A zero-sample remote sensing image land and sea segmentation method and system

Abstract

The present invention provides a zero-sample remote sensing image land and sea segmentation method and system. The zero-sample remote sensing image land and sea segmentation method extracts edge features from a received remote sensing image to generate an edge feature remote sensing image, segments the edge feature remote sensing image to generate a plurality of edge feature image blocks, and determines image attribute information of each edge feature image block in the plurality of edge feature image blocks. Based on the image attribute information, a mean clustering algorithm is used to determine foreground cue points, and the remote sensing image is further segmented and optimized based on the foreground cue points through an image segmentation model. By combining an open source image segmentation large model and an automatic cue point generation technology, land and sea segmentation without training data is achieved, thereby reducing dependence on manual annotation, improving segmentation accuracy and efficiency, and reducing application costs.

Description

Technical Field

[0001] The present invention relates to the field of image segmentation, and in particular to a method and system for zero-sample remote sensing image land and sea segmentation. Background Art

[0002] As a basic task in remote sensing image processing, land and sea segmentation is of great significance for application fields such as coastline extraction, maritime safety monitoring, and target detection. Traditional land and sea segmentation methods mainly rely on image features such as grayscale, texture, and spectrum, and are implemented through techniques such as threshold segmentation, edge detection, and region growing. However, these methods have obvious limitations when dealing with complex scenes. Traditional methods usually require manual adjustment of parameters such as thresholds and kernel sizes according to image characteristics, and have poor adaptability to different images. In addition, factors such as illumination changes, shadows, and cloud occlusions will significantly affect the segmentation effect, resulting in a decrease in segmentation accuracy. When processing large-scale remote sensing images, traditional methods have low computational efficiency and are difficult to meet real-time requirements.

[0003] With the rapid development of computer vision and deep learning technology, deep learning-based land and sea segmentation methods have gradually become a research hotspot. These methods can automatically learn the feature representation of images by building complex neural network models, significantly improving the accuracy and robustness of segmentation. In particular, the architecture based on convolutional neural network (CNN) has become the mainstream technology for land and sea segmentation. SeNet (Structured Edge Network) improves segmentation accuracy by combining local smoothing regularization and multi-task loss, but its adaptability to complex textures and illumination changes still needs to be improved. DeepUNet improves segmentation accuracy by introducing DownBlocks and UpBlocks as well as U-connection and Plus connection, but the computational efficiency and real-time performance when processing large-scale images still need to be improved. DBENet (Dual-Branch Ensemble Network) achieves effective feature extraction and information transmission through a dual-branch network architecture and integrated attention learning strategy, but its generalization ability on different remote sensing image datasets needs further verification. SANet (Scale-Adaptive Semantic Segmentation Network) improves the segmentation effect of complex coastlines through adaptive multi-scale feature learning and compression excitation modules, but its recognition ability for small targets and subtle structures is limited. TCUNet (A Lightweight Dual-Branch Parallel Network) combines the advantages of CNN and Transformer, improves segmentation accuracy through feature interaction modules and multi-scale multi-source feature fusion modules, but there is still room for optimization in terms of parameter number and computational complexity.

[0004] Although deep learning methods have made significant progress in land-sea segmentation, they also face some challenges. Deep learning methods usually require a large amount of labeled data for training, and obtaining a sufficient amount of high-quality labeled data is a daunting task in large-scale remote sensing image processing tasks. The deep learning model training process often requires a lot of computing resources, including high-performance hardware such as GPUs, which limits their application in resource-constrained environments. In application scenarios that require rapid response, such as maritime emergency response, the inference speed of deep learning methods may not meet real-time requirements.

[0005] Therefore, how to achieve land and sea segmentation of remote sensing images in a resource-constrained environment is of great significance. Summary of the invention

[0006] In order to solve the problems of strong dependence on training data, high image annotation cost and large consumption of computing resources in the image segmentation process of the prior art, the present invention proposes a zero-sample remote sensing image land and sea segmentation method and system.

[0007] In a first aspect, a zero-sample remote sensing image land and sea segmentation method is provided, comprising:

[0008] Extract edge features from the received remote sensing image to generate an edge feature remote sensing image;

[0009] Segmenting the edge feature remote sensing image to generate a plurality of edge feature image blocks, and determining image attribute information of each edge feature image block in the plurality of edge feature image blocks;

[0010] Based on the image attribute information, a mean clustering algorithm is used to determine a foreground prompt point;

[0011] The remote sensing image is segmented based on the foreground cue points using an image segmentation model.

[0012] Preferably, the method of determining the foreground cue point based on the image attribute information by using a mean value clustering algorithm includes:

[0013] Clustering is performed based on the image attribute information of each edge feature image block by using a mean value clustering algorithm, a cluster center point is determined, and the cluster center point is used as a foreground prompt point.

[0014] Preferably, said extracting edge features from the received remote sensing image to generate an edge feature remote sensing image comprises:

[0015] Down-sampling the received remote sensing image to generate an initial remote sensing image;

[0016] Performing partition processing on the initial remote sensing image based on a normalized cutting algorithm to generate a remote sensing image after partition processing;

[0017] Based on the edge detection algorithm, edge features are extracted from the remote sensing image after the partition processing to generate an edge feature remote sensing image.

[0018] Preferably, segmenting the edge feature remote sensing image to generate a plurality of edge feature image blocks comprises:

[0019] Acquire image pixel information of the edge feature remote sensing image;

[0020] The edge feature remote sensing image is equally divided into a plurality of edge feature image blocks based on the image pixel information.

[0021] Preferably, the determining of the image attribute information of each edge feature image block in the plurality of edge feature image blocks comprises:

[0022] Determine the number of white pixels in each edge feature image block in the plurality of edge feature image blocks;

[0023] Determine whether the number of white pixels is greater than a preset number threshold;

[0024] If yes, the edge feature image block is determined as land image attribute information;

[0025] If not, the edge feature image block is determined as ocean image attribute information.

[0026] Preferably, after performing image segmentation on the remote sensing image based on the foreground cue points by using an image segmentation model, the method further comprises:

[0027] Performing logical correction on the sea and land mask generated after image segmentation of the remote sensing image based on the position distribution information of the foreground prompt point in the edge feature image block;

[0028] The image denoising is performed on the logically corrected sea and land mask based on the median filtering technology.

[0029] In a second aspect, a zero-sample remote sensing image land and sea segmentation system is provided, comprising:

[0030] An extraction module is used to extract edge features from the received remote sensing image and generate an edge feature remote sensing image;

[0031] A generating module, used for segmenting the edge feature remote sensing image to generate a plurality of edge feature image blocks, and determining image attribute information of each edge feature image block in the plurality of edge feature image blocks;

[0032] A determination module, used to determine the foreground prompt point by using a mean value clustering algorithm based on the image attribute information;

[0033] The segmentation module is used to perform image segmentation on the remote sensing image based on the foreground cue points through an image segmentation model.

[0034] Preferably, the determining module is specifically used for:

[0035] Clustering is performed based on the image attribute information of each edge feature image block by using a mean value clustering algorithm, a cluster center point is determined, and the cluster center point is used as a foreground prompt point.

[0036] Preferably, the extraction module is specifically used for:

[0037] Down-sampling the received remote sensing image to generate an initial remote sensing image;

[0038] Performing partition processing on the initial remote sensing image based on a normalized cutting algorithm to generate a remote sensing image after partition processing;

[0039] Based on the edge detection algorithm, edge features are extracted from the remote sensing image after the partition processing to generate an edge feature remote sensing image.

[0040] Preferably, the generating module segments the edge feature remote sensing image to generate a plurality of edge feature image blocks, including:

[0041] Acquire image pixel information of the edge feature remote sensing image;

[0042] The edge feature remote sensing image is equally divided into a plurality of edge feature image blocks based on the image pixel information.

[0043] Preferably, the generating module determines the image attribute information of each edge feature image block in the plurality of edge feature image blocks, including:

[0044] Determine the number of white pixels in each edge feature image block in the plurality of edge feature image blocks;

[0045] Determine whether the number of white pixels is greater than a preset number threshold;

[0046] If yes, the edge feature image block is determined as land image attribute information;

[0047] If not, the edge feature image block is determined as ocean image attribute information.

[0048] Preferably, the system further comprises:

[0049] A correction module, used for performing logical correction on the land and sea mask generated after image segmentation of the remote sensing image based on the position distribution information of the foreground prompt point in the edge feature image block;

[0050] The image denoising is performed on the logically corrected sea and land mask based on the median filtering technology.

[0051] In another aspect, the present application further provides an electronic device, comprising: at least one processor and a memory; the memory and the processor are connected via a bus;

[0052] The memory is used to store one or more programs;

[0053] When the one or more programs are executed by the at least one processor, the zero-sample remote sensing image land and sea segmentation method as described above is implemented.

[0054] On the other hand, the present application also provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed, the zero-sample remote sensing image land and sea segmentation method as described above is implemented.

[0055] Compared with the prior art, the present invention has the following beneficial effects:

[0056] The present invention provides a zero-sample remote sensing image land-sea segmentation method and system. The zero-sample remote sensing image land-sea segmentation method extracts edge features from a received remote sensing image to generate an edge feature remote sensing image, segments the edge feature remote sensing image to generate a plurality of edge feature image blocks, and determines image attribute information of each edge feature image block in the plurality of edge feature image blocks. Based on the image attribute information, a mean clustering algorithm is used to determine foreground cue points, and the remote sensing image is further segmented based on the foreground cue points through an image segmentation model. By combining an open source image segmentation large model and an automatic cue point generation technology, land-sea segmentation without training data is achieved, the reliance on manual annotation is reduced, the segmentation accuracy and efficiency are improved, and the application cost is reduced. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 It is a flow chart of the zero-sample remote sensing image land and sea segmentation method of the present invention;

[0058] Figure 2 It is a specific flow chart of the zero-sample remote sensing image land and sea segmentation method of the present invention;

[0059] Figure 3 The intermediate results and final results of each step of the zero-sample remote sensing image land and sea segmentation method of the present invention;

[0060] Figure 4It is a structural schematic diagram of the zero-sample remote sensing image land and sea segmentation system of the present invention;

[0061] Figure 5 The figure is a schematic diagram of the structure of an electronic device of the present invention. DETAILED DESCRIPTION

[0062] The present invention provides a zero-sample remote sensing image land-sea segmentation method and system. The zero-sample remote sensing image land-sea segmentation method extracts edge features from a received remote sensing image to generate an edge feature remote sensing image, segments the edge feature remote sensing image to generate a plurality of edge feature image blocks, and determines image attribute information of each edge feature image block in the plurality of edge feature image blocks. Based on the image attribute information, a mean clustering algorithm is used to determine foreground cue points, and the remote sensing image is further segmented based on the foreground cue points through an image segmentation model. By combining an open source image segmentation large model and an automatic cue point generation technology, land-sea segmentation without training data is achieved, the reliance on manual annotation is reduced, the segmentation accuracy and efficiency are improved, and the application cost is reduced.

[0063] In order to better understand the present invention, the content of the present invention is further described below in conjunction with the accompanying drawings and embodiments.

[0064] Embodiment 1:

[0065] A zero-shot remote sensing image land and sea segmentation method, such as Figure 1 As shown, including:

[0066] Step S1: extract edge features from the received remote sensing image to generate an edge feature remote sensing image;

[0067] Step S2: segmenting the edge feature remote sensing image to generate a plurality of edge feature image blocks, and determining image attribute information of each edge feature image block in the plurality of edge feature image blocks;

[0068] Step S3: Based on the image attribute information, a mean clustering algorithm is used to determine the foreground prompt point;

[0069] Step S4: performing image segmentation on the remote sensing image based on the foreground cue points using an image segmentation model.

[0070] In this embodiment, in step S1, the process of extracting edge features from the received remote sensing image to generate an edge feature remote sensing image includes:

[0071] Down-sampling the received remote sensing image to generate an initial remote sensing image;

[0072] Performing partition processing on the initial remote sensing image based on a normalized cutting algorithm to generate a remote sensing image after partition processing;

[0073] The edge features of the remote sensing image after the partition processing are extracted based on the edge detection algorithm to generate an edge feature remote sensing image.

[0074] In one embodiment, the received remote sensing image is downsampled to reduce computational complexity, and then the remote sensing image is roughly partitioned using the NCuts algorithm to effectively shield the influence of noise, and then the Canny edge detection algorithm is used to extract the edge features of the remote sensing image to form an edge feature remote sensing image.

[0075] In this embodiment, in step S2, the process of segmenting the edge feature remote sensing image to generate a plurality of edge feature image blocks includes:

[0076] Acquire image pixel information of the edge feature remote sensing image;

[0077] The edge feature remote sensing image is equally divided into a plurality of edge feature image blocks based on the image pixel information.

[0078] In this embodiment, in step S2, the process of determining the image attribute information of each edge feature image block in the plurality of edge feature image blocks includes:

[0079] Determine the number of white pixels in each edge feature image block in the plurality of edge feature image blocks;

[0080] Determine whether the number of white pixels is greater than a preset number threshold;

[0081] If yes, the edge feature image block is determined as land image attribute information;

[0082] If not, the edge feature image block is determined as ocean image attribute information.

[0083] In this embodiment, in step S3, the process of determining the foreground prompt point by using the mean value clustering algorithm based on the image attribute information includes:

[0084] Clustering is performed based on the image attribute information of each edge feature image block by using a mean value clustering algorithm, a cluster center point is determined, and the cluster center point is used as a foreground prompt point.

[0085] In one embodiment, a K-Means clustering algorithm is used for each edge feature image block containing the sea surface and the land to find the cluster center point as the foreground cue point.

[0086] In this embodiment, after performing image segmentation on the remote sensing image based on the foreground cue points by using the image segmentation model in step S4, the following steps are also included:

[0087] Performing logical correction on the sea and land mask generated after image segmentation of the remote sensing image based on the position distribution information of the foreground prompt point in the edge feature image block;

[0088] The image denoising is performed on the logically corrected sea and land mask based on the median filtering technology.

[0089] Figure 2 is a specific flow chart of the zero-sample remote sensing image land and sea segmentation method of the present invention, Figure 3 The intermediate results and final results of each step of the zero-sample remote sensing image land and sea segmentation method of the present invention are as follows. Figure 2 and Figure 3 The zero-sample remote sensing image land and sea segmentation method of the present invention is introduced.

[0090] Step 1: Image preprocessing is the first step in the entire land-sea segmentation process. Its purpose is to provide high-quality input for subsequent image analysis and segmentation. The specific process is as follows:

[0091] Step 11: Receive an original remote sensing image (i.e., Original Image) and perform downsampling on it to reduce the resolution of the image. For example, downsample the image from the original resolution to 512×512 pixels. This step can significantly reduce the amount of computation for subsequent processing while retaining sufficient image details.

[0092] Step 12: Apply the Normalized Cuts (NCuts) algorithm to roughly partition the downsampled image. The NCuts algorithm can effectively divide the image into multiple regions, reduce the impact of noise, and retain the main features of the image. In this way, clearer image blocks can be provided for subsequent edge detection and feature extraction.

[0093] Step 13: Use the Canny edge detection algorithm to perform edge detection on the image block processed by NCuts to extract the edge features of the image. The Canny algorithm is an efficient edge detection method that can accurately identify the edge information in the image and provide key visual clues for subsequent image block classification and cue point generation.

[0094] Step 2: Classification and analysis of image blocks is one of the core steps of the present invention, and its purpose is to determine the sea and land attributes of each image block. The specific steps are as follows:

[0095] Step 21: Divide the edge feature map into multiple small blocks, for example, divide the image into 16 small blocks of 8×8 pixels. Each small block will be used as the basic unit for subsequent analysis.

[0096] Step 22: Attribute analysis is performed on each image block, mainly based on the number of white pixels in the block. By counting the number of white pixels in each block, it can be determined whether the block is land or ocean. For example, if the number of white pixels in a block exceeds a certain threshold, the block is determined to be land; otherwise, it is determined to be ocean.

[0097] Step 3: Automatic prompt point generation is one of the key innovations of the present invention, and its purpose is to provide accurate segmentation prompts for the image segmentation model. The specific steps are as follows:

[0098] Step 31: Perform K-Means clustering on each image block containing sea surface and land. K-Means algorithm is a commonly used clustering method that can divide data points into multiple clusters. In the present invention, the cluster center point will be used as a foreground prompt point and passed to the image segmentation model.

[0099] Step 32: Select cluster center points from the K-Means clustering results as foreground prompt points. These prompt points will guide the image segmentation model to more accurately identify and segment land areas. In order to improve segmentation accuracy, avoid generating background points, because background points may mislead the model and cause segmentation errors.

[0100] Step 4: Image segmentation based on cue points is another key step of the present invention, and its purpose is to use the automatically generated cue points to accurately segment the image. The specific steps of this process are as follows:

[0101] Step 41: Perform preliminary image segmentation using Segment Anything Model (SAM) and automatically generated prompt points. SAM is an advanced image segmentation model that can perform efficient image segmentation based on prompt points, where initial segmentation result represents the initial segmentation result, and refining the results with promptpoints represents refining the results with promptpoints, where S represents the segmented ocean image and L represents the segmented land image.

[0102] Step 5: Optimization and correction of segmentation results is the last step of the present invention, and its purpose is to further improve the accuracy and robustness of the segmentation results. The following are the specific implementation steps:

[0103] Step 51: Perform logical correction on the preliminary segmentation result according to the distribution of the prompt points. For example, if it is found that more than half of the prompt points fall in the land area, the segmentation result is negated to ensure the accuracy of the segmentation result.

[0104] Step 52: Apply image processing techniques such as median filtering to remove noise generated during the segmentation process. Median filtering is a commonly used image denoising method that can effectively smooth the image and reduce the impact of noise.

[0105] Embodiment 2:

[0106] The present invention based on the same inventive concept also provides a zero-sample remote sensing image land and sea segmentation system, such as Figure 4 As shown, including:

[0107] An extraction module is used to extract edge features from the received remote sensing image and generate an edge feature remote sensing image;

[0108] A generating module, used for segmenting the edge feature remote sensing image to generate a plurality of edge feature image blocks, and determining image attribute information of each edge feature image block in the plurality of edge feature image blocks;

[0109] A determination module, used to determine the foreground prompt point by using a mean value clustering algorithm based on the image attribute information;

[0110] The segmentation module is used to perform image segmentation on the remote sensing image based on the foreground cue points through an image segmentation model.

[0111] Preferably, the determining module is specifically used for:

[0112] Clustering is performed based on the image attribute information of each edge feature image block by using a mean value clustering algorithm, a cluster center point is determined, and the cluster center point is used as a foreground prompt point.

[0113] Preferably, the extraction module is specifically used for:

[0114] Down-sampling the received remote sensing image to generate an initial remote sensing image;

[0115] Performing partition processing on the initial remote sensing image based on a normalized cutting algorithm to generate a remote sensing image after partition processing;

[0116] Based on the edge detection algorithm, edge features are extracted from the remote sensing image after the partition processing to generate an edge feature remote sensing image.

[0117] Preferably, the generating module segments the edge feature remote sensing image to generate a plurality of edge feature image blocks, including:

[0118] Acquire image pixel information of the edge feature remote sensing image;

[0119] The edge feature remote sensing image is equally divided into a plurality of edge feature image blocks based on the image pixel information.

[0120] Preferably, the generating module determines the image attribute information of each edge feature image block in the plurality of edge feature image blocks, including:

[0121] Determine the number of white pixels in each edge feature image block in the plurality of edge feature image blocks;

[0122] Determine whether the number of white pixels is greater than a preset number threshold;

[0123] If yes, the edge feature image block is determined as land image attribute information;

[0124] If not, the edge feature image block is determined as ocean image attribute information.

[0125] Preferably, the system further comprises:

[0126] A correction module, used for performing logical correction on the land and sea mask generated after image segmentation of the remote sensing image based on the position distribution information of the foreground prompt point in the edge feature image block;

[0127] The image denoising is performed on the logically corrected sea and land mask based on the median filtering technology.

[0128] Example 3

[0129] like Figure 5 As shown, the present invention also provides an electronic device, which may be a computer device, a single-chip device, an intelligent mobile device, etc. The electronic device in this embodiment may include a processor, a memory, a transceiver component, etc. The memory, the processor, and the transceiver component are connected via a bus; the memory may be used to store an execution program, and an exemplary execution program may include instructions; the processor is used to execute the instructions stored in the memory. The memory may also be used to store data, which may be called and / or modified when the instructions are executed.

[0130] The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. It is the computing core and control core of the terminal, which is suitable for implementing one or more instructions, specifically suitable for loading and executing one or more instructions in a storage medium to implement the corresponding method flow or corresponding functions, so as to implement the steps of a zero-sample remote sensing image land and sea segmentation method in the above embodiment.

[0131] Example 4

[0132] Based on the same inventive concept, the present invention also provides a readable storage medium, specifically an electronic device readable storage medium (Memory), which is a memory device in the electronic device for storing programs and data. It can be understood that the storage medium here can include both the built-in storage medium in the electronic device and the extended storage medium supported by the electronic device. The storage medium provides a storage space, which stores the operating system of the terminal. In addition, one or more instructions suitable for being loaded and executed by the processor are also stored in the storage space, and these instructions can be one or more execution programs (including program codes). It should be noted that the storage medium here can be a high-speed RAM memory or a non-volatile memory, such as at least one disk memory. The processor loads and executes one or more instructions stored in the storage medium, which can implement the steps of a zero-sample remote sensing image sea and land segmentation method in the above embodiment.

[0133] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may 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.

[0134] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0135] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0136] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0137] The above are merely embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention are included in the scope of the claims of the present invention to be approved.

Claims

1. A zero-sample remote sensing image land and sea segmentation method, characterized in that: include: Extract edge features from the received remote sensing image to generate an edge feature remote sensing image; Segmenting the edge feature remote sensing image to generate a plurality of edge feature image blocks, and determining image attribute information of each edge feature image block in the plurality of edge feature image blocks; Based on the image attribute information, a mean clustering algorithm is used to determine the foreground prompt point; The remote sensing image is segmented based on the foreground cue points using an image segmentation model.

2. The method according to claim 1, characterized in that The method of determining the foreground prompt point based on the image attribute information by using a mean value clustering algorithm includes: Clustering is performed based on the image attribute information of each edge feature image block by using a mean value clustering algorithm, a cluster center point is determined, and the cluster center point is used as a foreground prompt point.

3. The method according to claim 1, characterized in that The step of extracting edge features from the received remote sensing image to generate an edge feature remote sensing image includes: Down-sampling the received remote sensing image to generate an initial remote sensing image; Performing partition processing on the initial remote sensing image based on a normalized cutting algorithm to generate a remote sensing image after partition processing; Based on the edge detection algorithm, edge features are extracted from the remote sensing image after the partition processing to generate an edge feature remote sensing image.

4. The method according to claim 1, characterized in that: The step of segmenting the edge feature remote sensing image to generate a plurality of edge feature image blocks comprises: Acquire image pixel information of the edge feature remote sensing image; The edge feature remote sensing image is equally divided into a plurality of edge feature image blocks based on the image pixel information.

5. The method according to claim 1, characterized in that The step of determining the image attribute information of each edge feature image block in the plurality of edge feature image blocks comprises: Determine the number of white pixels in each edge feature image block in the plurality of edge feature image blocks; Determine whether the number of white pixels is greater than a preset number threshold; If yes, the edge feature image block is determined as land image attribute information; If not, the edge feature image block is determined as ocean image attribute information.

6. The method according to claim 1, characterized in that After performing image segmentation on the remote sensing image based on the foreground prompting points by using the image segmentation model, the method further includes: Performing logical correction on the sea and land mask generated after image segmentation of the remote sensing image based on the position distribution information of the foreground prompt point in the edge feature image block; The image denoising is performed on the logically corrected sea and land mask based on the median filtering technology.

7. A zero-sample remote sensing image land and sea segmentation system, characterized in that: include: An extraction module is used to extract edge features from the received remote sensing image and generate an edge feature remote sensing image; A generating module, used for segmenting the edge feature remote sensing image to generate a plurality of edge feature image blocks, and determining image attribute information of each edge feature image block in the plurality of edge feature image blocks; A determination module, used to determine the foreground prompt point by using a mean value clustering algorithm based on the image attribute information; The segmentation module is used to perform image segmentation on the remote sensing image based on the foreground cue points through an image segmentation model.

8. The system according to claim 7, characterized in that The system further comprises: A correction module, used for performing logical correction on the land and sea mask generated after image segmentation of the remote sensing image based on the position distribution information of the foreground prompt point in the edge feature image block; The image denoising is performed on the logically corrected sea and land mask based on the median filtering technology.

9. An electronic device, characterized in that: include: at least one processor and memory; The memory and the processor are connected via a bus; The memory is used to store one or more programs; When the one or more programs are executed by the at least one processor, the zero-sample remote sensing image land and sea segmentation method according to any one of claims 1 to 6 is implemented.

10. A readable storage medium, characterized in that: An execution program is stored thereon, and when the execution program is executed, the zero-sample remote sensing image land and sea segmentation method as described in any one of claims 1 to 6 is implemented.

Images