A method and device for object-oriented remote sensing image geomorphic classification of coral reefs

Abstract

The present invention provides a method and device for object-oriented remote sensing image geomorphic classification of coral reefs, belonging to the technical field of remote sensing monitoring, including: constructing a geomorphic unit classification system for coral reefs and performing data preprocessing on the obtained original remote sensing images of coral reefs; using the multi-scale segmentation algorithm in object-oriented segmentation to segment the original remote sensing images, finding the optimal parameters, and finally obtaining the optimal segmentation result; performing feature extraction and classification on the preprocessed remote sensing images, calculating remote sensing index features and texture features for the coral reef geomorphic unit classification data set, and obtaining the coral reef geomorphic unit classification data set; using an object-oriented machine learning algorithm to train the coral reef geomorphic unit classification data set, optimizing the model parameters, and obtaining the trained recognition and classification model; using the trained classification model for prediction and making the corresponding coral reef remote sensing geomorphic unit classification result map. The present invention can improve the accuracy and efficiency of classification.

Description

Technical Field

[0001] The present invention belongs to the technical field of remote sensing monitoring, and particularly relates to a method and device for object-oriented remote sensing image geomorphic classification of coral reefs. Background Art

[0002] Coral reefs are biogenic limestones formed by the long-term accumulation and deposition of calcium substances generated by reef-building corals and other reef-building organisms, and are mainly distributed in tropical shallow waters. Coral reefs are known as "oases in the blue desert", and the coral reef ecosystem is one of the most species-rich and productive ecosystems on Earth. Coral reefs are rich in species, play an important role in maintaining marine biodiversity, and also provide many biological resources for human production and life, with great development potential, economic value and ecological environment value.

[0003] However, the sensitivity and vulnerability of coral reefs to global change have led to the high-biodiversity and high-economic-value coral reef ecosystem being in a state of rapid degradation in recent decades under the influence of global change. Coral reefs are crucial to the marine ecosystem. The fishery resources provided by coral reefs have made important contributions to people's livelihoods and the economic growth of countries. Moreover, coral reefs are important carriers for recording global change, and can accurately record the change process of tropical marine environments at different time scales and the mechanism of their interaction with global change. Therefore, protecting coral reefs is a key measure to maintain the balance of the marine ecosystem and sustainable development. In order to better understand and protect coral reefs, effective classification and monitoring means have become crucial.

[0004] With the progress of science and technology, remote sensing technology has achieved great success in the field of earth science, and remote sensing technology has been proven to be an effective technical means for coral reef classification and monitoring. In this context, remote sensing technology has become one of the important tools for studying coral reefs. Remote sensing technology has the advantages of large-area synchronous coverage, repeated detection of the same area, high precision, rich information, etc. Especially high-resolution remote sensing has become an important means for large-scale classification of coral islands and reefs, and it shows excellent performance in terms of spatial resolution and positioning accuracy. Remote sensing data with multiple platforms, multiple sensors, and various spatial and spectral resolutions have been widely used in the research of coral islands and reefs, providing a powerful tool for in-depth understanding and protection of this ecosystem.

[0005] Traditional coral reef classification methods are limited by the subjectivity and time-consuming nature of manual identification, and there is an urgent need for a more efficient and accurate classification means. At present, machine learning has unique advantages for the classification of coral reef geomorphic units due to its excellent performance in large-scale data processing and complex pattern recognition. However, how to improve the accuracy and running speed of machine learning in the classification results of coral reef geomorphic units is an urgent problem to be solved at present. Summary of the Invention

[0006] The object of the present invention is to overcome the deficiencies of the prior art. The present invention provides a method and device for object-oriented geomorphic classification of coral reef remote sensing images, which can improve the accuracy and efficiency of classification.

[0007] To achieve the above object, the present invention adopts the following technical solutions:

[0008] A method for object-oriented geomorphic classification of coral reef remote sensing images includes the following steps:

[0009] Step 1: Construct a geomorphic unit classification system for coral reefs, and perform data preprocessing on the obtained original coral reef remote sensing images.

[0010] Step 2: Use the multi-scale segmentation algorithm in object-oriented segmentation to segment the original remote sensing image, find the optimal parameters, and finally obtain the optimal segmentation result.

[0011] Step 3: Extract features and classify the preprocessed remote sensing images, calculate the remote sensing index features and texture features of the coral reef geomorphic unit classification data set, and obtain the coral reef geomorphic unit classification data set.

[0012] Step 4: Use the object-oriented machine learning algorithm to train the coral reef geomorphic unit classification data set, and use the k-fold cross-validation method with adaptive weight assignment to optimize the model parameters, and obtain the trained recognition and classification model.

[0013] Step 5: Use the trained classification model for prediction, and produce the corresponding coral reef remote sensing geomorphic unit classification result map.

[0014] Furthermore, the said Step 1 includes:

[0015] Step 11: According to the existing coral reef geomorphic unit classification system, and combining with the spectral characteristics of coral reef geomorphology on remote sensing images, establish a coral reef geomorphic unit classification system, and divide the coral reef geomorphic units into: deep reef front slope, shallow reef front slope, reef crest, coral dense area, coral sparse area, lagoon slope, lagoon bottom, patch reef, sandbank and land; each coral reef island will classify the system according to its unique geomorphic characteristics.

[0016] Step 12: The original remote sensing image of the coral reef adopts a World-View2 high-resolution remote sensing image, and the data preprocessing of the original remote sensing image includes image denoising, geometric correction, image fusion and image enhancement.

[0017] Furthermore, the said Step 2 includes:

[0018] Step 21: Segment the original remote sensing image using the multi-scale segmentation algorithm in object-oriented segmentation; the multi-scale segmentation algorithm segments according to the weights, compactness, and smoothness of different bands in the image, obtaining image objects of different sizes and shapes;

[0019] Step 22: Find the optimal segmentation parameters to finally obtain the best segmentation result;

[0020] The best segmentation result meets the following two conditions: (1) maximizing the homogeneity within the object, (2) maximizing the heterogeneity between objects.

[0021] Furthermore, the said step 3 includes:

[0022] Step 31: Extract and classify features from the preprocessed remote sensing image, calculate the remote sensing index features and texture features for the coral reef geomorphic unit classification data, obtaining the coral reef geomorphic unit classification data set;

[0023] Step 32: The division of the coral reef geomorphic unit classification data set adopts the stratified sampling method in random sampling. By dividing the population into different levels, the sampling data set has the same proportion as the original data set; for a data set , set the extraction ratio , according to the ratio divide the coral reef geomorphic unit data set into a training set and a test set .

[0024] Furthermore, in the said step 31, the remote sensing index features use a series of variant index features related to the normalized difference vegetation index NDVI and the normalized difference water index NDWI; the texture features are calculated using the gray-level co-occurrence matrix GLCM, define as the texture feature function, and the texture features included therein are contrast , dissimilarity , homogeneity , second-order moment , entropy , mean , variance , correlation , inverse difference degree .

[0025] Furthermore, the said step 4 includes:

[0026] By establishing a collection of multiple machine learning models , including algorithms such as random forest, support vector machine, K-nearest neighbor algorithm, XGBoost, etc., train each model in the model collection To predict the classification labels of coral reef geomorphic units; use the grid search method to optimize the hyperparameters of the model, and use the k-fold cross-validation method with adaptive weight assignment to adaptively adjust the size, and reassign the weights of the classes according to the class imbalance degree in each fold to determine the optimal parameter combination, so as to improve the accuracy of the model and retrain the model.

[0027] Furthermore, the k-fold cross-validation with adaptive weight assignment includes:

[0028] Divide the original dataset into subsets of equal size, called folds, and each subset will be used as the validation set in turn, and the remaining subsets are combined into the training set; on the basis of adaptively adjusting the value, dynamically adjust the weight of each class according to the class distribution in each fold ; for each fold, use the training set to train the model, and then use the validation set to evaluate the performance metrics of the model;

[0029] The weight is calculated as:

[0030] ;

[0031] where is the weight of the th class, is the total number of samples in the dataset, is the th class of the number of samples;

[0032] Repeat the steps so that each fold acts as a validation set once until each fold has been used as a validation set once; according to the results of each validation, calculate the average performance metric as the performance estimate of the model on the entire dataset.

[0033] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, the steps of the above-mentioned object-oriented coral reef remote sensing image geomorphic classification method are implemented.

[0034] The present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the above-mentioned object-oriented coral reef remote sensing image geomorphic classification method are implemented.

[0035] Beneficial effects:

[0036] The present invention is based on World-View2 high-resolution remote sensing images, and simultaneously trains multiple machine learning models - Random Forest model (RF), Support Vector Machine (SVM), K-Nearest Neighbor model (KNN), and Distributed Gradient Boosting model (XGBoost). In the process of predicting the categories of coral reef geomorphic units, different machine learning models have shown their respective characteristics and advantages. For example, in the classification task of patch reefs, the RF model and the XGBoost model have shown high accuracy and can better identify and distinguish the characteristics of patch reefs. The SVM model has a relatively high overall classification accuracy, and the KNN model has the fastest classification speed. After the four models are trained, the classification accuracy of each category of the geomorphic unit can be automatically extracted from the four models, and the overall classification accuracy and the classification accuracy of each category can be calculated. And the model can automatically generate a classification result map of the coral reef geomorphic unit based on the prediction results. Compared with traditional machine learning methods, the object-oriented classification method has shown significant advantages in the identification and classification of geomorphic units. This method can more accurately capture the essential characteristics of geomorphic units by identifying natural objects in the image, such as land cover types, buildings, water bodies, etc., rather than simply relying on pixel-level analysis. It can also better handle the redundancy problem of spatial data, reduce resource waste, and improve the practicality and reliability of classification results. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] With reference to the accompanying drawings, through the following detailed description of non-limiting embodiments, other features, objects, and advantages of the present invention will become more apparent. In the drawings:

[0038] Figure 1 is a flowchart of a method for object-oriented remote sensing image geomorphic classification of coral reefs according to the present invention;

[0039] Figure 2 is a schematic diagram of the multi-scale segmentation process according to the present invention;

[0040] Figure 3 is a schematic diagram of the accuracy distribution process of model training according to the present invention;

[0041] Figure 4 is a schematic diagram of the accuracy rate and kappa coefficient of model training according to the present invention;

[0042] Figure 5 is a schematic diagram of the accuracy of a single geomorphic category in model training according to the present invention;

[0043] Figure 6a , Figure 6b , Figure 6c , Figure 6d is a classification result map of the coral reef geomorphic unit according to the present invention; wherein, Figure 6a is Random Forest, Figure 6b is Support Vector Machine,Figure 6c is K nearest neighbor, Figure 6d is distributed gradient boosting. Specific implementation manners

[0044] In order to make the objectives, technical solutions and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention. In addition, the technical features involved in the various implementation manners of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0045] In the description of the present invention, it should be understood that the terms "upper", "lower", "top", "bottom", "inner", "outer", etc. adopted in the present invention indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the present invention. In addition, the terms "first", "second", etc. are only used for descriptive purposes and cannot be construed as indicating or implying relative importance.

[0046] In the present invention, unless otherwise clearly specified and limited, the terms "mounted", "connected", "connected", "fixed", etc. should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or integrated; it can be a mechanical connection or an electrical connection; it can be directly connected or indirectly connected through an intermediate medium, and it can be the internal connection of two elements or the interaction relationship between two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific situations.

[0047] The present invention discloses a method for object-oriented remote sensing image geomorphic classification of coral reefs. The method adopts a supervised classification idea and classifies coral reef geomorphic units based on a machine learning model. The method constructs classification quasi-samples of coral reef geomorphic units through visual interpretation, World-View2 satellite classification products or rough classification algorithms, etc.; by establishing a collection of multiple classification models , training each model in the collection , and thus predicting the classification labels of coral reef geomorphic units.

[0048] This method can ensure the operation speed under the condition of limited training samples and save the system operation cost; this method effectively samples the hyperparameters and model design space through a complete evaluation system, improves the advanced performance of the machine learning model and the quality of the machine learning model ensemble, so as to obtain the machine learning model with the optimal performance for coral reef geomorphic unit classification, greatly improving the operation accuracy. Therefore, this method can ensure high speed and accuracy in the process of coral reef geomorphic unit classification.

[0049] The details of the embodiments of the present invention will be introduced in detail below through specific embodiments.

[0050] As Figure 1 shown, a method for object-oriented coral reef remote sensing image geomorphic classification provided by the present invention includes the following steps:

[0051] Step S1, construct a coral reef geomorphic unit classification system, and perform data preprocessing on the obtained original coral reef remote sensing image, including:

[0052] Step S11, fully investigate the existing coral reef geomorphic classification system, and combine the spectral characteristics of coral reefs on remote sensing images to establish a coral reef geomorphic unit classification system, which divides the coral reef geomorphic units into: deep reef front slope, shallow reef front slope, reef crest, coral dense area, coral sparse area, lagoon slope, lagoon bottom, patch reef, sandbank and land. Since each coral reef has its unique distribution of coral reef geomorphic units, the established coral reef geomorphic unit classification system is the geomorphic unit characteristics of many coral reef islands. Referring to the existing coral reef geomorphic unit classification system, a general classification system for coral reef islands is proposed, and each coral reef island will modify its classification system according to its unique geomorphic characteristics.

[0053] Step S12, the original coral reef remote sensing image uses World-View2 high-resolution remote sensing image, and the data preprocessing includes image denoising, geometric correction, image fusion and image enhancement. These preprocessings help the machine learning model better capture the associations between different bands, better capture the details and features of geomorphic units, and improve the classification performance of the model for different geomorphic units.

[0054] Step S2, use the multi-scale segmentation algorithm in object-oriented segmentation to segment the original coral reef remote sensing image, and find the optimal segmentation scale parameter, and finally obtain the best segmentation result, including:

[0055] The multi-scale segmentation algorithm can segment according to the weights, compactness and smoothness of different bands in the original coral reef remote sensing image, and can obtain image objects of different sizes and shapes. This method can make full use of the spatial, spectral, texture and other information of the original coral reef remote sensing image, and is suitable for the extraction of complex areas or detailed land types.

[0056] As Figure 2 shown, the object-oriented segmentation is to obtain the optimal solution of the segmentation parameters by setting different segmentation parameters, and to obtain the optimal segmentation result by evaluating the results of the segmentation schemes with different parameters.

[0057] First, by setting a set of segmentation scale parameters , the original image is segmented by a multi-scale segmentation algorithm, the regional features of the segmentation result are calculated, the segmentation quality is evaluated, and the segmentation result is judged. If the segmentation result meets the requirements, the similar regions in the result are merged. The ESP tool is used to judge whether the segmentation parameters are optimal, and finally the optimal segmentation scale is determined and the segmentation result is obtained. The calculation process: calculating the local variance of different objects within one band to calculating the mean value of the local variances of multiple bands:

[0058] ;

[0059] Among them, represents the local variance of each band, i = 1, 2,..., 8. represents the calculation of the mean value of multiple bands.

[0060] If the segmentation scale is much larger than the objects in the scene, most of the objects obtained by segmentation are highly correlated, and the measured value of the local variance will be very low. If the segmentation scale is approximately the size of the target object, the heterogeneity between the target objects will increase and the local variance will increase. To obtain the optimal segmentation scale, the curve is used to represent that when the curve reaches the maximum value, this scale is the optimal segmentation scale, and the formula is as follows:

[0061] ;

[0062] Among them, represents the Receiver Operating Characteristic curve represents the local variance of the target layer; represents the local variance of the next layer with the local variance of the target layer as the benchmark. When the curve reaches the maximum value, this scale is the optimal segmentation scale.

[0063] Finally, the best result should basically meet the following two conditions: (1) maximizing the homogeneity within the object, (2) maximizing the heterogeneity between the objects.

[0064] The formula for calculating the average variance within the object is as follows:

[0065] ;

[0066] Among them, is the variance of the cell values within the object and is the area of the object . The average within-object variance is calculated as the area-weighted average, where n is the total number of objects. Among them, larger objects are given more weight, thus reducing the influence of instability caused by small objects. If has a higher value, the overall heterogeneity of the object is greater; if the value is lower, the homogeneity of the object in terms of the measured attribute is higher. Usually, there is a direct relationship between the average size of the object and the within-object variance: if the object is larger, then is higher, and vice versa.

[0067] Moran's I can be expressed as:

[0068] ;

[0069] Among them, is the average value of the object , is the average value of the object , is the average value of the scene, is a weight that measures the spatial adjacency of the object and the object . If the object and the object are adjacent, then has a value of 1; otherwise . The Moran's I has a value range from -1 to +1. An index close to zero and negative values indicates that the differences between the average values of adjacent objects are usually large. When both of these conditions are met, the optimal segmentation result of the remote sensing image can be obtained.

[0070] In this embodiment, the segmentation parameters determined by fitting the ROC curve through the ESP tool are: scale parameter: 80; shape parameter: 0.3; compactness: 0.5. In other embodiments, the segmentation parameters need to be adjusted according to the actual situation.

[0071] Step S3, perform feature extraction and classification from the preprocessed remote sensing image, calculate the remote sensing index features and texture features of the coral reef geomorphic unit classification data, and obtain the coral reef geomorphic unit classification data set, including:

[0072] Step S31: Since multi-scale segmentation was performed in step S2, the shape features of the remote sensing image are no longer calculated in step S31. The remote sensing index features mainly use a series of variant index features related to the Normalized Difference Vegetation Index (NDVI) and the Normalized Difference Water Index (NDWI). The texture features are calculated using the Gray Level Co-occurrence Matrix (GLCM), and define as the texture feature function, including texture features such as contrast , dissimilarity , homogeneity , second moment , entropy , mean , variance , correlation , inverse difference moment . The following gives the calculation methods of all texture feature functions:

[0073] ;

[0074] ;

[0075] ;

[0076] ;

[0077] ;

[0078] ;

[0079] ;

[0080] ;

[0081] Among them, is the number of rows or columns in the matrix (equal to the number of gray levels), is the row index of the matrix (equal to the gray level of the reference cell), is the column index of the matrix (equal to the gray level of the adjacent cell), has the probability (relative frequency) of adjacent cells with gray level and gray level . is the mean of the row or column, is the standard deviation of the row or column, is used as a condition.

[0082] ;

[0083] Among them, is expressed as the sum, and is the gray level and vector. The gray level and vector represents two gray levels based on the calculated matrix The sum of The probability distribution vector. The position in the vector (starting from 0) represents the value, and the value at this position corresponds to the gray level and the gray level The sum is equal to the probability of this value.

[0084] The gray level co-occurrence matrix in this embodiment The sliding window size is 3×3; the step , that is, the central pixel directly makes a comparison operation with its adjacent pixel points; direction selection: calculate the gray level co-occurrence matrix The direction is 45°.

[0085] In other embodiments, the sliding window size can be 5×5 or 7×7, or it can be set by itself according to the size of the data set; the direction selection is: after finding the eigenvalues of the four direction matrices, the average value of the four eigenvalues can be calculated as the final value.

[0086] Step S32: The coral reef geomorphic unit classification data set is divided by stratified sampling in random sampling. By dividing the population into different levels, the sampled data set has the same proportion as the original data set. Suppose there is a data set , set the extraction ratio , according to the extraction ratio The coral reef geomorphic unit classification data set is divided into a training set and a test set .

[0087] In this embodiment, the label data set is labeled by experts according to experience. The extraction ratio set in this embodiment is 30%. In other embodiments, the extraction ratio The best setting should be determined through experiments according to the actual situation in order to find the most suitable ratio for the current task.

[0088] Step S4: Use the training set to train the prediction model. The prediction model includes multiple sub-machine learning models; optimize the model parameters through the k-fold cross-validation method with adaptive weight allocation. The trained prediction model is used to identify the geomorphic unit classification in the data set, including:

[0089] By establishing an ensemble of multiple machine learning models (such as Random Forest (RF), Support Vector Machine (SVM), K-Nearest Neighbor model (KNN), Distributed Gradient Boosting model (XGBoost)) , training each model in the ensemble , so as to predict the classification labels of coral reef geomorphic units. The hyperparameters of the model are optimized and adjusted using the grid search method, and the k-fold cross-validation method with adaptive weight allocation is used to adaptively adjust the size of , and re-allocate the weights of the classes according to the degree of class imbalance in each fold, so as to improve the accuracy of the model and re-train the model to ensure that objective and stable training results can be obtained.

[0090] Grid search is a technique for determining the optimal hyperparameter combination of a machine learning model by traversing a predefined parameter grid. First, a set of limited parameter ranges is set , and by traversing all possible parameter combinations for training and validation, the parameter set with the highest cross-validation score is finally selected.

[0091] The specific process of k-fold cross-validation with adaptive weight allocation is as follows:

[0092] The original dataset is divided into subsets of equal size (usually called folds). Each subset will be used as the validation set in turn, and the remaining subsets are combined into the training set. On the basis of adaptively adjusting the value, the weights of each class are dynamically adjusted according to the class distribution in each fold . For each fold, the model is trained using the training set, and then the performance metrics of the model, such as accuracy, precision, recall, etc., are evaluated using the validation set. If the value is large and the dataset is small, the class weights can be adjusted more finely, so that the classes with fewer samples in each fold obtain higher weights , thus compensating for the bias caused by insufficient class sample size.

[0093] The weight is calculated as follows:

[0094] ;

[0095] where, where is the weight of the th class, is the total number of samples in the dataset, is the number of samples in the th class.

[0096] Repeat the steps so that each fold serves as a validation set once until each fold has been used as a validation set. Based on the results of each validation, calculate the average performance metric as an estimate of the model's performance on the entire dataset. This strategy helps reduce training time and provides more flexible weight allocation for imbalanced data across different classes.

[0097] The method for adjusting model parameters in this embodiment is not the only one. In this embodiment, the value is set to 10, and the class weights are calculated through formulas. Finally, the optimal parameter combination is obtained through validation. In other embodiments, other methods can be used for self-adjustment according to prior knowledge. In this embodiment, box plots are used to show the accuracy distributions obtained from different model trainings. Figure 3 shows a schematic diagram of the model training accuracy distribution according to an embodiment of the present invention. As Figure 3 shown, the model accuracy distribution is presented in this embodiment. The abscissa represents the models, and the ordinate represents the model accuracy. In other embodiments, the way to show the model accuracy distribution can be set according to the actual situation. In this embodiment, line charts are used to show the accuracy during the model training process. Figure 4 shows a schematic diagram of the model training accuracy according to an embodiment of the present invention. As Figure 4 shown, the model accuracy and kappa coefficient are presented in this embodiment. The abscissa represents the models, and the ordinate represents the model accuracy. In other embodiments, the way to show the model accuracy can be set according to the actual situation. In this embodiment, bar charts are used to show the accuracy of each geomorphic category during the model training process. Figure 5 shows a schematic diagram of the accuracy of a single category in model training according to an embodiment of the present invention. As Figure 5 shown, the model accuracy is presented in this embodiment. The abscissa represents the geomorphic categories, and the ordinate represents the model accuracy. In other embodiments, the way to show the accuracy of the model for each category can be set according to the actual situation.

[0098] Step S5: Use the trained classification model for prediction and automatically produce a corresponding classification result map of coral reef remote sensing geomorphic units. As Figure 6a , Figure 6b , Figure 6c , Figure 6d shown is the classification result map of coral reef geomorphic units of the present invention. Figure 6a is the random forest, Figure 6b is the support vector machine, Figure 6c is the K-nearest neighbor, Figure 6dIt is distributed gradient boosting. It can be found that the recognition accuracy of each model for each category of geomorphic units is different. For example, the random forest model and the distributed gradient boosting model based on decision trees are more suitable for distinguishing multiple categories of geomorphic units and can better capture the differences between different categories. The random forest model (RF) has the highest accuracy and is better than other models in identifying patch reefs; although the support vector machine (SVM) can overall identify the outlines of each geomorphic unit, there are cases of misclassification and missed classification; the K-nearest neighbor model (KNN) is insensitive to the two categories of lagoon slope and lagoon bottom, resulting in poor classification results; the distributed gradient boosting model (XGBoost) is overall good, but due to the limitations of its implementation principle, its running speed is relatively slow.

[0099] The present invention uses World-View2 remote sensing images as data sources, adopts a method combined with object-oriented machine learning, classifies coral reef geomorphic units based on the texture feature information and remote sensing indices of World-View2 remote sensing images, provides a specific implementation method for the classification of coral reef geomorphic units, and can automatically generate a classification result map, which is convenient for the analysis and investigation of coral reefs, etc.

[0100] The present invention also provides a device for object-oriented remote sensing image geomorphic classification of coral reefs, including the following modules:

[0101] System construction module, constructs a geomorphic unit classification system for coral reefs, and performs data preprocessing on the acquired original remote sensing images of coral reefs;

[0102] Segmentation module, uses the multi-scale segmentation algorithm in object-oriented segmentation to segment the original remote sensing image, finds the optimal parameters, and finally obtains the optimal segmentation result;

[0103] Data set acquisition module, extracts features and classifies the preprocessed remote sensing images, calculates the remote sensing index features and texture features of the coral reef geomorphic unit classification data set, and obtains the coral reef geomorphic unit classification data set;

[0104] Recognition and classification module, uses an object-oriented machine learning algorithm to train the coral reef geomorphic unit classification data set, and optimizes the model parameters using the k-fold cross-validation method with adaptive weight allocation to obtain a trained recognition and classification model;

[0105] Prediction module, uses the trained classification model for prediction and makes a corresponding classification result map of the coral reef remote sensing geomorphic unit.

[0106] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, it implements the steps of the above-mentioned method for object-oriented remote sensing image geomorphic classification of coral reefs.

[0107] The present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the above-mentioned object-oriented coral reef remote sensing image geomorphic classification method are implemented.

[0108] Those skilled in the art should understand that the embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of the present invention can be implemented using various computer languages. For example, object-oriented programming languages such as Java and interpreted scripting languages such as JavaScript.

[0109] The present invention is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to the embodiments of the present invention. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, as well as the combination of flows and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate for implementing in the process Figure 1 a process or multiple processes and / or blocks Figure 1 a device for the functions specified in one block or multiple blocks.

[0110] These computer program instructions can also be stored in a computer-readable memory that can guide a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory generate a manufactured product including an instruction device, and the instruction device implements in the process Figure 1 a process or multiple processes and / or blocks Figure 1 a device for the functions specified in one block or multiple blocks.

[0111] These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process, and thus the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in Figure 1 a process or multiple processes and / or blocks Figure 1 a device for the functions specified in one block or multiple blocks.

[0112] Although the preferred embodiments of the present invention have been described, additional changes and modifications can be made to these embodiments by those skilled in the art once they learn the basic creative concept. Therefore, the appended claims are intended to be construed to include the preferred embodiments as well as all changes and modifications that fall within the scope of the present invention.

[0113] 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 is also intended to include these modifications and variations.

Claims

1. An object-oriented method for landform classification of coral reef remote sensing images, characterized in that: The steps include: Step 1: Construct a classification system of coral reef geomorphic units and perform data preprocessing on the original remote sensing images of coral reefs; Step 2: Use the multi-scale segmentation algorithm in object-oriented segmentation to segment the original remote sensing image, find the optimal parameters, and finally obtain the optimal segmentation result; Step 3: Extract and classify features from the pre-processed remote sensing images, calculate remote sensing index features and texture features of the coral reef geomorphic unit classification dataset, and obtain the coral reef geomorphic unit classification dataset; Step 4: Use an object-oriented machine learning algorithm to train the coral reef geomorphic unit classification dataset, and use the k-fold cross-validation method with adaptive weight allocation to optimize the model parameters to obtain the trained recognition and classification model, including: By building a collection of multiple machine learning models , including random forest, support vector machine, K nearest neighbor algorithm, XGBoost algorithm, training model collection Each model in To predict the classification labels of coral reef geomorphic units; use the grid search method to tune the model's hyperparameters, and use the k-fold cross-validation method with adaptive weight distribution to adaptively adjust according to the size of the dataset The size of the class is adjusted, and the weights of the categories are redistributed according to the degree of class imbalance in each fold to determine the optimal parameter combination to improve the accuracy of the model and retrain the model; Step 5: Use the trained classification model to make predictions and produce the corresponding coral reef remote sensing landform unit classification result map.

2. The object-oriented method for coral reef remote sensing image landform classification according to claim 1, characterized in that: The step 1 comprises: Step 11. Based on the existing coral reef geomorphic unit classification system and combined with the spectral characteristics of coral reef geomorphology in remote sensing images, a coral reef geomorphic unit classification system is established. The coral reef geomorphic units are divided into: deep fore-reef slope, shallow fore-reef slope, reef ridge, coral-dense area, coral-sparse area, lagoon slope, lagoon bottom, point reef, sandbar and land. Each coral reef island will change the classification system according to its unique geomorphic characteristics. Step 12: The original remote sensing image of the coral reef uses the World-View2 high-resolution remote sensing image. The data preprocessing of the original remote sensing image includes image denoising, geometric correction, image fusion and image enhancement.

3. The object-oriented method for coral reef remote sensing image landform classification according to claim 1, characterized in that: The step 2 includes: Step 21: Segment the original remote sensing image using a multi-scale segmentation algorithm in object-oriented segmentation; the multi-scale segmentation algorithm performs segmentation based on the weights, compactness, and smoothness of different bands in the image to obtain image objects of different sizes and shapes; Step 22: Find the optimal segmentation parameters and finally obtain the best segmentation result; The optimal segmentation result satisfies the following two conditions: (1) maximizing homogeneity within objects and (2) maximizing heterogeneity between objects.

4. The object-oriented method for coral reef remote sensing image landform classification according to claim 1, characterized in that: The step 3 comprises: Step 31: extract and classify features from the pre-processed remote sensing image, calculate remote sensing index features and texture features of the coral reef geomorphic unit classification data, and obtain a coral reef geomorphic unit classification data set; Step 32: The classification dataset of coral reef geomorphic units is divided into stratified sampling in random sampling. By dividing the population into different layers, the sampled dataset and the original dataset have the same proportion. , set the extraction ratio , according to the proportion Divide the coral reef geomorphic unit dataset into training sets and test set .

5. The object-oriented method for coral reef remote sensing image landform classification according to claim 4, characterized in that: In step 31, the remote sensing index feature uses a series of variant index features related to the normalized vegetation index NDVI and the normalized difference water index NDWI; the texture feature is calculated using the gray level co-occurrence matrix GLCM, and is defined as is a texture feature function, which includes texture features such as contrast , dissimilarity , uniformity , second-order moment ,entropy , mean ,variance , correlation , inverse difference .

6. The object-oriented method for coral reef remote sensing image landform classification according to claim 1, characterized in that: The k-fold cross validation of the adaptive weight allocation includes: Divide the original dataset into Each subset is used as a validation set in turn, and the rest are subsets are combined into a training set; in the adaptive adjustment Based on the value, the weight of each category is dynamically adjusted according to the category distribution in each fold ; For each fold, the training set is used to train the model, and then the validation set is used to evaluate the performance indicators of the model; Weight The calculation formula is: ； Among them, It is The weight of the class, is the total number of samples in the dataset, It is The number of samples in the class; Repeat the steps so that each fold serves as a validation set once, until each fold has been used as a validation set once; based on the results of each validation, calculate the average performance index as the performance estimate of the model on the entire dataset.

7. An object-oriented device for coral reef remote sensing image landform classification, characterized in that: Includes the following modules: System construction module, builds a classification system for coral reef geomorphic units, and performs data preprocessing on the original remote sensing images of coral reefs; The segmentation module uses the multi-scale segmentation algorithm in object-oriented segmentation to segment the original remote sensing image, find the optimal parameters, and finally obtain the optimal segmentation result; The dataset acquisition module extracts and classifies the pre-processed remote sensing images, calculates the remote sensing index characteristics and texture characteristics of the coral reef geomorphic unit classification dataset, and obtains the coral reef geomorphic unit classification dataset; The recognition and classification module uses an object-oriented machine learning algorithm to train a coral reef geomorphic unit classification dataset and uses the k-fold cross-validation method with adaptive weight distribution to optimize model parameters. The trained recognition and classification model is obtained, including: By building a collection of multiple machine learning models , including random forest, support vector machine, K nearest neighbor algorithm, XGBoost algorithm, training model collection Each model in To predict the classification labels of coral reef geomorphic units; use the grid search method to tune the model's hyperparameters, and use the k-fold cross-validation method with adaptive weight distribution to adaptively adjust according to the size of the dataset The size of the class is adjusted, and the weights of the categories are redistributed according to the degree of class imbalance in each fold to determine the optimal parameter combination to improve the accuracy of the model and retrain the model; The prediction module uses the trained classification model to make predictions and produce the corresponding coral reef remote sensing landform unit classification result map.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the steps of the object-oriented method for coral reef remote sensing image landform classification according to any one of claims 1 to 6 are implemented.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the object-oriented method for coral reef remote sensing image landform classification according to any one of claims 1 to 6 are implemented.

Images