Mangrove forest carbon sink potential dynamic evaluation method and system based on AI model quantification

Abstract

The invention discloses a mangrove forest carbon sink potential dynamic evaluation method and system based on AI model quantification. The method comprises the steps of obtaining air data of a mangrove forest based on an unmanned aerial vehicle; acquiring ground data of the mangrove forest based on a sensor arranged on the ground; performing integration processing on the air data and the ground data to obtain detection data; and inputting the detection data into a pre-trained AI model for quantitative processing to obtain a mangrove forest carbon sink potential dynamic evaluation result. According to the mangrove forest carbon sink potential dynamic evaluation method, the air data of the mangrove forest and the ground data of the mangrove forest are collected based on the unmanned aerial vehicle and are integrated, so that the detection data are more accurate and comprehensive, and the accuracy of the mangrove forest carbon sink potential dynamic evaluation is improved.

Description

Technical Field

[0001] The present invention relates to the field of evaluation technologies, and particularly to a dynamic evaluation method and system for mangrove carbon sink potential based on AI model quantification. Background Art

[0002] Mangroves are tropical and subtropical thermophilic plants that are beneficial to the improvement of the environmental climate. Currently, when using drones to obtain corresponding detection data for dynamic evaluation of mangrove carbon sink potential, the collection and collation of ground data are lacking, resulting in inaccurate and incomplete final detection data, and thus errors in the dynamic evaluation results of mangrove carbon sink potential. Summary of the Invention

[0003] The present invention aims to solve at least one of the technical problems in the above technologies to some extent. For this purpose, the object of the present invention is to propose a dynamic evaluation method and system for mangrove carbon sink potential based on AI model quantification, which not only collects aerial data of mangroves based on drones, but also collects ground data of mangroves and integrates and processes them, making the detection data more accurate and comprehensive, and thus improving the accuracy of the dynamic evaluation of mangrove carbon sink potential.

[0004] To achieve the above object, an embodiment of the present invention proposes a dynamic evaluation method for mangrove carbon sink potential based on AI model quantification, including:

[0005] Obtaining aerial data of mangroves based on drones;

[0006] Obtaining ground data of mangroves based on sensors set on the ground;

[0007] Integrating and processing the aerial data and the ground data to obtain detection data;

[0008] Inputting the detection data into a pre-trained AI model for quantification processing to obtain a dynamic evaluation result of mangrove carbon sink potential.

[0009] According to some embodiments of the present invention, obtaining aerial data of mangroves based on drones includes:

[0010] Obtaining first image data of mangroves based on a multispectral camera carried by the drone;

[0011] Obtaining radar data of mangroves based on a lidar carried by the drone;

[0012] Determining the aerial data of mangroves according to the first image data and the radar data.

[0013] According to some embodiments of the present invention, obtaining ground data of mangroves based on sensors set on the ground includes:

[0014] Obtain the soil data of the mangrove forest based on the soil analyzer set on the ground;

[0015] Obtain the rainfall data of the mangrove forest based on the rainfall sensing sensor set on the ground;

[0016] Obtain the environmental data of the mangrove forest based on the temperature and humidity sensor set on the ground;

[0017] Obtain the second image data of the mangrove forest based on the camera sensor set on the ground;

[0018] Determine the ground data of the mangrove forest according to the soil data, rainfall data, environmental data and second image data.

[0019] According to some embodiments of the present invention, integrate and process the aerial data and the ground data to obtain the detection data, including:

[0020] Perform image sorting and summarization on the aerial data and the second image data to determine the vegetation biological data and the litter data;

[0021] Determine the detection data according to the vegetation biological data, litter data, soil data, rainfall data and environmental data.

[0022] According to some embodiments of the present invention, obtain the first image data of the mangrove forest based on the multispectral camera carried by the unmanned aerial vehicle, including:

[0023] Obtain two captured images at the same moment based on the multispectral camera carried by the unmanned aerial vehicle;

[0024] Match the two captured images, and after the matching is completed, determine the abnormal area;

[0025] Determine the pixel change amount between the two captured images according to the abnormal area;

[0026] Determine the change angle corresponding to the pixel change amount according to the shooting angle and shooting resolution parameters of the multispectral camera at the current moment;

[0027] Compare the change angle with the preset change angle threshold, and determine the shooting jitter parameter according to the comparison result;

[0028] Query the preset data table according to the jitter parameter to determine the image correction parameter, and perform image correction processing on the two captured images according to the image correction parameter to obtain the first image data.

[0029] According to some embodiments of the present invention, match the two captured images, and after the matching is completed, determine the abnormal area, including:

[0030] Determine the pixel value of each pixel point in each captured image;

[0031] Compare the pixel values of the pixel points at the same position in the two captured images after matching is completed to obtain a pixel difference value;

[0032] Determine the pixel point corresponding to the largest pixel difference value as an abnormal pixel point, and determine a preset range centered on the abnormal pixel point and with a preset radius r as a suspected area;

[0033] Determine whether the suspected area is an abnormal area;

[0034] When it is determined that the suspected area is not an abnormal area, divide the suspected area of each captured image into two sub-areas; perform corresponding comparison according to the two sub-areas of each captured image, and determine the abnormal area according to the comparison result.

[0035] According to some embodiments of the present invention, determining whether the suspected area is an abnormal area includes:

[0036] Calculate the distance information between each pixel point in the suspected area of the captured image and other pixel points in the corresponding neighborhood;

[0037] Calculate the gray difference information between each pixel point in the suspected area of the captured image and other pixel points in the corresponding neighborhood;

[0038] Calculate the ratio of the gray difference information to the distance information as the characteristic value of the pixel point;

[0039] Calculate the characteristic values of all pixel points in the suspected area as the characteristic value of the suspected area;

[0040] Compare the characteristic values of the suspected areas of the two captured images. When it is determined that the difference between the characteristic values of the suspected areas of the two captured images is less than a preset difference value, it indicates that the suspected area is an abnormal area; otherwise, it indicates that the suspected area is not an abnormal area.

[0041] According to some embodiments of the present invention, when it is determined that the suspected area is not an abnormal area, divide the suspected area of each captured image into two sub-areas; perform corresponding comparison according to the two sub-areas of each captured image, and determine the abnormal area according to the comparison result, including:

[0042] Obtain each texture line in the neighborhood of each pixel point in the suspected area of the captured image; obtain the gradient variance of each texture line according to the gradient amplitude of each pixel point on each texture line; obtain the texture line with the smallest gradient variance as the dividing line, and divide the suspected area into a first sub-area and a second sub-area based on the dividing line;

[0043] Calculate the characteristic values of the corresponding sub-areas of the two captured images, and perform comparison, and determine the abnormal area according to the comparison result.

[0044] According to some embodiments of the present invention, determining the pixel change amount between two captured images based on the abnormal area includes:

[0045] Re - matching the abnormal areas of the two captured images through the FAST matching algorithm to determine the pixel change amounts in the horizontal axis direction and the vertical axis direction.

[0046] According to some embodiments of the present invention, an evaluation system applying the above - mentioned dynamic evaluation method for mangrove carbon sink potential quantified based on an AI model includes:

[0047] A first acquisition module, configured to acquire aerial data of the mangrove based on a drone;

[0048] A second acquisition module, configured to acquire ground data of the mangrove based on sensors set on the ground;

[0049] An integration module, configured to integrally process the aerial data and the ground data to obtain detection data;

[0050] A quantification module, configured to input the detection data into a pre - trained AI model for quantification processing to obtain a dynamic evaluation result of the mangrove carbon sink potential.

[0051] The present invention proposes a dynamic evaluation method and system for mangrove carbon sink potential quantified based on an AI model. It not only acquires aerial data of the mangrove based on a drone, but also acquires ground data of the mangrove and integrally processes them, making the detection data more accurate and comprehensive, and further improving the accuracy of the dynamic evaluation of the mangrove carbon sink potential.

[0052] Other features and advantages of the present invention will be described in the following specification, and, in part, will be obvious from the specification, or will be understood by implementing the present invention. The objectives and other advantages of the present invention can be achieved and obtained through the structures specifically pointed out in the written specification and the drawings.

[0053] Next, through the drawings and embodiments, the technical solutions of the present invention will be further described in detail. Description of the Drawings

[0054] The drawings are used to provide a further understanding of the present invention and constitute a part of the specification. They are used together with the embodiments of the present invention to explain the present invention and do not constitute a limitation to the present invention. In the drawings:

[0055] Figure 1 is a flowchart of a dynamic evaluation method for mangrove carbon sink potential quantified based on an AI model according to an embodiment of the present invention;

[0056] Figure 2 is a flowchart of a method for acquiring aerial data according to an embodiment of the present invention;

[0057] Figure 3 This is a block diagram of a dynamic assessment system for mangrove carbon sequestration potential based on AI model quantification according to an embodiment of the present invention. DETAILED DESCRIPTION

[0058] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0059] like Figure 1 As shown, the embodiment of the present invention proposes a dynamic assessment method of mangrove carbon sequestration potential based on AI model quantification, including steps S1-S4:

[0060] S1, obtaining aerial data of mangroves based on drones;

[0061] S2. acquiring ground data of the mangrove forest based on sensors set on the ground;

[0062] S3, integrating and processing the aerial data and the ground data to obtain detection data;

[0063] S4. Input the detection data into the pre-trained AI model for quantitative processing to obtain the dynamic assessment results of the mangrove carbon sequestration potential.

[0064] The working principle of the above technical solution is as follows: The drone's flight route, altitude, and speed are planned based on the distribution range and terrain characteristics of the mangroves to ensure comprehensive coverage of the mangrove area and obtain high-quality image data. Data collection is carried out according to the flight plan, obtaining orthophotos and laser point cloud data of the mangroves. Sensors are strategically deployed within the mangrove area to ensure comprehensive and accurate ground-based monitoring of the mangroves. Data fusion technology is used to fuse aerial data acquired by the drone with ground-based data acquired by ground sensors to form a unified dataset. Features related to the mangrove carbon sequestration potential, such as vegetation index, soil quality, and temperature, are extracted from this fused dataset. These features serve as input data for the AI model. The extracted feature data is standardized to eliminate dimensional differences between different features, improving the training effectiveness and evaluation accuracy of the AI model. AI models such as neural networks and support vector machines (SVMs) are used. The integrated and processed detection data is input into the trained AI model, which then outputs a quantitative assessment of the mangrove carbon sequestration potential based on the input data. The assessment results include information such as the spatial distribution and temporal trends of carbon sequestration potential. The assessment results will be displayed visually, such as generating carbon sink potential distribution maps, time change curves, etc.

[0065] The beneficial effects of the above technical solution include: By combining drone aerial photogrammetry technology with ground-based sensor monitoring technology, multi-source data on mangroves can be obtained from both the air and the ground. Drones can provide high-resolution imagery, visually reflecting information such as the distribution and vegetation cover of mangroves; ground-based sensors can accurately monitor mangrove environmental parameters in real time. After integrating and processing this multi-source data, it is fed into a pre-trained AI model. Leveraging the AI model's powerful data processing and analysis capabilities, a quantitative assessment of the mangroves' carbon sequestration potential is conducted and dynamically updated. The combined use of drones and ground-based sensors enables the acquisition of multi-source, high-precision data on mangroves, providing a rich information base for carbon sequestration potential assessment. By regularly collecting data and feeding it into the AI model for evaluation, dynamic monitoring and assessment of mangrove carbon sequestration potential can be achieved, allowing for timely understanding of changes in mangrove carbon sequestration capacity. The application of the AI model automates data processing and analysis, improving assessment efficiency and accuracy while reducing the impact of human intervention and subjective factors.

[0066] like Figure 2 As shown, according to some embodiments of the present invention, obtaining aerial data of mangroves based on a drone includes steps S11-S13:

[0067] S11, obtaining first image data of the mangrove forest using a multispectral camera carried by the UAV;

[0068] S12, obtaining radar data of mangroves based on the lidar carried by the drone;

[0069] S13. Determine aerial data of the mangrove forest based on the first image data and the radar data.

[0070] The working principle and beneficial effects of the above technical solution are as follows: First, both the image data and the radar data reflect the tree height and crown width data of the mangrove forest, and accurately determine the aerial data of the mangrove forest.

[0071] According to some embodiments of the present invention, obtaining ground data of mangroves based on sensors disposed on the ground includes:

[0072] Obtain soil data of mangroves based on a soil analyzer set up on the ground;

[0073] Rainfall data of mangroves is obtained based on rain sensing sensors installed on the ground;

[0074] Obtaining environmental data of mangroves based on temperature and humidity sensors installed on the ground;

[0075] acquiring second image data of the mangrove forest based on a camera sensor disposed on the ground;

[0076] Determine the ground data of mangroves based on soil data, rainfall data, environmental data, and second image data.

[0077] The working principle and beneficial effects of the above technical solution: The soil analyzer can simultaneously measure parameters such as the soil pH value, electrical conductivity (EC value), organic matter content, nutrient contents such as nitrogen, phosphorus, and potassium, and soil moisture content. These parameters can comprehensively reflect the fertility status and physical and chemical properties of mangrove soil. Soil sampling points are set at certain intervals (such as every 50 - 100 meters) within the mangrove area, and soil analyzers are installed at each sampling point. A rainfall sensing sensor is set at a high or open area within the mangrove area to avoid the influence of tree shading on the accuracy of rainfall data. Temperature and humidity sensors are set at different height levels such as the canopy layer, understory layer, and soil surface of the mangroves to obtain the vertical distribution characteristics of the mangrove environment. At the same time, sensors are also appropriately arranged at the edges and central areas of the mangroves to reflect environmental differences at different locations. Camera sensors are set at key positions within the mangrove area, such as main channels and important ecological areas, to achieve all-round monitoring of the mangroves. At the same time, the angle and focal length of the camera sensor can be adjusted as needed to obtain second image data from different perspectives. Based on the sensors set on the ground, comprehensive, accurate, and real-time ground data of the mangroves are obtained, providing strong data support for the dynamic assessment of the mangrove carbon sink potential.

[0078] According to some embodiments of the present invention, the aerial data and the ground data are integrated and processed to obtain detection data, including:

[0079] Carry out image sorting and summarization on the aerial data and the second image data to determine vegetation biological data and litter data;

[0080] Determine the detection data based on the vegetation biological data, litter data, soil data, rainfall data, and environmental data.

[0081] Working principle of the above technical solution: Image registration technology is adopted to match the multi-spectral and hyperspectral images obtained in the air with the second image obtained by the ground camera sensor in terms of spatial position, ensuring the spatial consistency of image data from different sources. At the same time, based on the combination of radar data and the matched image data, vegetation biological data and litter data are determined. The matched images are classified using image classification algorithms (such as support vector machines, random forests, etc.), and the mangrove vegetation is divided into different types, such as mangrove plants, semi-mangrove plants, etc. The contour and area information of the vegetation are extracted through image segmentation technology, and parameters such as vegetation coverage and canopy density are calculated. The spectral characteristics of the vegetation are extracted, such as the normalized difference vegetation index (NDVI), enhanced vegetation index (EVI), etc., and the growth status and health of the vegetation are evaluated based on the changes in these indices. Combining the classification and growth status information of the vegetation, vegetation biological data are determined, including vegetation type, coverage, canopy density, growth status, etc. The distribution areas of mangrove litter (such as fallen leaves, fallen flowers, fallen fruits, etc.) are identified and extracted. The areas of the extracted litter areas are measured and the quantities are counted, and the distribution density and total amount of the litter are calculated. The characteristics such as the color and morphology of the litter are analyzed to judge the decomposition degree and freshness of the litter. Combining the distribution and status information of the litter, litter data are determined, including litter distribution density, total amount, decomposition degree, etc. Detection data are determined based on vegetation biological data, litter data, soil data, rainfall data and environmental data.

[0082] Beneficial effects of the above technical solution: The air data and ground data are effectively integrated and processed to obtain comprehensive and accurate detection data, providing a scientific basis for the protection, management and ecological research of mangroves.

[0083] According to some embodiments of the present invention, first image data of mangroves are obtained based on a multi-spectral camera carried by a drone, including:

[0084] Two captured images at the same moment are obtained based on a multi-spectral camera carried by a drone;

[0085] The two captured images are matched, and after the matching is completed, abnormal areas are determined;

[0086] According to the abnormal areas, the pixel change amount between the two captured images is determined;

[0087] According to the shooting angle and shooting resolution parameters of the multi-spectral camera at the current moment, the change angle corresponding to the pixel change amount is determined;

[0088] The change angle is compared with a preset change angle threshold, and the shooting jitter parameter is determined according to the comparison result;

[0089] Query a preset data table according to the jitter parameter, determine the image correction parameter, and perform image correction processing on the two captured images according to the image correction parameter to obtain the first image data.

[0090] The working principle of the above technical solution: Two captured images are obtained at the same moment based on the multispectral camera carried by the drone; the two captured images are matched, and after the matching is completed, the abnormal area is determined; the abnormal area is the difference area between the two captured images at the same moment determined due to jitter during the shooting process of the drone. Based on the abnormal area, it is convenient to narrow the data processing range and determine the pixel change amount between the two captured images; the pixel change amount is represented as a two-dimensional vector, including the change amounts in the horizontal and vertical directions. According to the shooting angle and shooting resolution parameters of the multispectral camera at the current moment, determine the change angle corresponding to the pixel change amount; the change angle can reflect the attitude change of the camera during the shooting process. The preset change angle threshold is a preset attitude change that does not affect the shooting too much, and is set according to actual application requirements and experience, and is used to judge whether there is obvious jitter during the shooting process. If the change angle exceeds the preset threshold, it is considered that there is jitter, and the specific value of the shooting jitter parameter is determined according to the magnitude of the change angle. Query the preset data table according to the determined jitter parameter. The data table stores the image correction parameters corresponding to different jitter parameters, and these correction parameters are obtained through experiments or simulation calculations. Obtain the image correction parameters corresponding to the current jitter parameter from the data table, including translation amount, rotation angle, scaling ratio, etc. Perform image correction processing on the two captured images according to the obtained image correction parameters. Use image transformation algorithms (such as affine transformation, perspective transformation, etc.) to perform operations such as translation, rotation, and scaling on the image to eliminate the influence of shooting jitter on the image. After the image correction processing, two corrected images are obtained, and these two images are used as the first image data. The first image data has high accuracy and stability.

[0091] The beneficial effects of the above technical solution: Eliminate the influence of jitter during the shooting process, obtain accurate first image data, and ensure that the obtained image data can truly reflect the actual situation of the mangrove forest.

[0092] According to some embodiments of the present invention, matching the two captured images, and after the matching is completed, determining the abnormal area, including:

[0093] Determine the pixel values of each pixel point in each captured image;

[0094] Compare the pixel values of the pixel points at the same position of the two captured images after the matching is completed to obtain the pixel difference;

[0095] Determine the pixel point corresponding to the largest pixel difference as the abnormal pixel point, and the preset range determined with the abnormal pixel point as the center and the preset radius r as the suspected area;

[0096] Determine whether the suspected area is an abnormal area;

[0097] When it is determined that the suspected area is not an abnormal area, divide the suspected area of each captured image into two sub-areas; perform corresponding comparisons based on the two sub-areas of each captured image, and determine the abnormal area according to the comparison results.

[0098] Working principle of the above technical solution: Determine the pixel values of each pixel point in each captured image; compare the pixel values of the pixel points at the same position of the two captured images after matching to obtain the pixel difference; determine the pixel point corresponding to the largest pixel difference as the abnormal pixel point, and use the preset range determined with the abnormal pixel point as the center and the preset radius r as the suspected area; the size of the preset radius r can be adjusted according to the actual application scenario and image resolution. The suspected area includes the abnormal area. Determine whether the suspected area is an abnormal area; when it is determined that the suspected area is not an abnormal area, divide the suspected area of each captured image into two sub-areas; perform corresponding comparisons based on the two sub-areas of each captured image, and determine the abnormal area according to the comparison results.

[0099] Beneficial effects of the above technical solution: Determine a relatively large range of suspected areas, compare the suspected areas in the two captured images, and determine whether the suspected area is an abnormal area, that is, whether the two are basically the same. When it is determined that the suspected area is not an abnormal area, divide the suspected area of each captured image into two sub-areas for more accurate positioning, which is convenient for accurately determining the abnormal area.

[0100] According to some embodiments of the present invention, determining whether the suspected area is an abnormal area includes:

[0101] Calculate the distance information between each pixel point in the suspected area of the captured image and other pixel points in the corresponding neighborhood;

[0102] Calculate the gray level difference information between each pixel point in the suspected area of the captured image and other pixel points in the corresponding neighborhood;

[0103] Calculate the ratio of the gray level difference information to the distance information as the characteristic value of the pixel point;

[0104] Calculate the characteristic values of all pixel points in the suspected area as the characteristic value of the suspected area;

[0105] Compare the characteristic values of the suspected areas of the two captured images. When it is determined that the difference between the characteristic values of the suspected areas of the two captured images is less than the preset difference, it indicates that the suspected area is an abnormal area; otherwise, it indicates that the suspected area is not an abnormal area.

[0106] Working principle and beneficial effects of the above technical solution: For each pixel point in the suspected area, calculate the distance information and gray-scale difference information between it and other pixel points in the corresponding neighborhood. The distance information can be Euclidean distance information. Calculate the ratio of the gray-scale difference information to the distance information as the eigenvalue of the pixel point. Calculate the eigenvalues of all pixel points in the suspected area as the eigenvalue of the suspected area; compare the eigenvalues of the suspected areas of the two captured images. When it is determined that the difference between the eigenvalues of the suspected areas of the two captured images is less than the preset difference, it indicates that the suspected area is an abnormal area; otherwise, it indicates that the suspected area is not an abnormal area. This facilitates accurately determining whether the suspected area is an abnormal area.

[0107] According to some embodiments of the present invention, when it is determined that the suspected area is not an abnormal area, divide the suspected area of each captured image into two sub-areas; perform corresponding comparisons according to the two sub-areas of each captured image, and determine the abnormal area according to the comparison results, including:

[0108] Obtain each texture line in the neighborhood of each pixel point in the suspected area of the captured image; obtain the gradient variance of each texture line according to the gradient amplitude of each pixel point on each texture line; obtain the texture line with the minimum gradient variance as the dividing line, and divide the suspected area into the first sub-area and the second sub-area based on the dividing line;

[0109] Calculate the eigenvalues of the corresponding sub-areas of the two captured images, and compare them. Determine the abnormal area according to the comparison results.

[0110] Working principle and beneficial effects of the above technical solution: Perform texture analysis on the suspected area to extract each texture line. Calculate the gradient amplitude of each pixel point on each texture line. Calculate the gradient variance of each texture line. Find the texture line with the minimum gradient variance as the dividing line. Use the dividing line to divide the suspected area into two sub-areas. Calculate the eigenvalue of each sub-area. Compare the eigenvalues of the corresponding sub-areas of the two captured images. When comparing the eigenvalues of the corresponding sub-areas of the two captured images, compare them at the corresponding positions. For example, the left side of the first captured image is the first sub-area; the left side of the second captured image is the first sub-area; compare the left side of the first captured image with the left side of the second captured image. Determine the sub-area with a difference value less than the preset difference as the abnormal area. In the actual comparison process, it is possible that the corresponding (sub-areas) have different sizes. Take the sub-area with a larger area as the abnormal area to improve the redundancy and accuracy of the data. Based on the above method, it is convenient to accurately determine the abnormal area.

[0111] According to some embodiments of the present invention, determine the pixel change amount between the two captured images according to the abnormal area, including:

[0112] The abnormal regions of the two captured images are matched again through the FAST matching algorithm to determine the pixel change amounts in the horizontal and vertical directions.

[0113] The working principle and beneficial effects of the above technical solution: Use the FAST algorithm to detect feature points in the abnormal region respectively. Use feature descriptors (such as BRIEF, ORB descriptors) to calculate the similarity of feature points. Screen out reliable matching points through nearest neighbor matching (such as FLANN matcher) or RANSAC algorithm. For each pair of matching points, calculate the coordinate difference between them in the two images. Take the average value of the differences of all matching points to obtain the overall pixel change amount. It is convenient to determine the image changes caused by jitter.

[0114] According to some embodiments of the present invention, according to the shooting angle and shooting resolution parameters of the multispectral camera at the current moment, determine the change angle corresponding to the pixel change amount, including:

[0115] θ1 = d1 / w * θ w

[0116] θ2 = d2 / h * θ h

[0117]

[0118] Where, θ is the change angle corresponding to the pixel change amount; θ1 is the change angle corresponding to the pixel change amount in the horizontal direction; θ2 is the change angle corresponding to the pixel change amount in the vertical direction; d1 is the pixel change amount in the horizontal direction; w is the vertical shooting angle of the multispectral camera at the current moment; θ w is the resolution parameter in the vertical axis; h is the horizontal shooting angle of the multispectral camera at the current moment; θ h is the resolution parameter in the horizontal axis.

[0119] The working principle and beneficial effects of the above technical solution: According to the shooting angle and shooting resolution parameters of the multispectral camera at the current moment, the pixel change amount can be accurately converted into the actual change angle.

[0120] Such as Figure 3 shown, according to some embodiments of the present invention, an evaluation system applying the above-mentioned dynamic evaluation method for mangrove carbon sink potential quantified based on an AI model includes:

[0121] The first acquisition module is used to acquire aerial data of mangroves based on a drone;

[0122] The second acquisition module is used to acquire ground data of mangroves based on sensors arranged on the ground;

[0123] The integration module is used to integrally process the aerial data and the ground data to obtain detection data;

[0124] A quantization module for inputting detection data into a pre-trained AI model for quantization processing to obtain a dynamic assessment result of the mangrove carbon sink potential.

[0125] Advantages of the above technical solution: By comprehensively applying the drone aerial photogrammetry technology and the ground sensor monitoring technology, multi-source data of mangroves are obtained from two dimensions, namely, the air and the ground. The drone can provide high-resolution image data, intuitively reflecting information such as the distribution range and vegetation coverage of mangroves; the ground sensors can monitor the mangrove environmental parameters in real time and accurately. After integrating and processing these multi-source data, they are input into a pre-trained AI model. Using the powerful data processing and analysis capabilities of the AI model, the mangrove carbon sink potential is quantitatively evaluated and dynamically updated. The combined use of drones and ground sensors can obtain multi-source and high-precision data of mangroves, providing a rich information basis for the assessment of carbon sink potential. By regularly collecting data and inputting them into the AI model for evaluation, the dynamic monitoring and assessment of the mangrove carbon sink potential can be realized, and the changes in the mangrove carbon sink capacity can be grasped in a timely manner. The application of the AI model realizes the automated processing and analysis of data, improves the evaluation efficiency and accuracy, and reduces the influence of manual intervention and subjective factors.

[0126] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalent technologies, the present invention also intends to include these changes and modifications.

Claims

1. A dynamic assessment method for the carbon sequestration potential of mangroves based on the quantization of an AI model, characterized in that, Including: Obtaining aerial data of mangroves based on drones; Obtaining ground data of mangroves based on sensors set on the ground; Integrating and processing the aerial data and the ground data to obtain detection data; Inputting the detection data into a pre-trained AI model for quantization processing to obtain the dynamic evaluation result of the mangrove carbon sink potential.

2. The dynamic assessment method for mangrove carbon sink potential based on AI model quantization according to claim 1, wherein Obtaining aerial data of mangroves based on drones, including: Obtaining first image data of mangroves based on a multispectral camera carried by a drone; Obtaining lidar data of mangroves based on a lidar carried by a drone; Determining the aerial data of mangroves according to the first image data and the lidar data.

3. The dynamic assessment method for the mangrove carbon sink potential based on AI model quantization according to claim 2, characterized in that Obtaining ground data of mangroves based on sensors set on the ground, including: Obtaining soil data of mangroves based on a soil analyzer set on the ground; Obtaining rainfall data of mangroves based on a rainfall sensing sensor set on the ground; Obtaining environmental data of mangroves based on a temperature and humidity sensor set on the ground; Obtaining second image data of mangroves based on a camera sensor set on the ground; Determining the ground data of mangroves according to the soil data, the rainfall data, the environmental data and the second image data.

4. The dynamic assessment method for the mangrove carbon sink potential based on the quantization of the AI model according to claim 3, characterized in that Integrating and processing the aerial data and the ground data to obtain detection data, including: Collating and summarizing the aerial data and the second image data to determine vegetation biological data and litter data; Determining the detection data according to the vegetation biological data, the litter data, the soil data, the rainfall data and the environmental data.

5. The dynamic assessment method for the mangrove carbon sink potential based on the quantization of the AI model according to claim 2, wherein Obtaining first image data of mangroves based on a multispectral camera carried by a drone, including: Obtaining two captured images at the same moment based on a multispectral camera carried by a drone; Matching the two captured images, and after the matching is completed, determining the abnormal area; Determining the pixel change amount between the two captured images according to the abnormal area; Determining the change angle corresponding to the pixel change amount according to the shooting angle and the shooting resolution parameters of the multispectral camera at the current moment; Comparing the change angle with a preset change angle threshold, and determining the shooting jitter parameter according to the comparison result; Querying a preset data table according to the jitter parameter to determine an image correction parameter, and performing image correction processing on the two captured images according to the image correction parameter to obtain the first image data.

6. The dynamic assessment method for mangrove carbon sink potential based on AI model quantization as described in claim 5, characterized in that, Matching the two captured images, and after the matching is completed, determining the abnormal area, including: Determining the pixel value of each pixel point in each captured image; Comparing the pixel values of the pixel points at the same position of the two captured images after the matching is completed to obtain a pixel difference; Determining the pixel point corresponding to the maximum pixel difference as the abnormal pixel point, and taking the preset range determined with the abnormal pixel point as the center and a preset radius r as the suspected area; Judging whether the suspected area is an abnormal area; When it is determined that the suspected area is not an abnormal area, dividing the suspected area of each captured image into two sub-areas; making corresponding comparisons according to the two sub-areas of each captured image, and determining the abnormal area according to the comparison result.

7. The dynamic assessment method for mangrove carbon sink potential based on AI model quantization according to claim 6, wherein Judging whether the suspected area is an abnormal area, including: Calculating the distance information between each pixel point in the suspected area of the captured image and other pixel points in the corresponding neighborhood; Calculate the grayscale difference information between each pixel point in the suspected area of the captured image and other pixel points in the corresponding neighborhood; Calculate the ratio of the grayscale difference information to the distance information as the eigenvalue of the pixel point; Calculate the eigenvalues of all pixel points in the suspected area as the eigenvalue of the suspected area; Compare the eigenvalues of the suspected areas of the two captured images. When it is determined that the difference between the eigenvalues of the suspected areas of the two captured images is less than the preset difference, it indicates that the suspected area is an abnormal area; otherwise, it indicates that the suspected area is not an abnormal area.

8. The dynamic evaluation method for the mangrove carbon sink potential based on the quantization of the AI model according to claim 7, wherein When it is determined that the suspected area is not an abnormal area, divide the suspected area of each captured image into two sub-areas; perform corresponding comparisons based on the two sub-areas of each captured image, and determine the abnormal area according to the comparison results, including: Obtain each texture line in the neighborhood of each pixel point in the suspected area of the captured image; obtain the gradient variance of each texture line based on the gradient amplitude of each pixel point on each texture line; obtain the texture line with the minimum gradient variance as the segmentation line, and divide the suspected area into the first sub-area and the second sub-area based on the segmentation line; Calculate the eigenvalues of the corresponding sub-areas of the two captured images, compare them, and determine the abnormal area according to the comparison results.

9. The dynamic assessment method for mangrove carbon sink potential based on AI model quantization according to claim 5, characterized in that, Determine the pixel change amount between the two captured images according to the abnormal area, including: Match the abnormal areas of the two captured images again through the FAST matching algorithm to determine the pixel change amounts in the horizontal and vertical directions.

10. An evaluation system applying the dynamic evaluation method for mangrove carbon sink potential quantified based on the AI model as described in any one of claims 1-9, characterized in that, Including: The first acquisition module is used to acquire the aerial data of the mangrove forest based on the unmanned aerial vehicle; The second acquisition module is used to acquire the ground data of the mangrove forest based on the sensors set on the ground; The integration module is used to integrate and process the aerial data and the ground data to obtain the detection data; The quantization module is used to input the detection data into a pre-trained AI model for quantization processing to obtain the dynamic evaluation result of the mangrove forest carbon sink potential.

Images