基于3D-CAE的高光谱解混及小样本分类方法

黄川, 李雅琴, 祁越然, 魏晓燕, 邵远征

A hyperspectral unmixing and few-shot classification method based on 3DCAE network

HUANG Chuan, LI Yaqin, QI Yueran, WEI Xiaoyan, SHAO Yuanzheng

表1 网络组成及相关参数设置

Tab.1 Network composition and related parameter settings

特征层 卷积核 大小 卷积核 数量/个 激活 函数 特征 尺寸 Conv3D-1 (1, 1, 7) 32 ReLU (1, 1, 155, 32) Conv3D-2 (1, 1, 7) 16 ReLU (1, 1, 148, 16) Conv3D-3 (1, 1, 7) 8 ReLU (1, 1, 141, 8) Conv3D-4 (1, 1, 7) 2 ReLU (1, 1, 134, 2) Flatten — — — 134×2 Dense-1 — 32 ReLU 32 Dense-2 — 6 Softmax 6 Dense-3 — 162 ReLU 162