研究目的
To obtain a robust classification methodology to generate accurate benthic habitat maps applying object-oriented and pixel-based classification methods in shallow waters using WorldView-2 and AHS (Airborne Hyperspectral Scanner) images.
研究成果
The study concludes that SVM and the object-based classification approach (OBIA) provide the best results for generating benthic habitat maps, with ML also being a viable option. Hyperspectral data (AHS) offers slightly better accuracy than multispectral data (WV-2), but the improvement is not significant for the coarse number of classes considered. The inclusion of texture information improves classification accuracy. Future work includes applying water column correction algorithms and using HS imagery from drone platforms for more precise benthic maps.
研究不足
The study faces challenges due to the low signal-to-noise ratio at the sensor level, atmospheric and water column disturbances, and the complexity of discriminating mixed classes in seagrass meadows. The reference ecocartographic map from 2000 may not fully represent current conditions.
1：Experimental Design and Method Selection:
The study compares hyperspectral (HS) and multispectral (MS) imagery for seafloor mapping, focusing on seagrass meadows in shallow waters. It employs object-oriented and pixel-based classification methods, including Maximum Likelihood (ML), Spectral Angle Mapper (SAM), and Support Vector Machine (SVM) algorithms.
2：Sample Selection and Data Sources:
The study area is the Natural Reserve of Maspalomas (Gran Canaria, Spain), known for its complex coastal-dune ecosystem and seagrass beds of Cymonocea nodosa. Data sources include INTA Airborne Hyperspectral Scanner (AHS) and WorldView-2 (WV-2) satellite imagery, along with in-situ field data.
3：List of Experimental Equipment and Materials:
Equipment includes the INTA Airborne Hyperspectral Scanner (AHS), WorldView-2 satellite, ecosounder Reson Navisound 110, Neptune and Go Pro Hero 3+ cameras, and a differential GPS receiver (Trimble DSM132).
4：2). Experimental Procedures and Operational Workflow:
4. Experimental Procedures and Operational Workflow: The workflow involves corrections (radiometric, atmospheric, sunglint, and water column), feature extraction (including PCA, ICA, MNF transforms, and textural features), and classification using ML, SAM, and SVM algorithms.
5：Data Analysis Methods:
Accuracy assessment is performed using test regions of interest (ROIs), computing the kappa coefficient and confusion matrix.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
