spectral characteristics of the mangroves, guaranteeing the integrity and accuracy of the time-series data.

[0046] Based on the sampling time intervals determined above, the satellite remote sensing equipment continuously collects spectral data from multiple time phases, generating a raw dataset containing changes in spectral characteristics. The spectral changes of each time phase are recorded. The raw dataset covers the changes in spectral characteristics of the mangrove area at different time points, which can accurately reflect the dynamic impact of environmental factors. In order to eliminate the interference of noise data on the analysis results, the collected raw dataset is preprocessed. First, a mean-mode filtering algorithm is applied to remove cloud interference and other transient noise, thereby optimizing data quality. Then, the data is standardized using a min-max normalization method, converting all spectral data into a uniform numerical range. This makes data processing and analysis more stable and consistent, further improving the reliability of subsequent analyses. The aforementioned spectral dataset includes not only the spectral features of each pixel but also the spatial location information of each pixel, thus establishing a one-to-one correspondence between the geographic coordinates of each data point and its spectral feature value. The integration and preprocessing of time-series data ensures that the collected spectral feature data can truly reflect changes in the dynamic environment, and the standardization process provides a foundation for further analysis and processing. This technical solution ensures accurate positioning of mangrove boundaries and precise capture of environmental changes, avoiding data deviations caused by the inability of traditional data acquisition methods to effectively cope with dynamic environmental factors such as tides and storm surges, thereby improving the overall accuracy and stability of the mangrove intelligent identification method.

[0047] Step S2: Calculate the multi-band differences between adjacent pixels based on the spectral feature dataset, and construct a difference matrix to determine a quantitative index reflecting changes in spectral features; specifically including:

[0048] Based on the multi-band spectral values ​​of each pixel and its neighboring pixels in the spectral feature dataset; calculate the difference between the pixel and its neighboring pixels in each band to generate a multi-band difference set; construct a difference matrix based on the multi-band difference set; perform statistical analysis on the difference matrix to extract quantitative indicators that reflect the spectral jump pattern, wherein the quantitative indicators include at least the mean and variance of the differences.

[0049] Specifically, in this embodiment, based on the multi-band spectral values ​​of each pixel and its neighboring pixels in the aforementioned spectral feature dataset, that is, traversing the position of each pixel in the original dataset and identifying its neighboring pixels in the image grid, the neighboring pixels typically include pixels in the four directions of up, down, left, and right. The spectral values ​​of these neighboring pixels in the near-infrared, red, and green bands are extracted from the multi-band spectral data collected by satellite remote sensing equipment to form a preliminary list of the aforementioned spectral values. To capture the influence of natural factors such as spectral changes, tidal cycles, and storm surges in dynamic environments on the spectral characteristics of mangrove areas, the numerical difference between each pixel and its neighboring pixels is calculated in each band, generating a multi-band difference set. The differences of all bands are then summarized into a single set, and the above process is repeated. The process covers all adjacent pixel relationships, ensuring the integrity of the difference matrix and the consistency of the data. Based on the generated multi-band difference set, it is further organized into a matrix, where rows represent pixel positions and columns represent combinations of differences in different bands and adjacent directions. The construction of the difference matrix allows the spectral differences of each pixel to be presented in a structured manner, facilitating subsequent statistical analysis and feature extraction. In order to better reflect the impact of tidal periodic disturbances on the data, the matrix dimension is adjusted for time-series sampling data, and time series are incorporated. The differences of multiple time phases are stacked into a three-dimensional matrix, thereby enhancing the responsiveness to dynamic environments such as storm surges. This allows the capture of spectral jump phenomena under the influence of storm surges and the identification of systematic shifts caused by environmental changes.

[0050] After obtaining the difference matrix, statistical analysis is performed to extract quantitative indicators reflecting spectral jump patterns. These quantitative indicators include at least the mean and variance of the differences. By calculating the mean and variance of each element in the matrix, the intensity and stability of spectral differences can be comprehensively evaluated, thus providing a quantitative basis for subsequent analysis. In particular, during the time-series fusion process, the quantitative indicators are weighted and averaged based on the characteristics of the tidal cycle to comprehensively consider data from multiple time phases, further enhancing the description of spectral feature changes. For areas that show significant changes in the time-series sampling, additional standard deviation indicators are extracted to enhance the response to spectral jumps, thereby improving the model's sensitivity to dynamic environmental changes. This ensures accurate identification of boundary transition zones under the influence of factors such as storm surges or tidal cycles, and improves the accuracy and continuity of the final mangrove boundary location results. The above technical solutions not only guarantee the comprehensiveness and accuracy of the data but also provide a solid data foundation for subsequent clustering algorithms and boundary identification, thus significantly improving the stability and accuracy of the intelligent identification method for coastal mangroves.

[0051] Step S3: Use a threshold segmentation algorithm to process the quantization index in the difference matrix, mark potential jump points, and obtain a preliminary jump position set; integrate the preliminary jump position set with multi-temporal spectral data through a time series fusion method, adjust the dynamic offset, and obtain the corrected jump features;

[0052] In step S3, a preliminary set of jump positions is obtained, such as... Figure 2 As shown, it includes:

[0053] Based on the quantization index in the difference matrix, a preset threshold is set; if the quantization index exceeds the preset threshold, the corresponding pixel is marked as a potential jump point; for the potential jump point, the difference distribution characteristics of its adjacent regions are calculated; the spatial consistency of the difference distribution characteristics is compared by an iterative method; if the difference distribution characteristics meet the preset spatial connectivity conditions, the potential jump point is included in the preliminary jump position set, which includes the spatial coordinates and spectral feature values ​​of the jump point.

[0054] Specifically, in this embodiment, based on the quantization indicators in the difference matrix, a preset threshold is first set. Statistical methods are used to calculate the mean and standard deviation of all quantization indicators in the difference matrix to determine the benchmark for the threshold. Based on the statistical results, the preset threshold is set as a multiple of the mean to adapt to the dynamic changes in the spectral data of the mangrove region, ensuring effective noise filtering during data analysis and providing an accurate basis for subsequent labeling of potential transition points. Furthermore, by traversing each quantization indicator in the difference matrix and comparing its relationship with the preset threshold, if a quantization indicator exceeds the threshold, the corresponding pixel is labeled as a potential transition point, and its location coordinates are recorded. Through this simple and efficient labeling method, pixels with significant changes in spectral characteristics are initially screened. The system identifies potential transition points, providing initial data support for subsequent boundary localization. After marking potential transition points, it iteratively compares the difference distribution of adjacent regions to determine the spatial connectivity of these points. Specifically, it selects eight neighboring pixels of each potential transition point as adjacent regions and calculates the mean and variance of the difference distribution within these regions. It then iteratively compares these values ​​with the differences of the potential transition points. If the similarity is higher than the set similarity, the region is confirmed as a connected region. This process is repeated until all potential transition points are compared and connected components are formed. This method is particularly suitable for mangrove boundary identification under dynamic environmental influences, such as storm surges or tidal disturbances. It can effectively capture spectral shifts caused by environmental changes, thereby improving the accuracy of boundary transition zone identification.

[0055] Based on the spatial connectivity analysis results, a preliminary set of transition locations is generated, including the coordinates and spectral features of potential transition points. Specifically, the center coordinates and average spectral features of each component are extracted from the connected components. The extracted results are integrated into a set, ensuring that each element contains the corresponding coordinates and spectral feature values. This preliminary set of transition locations provides the basic data for subsequent time-series data fusion and refined boundary map generation. The above technical solution ensures the efficiency and accuracy of mangrove boundary identification, especially in dynamically changing environments, where the boundary identification results can be updated and adjusted in real time.

[0056] Further, in step S3, the corrected transition feature is obtained, such as... Figure 3 As shown, it includes:

[0057] Obtain the coordinates of the transition points in the preliminary transition position set; analyze the impact of tidal periodic disturbances on the transition points based on the multi-temporal spectral data; dynamically offset and adjust the coordinates of the transition points using a time-series fusion method to generate a set of transition points after offset correction; extract corrected transition features based on the set of transition points after offset correction, the corrected transition features including the spectral features and positional stability of the transition points.

[0058] Specifically, in this embodiment, firstly, by traversing the coordinates of each jump point in the preliminary jump location set, the row and column indices corresponding to each jump point are extracted from the difference matrix processed by the threshold segmentation algorithm. These indices represent the pixel positions of the jump points in the image. Using these indices, they are converted into latitude and longitude values ​​in the geographic coordinate system to ensure alignment with the multi-temporal spectral data, thereby providing accurate spatial positioning information for subsequent data analysis and adjustment. Based on the aforementioned multi-temporal spectral data, the impact of tidal cycle disturbances on the jump points is analyzed. Tidal cycle disturbances, caused by the rise and fall of seawater, directly affect the soil moisture and vegetation reflectance of the mangrove area, leading to dynamic changes in the spectral jump points over time. To quantitatively analyze the impact of tidal disturbances, the spectral differences of each jump point at different temporal phases are calculated. For example, the difference between different bands is calculated using Euclidean distance to further assess the offset caused by tidal disturbances, thereby identifying the impact of dynamic environmental factors such as tidal cycles and storm surges on the spectral jump points. Especially during peak tides or storm surges, the spectral jump points may experience positional shifts, thus affecting the accurate positioning of mangrove boundaries.

[0059] Then, a time-series fusion method is used to dynamically adjust the coordinates of the aforementioned jump points. This method primarily utilizes the Kalman filter algorithm, which smooths the shifts caused by tidal disturbances by weighted averaging of multi-temporal spectral data. Kalman filtering is a recursive estimation method that minimizes the shift estimation error through prediction and update steps. Specifically, Kalman filtering adjusts the positions of jump points by combining historical observation data and predicted jump point locations, generating a set of shift-corrected jump points to ensure positional stability and improve the accuracy of subsequent analysis. The shift-corrected jump point set obtained after time-series fusion is then used to further extract corrected jump features. Specifically, for each corrected jump point, the shift features are further extracted through… Multi-temporal spectral data are used to calculate spectral feature vectors and obtain spectral characteristics of each band. Then, the standard deviation of the location is calculated to quantify the positional stability of the point in the time series. If the standard deviation is less than a set value, it indicates that the positional stability of the transition point is high and the reliability is strong. The above-mentioned corrected transition features are used as input data for subsequent boundary grouping and clustering algorithms to ensure that the clustering process can take into account the influence of dynamic environmental factors, thereby accurately identifying the boundary transition zone. The above technical solution can effectively identify and correct the dynamic influence of factors such as tides and storm surges on mangrove boundaries, providing accurate and stable mangrove boundary positioning results, thereby providing reliable data support for subsequent ecological monitoring and environmental protection.

[0060] Step S4: Based on the corrected jump features, apply a clustering algorithm to group the boundary region pixels and determine the critical position coordinates; overlay the geographic information system data with the critical position coordinates to generate a boundary vector layer and obtain a refined boundary map;

[0061] In step S4, determining the critical position coordinates includes:

[0062] Based on the corrected abrupt change features, the spectral feature vectors of the pixels in the boundary region are extracted; the spectral feature vectors are grouped using a clustering algorithm to generate mutation clusters; if the mutation clusters show significant changes in spectral features, they are confirmed as mangrove boundary transition zones; based on the center point of the mutation clusters, the critical position coordinates are calculated, and the critical position coordinates include the spatial location information of the boundary region.

[0063] Specifically, in this embodiment, spectral feature vectors of boundary region pixels are extracted based on corrected jump features. These corrected jump features are multi-temporal spectral data integrated using a temporal fusion method. For the jump positions adjusted after tidal periodic disturbances, the spectral feature vectors include reflectance values ​​of each boundary region pixel in different bands, such as near-infrared, red, and green bands. The feature vectors formed by these features provide basic data for subsequent boundary detection and grouping. A clustering algorithm is used to group the spectral feature vectors to generate mutation clusters. Specifically, the K-means clustering algorithm is selected, which can assign feature vectors to different clusters based on their similarity. By calculating the Euclidean distance, K-means clustering can determine the similarity between each spectral feature vector and the cluster center. The cluster centers are updated iteratively multiple times until the clustering converges, thus forming multiple groups of spectral feature vectors. In these groups, mutation clusters typically represent spectral jumps caused by dynamic environments such as tides or storm surges.

[0064] If the generated mutant clusters show significant changes in spectral characteristics, these clusters can be identified as mangrove boundary transition zones. The variance of pixel spectral changes within each mutant cluster is calculated; if the variance exceeds a preset threshold, it is considered a significant change. Furthermore, by comparing the differences in spectral distribution between clusters, a t-test statistic is used to assess their significance level. If this significance level is higher than a preset threshold, the cluster is confirmed as a mangrove boundary transition zone. The t-test calculates the t-value by comparing the spectral vectors of the mutant cluster with those of adjacent non-mutant regions. If the t-value is greater than a critical value, it indicates that the differences between clusters are statistically significant, thus ensuring accurate identification of the boundary transition zone.

[0065] Based on the geometric center of the identified mutation cluster, the critical position coordinates are calculated. These coordinates represent the spatial location information of the boundary area. Specifically, the critical position coordinates are obtained by overlaying the center point of the mutation cluster with the geographic coordinate system, expressed in latitude and longitude. These coordinates not only provide the precise location of the mangrove boundary but also provide spatial data support for subsequent boundary map generation, ensuring the accurate positioning of the mangrove boundary. The above technical solution can effectively extract stable and accurate mangrove boundary information from a complex dynamic environment, providing reliable technical support for ecological monitoring and environmental protection.

[0066] Further, in step S4, a refined boundary map is obtained, including:

[0067] The spatial location information in the critical location coordinates is obtained; the spatial location information and terrain data are superimposed using geographic information system data; a boundary vector layer is generated based on the superposition result; the positioning accuracy of the boundary vector layer is verified for the spectral jump phenomenon under the influence of storm surge, and a refined boundary map is generated, the refined boundary map including the spatial distribution of mangrove boundaries.

[0068] Specifically, in this embodiment, spatial location information in the critical position coordinates is obtained from the cluster center offset calculation obtained by correcting jump feature clustering. The corresponding latitude and longitude values ​​are extracted using these coordinates as spatial location information to generate two-dimensional coordinate pairs. Geographic Information System (GIS) data is used to overlay the spatial location information with terrain data to ensure correct alignment between the coordinates and the terrain data. Specifically, terrain data, including elevation and slope, is retrieved from the GIS database, and the spatial location information is spatially overlaid with the terrain data using a rasterization method. To eliminate errors in coordinate transformation, all data is processed in the same projection system and then matched point-by-point to ensure that the geographic coordinates and terrain data are correctly aligned. Precise integration of terrain data; based on the superimposed spatial location information and terrain data, a boundary vector layer is generated using a vector generation algorithm. Boundary lines are generated by connecting points, thus obtaining a spatial representation of the mangrove boundary. Furthermore, for spectral jump phenomena under dynamic environments such as storm surges, multi-temporal spectral data is collected to calculate the offset of the jump phenomenon, thereby verifying the positioning accuracy of the generated boundary vector layer. Specifically, when calculating the spectral jump phenomenon, the offset of the jump is obtained by comparing the difference matrix before and after the tide, and the offset is compared with the boundary vector layer. Accuracy indicators such as root mean square error are used to evaluate the accuracy of boundary positioning. If the error is lower than a preset threshold, the positioning accuracy is considered to meet the requirements; otherwise, it is not.

[0069] Furthermore, under dynamic disturbances such as storm surges, the verification process will consider the influence of tidal cycles. By selecting samples from the peak of storm surges in time-series sampling data, the spectral jump shift pattern will be analyzed, and the position of the boundary vector layer will be adjusted accordingly. This ensures that the mangrove boundary in the refined boundary map can accurately reflect the actual environmental changes under the influence of storm surges, etc. This not only reduces the impact of dynamic disturbances on boundary positioning but also improves the robustness of the map in changing environments, ensuring the stability of the boundary. The refined boundary map shown above demonstrates the spatial distribution of mangrove boundaries and can accurately reflect the dynamic changes and boundary transitions of mangrove areas. The above technical solution ensures that the generated mangrove boundary map has high accuracy and high reliability, and can provide accurate data support for subsequent ecological monitoring and environmental protection, especially under complex environmental change conditions, such as the impact of natural factors like storm surges on mangrove areas.

[0070] Step S5: Extract the edge contours based on the refined boundary map, fill the missing segments of the edge contours using an interpolation method, and determine the continuity of the contours to obtain the final mangrove boundary positioning result; specifically, this includes: extracting the edge contours of the mangrove boundary based on the refined boundary map; generating continuous contour lines using an interpolation method for missing contour segments caused by dynamic environmental influences; determining the continuity of the contours by analyzing the spatial connectivity of the continuous contour lines; if the continuity of the contours meets preset conditions, generating the final mangrove boundary positioning result, which includes the spatial coordinates of the boundary contours.

[0071] Specifically, in this embodiment, the edge contour of the mangrove boundary is extracted based on the refined boundary map. This is achieved by processing the refined boundary map using the Canny edge detection algorithm. The algorithm first smooths the image using Gaussian filtering to reduce noise interference, then calculates the gradient intensity and direction of the image, uses non-maximum suppression to remove non-edge pixels, and finally applies double threshold detection to extract edge pixels, thus obtaining the edge contour of the mangrove boundary. These edge lines reflect abrupt changes in spectral characteristics, marking the transition zone of the mangrove boundary. For contour gaps caused by dynamic environmental influences, interpolation methods are used to generate continuous contour lines. Specifically, the missing contour segments are first identified. These missing spectral data segments are usually caused by tidal periodic disturbances or storm surges. They are identified by calculating the spacing between breakpoints on the contour line. When the spacing exceeds a preset pixel threshold, it is marked as... Missing segments are filled using spline interpolation. Spline interpolation uses a polynomial function to fit adjacent known contour points to ensure that the generated curve smoothly connects the two missing endpoints, forming a continuous contour line. During the interpolation process, dynamic offsets in the time-series data are also considered to adapt to spectral jumps caused by factors such as storm surges, ensuring the integrity of the contour line and avoiding the impact of dynamic disturbances on positioning accuracy. To further improve the accuracy of the continuous contour line, linear interpolation can be used to quickly repair small missing segments in the time-series spectral data. If the missing segment is large and the data changes are complex, spline interpolation is used to capture nonlinear changes, thereby ensuring the stability of the contour line under environmental conditions such as high tides and high water levels. This ensures that the connection of the contour line remains stable even under significant environmental changes, ensuring that the generated mangrove boundary map has high usability.

[0072] A depth-first search algorithm is also used to perform spatial connectivity analysis on the generated continuous contour lines. This algorithm recursively visits all adjacent pixels starting from the beginning of the contour line, marking visited points and determining whether all pixels on the contour line can be traversed. If all pixels of the entire contour line can be visited, the contour line is considered connected, thus determining the integrity of the contour's continuity. This ensures that the generated mangrove boundary contour lines have no isolated segments and that the boundary positioning results have high reliability. After confirming that the contour continuity meets preset conditions, the final mangrove boundary positioning result is generated. Specifically, this is achieved by checking continuity indicators such as the number of connected components. If the number of components is 1 and there are no breakpoints, all spatial coordinates of the boundary contour are output. These coordinates are recorded in latitude and longitude format, representing the spatial distribution of the mangrove boundary, forming the final mangrove boundary positioning result. This technical solution ensures that the generated mangrove boundary positioning result has high accuracy and can effectively reflect the impact of dynamic environmental changes on the mangrove boundary, providing reliable data support for ecological monitoring and protection.

[0073] This invention also provides an intelligent identification system for coastal mangroves based on UAV remote sensing, used to implement the above-mentioned methods, such as... Figure 4 As shown, the system includes:

[0074] The data acquisition unit is used to acquire multi-band spectral data of the coastal mangrove area through remote sensing equipment, obtain a raw dataset containing spectral feature changes, and standardize the raw dataset to obtain a standardized spectral feature dataset.

[0075] The difference calculation unit is used to calculate the multi-band difference between adjacent pixels based on the spectral feature dataset, and construct the difference matrix to determine the quantitative index reflecting the change of spectral features;

[0076] The threshold segmentation unit is used to process the quantization index in the difference matrix through the threshold segmentation algorithm, mark potential jump points, and obtain a preliminary jump position set; the preliminary jump position set is integrated with multi-temporal spectral data through a time series fusion method, and the dynamic offset is adjusted to obtain the corrected jump features;

[0077] The clustering analysis unit is used to apply a clustering algorithm to group boundary region pixels according to the corrected jump features, determine critical position coordinates, and generate a boundary vector layer by overlaying geographic information system data with the critical position coordinates to obtain a refined boundary map.

[0078] The contour extraction unit is used to extract edge contours based on the refined boundary map, fill in the missing segments of the edge contours using an interpolation method, and determine the continuity of the contours to obtain the final mangrove boundary positioning result.

[0079] The present invention also provides a computer-readable storage medium storing instructions that, when executed by a processor, implement the above-described method.

[0080] In summary, this invention collects multi-band spectral data of coastal mangrove areas using remote sensing equipment, and standardizes the original dataset to generate a standardized spectral feature dataset, providing a consistent and stable data foundation for subsequent analysis. By calculating the multi-band differences between adjacent pixels, a difference matrix is ​​constructed to extract quantitative indicators reflecting changes in spectral features, providing spectrally sensitive quantitative data for subsequent boundary localization, effectively capturing spectral jumps caused by environmental changes. A threshold segmentation algorithm is used to process the quantitative indicators in the difference matrix, marking potential jump points and forming a preliminary set of jump locations. Through temporal fusion with multi-temporal spectral data, the potential jump points are dynamically offset and adjusted to obtain corrected jump features. This not only captures the trend of environmental changes but also effectively addresses changes caused by dynamic environmental factors such as tides and storm surges, ensuring the stability and accuracy of boundary identification. After correcting the abrupt change features, the pixels in the boundary region are grouped using a clustering algorithm to generate a mutation cluster. Critical position coordinates are determined through spatial location calculations, ultimately generating a boundary vector layer and a refined boundary map. This effectively identifies and groups regions exhibiting significant spectral changes, providing strong support for the precise location of mangrove boundaries. By analyzing the refined boundary map, interpolation methods are used to repair missing edge contour segments caused by dynamic environmental changes, ensuring contour continuity and determining spatial connectivity. Finally, a complete mangrove boundary location result is obtained, ensuring accurate identification and precise location of mangrove boundaries even under the influence of environmental factors such as storm surges. This significantly improves the accuracy and robustness of boundary location. Through the synergy of these technical solutions, the problem of traditional methods being unable to cope with dynamic environmental changes is overcome, providing technical support for the precise monitoring and protection of coastal mangroves.

[0081] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0082] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, 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 steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0083] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for intelligent identification of coastal mangroves based on UAV remote sensing, characterized in that, include: Step S1: Collect multi-band spectral data of the coastal mangrove area using remote sensing equipment to obtain a raw dataset containing spectral feature changes. Standardize the raw dataset to obtain a standardized spectral feature dataset. Step S2: Calculate the multi-band difference between adjacent pixels based on the spectral feature dataset, and construct the difference matrix to determine the quantitative index reflecting the change of spectral features; Step S3: Use a threshold segmentation algorithm to process the quantization indexes in the difference matrix, mark potential jump points, and obtain a preliminary set of jump positions; This study integrates a preliminary jump location set with multi-temporal spectral data using a temporal fusion method, adjusts the dynamic offset, and obtains corrected jump features. The process includes: setting a preset threshold based on the quantization index in the difference matrix; marking the corresponding pixel as a potential jump point if the quantization index exceeds the preset threshold; calculating the difference distribution characteristics of the adjacent regions for each potential jump point; comparing the spatial consistency of the difference distribution characteristics using an iterative method; if the difference distribution characteristics meet a preset spatial connectivity condition, incorporating the potential jump point into the preliminary jump location set, which includes the spatial coordinates and spectral feature values ​​of the jump point; obtaining the coordinates of the jump points in the preliminary jump location set; analyzing the impact of tidal periodic disturbances on the jump points based on multi-temporal spectral data; dynamically adjusting the jump point coordinates using a temporal fusion method to generate a set of offset-corrected jump points; and extracting corrected jump features from the offset-corrected jump point set, which includes the spectral features and positional stability of the jump points. Step S4: Based on the corrected jump characteristics, apply a clustering algorithm to group the pixels in the boundary area and determine the critical position coordinates; generate a boundary vector layer by overlaying GIS data with the critical position coordinates to obtain a refined boundary map; this includes: extracting the spectral feature vectors of the pixels in the boundary area based on the corrected jump characteristics; grouping the spectral feature vectors using a clustering algorithm to generate abrupt change clusters; if the abrupt change clusters show significant changes in spectral features, they are confirmed as mangrove boundary transition zones; calculating the critical position coordinates based on the center point of the abrupt change clusters, which include the spatial location information of the boundary area; obtaining the spatial location information from the critical position coordinates; overlaying the spatial location information with topographic data using GIS data; generating a boundary vector layer based on the overlay result; verifying the positioning accuracy of the boundary vector layer for the spectral jump phenomenon under the influence of storm surge, and generating a refined boundary map, which includes the spatial distribution of the mangrove boundary; Step S5: Extract the edge contours based on the refined boundary map, fill the missing segments of the edge contours using interpolation methods, and determine the continuity of the contours to obtain the final mangrove boundary positioning result.

2. The method as described in claim 1, characterized in that, Step S1, obtain a standardized spectral feature dataset, including: Multi-band spectral data of coastal mangrove areas are acquired using satellite remote sensing equipment. The multi-band spectral data includes at least near-infrared, red, and green bands. Time-series sampling is performed based on the dynamic environment under the influence of tidal cycle disturbances and storm surges to determine the sampling time interval. Multi-temporal spectral data are collected according to the sampling time interval to generate a raw dataset containing spectral feature changes. The raw dataset is preprocessed to remove noisy data, resulting in a standardized spectral feature dataset, which includes the spectral features and spatial location information of each pixel.

3. The method as described in claim 1, characterized in that, Step S2, construct the difference matrix to determine the quantitative indicators reflecting changes in spectral characteristics, including: Based on the multi-band spectral values ​​of each pixel and its neighboring pixels in the spectral feature dataset; calculate the difference between the pixel and its neighboring pixels in each band to generate a multi-band difference set; construct a difference matrix based on the multi-band difference set; perform statistical analysis on the difference matrix to extract quantitative indicators that reflect the spectral jump pattern, wherein the quantitative indicators include at least the mean and variance of the differences.

4. The method as described in claim 1, characterized in that, Step S5 yields the final mangrove boundary localization result, including: Based on the refined boundary map, the edge contour of the mangrove boundary is extracted; for the missing contour segments caused by the dynamic environment, an interpolation method is used to generate continuous contour lines; the spatial connectivity of the continuous contour lines is analyzed to determine the contour continuity; if the contour continuity meets the preset conditions, the final mangrove boundary positioning result is generated, and the final mangrove boundary positioning result includes the spatial coordinates of the boundary contour.

5. A coastal mangrove intelligent identification system based on UAV remote sensing, used to implement the method as described in any one of claims 1-4, characterized in that, The system includes: The data acquisition unit is used to acquire multi-band spectral data of the coastal mangrove area through remote sensing equipment, obtain a raw dataset containing spectral feature changes, and standardize the raw dataset to obtain a standardized spectral feature dataset. The difference calculation unit is used to calculate the multi-band difference between adjacent pixels based on the spectral feature dataset, and construct the difference matrix to determine the quantitative index reflecting the change of spectral features; The threshold segmentation unit is used to process the quantization index in the difference matrix through the threshold segmentation algorithm, mark potential jump points, and obtain a preliminary jump position set; the preliminary jump position set is integrated with multi-temporal spectral data through a time series fusion method, and the dynamic offset is adjusted to obtain the corrected jump features; The clustering analysis unit is used to apply a clustering algorithm to group boundary region pixels according to the corrected jump features, determine critical position coordinates, and generate a boundary vector layer by overlaying geographic information system data with the critical position coordinates to obtain a refined boundary map. The contour extraction unit is used to extract edge contours based on the refined boundary map, fill in the missing segments of the edge contours using an interpolation method, and determine the continuity of the contours to obtain the final mangrove boundary positioning result.

6. A computer-readable storage medium storing instructions thereon, characterized in that, When the instructions are executed by the processor, they implement the method as described in any one of claims 1-4.

Images