Mangrove inter-specific classification method and device based on unmanned aerial vehicle remote sensing

Abstract

The present application relates to the field of remote sensing monitoring, in particular to a mangrove inter-specific classification method and device based on unmanned aerial vehicle remote sensing, adopting a decision tree learning algorithm to update feature threshold values in a feature corresponding relationship of a plurality of mangrove plant categories in an initial multi-modal plant knowledge graph, obtaining an updated multi-modal plant knowledge graph, performing inter-specific classification reasoning according to a plurality of types of feature data in the extracted feature data set of the region to be classified, feature values corresponding to the feature data, and the updated multi-modal plant knowledge graph, obtaining a mangrove inter-specific classification result, which integrates classification knowledge, overcomes the shortcomings of a method that simply relies on data, and improves the accuracy, efficiency and reliability of mangrove inter-specific classification.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of remote sensing monitoring, and in particular to a mangrove inter-specific classification method and device based on unmanned aerial vehicle remote sensing, a computer device and a storage medium. BACKGROUND

[0002] Mangrove plants are the basis of mangrove plant ecosystems, and the species composition, community structure and diversity of mangrove plant communities have an important influence on the productivity and stability of the ecosystem, determining the energy flow and material circulation processes of the mangrove plant ecosystem. Therefore, efficient and accurate monitoring of the spatial distribution and dynamics of mangrove plant species is also an important challenge and key scientific issue in the field of mangrove plant remote sensing. However, mangrove plants have unevenness and complexity in spatial distribution, resulting in great uncertainty in current mangrove plant remote sensing inter-specific classification. SUMMARY

[0003] Therefore, the purpose of the present application is to provide a mangrove inter-specific classification method and device based on unmanned aerial vehicle remote sensing, a computer device and a storage medium, which updates the feature threshold values in the feature correspondence relationship of a plurality of mangrove plant categories in an initial multi-modal plant knowledge graph using a decision tree learning algorithm, obtains an updated multi-modal plant knowledge graph, and performs inter-specific classification reasoning based on the feature data of a plurality of types in the extracted feature data set of the region to be classified, the feature values corresponding to the feature data, and the updated multi-modal plant knowledge graph, to obtain a mangrove inter-specific classification result. The method overcomes the shortcomings of methods that rely solely on data and improves the accuracy, efficiency and reliability of mangrove inter-specific classification.

[0004] In a first aspect, the present application provides a mangrove inter-specific classification method based on unmanned aerial vehicle remote sensing, comprising the following steps:

[0005] obtaining a plurality of mangrove plant category multi-source remote sensing sample data sets and an initial multi-modal plant knowledge graph, wherein the initial multi-modal plant knowledge graph includes a plurality of mangrove plant category feature correspondence relationships, and the feature correspondence relationship includes a plurality of types of feature data and corresponding feature threshold values;

[0006] extracting mangrove plant features based on unmanned aerial vehicle remote sensing data and elevation data in the multi-source remote sensing sample data set to obtain a plurality of mangrove plant category sample feature data sets, wherein the sample feature data set includes a plurality of types of feature data;

[0007] According to the sample feature data set of several mangrove plant categories, the feature threshold corresponding to the feature data in the feature corresponding relationship of the several mangrove plant categories in the initial multi-modal plant knowledge graph is updated by using a decision tree learning algorithm, and an updated multi-modal plant knowledge graph is obtained.

[0008] Obtain multi-source remote sensing data of the region to be classified, and perform feature extraction according to the unmanned aerial vehicle remote sensing data and the elevation data in the multi-source remote sensing data to obtain a feature data set of the region to be classified.

[0009] According to the feature data of several types in the feature data set of the region to be classified, the feature values corresponding to the feature data, and the feature data of several types in the feature corresponding relationship of the several mangrove plant categories in the updated multi-modal plant knowledge graph, the feature threshold corresponding to the feature data is inferred between species, and a mangrove forest interspecific classification result of the region to be classified is obtained.

[0010] In a second aspect, an embodiment of the present application provides a mangrove forest interspecific classification device based on unmanned aerial vehicle remote sensing, comprising:

[0011] A data obtaining module is configured to obtain a multi-source remote sensing sample data set of several mangrove plant categories and an initial multi-modal plant knowledge graph, wherein the initial multi-modal plant knowledge graph comprises a feature corresponding relationship of several mangrove plant categories, and the feature corresponding relationship comprises feature data of several types and corresponding feature thresholds.

[0012] A first feature extraction module is configured to perform mangrove plant feature extraction according to unmanned aerial vehicle remote sensing data and elevation data in the multi-source remote sensing sample data set to obtain a sample feature data set of several mangrove plant categories, wherein the sample feature data set comprises feature data of several types.

[0013] A feature threshold updating module is configured to update the feature threshold corresponding to the feature data in the feature corresponding relationship of the several mangrove plant categories in the initial multi-modal plant knowledge graph according to the sample feature data set of the several mangrove plant categories by using a decision tree learning algorithm, and obtain an updated multi-modal plant knowledge graph.

[0014] A second feature extraction module is configured to obtain multi-source remote sensing data of the region to be classified, and perform feature extraction according to unmanned aerial vehicle remote sensing data and elevation data in the multi-source remote sensing data to obtain a feature data set of the region to be classified.

[0015] The inter-specific classification result module is configured to perform inter-specific classification reasoning according to the feature data of the plurality of types in the feature data set of the region to be classified, the feature values corresponding to the feature data, and the feature data of the plurality of types in the feature corresponding relationship of the plurality of mangrove plant categories in the updated multi-modal plant knowledge graph and the feature threshold values corresponding to the feature data, to obtain the inter-specific classification result of the mangrove forest of the region to be classified.

[0016] In a third aspect, an embodiment of the present application provides a computer device, comprising a processor, a memory, and a computer program stored in the memory and executable on the processor; when the computer program is executed by the processor, the steps of the method for inter-specific classification of mangrove forests based on unmanned aerial vehicle remote sensing are implemented.

[0017] In a fourth aspect, an embodiment of the present application provides a storage medium, which stores a computer program, and when the computer program is executed by a processor, the steps of the method for inter-specific classification of mangrove forests based on unmanned aerial vehicle remote sensing are implemented.

[0018] In the embodiments of the present application, a method, device, computer device and storage medium for inter-specific classification of mangrove forests based on unmanned aerial vehicle remote sensing are provided, a decision tree learning algorithm is used to update the feature threshold values in the feature corresponding relationship of the plurality of mangrove plant categories in the initial multi-modal plant knowledge graph, an updated multi-modal plant knowledge graph is obtained, inter-specific classification reasoning is performed according to the feature data of the plurality of types in the feature data set of the region to be classified, the feature values corresponding to the feature data, and the updated multi-modal plant knowledge graph, and the inter-specific classification result of the mangrove forest is obtained. The method combines classification knowledge, overcomes the shortcomings of a method that simply relies on data, and improves the accuracy, efficiency and reliability of inter-specific classification of mangrove forests.

[0019] For better understanding and implementation, the present application is described in detail below with reference to the accompanying drawings. BRIEF DESCRIPTION OF DRAWINGS

[0020] Figure 1 A flowchart of the method for inter-specific classification of mangrove forests based on unmanned aerial vehicle remote sensing provided by an embodiment of the present application is shown;

[0021] Figure 2 A flowchart of S3 in the method for inter-specific classification of mangrove forests based on unmanned aerial vehicle remote sensing provided by an embodiment of the present application is shown;

[0022] Figure 3 A flowchart of S31 in the method for inter-specific classification of mangrove forests based on unmanned aerial vehicle remote sensing provided by an embodiment of the present application is shown;

[0023] Figure 4A flowchart of S32 in a mangrove inter-specific classification method based on unmanned aerial vehicle remote sensing provided by an embodiment of the present application is shown in FIG. 6;

[0024] Figure 5 A flowchart of S4 in a mangrove inter-specific classification method based on unmanned aerial vehicle remote sensing provided by an embodiment of the present application is shown in FIG. 3;

[0025] Figure 6 A flowchart of S5 in a mangrove inter-specific classification method based on unmanned aerial vehicle remote sensing provided by an embodiment of the present application is shown in FIG. 4;

[0026] Figure 7 A structural diagram of a mangrove inter-specific classification device based on unmanned aerial vehicle remote sensing provided by an embodiment of the present application is shown in FIG. 5;

[0027] Figure 8 A structural diagram of a computer device provided by an embodiment of the present application is shown in FIG. 7. DETAILED DESCRIPTION

[0028] The exemplary embodiments will be described in detail herein with reference to the attached drawings. The description of the exemplary embodiments is intended to apply to various alternative embodiments as well. The following description is not limited to the exemplary embodiments, but rather, is applicable to any embodiment within the scope of the present application. In addition, those skilled in the art will appreciate that the exemplary embodiments described herein can be practiced with a variety of computer-system configurations, including personal computers, server computers, hand-held computing devices, multi-core computing devices, microprocessor-based or programmable-consumer electronics, network PCs, minicomputers, mainframe computers, computing kiosks, tablet computers, cell phones, wearable devices, gaming devices, and the like.

[0029] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the present application. As used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms "comprises" and / or "comprising," when used in this specification, specify the presence of stated features, integers, steps, operations, elements, and / or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof.

[0030] It will be understood that, although the terms first, second, third, etc. can be used herein to describe various information, these terms are not intended to denote a temporal or chronological order. Rather, these terms are used only to distinguish one from another. For example, a first information can be termed a second information, and, similarly, a second information can be termed a first information, without departing from the scope of the present application. As used herein, the word "if' can be interpreted to mean "when" or "upon" or "in response to determining" taking into account the context in which the term is used.

[0031] The data sending end can be a computer device or a mobile terminal device, which is used to establish a network connection with the data receiving end, can encode data information sent to the data receiving end, and can analyze data information sent from the data receiving end.

[0032] The data receiving end can be a computer device or a mobile terminal device, which is used to establish a network connection with the data sending end, can encode data information sent to the data sending end, and can analyze data information sent from the data sending end.

[0033] Please refer to Figure 1 , Figure 1 A flowchart of a mangrove inter-specific classification method based on unmanned aerial vehicle remote sensing is provided for an embodiment of the application, and the method comprises the following steps:

[0034] S1: Obtain a plurality of multi-source remote sensing sample data sets of mangrove plant categories and an initial multi-modal plant knowledge graph.

[0035] The execution subject of the mangrove inter-specific classification method based on unmanned aerial vehicle remote sensing is a classification device (hereinafter referred to as a classification device) of the mangrove inter-specific classification method based on unmanned aerial vehicle remote sensing. In an optional embodiment, the classification device can be a computer device, which can be a server or a server cluster formed by a plurality of computer devices.

[0036] In this embodiment, the classification device obtains a plurality of multi-source remote sensing sample data sets of mangrove plant categories, wherein the multi-source remote sensing sample data sets comprise unmanned aerial vehicle remote sensing data and elevation data. The unmanned aerial vehicle remote sensing data is a plurality of different types of remote sensing data collected by different sensors carried by unmanned aerial vehicles, including multispectral, LiDAR, hyperspectral, video, etc. The elevation data is a digital simulation of the ground terrain realized by limited terrain elevation data.

[0037] The classification device obtains an initial multi-modal plant knowledge graph, wherein the initial multi-modal plant knowledge graph comprises a plurality of feature correspondence relationships of mangrove plant categories, and the feature correspondence relationships comprise a plurality of types of feature data and corresponding feature threshold values.

[0038] In an optional embodiment, the classification device can obtain interspecific classification knowledge data including natural language description of expert experience from a preset database, the interspecific classification knowledge data including classification information, morphological characteristics, habitat characteristics and spatial distribution characteristics of all mangrove plant categories obtained from a flora and literature, the morphological characteristics including tree height, crown width, life form (shrub or tree, etc.), the habitat characteristics including growth tide level, cold tolerance ability (the coldest monthly mean temperature that can be tolerated), etc., and the spatial distribution characteristics referring to the place names of the present and historical distribution areas of the species.

[0039] The classification device converts the interspecific classification knowledge data into rules described by the above characteristics and corresponding characteristic thresholds, and expresses the rules according to the syntax of the Semantic Web Rule Language (SWRL), to obtain a plurality of feature corresponding relationships of mangrove plant categories, and construct the initial multi-modal plant knowledge graph.

[0040] S2: Extracting mangrove plant features according to the unmanned aerial vehicle remote sensing data and elevation data in the multi-source remote sensing sample data set to obtain a plurality of sample feature data sets of mangrove plant categories.

[0041] In this embodiment, the classification device extracts features according to the unmanned aerial vehicle remote sensing data and elevation data in the multi-source remote sensing sample data set of a plurality of mangrove plant categories to obtain a plurality of sample feature data sets of mangrove plant categories, wherein the sample feature data set includes a plurality of types of feature data.

[0042] Specifically, the classification device uses OTB (Orfeo Toolbox), which is an open-source high-performance image processing system for the field of remote sensing, can process high-resolution optical, multi-spectral and radar images, and the classification device uses the feature extraction algorithm in OTB to perform feature extraction on the unmanned aerial vehicle remote sensing data and the elevation data, obtain feature data of several types of several mangrove plant categories, and use fragtats software to calculate the connectivity features of the objects. Each feature extraction result is saved as an independent layer for subsequent use. The feature data includes several feature vectors, the feature data includes spectral feature data, texture feature data, shape feature data and height feature data, the spectral feature data includes hue feature vector, saturation feature vector, brightness feature vector, spectral mean feature vector and spectral standard deviation feature vector; the texture feature data includes homogeneity feature vector, contrast feature vector, dissimilarity feature vector, entropy feature vector, angular second moment feature vector, texture mean feature vector, texture standard deviation feature vector and autocorrelation feature vector; the shape feature data includes asymmetry feature vector, border index feature vector, compactness feature vector, density feature vector, ellipse fitting degree feature vector, main direction feature vector, maximum closed ellipse radius feature vector, minimum closed ellipse radius feature vector, rectangular fitting degree feature vector, circularity feature vector and shape index feature vector; and the height feature data includes elevation standard deviation feature vector, elevation variance feature vector, elevation feature vector, elevation mean absolute deviation feature vector, elevation L2 feature vector, crown roughness rate feature vector and crown height model feature vector.

[0043] In an optional embodiment, before the classification device performs feature extraction on the unmanned aerial vehicle remote sensing data in the multi-source remote sensing sample data set of several mangrove plant categories, the classification device pre-processes the unmanned aerial vehicle remote sensing data, including image stitching, data cropping, geometric correction, radiation correction, point cloud denoising and filtering, etc., to improve the accuracy of feature extraction.

[0044] S3: According to the sample feature data set of several mangrove plant categories, a decision tree learning algorithm is used to update the feature threshold corresponding to the feature data in the feature corresponding relationship of the several mangrove plant categories in the initial multi-modal plant knowledge graph, and an updated multi-modal plant knowledge graph is obtained.

[0045] The decision tree learning algorithm uses the CART (Classification and Regression Trees) algorithm based on classification and regression trees. The CART algorithm is called a machine learning or expert system and provides a non-parametric discriminant to determine the statistical relationship between multiple data layers to produce a binary decision tree.

[0046] In the embodiment, the classification device updates the feature threshold corresponding to the feature data in the feature corresponding relationship of the plurality of mangrove plant categories in the initial multi-modal plant knowledge graph according to the sample feature data set of the plurality of mangrove plant categories, and obtains an updated multi-modal plant knowledge graph.

[0047] Referring to Figure 2 , Figure 2 The flowchart of S3 in the mangrove inter-specific classification method based on unmanned aerial vehicle remote sensing provided by an embodiment of the present application includes steps S31-S32, and specifically as follows.

[0048] S31: According to the sample feature data set of the plurality of mangrove plant categories, a plurality of feature vectors are extracted from the sample feature data set as training features of the plurality of mangrove plant categories. According to the sample feature data set of the plurality of mangrove plant categories and the training features, a classification tree of the plurality of mangrove plant categories is constructed by using a Gini index minimization criterion method.

[0049] In the embodiment, the classification device extracts a plurality of feature vectors from the sample feature data set of the plurality of mangrove plant categories as training features of the plurality of mangrove plant categories. Specifically, the classification device extracts a plurality of feature data from the sample feature data set of the plurality of mangrove plant categories in a sampling manner with replacement, and obtains a plurality of feature vectors from the feature data as training features of the plurality of mangrove plant categories in a sampling manner with replacement.

[0050] The classification device constructs a classification tree of the plurality of mangrove plant categories by using a Gini index minimization criterion method according to the sample feature data set of the plurality of mangrove plant categories and the training features.

[0051] Referring to Figure 3 , Figure 3 The flowchart of S31 in the mangrove inter-specific classification method based on unmanned aerial vehicle remote sensing provided by an embodiment of the present application includes steps S311-S312, and specifically as follows.

[0052] S311: According to the plurality of training features, the sample feature data set is respectively binary classified to obtain a first sub-sample feature data set and a second sub-sample feature data set of the plurality of training features. According to a preset Gini index algorithm, the Gini index of the binary classification of the sample feature data set under any feature value is calculated, and the training feature with the minimum Gini index and the corresponding feature value are taken as the optimal split point of the current round.

[0053] In this embodiment, the classification device respectively classifies the sample feature data set according to a plurality of training features, and obtains a first sub-sample feature data set and a second sub-sample feature data set of the plurality of training features.

[0054] For the training features and the corresponding first sub-sample feature data set and the second sub-sample feature data set, the classification device calculates the Gini index of the sample feature data set after being classified according to a preset Gini index algorithm at any feature value, and the classification device takes the training feature with the minimum Gini index and the corresponding feature value as the optimal split point of the current round, wherein the Gini index algorithm is:

[0055]

[0056] In the formula, Gini represents the Gini index, Z is the training feature, T is the sample feature data set, T1 is the first sub-sample feature data set, T2 is the second sub-sample feature data set, K represents the number of classification categories, p i represents the probability of belonging to the i-th category.

[0057] S312: The first sub-sample feature data set and the second sub-sample feature data set of the training feature with the minimum Gini index are respectively taken as the sample feature data sets of the left and right nodes, and the optimal split point acquisition of the next round is performed until all nodes no longer satisfy the node re-partition condition, and a plurality of mangrove plant classification trees are obtained.

[0058] In this embodiment, the classification device respectively classifies the sample feature data set according to a plurality of training features, and obtains a first sub-sample feature data set and a second sub-sample feature data set of the plurality of training features.

[0059] S32: According to the plurality of mangrove plant classification trees, a plurality of target feature vectors and corresponding target feature thresholds of a plurality of mangrove plant categories are obtained, the feature threshold corresponding to the feature vector in the initial feature correspondence relationship of the corresponding mangrove plant category in the initial multi-modal plant knowledge graph is updated, and an updated multi-modal plant knowledge graph is obtained.

[0060] In this embodiment, the classification device respectively classifies the sample feature data set according to a plurality of training features, and obtains a first sub-sample feature data set and a second sub-sample feature data set of the plurality of training features.

[0061] Please refer toFigure 4 , Figure 4 A flowchart of S32 in the mangrove inter-specific classification method based on unmanned aerial vehicle remote sensing provided by an embodiment of the present application is shown in FIG. 32. The method comprises the following steps S321.

[0062] S321: According to the classification tree of the plurality of mangrove plant categories, the training features corresponding to the optimal split points of all rounds of the classification tree of the plurality of mangrove plant categories and the feature values are obtained. The training features are taken as the target feature vectors, and the feature values are taken as the target feature thresholds. A plurality of target feature vectors of the plurality of mangrove plant categories and the corresponding target feature thresholds are obtained.

[0063] In this embodiment, the classification device obtains the training features corresponding to the optimal split points of all rounds of the classification tree of the plurality of mangrove plant categories and the feature values according to the classification tree of the plurality of mangrove plant categories. The training features are taken as the target feature vectors, and the feature values are taken as the target feature thresholds. A plurality of target feature vectors of the plurality of mangrove plant categories and the corresponding target feature thresholds are obtained.

[0064] S4: Obtain multi-source remote sensing data of the region to be classified. Feature extraction is performed according to the unmanned aerial vehicle remote sensing data and the elevation data in the multi-source remote sensing data to obtain a feature data set of the region to be classified.

[0065] In this embodiment, the classification device obtains multi-source remote sensing data of the region to be classified. Feature extraction is performed according to the unmanned aerial vehicle remote sensing data and the elevation data in the multi-source remote sensing data to obtain a feature data set of the region to be classified.

[0066] Please refer to Figure 5 , Figure 5 A flowchart of S4 in the mangrove inter-specific classification method based on unmanned aerial vehicle remote sensing provided by an embodiment of the present application is shown in FIG. 41. The method comprises the following steps S41.

[0067] S41: Obtain a mangrove vegetation index of the region to be classified. According to the mangrove vegetation index, the mangrove region in the region to be classified is confirmed. Feature extraction is performed according to the unmanned aerial vehicle remote sensing data and the elevation data in the multi-source remote sensing data corresponding to the mangrove region in the region to be classified to obtain a feature data set of the region to be classified.

[0068] In this embodiment, the classification device obtains a mangrove vegetation index of the region to be classified. According to the mangrove vegetation index, the mangrove region in the region to be classified is confirmed to improve the accuracy of feature extraction and the efficiency of reasoning.

[0069] The classification device extracts features from the unmanned aerial vehicle remote sensing data and the elevation data in the multi-source remote sensing data corresponding to the mangrove forest region in the region to be classified, to obtain a feature data set of the region to be classified. For details, refer to step S2, which will not be described here.

[0070] S5: According to the feature data of several types in the feature data set of the region to be classified, the feature values corresponding to the feature data, and the feature threshold values of several types of feature data in the updated feature corresponding relationship of several mangrove plant categories, inter-specific classification reasoning is performed to obtain the inter-specific classification result of the mangrove forest in the region to be classified.

[0071] In this embodiment, the classification device performs inter-specific classification reasoning according to the feature data of several types in the feature data set of the region to be classified, the feature values corresponding to the feature data, and the feature threshold values of several types of feature data in the updated feature corresponding relationship of several mangrove plant categories, to obtain the inter-specific classification result of the mangrove forest in the region to be classified.

[0072] The feature threshold values in the feature corresponding relationship of several mangrove plant categories in the initial multi-modal plant knowledge graph are updated using a decision tree learning algorithm to obtain an updated multi-modal plant knowledge graph. According to the extracted feature data of several types in the feature data set of the region to be classified, the feature values corresponding to the feature data, and the updated multi-modal plant knowledge graph, inter-specific classification reasoning is performed to obtain the inter-specific classification result of the mangrove forest, which integrates classification knowledge, overcomes the shortcomings of methods that simply rely on data, and improves the accuracy, efficiency and reliability of inter-specific classification of mangrove forests.

[0073] Please refer to Figure 6 , Figure 6 The flowchart of S5 in the method for inter-specific classification of mangrove forests based on unmanned aerial vehicle remote sensing provided by an embodiment of the present application is shown in FIG. 5, which includes steps S51-S52, as follows:

[0074] S51: Determine several candidate feature corresponding relationships of the region to be classified from the updated multi-modal plant knowledge graph, and determine the target feature corresponding relationship of the mangrove plant category that has a corresponding feature vector according to all feature vectors in all types of feature data in the feature data set of the region to be classified and the several candidate feature corresponding relationships;

[0075] In this embodiment, the classification device determines the plurality of candidate feature correspondence relationships of the to-be-classified region from the updated multi-modal plant knowledge graph. Specifically, the classification device can obtain geographic information of the mangrove region of the to-be-classified region, the geographic information indicating possible mangrove plant categories in the mangrove region, and determine the plurality of candidate feature correspondence relationships of the to-be-classified region according to the geographic information and spatial distribution features corresponding to the plurality of feature correspondence relationships of the mangrove plant categories in the updated multi-modal plant knowledge graph.

[0076] The classification device determines the target feature correspondence relationship of the mangrove plant category whose corresponding feature vector exists at the same time according to all feature vectors in all types of feature data in the feature data set of the to-be-classified region and the plurality of candidate feature correspondence relationships, so as to perform interspecific classification, thereby reducing the search space and the probability of misclassification and omission, and improving the reasoning efficiency and accuracy.

[0077] S52: According to the feature values corresponding to all feature vectors in all types of feature data in the feature data set of the to-be-classified region, if the feature values all satisfy the feature threshold values corresponding to the corresponding feature vectors in the target feature correspondence relationship, the mangrove plant category corresponding to the target feature correspondence relationship is taken as the mangrove interspecific classification result, and the mangrove interspecific classification result of the to-be-classified region is obtained.

[0078] In this embodiment, the classification device determines the target feature correspondence relationship of the mangrove plant category whose corresponding feature vector exists at the same time according to all feature vectors in all types of feature data in the feature data set of the to-be-classified region and the plurality of candidate feature correspondence relationships, so as to perform interspecific classification, thereby reducing the search space and the probability of misclassification and omission, and improving the reasoning efficiency and accuracy.

[0079] For example, Figure 7 , Figure 7 The structure diagram of the mangrove interspecific classification device based on unmanned aerial vehicle remote sensing provided by an embodiment of the present application, which can realize all or part of the mangrove interspecific classification device based on unmanned aerial vehicle remote sensing through software, hardware or combination of both. The device 7 comprises:

[0080] The data obtaining module 71 is configured to obtain a plurality of multi-source remote sensing sample data sets of mangrove plant categories and an initial multi-modal plant knowledge graph, wherein the initial multi-modal plant knowledge graph comprises a plurality of feature correspondence relationships of mangrove plant categories, and the feature correspondence relationship comprises a plurality of types of feature data and corresponding feature threshold values.

[0081] The first feature extraction module 72 is configured to perform mangrove feature extraction according to the unmanned aerial vehicle remote sensing data and the elevation data in the multi-source remote sensing sample data set, and obtain a plurality of sample feature data sets of mangrove categories. The sample feature data set includes a plurality of types of feature data.

[0082] The feature threshold updating module 73 is configured to update the feature threshold corresponding to the feature data in the feature corresponding relationship of the plurality of mangrove categories in the initial multi-modal plant knowledge graph by using a decision tree learning algorithm according to the sample feature data set of the plurality of mangrove categories, and obtain an updated multi-modal plant knowledge graph.

[0083] The second feature extraction module 74 is configured to obtain multi-source remote sensing data of a region to be classified, and perform feature extraction according to the unmanned aerial vehicle remote sensing data and the elevation data in the multi-source remote sensing data, and obtain a feature data set of the region to be classified.

[0084] The inter-specific classification result module 75 is configured to perform inter-specific classification reasoning according to the plurality of types of feature data in the feature data set of the region to be classified, the feature values corresponding to the feature data, and the plurality of types of feature data in the feature corresponding relationship of the plurality of mangrove categories in the updated multi-modal plant knowledge graph, and the feature thresholds corresponding to the feature data, and obtain an inter-specific classification result of the mangrove forest in the region to be classified.

[0085] In the embodiment, a plurality of mangrove plant category multi-source remote sensing sample data sets and an initial multi-modal plant knowledge graph are obtained by a data obtaining module, the initial multi-modal plant knowledge graph includes a plurality of mangrove plant category feature corresponding relationships, the feature corresponding relationships include a plurality of types of feature data and corresponding feature thresholds; a plurality of mangrove plant category sample feature data sets are obtained by a first feature extraction module according to unmanned aerial vehicle remote sensing data and elevation data in the multi-source remote sensing sample data sets for mangrove plant feature extraction, wherein the sample feature data sets include a plurality of types of feature data; the feature threshold updating module updates the feature thresholds corresponding to the feature data in the feature corresponding relationships of the plurality of mangrove plant categories in the initial multi-modal plant knowledge graph according to the plurality of mangrove plant category sample feature data sets by using a decision tree learning algorithm to obtain an updated multi-modal plant knowledge graph; the second feature extraction module obtains multi-source remote sensing data of a region to be classified, and the feature data set of the region to be classified is obtained by feature extraction according to unmanned aerial vehicle remote sensing data and elevation data in the multi-source remote sensing data; the interspecific classification result module performs interspecific classification reasoning according to the plurality of types of feature data in the feature data set of the region to be classified, the feature values corresponding to the feature data, and the plurality of types of feature data in the feature corresponding relationships of the plurality of mangrove plant categories in the updated multi-modal plant knowledge graph, and the feature thresholds corresponding to the feature data, to obtain the mangrove interspecific classification result of the region to be classified. The decision tree learning algorithm is used to update the feature thresholds in the feature corresponding relationships of the plurality of mangrove plant categories in the initial multi-modal plant knowledge graph to obtain the updated multi-modal plant knowledge graph, and the interspecific classification reasoning is performed according to the plurality of types of feature data in the extracted feature data set of the region to be classified, the feature values corresponding to the feature data, and the updated multi-modal plant knowledge graph, to obtain the mangrove interspecific classification result. The classification knowledge is fused, the shortcomings of the method depending on data alone are overcome, and the accuracy, efficiency and reliability of the mangrove interspecific classification are improved.

[0086] Please refer to Figure 8 , Figure 8 The structure schematic diagram of the computer device provided in an embodiment of the present application, the computer device 8 includes: a processor 81, a memory 82, and a computer program 83 stored in the memory 82 and executable on the processor 81; the computer device can store a plurality of instructions, the instructions are suitable for being loaded and executed by the processor 81 to perform the method steps of the above Figures 1 to 6 , and the specific execution process can be referred to the specific description of the above Figures 1 to 6 , which will not be described here.

[0087] The processor 81 can include one or more processing cores. The processor 81 connects various parts within the server through various interfaces and lines, executes various functions and processes data of the mangrove inter-specific classification device 7 based on unmanned aerial vehicle remote sensing by running or executing instructions, programs, code sets or instruction sets stored in the memory 82, and calling data in the memory 82. Optionally, the processor 81 can be implemented in at least one of a hardware form of a digital signal processing (DSP), a field-programmable gate array (FPGA), and a programable logic array (PLA). The processor 81 can be integrated with one or a combination of a central processing unit (CPU), a graphics processing unit (GPU), and a modem. Among them, the CPU mainly processes operating systems, user interfaces, and application programs; the GPU is responsible for rendering and drawing the content displayed on the touch display screen; and the modem is used for processing wireless communication. It can be understood that the above-mentioned modem can also not be integrated into the processor 81, but can be realized by a separate chip.

[0088] The memory 82 can include a random access memory (RAM) and a read-only memory (ROM). Optionally, the memory 82 includes a non-transitory computer-readable storage medium. The memory 82 can be used to store instructions, programs, codes, code sets or instruction sets. The memory 82 can include a program storage area and a data storage area, wherein the program storage area can store instructions for implementing an operating system, instructions for at least one function (such as touch instructions, etc.), instructions for implementing the above-mentioned various method embodiments, etc.; the data storage area can store data involved in the above-mentioned various method embodiments, etc. The memory 82 can also be at least one storage device located away from the aforementioned processor 81.

[0089] The embodiment of the present application further provides a storage medium, which can store a plurality of instructions. The instructions are suitable for being loaded and executed by a processor to perform the method steps of the above-mentioned Figures 1 to 6 . For specific implementation processes, refer to the specific description of the above-mentioned Figures 1 to 6 , which will not be described here in detail.

[0090] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the above-mentioned division of each functional unit and module is exemplified, and in actual application, the above-mentioned functions can be completed by different functional units and modules according to needs, that is, the internal structure of the device is divided into different functional units or modules to complete all or part of the functions described above. Each functional unit and module in the embodiment can be integrated in one processing unit, or each unit can be physically present separately, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be realized in the form of hardware or software. In addition, the specific name of each functional unit and module is only for the convenience of mutual distinction, and does not limit the protection scope of the present application. The specific working process of the unit and module in the above system can be referred to the corresponding process in the foregoing method embodiment, which will not be described here.

[0091] In the above embodiments, the description of each embodiment has its own emphasis, and the parts not described or recorded in detail in a certain embodiment can be referred to the related description of other embodiments.

[0092] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be realized in electronic hardware or a combination of computer software and electronic hardware. Whether the functions are executed in hardware or software depends on the specific application and design constraints of the algorithm. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0093] In the embodiments provided by the present application, it should be understood that the disclosed apparatus / terminal device and method can be implemented in other ways. For example, the above-mentioned apparatus / terminal device embodiments are only schematic, and the division of the modules or units is only a logical function division, and there can be another division manner in actual implementation. For example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed mutual coupling or direct coupling or communication connection can be indirect coupling or communication connection through some interface, device or unit, and can be electrical, mechanical or other forms.

[0094] The units described as separate components can or can not be physically separated, and the components shown as units can or can not be physical units, that is, they can be located in one place, or can be distributed on multiple network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the embodiment.

[0095] In addition, each function unit in each embodiment of the present application can be integrated in one processing unit, or each unit can exist physically separately, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of a software function unit.

[0096] The integrated module / unit, if realized in the form of a software function unit and sold or used as an independent product, can be stored in a computer readable storage medium. Based on such understanding, all or part of the processes in the above-mentioned embodiment methods can also be completed by a computer program instructing related hardware, and the computer program can be stored in a computer readable storage medium. The computer program, when executed by a processor, can implement the steps of each method embodiment. The computer program includes computer program code, which can be in the form of source code, object code, an executable file, or some intermediate form.

[0097] The present application is not limited to the above-described embodiments, and various modifications or changes can be made to the present application without departing from the spirit and scope of the present application. Therefore, it is intended that the present application encompass all such modifications and changes and fall within the scope of the appended claims and their equivalents.

Claims

1. A method for interspecific classification of mangroves based on UAV remote sensing, characterized in that, Includes the following steps: A multi-source remote sensing sample dataset of several mangrove plant categories and an initial multimodal plant knowledge graph are obtained. The initial multimodal plant knowledge graph includes feature correspondences of several mangrove plant categories, and the feature correspondences include feature data of several types and corresponding feature thresholds. Based on the UAV remote sensing data and elevation data in the multi-source remote sensing sample dataset, mangrove plant features are extracted to obtain sample feature datasets for several mangrove plant categories, wherein the sample feature datasets include several types of feature data. Based on the sample feature datasets of several mangrove plant categories, several feature vectors are extracted from the sample feature datasets as training features for several mangrove plant categories. Based on the sample feature datasets of several mangrove plant categories and the training features, the Gini index minimization criterion method is used to construct classification trees for several mangrove plant categories. Based on the classification trees of several mangrove plant categories, several target feature vectors and corresponding target feature thresholds of several mangrove plant categories are obtained. The feature thresholds corresponding to the feature vectors in the initial feature correspondence of the corresponding mangrove plant categories in the initial multimodal plant knowledge graph are updated to obtain the updated multimodal plant knowledge graph. Obtain multi-source remote sensing data of the region to be classified, and extract features based on UAV remote sensing data and elevation data in the multi-source remote sensing data to obtain the feature dataset of the region to be classified. Based on the feature data of several types in the feature dataset of the region to be classified, the feature values ​​corresponding to the feature data, and the feature thresholds corresponding to the feature values ​​of several types in the feature correspondence of several mangrove plant categories in the updated multimodal plant knowledge graph, interspecific classification reasoning is performed to obtain the interspecific classification results of the mangroves in the region to be classified.

2. The mangrove interspecific classification method based on UAV remote sensing according to claim 1, characterized in that: The feature data comprises several feature vectors, including spectral feature data, texture feature data, shape feature data, and height feature data. The spectral feature data includes hue feature vector, saturation feature vector, brightness feature vector, spectral mean feature vector, and spectral standard deviation feature vector. The texture feature data includes homogeneity feature vectors, contrast feature vectors, dissimilarity feature vectors, entropy feature vectors, dihedral matrix feature vectors, texture mean feature vectors, texture standard deviation feature vectors, and autocorrelation feature vectors. The shape feature data includes asymmetry feature vector, boundary index feature vector, compactness feature vector, density feature vector, ellipse fit feature vector, principal direction feature vector, maximum closed ellipse radius feature vector, minimum closed ellipse radius feature vector, rectangle fit feature vector, circularity feature vector, and shape index feature vector. The height feature data includes the elevation standard deviation feature vector, elevation variance feature vector, altitude elevation feature vector, elevation mean absolute deviation feature vector, elevation L2 feature vector, canopy undulation feature vector, and canopy height model feature vector.

3. The method for interspecific classification of mangroves based on UAV remote sensing according to claim 2, characterized in that, The method of constructing a classification tree for several mangrove plant categories based on sample feature datasets and training features of several mangrove plant categories using the Gini index minimization criterion includes the following steps: Based on several training features, the sample feature dataset is divided into two categories to obtain a first sub-sample feature dataset and a second sub-sample feature dataset. According to the preset Gini index algorithm, the Gini index after the sample feature dataset is divided into two categories under any feature value is calculated. The training feature with the smallest Gini index and its corresponding feature value are taken as the optimal split point for the current round. The first and second subsample feature datasets of the training features with the smallest Gini index are used as the sample feature datasets of the left and right nodes, respectively. The optimal split point is obtained in the next round until all nodes no longer meet the condition that nodes can be further divided, thus obtaining a classification tree of several mangrove plant categories.

4. The method for interspecific classification of mangroves based on UAV remote sensing according to claim 3, characterized in that, The step of obtaining several target feature vectors and corresponding target feature thresholds for several mangrove plant categories based on a classification tree of several mangrove plant categories includes the following steps: Based on the classification trees of several mangrove plant categories, the training features and feature values ​​corresponding to the optimal split points of the classification trees of several mangrove plant categories in all rounds are obtained. The training features are used as target feature vectors and the feature values ​​are used as target feature thresholds to obtain several target feature vectors and corresponding target feature thresholds for several mangrove plant categories.

5. The mangrove interspecific classification method based on UAV remote sensing according to claim 4, characterized in that, The step of extracting features from the UAV remote sensing data and elevation data in the multi-source remote sensing data to obtain the feature dataset of the region to be classified includes the following steps: Obtain the mangrove vegetation index of the region to be classified, identify the mangrove areas in the region to be classified based on the mangrove vegetation index, and extract features from the UAV remote sensing data and elevation data in the multi-source remote sensing data corresponding to the mangrove areas in the region to be classified to obtain the feature dataset of the region to be classified.

6. The method for interspecific classification of mangroves based on UAV remote sensing according to claim 5, characterized in that, The process of performing interspecific classification inference based on several types of feature data and corresponding feature values ​​in the feature dataset of the region to be classified, and several types of feature data and corresponding feature thresholds in the feature correspondence of several mangrove plant categories in the updated multimodal plant knowledge graph, to obtain the interspecific classification result of mangroves in the region to be classified, includes the following steps: From the updated multimodal plant knowledge graph, determine several candidate feature correspondences for the region to be classified. Based on all feature vectors in all types of feature data in the feature dataset of the region to be classified and several candidate feature correspondences, determine the target feature correspondences for mangrove plant categories that simultaneously have corresponding feature vectors. Based on the feature values ​​corresponding to all feature vectors in all types of feature data in the feature dataset of the region to be classified, if all feature values ​​satisfy the feature thresholds corresponding to the feature vectors in the target feature correspondence, then the mangrove plant category corresponding to the target feature correspondence is taken as the mangrove interspecific classification result, and the mangrove interspecific classification result of the region to be classified is obtained.

7. A mangrove interspecific classification device based on UAV remote sensing, characterized in that, include: The data acquisition module is used to acquire multi-source remote sensing sample datasets of several mangrove plant categories and an initial multimodal plant knowledge graph. The initial multimodal plant knowledge graph includes feature correspondences of several mangrove plant categories, and the feature correspondences include feature data of several types and corresponding feature thresholds. The first feature extraction module is used to extract mangrove plant features based on UAV remote sensing data and elevation data in the multi-source remote sensing sample dataset, and obtain sample feature datasets of several mangrove plant categories, wherein the sample feature datasets include several types of feature data. The feature threshold update module is used to extract several feature vectors from the sample feature dataset of several mangrove plant categories as training features for several mangrove plant categories, and to construct a classification tree for several mangrove plant categories based on the sample feature dataset of several mangrove plant categories and the training features, using the Gini index minimization criterion method. Based on the classification trees of several mangrove plant categories, several target feature vectors and corresponding target feature thresholds of several mangrove plant categories are obtained. The feature thresholds corresponding to the feature vectors in the initial feature correspondence of the corresponding mangrove plant categories in the initial multimodal plant knowledge graph are updated to obtain the updated multimodal plant knowledge graph. The second feature extraction module is used to obtain multi-source remote sensing data of the region to be classified, and to perform feature extraction based on the UAV remote sensing data and elevation data in the multi-source remote sensing data to obtain the feature dataset of the region to be classified. The interspecific classification result module is used to perform interspecific classification reasoning based on several types of feature data and corresponding feature values ​​in the feature dataset of the region to be classified, as well as several types of feature data and corresponding feature thresholds in the feature correspondence of several mangrove plant categories in the updated multimodal plant knowledge graph, to obtain the interspecific classification result of mangroves in the region to be classified.

8. A computer device, characterized in that, It includes a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the mangrove interspecific classification method based on UAV remote sensing as described in any one of claims 1 to 6.

9. A storage medium, characterized in that, The storage medium stores a computer program, which, when executed by a processor, implements the steps of the mangrove interspecific classification method based on UAV remote sensing as described in any one of claims 1 to 6.

Images