Hybrid texture-structural learning for hyperspectral image classification

doi:10.1016/j.patcog.2025.112526

Hybrid texture-structural learning for hyperspectral image classification

2025-10-04

0

PRE

AI

DOI：10.1016/j.patcog.2025.112526

Original

Original request for help

Save

Abstract

En 中文

• TDConv aggregates multiple convolution kernels by the texture structure information. • TDConv enhances the ability to capture local features. • MHTS2A incorporates texture and spectral information into self-attention mechanism. • MHTS2A can conduct more comprehensive feature characterization. • HTLNet can effectively aggregate frequency-domain and spatial-domain features.

Hybrid texture-structural learning for hyperspectral image classification

Abstract

Journal

Organization