Method and system for identifying macroalgae culture area based on Otsu feature enhancement

Abstract

The invention discloses an Otsu feature enhancement-based macroalgae culture area identification method and system. The method comprises the following steps: acquiring multi-spectral remote sensing image data of a target area and preprocessing the multi-spectral remote sensing image data; generating a spatial constraint binary mask for the preprocessed image by adopting an Otsu algorithm; constructing a multi-dimensional feature space containing a spectral band, a normalized index and an Otsu segmentation result; and inputting the multi-dimensional features into a classifier for classification. According to the method, the Otsu binarization result is segmented and reconstructed into the spatial distribution features from the terminal, the classification precision is remarkably improved in multiple test scenes, the area error of seaweed monitoring is effectively reduced, the defects of feature solidification and noise sensitivity in a traditional method are overcome, and the method is suitable for near-real-time monitoring requirements in multiple scale scenes.

Description

Technical Field

[0001] The present invention relates to the field of marine remote sensing information, and particularly relates to a method for identifying large-scale seaweed aquaculture areas by fusing Otsu segmentation results and supervised classification. Background Art

[0002] The accurate identification of large-scale seaweed aquaculture areas is a core requirement for marine ecological monitoring and fishery management. Existing remote sensing monitoring technologies generally face the bottleneck problem of insufficient feature characterization ability in complex marine environments, mainly manifested as the fluctuation of misjudgment rate caused by spectral confusion effect and spatial heterogeneity interference. In traditional methods, although the binary segmentation technology based on the Otsu algorithm is widely used for the coarse-grained separation of water bodies and land-based targets, its application dimension still has significant limitations:

[0003] Insufficient development of feature value: In the current technical system, the Otsu segmentation results are mostly used as the terminal output, only for the division of land-sea boundaries, and the spatial distribution characteristics and neighborhood correlation characteristics contained therein are not deeply mined, resulting in the potential of the algorithm not being fully released at the levels of texture feature modeling and morphological optimization.

[0004] Lack of spatial information: Existing feature construction methods mostly use simple stacking of spectral indices and original bands, ignoring the topological attributes such as gradient changes and spatial continuity hidden in the Otsu segmentation results, making it difficult for the classification model to capture the spatial distribution difference characteristics between algae and interference objects.

[0005] The core improvement of the present invention lies in breaking through the single application mode of the traditional technology for the Otsu algorithm. By reconstructing the Otsu binary result into spatial features and inputting them into the classification model, the collaborative enhancement of segmentation features and spectral features is realized, so as to improve the recognition accuracy in complex scenarios at a lower computational cost. Summary of the Invention

[0006] The purpose of the present invention is to provide a method for identifying large-scale seaweed aquaculture areas based on Otsu feature enhancement. By fusing the Otsu segmentation results and the collaborative modeling of multi-dimensional feature spaces, the problems of insufficient utilization of spatial features and weak anti-interference ability in traditional methods are solved, and high-precision and high-efficiency dynamic monitoring of algae aquaculture is realized.

[0007] The specific technical solution adopted by the present invention is as follows:

[0008] In the first aspect, the present invention provides a method for identifying large-scale seaweed aquaculture areas based on Otsu feature enhancement, which includes the following steps:

[0009] S1. Obtain multi-spectral remote sensing image data including blue, green, red, and near-infrared bands during the seaweed growth period in the target sea area, and perform image preprocessing operations to obtain preprocessed image data;

[0010] S2. Extract spectral features, index features, and spatial constraint features from the preprocessed image data to form multi-dimensional features. The spectral features consist of the blue light band, green light band, red light band, and near-infrared band. The index features consist of at least one of the normalized difference vegetation index and the normalized difference water index. The spatial constraint feature is a binary mask of the target sea area generated by the Otsu algorithm based on the green light band.

[0011] S3. Input the multi-dimensional features into a supervised trained classification model. The classification model performs pattern recognition based on the multi-dimensional feature values at each pixel position to predict whether the pixel position belongs to the seaweed cultivation area, and finally outputs a spatial distribution map of the seaweed cultivation area.

[0012] As a preference of the first aspect above, the spatial resolution of the multi-spectral remote sensing image data is not less than 10 meters, and it needs to include multi-spectral data of the blue light, green light, red light, and near-infrared bands.

[0013] As a preference of the first aspect above, the image preprocessing operations include one or a combination of geometric correction, radiometric correction, atmospheric calibration, cloud masking, and land masking.

[0014] As a preference of the first aspect above, the classification model uses a machine learning classifier, preferably a random forest classifier.

[0015] As a preference of the first aspect above, before being used for actual recognition, the classification model is pre-supervised trained using labeled sample data. The labeled sample data is randomly collected from remote sensing images that have completed visual interpretation of the seaweed cultivation area, and the visual interpretation result is used as the true value label.

[0016] As a preference of the first aspect above, the parameter settings during the supervised training of the classification model are dynamically adjusted through a preset parameter combination. The adjusted parameters include the classifier type selection strategy, the model training iteration control method, and the accuracy optimization mechanism, and the best parameter combination that can achieve the optimal classification performance is selected.

[0017] In the second aspect, the present invention provides a large-scale seaweed cultivation area recognition system based on Otsu feature enhancement, which includes:

[0018] An image preprocessing module for obtaining multi-spectral remote sensing image data of the target sea area during the seaweed growth period, including the blue light, green light, red light, and near-infrared bands, and performing image preprocessing operations to obtain preprocessed image data;

[0019] A feature extraction module, configured to extract spectral features, index features, and spatial constraint features from the preprocessed image data to form multi-dimensional features; the spectral features are composed of a blue light band, a green light band, a red light band, and a near-infrared band, and the index features are composed of at least one of a normalized difference vegetation index and a normalized difference water index; the spatial constraint feature is a binary mask of the target sea area generated by the Otsu algorithm based on the green light band.

[0020] A pattern recognition module, configured to input the multi-dimensional features into a supervised trained classification model, and the classification model performs pattern recognition according to the multi-dimensional feature values at each pixel position to predict whether the pixel position belongs to a seaweed aquaculture area, and finally outputs a spatial distribution map of the seaweed aquaculture area.

[0021] In a third aspect, the present invention provides a computer program product, including a computer program / instructions, when the computer program / instructions are executed by a processor, it can implement the method for identifying a large-scale seaweed aquaculture area based on Otsu feature enhancement as described in any item of the first aspect above.

[0022] In a fourth aspect, the present invention provides a computer-readable storage medium, characterized in that a computer program is stored on the storage medium, and when the computer program is executed by a processor, it can implement the method for identifying a large-scale seaweed aquaculture area based on Otsu feature enhancement as described in any item of the first aspect above.

[0023] In a fifth aspect, the present invention provides a computer electronic device, characterized in that it includes a memory and a processor;

[0024] The memory is used to store a computer program;

[0025] The processor is configured to, when executing the computer program, be able to implement the method for identifying a large-scale seaweed aquaculture area based on Otsu feature enhancement as described in any item of the first aspect above.

[0026] The present invention has the following beneficial effects compared with the prior art:

[0027] 1. By constructing a cascaded processing architecture of "Otsu spatial constraint generation → multi-dimensional feature classification", the present invention breaks through the technical limitation that segmentation and classification are mutually separated in traditional methods. The Otsu segmentation result is dynamically reconstructed into a spatial constraint feature and input into the classification model, effectively suppressing the spectral confusion effect of background interference objects (such as suspended matter and aquaculture facilities).

[0028] 2. The present invention innovatively fuses the Otsu binary mask with multi-spectral features and vegetation indices in the spatial-spectral dimension to form a feature expression system with environmental adaptability. This multi-source feature collaboration mechanism has the following technical advantages:

[0029] Enhance the expression of algal morphological features through spatial constraint features, and reduce the missed detection rate of fragmented small targets;

[0030] Dynamically balance spectral sensitivity and spatial continuity, and improve classification stability in complex scenarios such as solar flares and thin cloud interference.

[0031] 3. The present invention has the advantages of accurate results and good versatility. It supports the input of multi-spectral images with resolutions from 10 meters to sub-meter level, and adapts to mainstream satellite data such as Sentinel-2 and high-resolution series; in addition to seaweed aquaculture monitoring, after feature adaptation and adjustment, it can be migrated to monitoring scenarios of blue carbon ecosystems such as mangroves, salt marshes, and seagrass beds, providing core technical support for coastal zone environmental resource management. Brief Description of the Drawings

[0032] Figure 1 It is a schematic diagram of the steps of a method for identifying large seaweed aquaculture areas based on Otsu feature enhancement;

[0033] Figure 2 It is a schematic diagram of the complete process of training and validating the method of the present invention;

[0034] Figure 3 It is a schematic diagram of the modules of a system for identifying large seaweed aquaculture areas based on Otsu feature enhancement;

[0035] Figure 4 It is a schematic diagram of the structure of a computer electronic device;

[0036] Figure 5 It is a schematic diagram of the detailed process of an embodiment of the present invention;

[0037] Figure 6 It is a diagram of the visual interpretation results of seaweed aquaculture areas, the classification results of the experimental group, and the classification results of the control group in an embodiment of the present invention. Detailed Embodiments

[0038] To make the above objects, features, and advantages of the present invention more obvious and understandable, the following describes the detailed embodiments of the present invention with reference to the accompanying drawings. Many specific details are set forth in the following description to fully understand the present invention. However, the present invention can be implemented in many other ways different from those described herein, and those skilled in the art can make similar improvements without departing from the connotation of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below. The technical features in each embodiment of the present invention can be combined correspondingly without conflict.

[0039] As Figure 1 shown, in a preferred embodiment of the present invention, a method for identifying large seaweed aquaculture areas based on Otsu feature enhancement is provided, which includes the following steps:

[0040] S1. Obtain multispectral remote sensing image data including blue, green, red, and near-infrared bands during the seaweed growth period in the target sea area, and perform image preprocessing operations to obtain preprocessed image data.

[0041] It should be noted that since there is a large amount of multispectral remote sensing image data, it is necessary to pre-screen the data suitable for remote sensing classification and interpretation. The image screening follows the following principles: The imaging time of the image conforms to the seaweed growth period, that is, from December of each year to April of the following year. At this time, the seaweed grows vigorously, with stronger identifiability, which is beneficial to subsequent remote sensing classification and interpretation.

[0042] In the embodiment of the present invention, the spatial resolution of the above-mentioned multispectral remote sensing image data is not less than 10 meters, and it needs to include multispectral data of blue, green, red, and near-infrared bands. Therefore, the present invention supports the input of multispectral images with a resolution of 10 meters to sub-meter level, and can adapt to mainstream satellite data such as Sentinel-2 and high-resolution series.

[0043] In addition, the image preprocessing operations adopted in the present invention include one or a combination of geometric correction, radiometric correction, atmospheric calibration, cloud masking, and land masking. Among them, land masking is used to remove the land area in the image, reduce the influence of irrelevant ground objects, and narrow the research scope; while cloud masking is used to remove cloud interference. The specific image preprocessing operations need to be selected according to the actual quality of the remote sensing image.

[0044] S2. Extract spectral features, index features, and spatial constraint features from the preprocessed image data to form multi-dimensional features; the spectral features are composed of blue, green, red, and near-infrared bands, and the index features are composed of at least one of the normalized difference vegetation index and the normalized difference water index; the spatial constraint feature is a binary mask of the target sea area generated by the Otsu algorithm based on the green band.

[0045] The key of the present invention is to perform adaptive threshold segmentation on the preprocessed image based on the Otsu algorithm to generate a binary mask representing the potential seaweed distribution. The Otsu algorithm is also known as the maximum inter-class variance method (Otsu algorithm), which was proposed by Japanese scholar Nobuyuki Otsu in 1979. This algorithm is mainly used for automatic threshold selection in the field of image processing, especially widely used in the binaryzation process of images. The core idea of the Otsu algorithm is to divide the image into foreground and background parts by traversing all possible thresholds, so that the inter-class variance between these two parts is maximized. The specific implementation of the Otsu algorithm belongs to the prior art and will not be elaborated in the present invention. The present invention extracts the binary mask of the target area from the green band through the Otsu algorithm, and can dynamically determine the boundary between the seaweed area and the non-seaweed area through gray-scale statistics, providing a spatial constraint benchmark for subsequent feature space construction.

[0046] S3. Input the multi-dimensional features into a supervised trained classification model. The classification model performs pattern recognition based on the multi-dimensional feature values at each pixel position to predict whether the pixel position belongs to the seaweed cultivation area, and finally outputs the spatial distribution map of the seaweed cultivation area.

[0047] It should be noted that in theory, any model capable of implementing pattern recognition and classification can be used for the classification model in the present invention. Preferably, a machine learning classifier with high-dimensional data processing capabilities is used, such as a logistic regression model, support vector machine, random forest, etc. In the embodiments of the present invention, a random forest algorithm is preferably used to construct the classifier. The parameter optimization mechanism can be set to dynamically adjust the model parameter combination through a preset number of iterations and accuracy evaluation indicators.

[0048] It should be noted that the above steps S1 to S3 describe the identification process of the large-scale seaweed cultivation area in actual application. However, those skilled in the art should know that the classification model is pre-trained using labeled sample data before being used for actual identification. The labeled sample data used in the present invention can be randomly collected from remote sensing images that have completed visual interpretation of the seaweed cultivation area, and each sample corresponds to a pixel. The visual interpretation result of the pixel position can be used as the true value label. After the model completes supervised training, accuracy verification is required, and it can only be used for actual classification after meeting the detection accuracy requirements. As Figure 2 shown, the specific process of the training and verification process of the above classification model is exemplarily shown.

[0049] In addition, different classification models need to be supervised and trained according to their respective training requirements, and the parameter settings during the supervised training of the classification model are dynamically adjusted through a preset parameter combination. The adjusted parameters include the classifier type selection strategy, the model training iteration control method, and the accuracy optimization mechanism, and the best parameter combination that can achieve the optimal classification performance is selected.

[0050] It should be noted that the method steps shown in the above S1 to S3 can essentially be implemented in the form of a computer program or a functional module.

[0051] Thus, based on the same inventive concept, as Figure 3 shown, the present invention also provides a large-scale seaweed cultivation area identification system based on Otsu feature enhancement corresponding to the large-scale seaweed cultivation area identification method based on Otsu feature enhancement provided in the above embodiment. It includes the following functional modules:

[0052] An image preprocessing module, configured to obtain multi-spectral remote sensing image data including blue, green, red and near-infrared bands during the seaweed growth period in the target sea area, and perform image preprocessing operations to obtain preprocessed image data;

[0053] A feature extraction module, which is used to extract spectral features, index features and spatial constraint features from the preprocessed image data to form multi-dimensional features; the spectral features are composed of a blue light band, a green light band, a red light band and a near-infrared band, and the index features are composed of at least one of a normalized difference vegetation index and a normalized difference water index; the spatial constraint feature is a binary mask of the target sea area generated by the Otsu algorithm based on the green light band.

[0054] A pattern recognition module, which is used to input the multi-dimensional features into a supervised trained classification model, and the classification model performs pattern recognition according to the multi-dimensional feature values at each pixel position to predict whether the pixel position belongs to a seaweed aquaculture area, and finally outputs a spatial distribution map of the seaweed aquaculture area.

[0055] In addition, based on the same inventive concept, as Figure 4 shown, the present invention also provides a computer electronic device corresponding to a method for identifying a large seaweed aquaculture area based on Otsu feature enhancement provided in the above embodiment, which includes a memory and a processor;

[0056] The memory is used to store a computer program;

[0057] The processor is used to implement the method for identifying a large seaweed aquaculture area based on Otsu feature enhancement as described above when executing the computer program;

[0058] In addition, when the logical instructions in the above-mentioned memory are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention.

[0059] Therefore, based on the same inventive concept, the present invention provides a computer-readable storage medium corresponding to a method for identifying a large seaweed aquaculture area based on Otsu feature enhancement. A computer program is stored on the storage medium, and when the computer program is executed by a processor, it can implement the method for identifying a large seaweed aquaculture area based on Otsu feature enhancement as described above.

[0060] Therefore, based on the same inventive concept, the present invention provides a computer program product, including a computer program / instructions, and when the computer program / instructions are executed by a processor, it can implement the method for identifying a large seaweed aquaculture area based on Otsu feature enhancement as described above.

[0061] Specifically, in the computer-readable storage media of the above three embodiments, the stored computer program is executed by a processor, and the steps of S1 to S3 can be executed.

[0062] It can be understood that the above storage media may include a random access memory (RAM), and may also include a non-volatile memory (NVM), such as at least one disk memory. At the same time, the storage media may also be various media such as a USB flash drive, a mobile hard disk, a magnetic disk, or an optical disc that can store program codes.

[0063] It can be understood that the above-mentioned processor may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it may also be a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

[0064] In addition, it should be noted that those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working process of the above-described system can refer to the corresponding process in the foregoing method embodiments, and will not be elaborated herein. In each of the embodiments provided in the present application, the division of steps or modules in the system and method is only a logical function division, and there may be other division methods in actual implementation. For example, multiple modules or steps may be combined or integrated together, and one module or step may also be split.

[0065] Next, the present invention will further show the detailed implementation process and technical effects of the large-scale seaweed aquaculture area recognition method based on Otsu feature enhancement shown in the above S1 to S3 steps on a specific dataset through a specific embodiment, so as to facilitate understanding the essence of the present invention.

[0066] Embodiment

[0067] As Figure 5As shown, in this embodiment, with reference to the large seaweed cultivation area classification method based on Otsu feature enhancement described in S1 to S3 above, the model construction, training and verification process are specifically carried out. In order to systematically verify the effect of Otsu feature enhancement on the classification accuracy of seaweed cultivation areas, this embodiment selects three typical experimental areas in large seaweed cultivation areas to carry out experiments. At the same time, two groups of experiments with spatial constraint features (experimental group) and without spatial constraint features (control group) are set up to classify and identify large seaweed cultivation areas. The technical implementation process is described in detail below in combination with the embodiment:

[0068] Step (1), obtaining and screening Sentinel-2 multispectral remote sensing image data during the seaweed growth period of the target sea area, and performing image cropping and cloud masking operations to obtain pre-processed image data.

[0069] In this embodiment, three typical experimental areas in the waters of South Jeolla Province, South Korea are selected as target sea areas to be identified. Sentinel-2MSI Level-2A data (which contains blue light, green light, red light and near-infrared bands) is called through Google Earth Engine, and a scene image that meets the conditions is screened out by boundary, time (must be within the seaweed growth period, that is, from December of each year to April of the following year) and cloud cover, and the image is cropped according to the three experimental areas. Since the image selected in this embodiment is not affected by clouds, cloud mask processing is not performed in the image preprocessing operation, but if there is thin cloud interference, it is necessary to use the QA60 band to mark the cloud pixels and eliminate them, so as to obtain the preprocessed image data of each experimental area.

[0070] Step (2): Select the green band image from the preprocessed image data of each experimental area, input the Otsu algorithm, dynamically calculate the optimal segmentation threshold, and generate a binary mask for each experimental area. In the binary mask, the potential algae area is marked as 1, and the background area is marked as 0.

[0071] Step (3): Multidimensional feature construction and comparative experimental design

[0072] a) The experimental group design is as follows:

[0073] The experimental group needs to construct a multidimensional feature space for the experimental area that includes spectral features, index features, and spatial constraint features, where:

[0074] The spectral characteristics are composed of blue light band, green light band, red light band and near infrared band;

[0075] The index characteristics adopt the combination of normalized vegetation index and normalized water index;

[0076] The spatial constraint feature is the binary mask generated in step (2).

[0077] b) The control group design is as follows:

[0078] In the control group, the Otsu spatial constraint feature is not introduced, that is, it only includes spectral features and index features.

[0079] All features in the experimental group and the control group are standardized before being input into the classification model.

[0080] S3. Input the multi-dimensional features into the supervised-trained classification model. The classification model performs pattern recognition based on the multi-dimensional feature values at each pixel position to predict whether the pixel position belongs to the seaweed cultivation area, and finally outputs the spatial distribution map of the seaweed cultivation area.

[0081] Step (4). Conduct visual interpretation based on the high-resolution reference image to determine the actual seaweed cultivation areas in each experimental area. Then, uniformly select sample points in the seaweed areas and background areas in each experimental area to construct a sample data set, where each sample point corresponds to a pixel. It should be noted that since the feature spaces in the experimental group and the control group are different, corresponding sample inputs need to be constructed separately in this embodiment. The sample input for each sample in the experimental group is spectral features, index features, and spatial constraint features, while the sample input for each sample in the control group is spectral features and index features. Each sample is labeled with a 0-1 label, where the seaweed area is labeled as 1 and the background area is labeled as 0. The sample data set is divided into a training set and a validation set according to a preset ratio. Note that the collected sample distribution covers different growth stages and typical interference scenarios.

[0082] Supervised classification mode is used to train different machine learning models, and their pattern recognition effects are tested separately. After comparison in this embodiment, the random forest classifier is finally selected to construct the large seaweed cultivation area recognition model. In the experimental group and the control group, supervised training classification is performed using the random forest respectively. The number of decision trees is determined through cross-validation, and the parameter combination is dynamically configured based on the criterion of optimal accuracy.

[0083] (5) After training is completed, use the trained random forest classifier on the validation set to output the recognition results of the large seaweed cultivation areas corresponding to the experimental group and the control group in the three regions, and conduct visual interpretation on the three cultivation areas as a reference benchmark for visual comparison. The comparison results are as Figure 6 shown. When comparing the classification results of the experimental group and the control group, the verification indicators include the Kappa coefficient, overall accuracy, and consistency coefficient. The results show that the experimental group shows significant advantages in all experimental areas, and the improvement amplitude of the classification accuracy reaches an observable level. At the same time, in the typical experimental area, the experimental group effectively suppresses background misjudgment through the spatial constraint feature, and the area error is significantly reduced compared with the control group.

[0084] In summary, the present invention reconstructs the Otsu binarization result from the terminal into spatial distribution features, significantly improves the classification accuracy in multiple test scenarios, effectively reduces the area error of seaweed monitoring, overcomes the defects of feature solidification and noise sensitivity in traditional methods, and is applicable to the near-real-time monitoring requirements in various scale scenarios.

[0085] The embodiments described above are only some preferred implementation solutions of the present invention, but are not intended to limit the present invention. Those of ordinary skill in the relevant technical fields can still make various changes and modifications without departing from the spirit and scope of the present invention. Therefore, all technical solutions obtained by adopting the means of equivalent replacement or equivalent transformation fall within the protection scope of the present invention.

Claims

1. A method for identifying large seaweed aquaculture areas based on Otsu feature enhancement, characterized in that, It includes the following steps: S1. Obtain multispectral remote sensing image data including blue light, green light, red light, and near-infrared bands during the seaweed growth period in the target sea area, and perform image preprocessing operations to obtain preprocessed image data; S2. Extract spectral features, index features, and spatial constraint features from the preprocessed image data to form multi-dimensional features; the spectral features are composed of blue light band, green light band, red light band, and near-infrared band, and the index features are composed of at least one of the normalized difference vegetation index and the normalized difference water index; the spatial constraint feature is a binary mask of the target sea area generated by the Otsu algorithm based on the green light band; S3. Input the multi-dimensional features into a supervised trained classification model, and the classification model performs pattern recognition according to the multi-dimensional feature values at each pixel position to predict whether the pixel position belongs to the seaweed cultivation area, and finally outputs a spatial distribution map of the seaweed cultivation area.

2. The method for identifying large seaweed aquaculture areas based on Otsu feature enhancement according to claim 1, wherein, The spatial resolution of the multispectral remote sensing image data is not less than 10 meters, and it needs to include multispectral data of blue light, green light, red light, and near-infrared bands.

3. The method for identifying large-scale seaweed aquaculture areas based on Otsu feature enhancement according to claim 1, wherein The image preprocessing operations include one or a combination of geometric correction, radiometric correction, atmospheric calibration, cloud masking, and land masking.

4. The method for identifying large seaweed cultivation areas based on Otsu feature enhancement according to claim 1, wherein The classification model uses a machine learning classifier, preferably a random forest classifier.

5. The method for identifying a large-scale seaweed cultivation area based on Otsu feature enhancement according to claim 1, wherein, Before being used for actual recognition, the classification model is pre-supervised trained using labeled sample data; the labeled sample data is randomly collected from remote sensing images that have completed visual interpretation of the seaweed cultivation area, and the visual interpretation result is used as the true value label.

6. The method for identifying large seaweed aquaculture areas based on Otsu feature enhancement according to claim 1, wherein, The parameter settings during the supervised training of the classification model are dynamically adjusted through a preset parameter combination. The adjusted parameters include the classifier type selection strategy, the model training iteration control method, and the accuracy optimization mechanism, and the best parameter combination that can achieve the optimal classification performance is selected.

7. A large-scale seaweed aquaculture area recognition system based on Otsu feature enhancement, characterized in that, It includes: An image preprocessing module for obtaining multispectral remote sensing image data including blue light, green light, red light, and near-infrared bands during the seaweed growth period in the target sea area, and performing image preprocessing operations to obtain preprocessed image data; A feature extraction module for extracting spectral features, index features, and spatial constraint features from the preprocessed image data to form multi-dimensional features; the spectral features are composed of blue light band, green light band, red light band, and near-infrared band, and the index features are composed of at least one of the normalized difference vegetation index and the normalized difference water index; the spatial constraint feature is a binary mask of the target sea area generated by the Otsu algorithm based on the green light band; A pattern recognition module for inputting the multi-dimensional features into a supervised trained classification model, and the classification model performs pattern recognition according to the multi-dimensional feature values at each pixel position to predict whether the pixel position belongs to the seaweed cultivation area, and finally outputs a spatial distribution map of the seaweed cultivation area.

8. A computer program product comprising a computer program / instructions, characterized in that, When the computer program / instructions are executed by a processor, it can implement the method for identifying large-scale seaweed cultivation areas based on Otsu feature enhancement according to any one of claims 1 to 7.

9. A computer-readable storage medium, characterized in that, A computer program is stored on the storage medium, and when the computer program is executed by a processor, the method for identifying large seaweed aquaculture areas based on Otsu feature enhancement as described in any one of claims 1 to 7 is implemented.

10. A computer electronic device, characterized in that, It includes a memory and a processor; The memory is used for storing a computer program; The processor is used for implementing the method for identifying large seaweed aquaculture areas based on Otsu feature enhancement as described in any one of claims 1 to 7 when executing the computer program.

Images