Search

CN-122023993-A - Coastal region coverage mapping optimization method and device based on artificial intelligence

CN122023993ACN 122023993 ACN122023993 ACN 122023993ACN-122023993-A

Abstract

The invention provides an artificial intelligence-based coastal region coverage mapping optimization method and device, and relates to the technical field of remote sensing mapping. The method comprises the steps of obtaining multi-source remote sensing data of coastal areas and preprocessing the multi-source remote sensing data to obtain standardized processing data, inputting the standardized processing data into a preset artificial intelligent model, fusing spectral features and spatial features through a feature extraction module based on a shift window attention mechanism to obtain layered fusion features, processing the layered fusion features through a classification detection module, and outputting coastal area coverage classification results and target detection results to achieve mapping optimization. According to the invention, through layered feature extraction and multi-source data fusion of the artificial intelligent model, the problems of spectrum and spatial feature separation, poor multi-source data adaptability and weak anti-interference capability in the traditional mapping method are solved, the coverage classification precision of coastal areas and the target detection efficiency of ships and the like are remarkably improved, and technical support is provided for coastal ecological protection and sustainable management.

Inventors

  • PENG MIN

Assignees

  • 广东省轻工业技师学院

Dates

Publication Date
20260512
Application Date
20260209

Claims (10)

  1. 1. The coastal region coverage mapping optimization method based on artificial intelligence is characterized by comprising the following steps of: Acquiring multi-source remote sensing data in coastal areas, and preprocessing the multi-source remote sensing data to obtain standardized processing data; inputting the standardized processing data into a preset artificial intelligent model, and fusing spectral features and spatial features through a feature extraction module of the preset artificial intelligent model to obtain layered fusion features, wherein the feature extraction module is constructed based on a shift window attention mechanism; And processing the layered fusion characteristics through a classification detection module of the preset artificial intelligent model, and outputting coastal region coverage classification results and target detection results to realize coastal region coverage mapping optimization.
  2. 2. The method of optimizing coverage mapping in coastal areas based on artificial intelligence according to claim 1, wherein the multi-source remote sensing data comprises at least two of hyperspectral data, optical aviation data and multi-temporal satellite data, the hyperspectral data has a spectral coverage of 400-2500nm, and the optical aviation data has an image resolution of not less than 512 x 512 pixels.
  3. 3. The method of artificial intelligence based coastal region coverage mapping optimization of claim 1, wherein preprocessing the multi-source remote sensing data comprises: Noise reduction processing is respectively carried out on each remote sensing data, and a low signal-to-noise ratio wave band and an invalid interference wave band are removed; The standardized processing is carried out through the calibration of the detector, and the system error brought by the data acquisition equipment is eliminated; Adopting a dimension reduction technology to reduce high-dimension data redundancy and reserving key characteristic wave bands; and carrying out format unification and space-time registration on the processed different source data to obtain the standardized processing data.
  4. 4. The coastal region coverage mapping optimization method based on artificial intelligence according to claim 1, wherein the feature extraction module comprises a block embedding unit, a Swin transform tandem module and a block merging unit which are sequentially connected, the Swin transform tandem module comprises two continuous sub-modules, and a multi-head self-attention mechanism based on windows and a multi-head self-attention mechanism based on shift windows are respectively adopted.
  5. 5. The method of optimizing coverage mapping in coastal areas based on artificial intelligence according to claim 4, wherein the process of fusing spectral features with spatial features comprises: dividing standardized processing data into non-overlapping blocks and converting the non-overlapping blocks into preset dimension vectors through a block embedding unit; Extracting local features and global features through a Swin transducer series module, reducing computational complexity through a multi-head self-attention mechanism, and realizing cross-window information interaction through the multi-head self-attention mechanism; And downsampling the feature map through a block merging unit, enhancing feature layering and completing depth fusion of the spectral features and the spatial features.
  6. 6. The artificial intelligence based coastal region coverage mapping optimization method of claim 1, further comprising: Aiming at the dynamic tidal influence of coastal areas, mangrove inundation indexes are introduced as tidal characterization parameters, and a tidal adaptation model is built by combining multi-temporal remote sensing data; and embedding the tide adaptation model into a classification detection module, correcting spectral characteristic distortion caused by tide change, and improving the coverage classification precision of the intertidal zone.
  7. 7. The coastal region coverage mapping optimization method based on artificial intelligence according to claim 1, wherein the coverage classification result output by the classification detection module comprises one or more types of coastal coverage types in mangrove, salt marsh, seaweed, river, land and ocean, the target detection result comprises position, quantity and type information of ship targets, and the ship target detection accuracy is not lower than 90.8%.
  8. 8. An artificial intelligence-based coastal region coverage mapping optimization device, which is characterized by comprising: The data preprocessing unit is used for acquiring multi-source remote sensing data in coastal areas, and performing noise reduction, standardization, dimension reduction and time-space registration processing on the multi-source remote sensing data to obtain standardized processing data; The feature fusion unit is used for inputting the standardized processing data into a preset artificial intelligent model, and obtaining layered fusion features by fusing spectral features and spatial features through a feature extraction module based on a shift window attention mechanism; The classification detection unit is used for processing the layered fusion characteristics through the classification detection module and outputting coastal area coverage classification results and target detection results; and the optimization and adjustment unit is used for introducing a tide adaptation model to correct the classification detection deviation and improving the anti-interference capability and the accuracy of the mapping result.
  9. 9. A computer device comprising a memory storing a computer program and a processor configured to execute the computer program to implement the artificial intelligence based coastal region coverage mapping optimization method of any one of claims 1-7.
  10. 10. A computer readable storage medium, characterized in that it stores a computer program, which when executed by a processor implements the artificial intelligence based coastal region coverage mapping optimization method of any one of claims 1-7.

Description

Coastal region coverage mapping optimization method and device based on artificial intelligence Technical Field The invention relates to the technical field of remote sensing mapping, in particular to a coastal region coverage mapping optimization method and device based on artificial intelligence. Background The coastal area is used as a key ecological zone for connecting land and sea, and plays important ecological functions of biodiversity protection, carbon storage, coastal protection and the like. However, the area faces multiple threats such as human activity interference, sea level rise, tidal dynamic change and the like, and accurate coverage mapping and dynamic monitoring become core preconditions for coastal ecological protection and sustainable management. The existing coastal area coverage mapping method mainly depends on a traditional machine learning algorithm (such as KNN and SVM) and a conventional Convolutional Neural Network (CNN), but has obvious technical limitations that firstly, the traditional machine learning algorithm can only rely on single spectral features or spatial features for classification, deep fusion of the traditional machine learning algorithm and the coastal area coverage mapping method cannot be achieved, so that classification accuracy is low (OA is generally lower than 80%) in a complex coastal environment, secondly, a short board exists in long-distance information interaction of a conventional CNN model, heterogeneous features of multi-source remote sensing data are difficult to process, suitability is poor, thirdly, tidal changes can cause dynamic changes of spectral features and visibility of coastal vegetation (such as mangrove), the existing method lacks a targeted anti-interference mechanism, the precision of intertidal area mapping is unstable, fourthly, detection of moving targets such as ships and the like and the coastal coverage classification tasks are mutually independent, a cooperative optimization mechanism is not formed, and the overall mapping efficiency is low. Therefore, how to construct an intelligent mapping method capable of deeply fusing spectrum and spatial characteristics of multi-source remote sensing data, adapting to complex environmental interference such as tides and the like and simultaneously considering coverage classification and target detection becomes a technical problem to be solved in the current coastal area coverage mapping field. Disclosure of Invention The invention provides an artificial intelligence-based coastal region coverage mapping optimization method and device, which solve the problems of spectrum and spatial feature separation, poor multi-source data adaptability and weak anti-interference capability in the traditional mapping method. In order to achieve the above object, the present invention is realized by the following technical scheme: in a first aspect, the invention provides an artificial intelligence-based coastal region coverage mapping optimization method, which comprises the following steps: Acquiring multi-source remote sensing data in coastal areas, and preprocessing the multi-source remote sensing data to obtain standardized processing data; inputting the standardized processing data into a preset artificial intelligent model, and fusing spectral features and spatial features through a feature extraction module of the preset artificial intelligent model to obtain layered fusion features, wherein the feature extraction module is constructed based on a shift window attention mechanism; And processing the layered fusion characteristics through a classification detection module of the preset artificial intelligent model, and outputting coastal region coverage classification results and target detection results to realize coastal region coverage mapping optimization. Further as an improvement of the technical scheme of the invention, the multi-source remote sensing data comprises at least two of hyperspectral data, optical aviation data and multi-temporal satellite data, the spectrum coverage range of the hyperspectral data is 400-2500nm, and the image resolution of the optical aviation data is not lower than 512 multiplied by 512 pixels. Further as an improvement of the technical scheme of the invention, the preprocessing of the multi-source remote sensing data comprises the following steps: Noise reduction processing is respectively carried out on each remote sensing data, and a low signal-to-noise ratio wave band and an invalid interference wave band are removed; The standardized processing is carried out through the calibration of the detector, and the system error brought by the data acquisition equipment is eliminated; Adopting a dimension reduction technology to reduce high-dimension data redundancy and reserving key characteristic wave bands; and carrying out format unification and space-time registration on the processed different source data to obtain the standardized processing data. Further as