A Spatial-spectral Multi-view Contrastive Learning Framework for Scene Representation Learning

doi:10.1016/j.inffus.2025.103889

A Spatial-spectral Multi-view Contrastive Learning Framework for Scene Representation Learning

2025-11-01

0

PRE

AI

DOI：10.1016/j.inffus.2025.103889

Original

Original request for help

Save

Abstract

En 中文

• Combines FFT with contrastive learning to enhance spectral feature extraction. • R-MSA mechanism improves feature updating and 3D space information processing. • Complex scene representation by integrating spatial and spectral domain features. • Expands scope by integrating point clouds and images into a unified framework.

A Spatial-spectral Multi-view Contrastive Learning Framework for Scene Representation Learning

Abstract

Journal

Organization