Answers are grounded in this paper and cite their sources.
Hyperspectral remote sensing technology in wetland precise mapping plays a crucial role in ecological monitoring. Currently, hyperspectral image (HSI) wetland mapping achieves remarkable progress with the development of deep learning (DL) technology. Nevertheless, the existing approaches focus on establishing the elaborate structure of the DL model and overlook the inherent characteristic of abundant vegetation information of wetland imageries. In this paper, we present a novel multi-view vegetation information perception (V3IP) network for HSI wetland mapping, which significantly enhances the discriminative representation of feature extraction by incorporating three presented vegetation perception components. Specifically, the vegetation index fusion module involves explicit vegetation index information in the original HSI data and highlights the vegetation information of the fed training samples. The presented vegetation form perception module incorporates the perception of vegetation form in the classification model, which is beneficial for the discriminative representation of vegetation features in the wetland. Moreover, the vegetation spectral saliency perception module adequately captures the perception of vegetation spectral saliency and mitigates the impact of spectral variations of HSI. In the experimental part, we evaluate the performance of the proposed V3IP model with two Yellow River estuary wetland datasets compared with existing DL methods for wetland mapping.
No takes yet. Share an insight, caveat, or question.
Yu et al. (2025) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: