Remote Sensing for Natural Resources >
A hyperspectral unmixing and few-shot classification method based on 3DCAE network
Received date: 2023-08-28
Revised date: 2024-09-01
Online published: 2026-06-03
The rapid development of hyperspectral remote sensing technology in China fully ensures the effective application of large-scale surface feature classification. However, achieving high-precision classification under few-spot conditions while fully leveraging hyperspectral spatial-spectral information remains challenging. This study developed a 3D convolutional autoencoder (3D-CAE) network guided by physical constraints from mixed pixel decomposition. This approach enables accurate estimation of endmember abundance while effectively expressing regularized spatial-spectral features of hyperspectral data. In combination with a support vector machine (SVM) classifier, the method achieves hyperspectral classification under few-spot conditions. The classification performance of various models was evaluated at different sampling rates. To validate the proposed method, this study conducted experiments including comparisons with traditional hyperspectral feature extraction and classification methods, such as supervised classification approaches. The classification performance of various models was also evaluated at different sampling rates. The experimental results demonstrate that the proposed hyperspectral classification method has a significant advantage of accuracy, achieving a mean intersection over union (mIoU) of 0.829, which was close to 0.8 even at a low sampling rate of 1/200, surpassing its counterparts. These results confirm that the proposed method exhibits robustness under few-spot conditions. This study provides a valuable technical reference for addressing hyperspectral classification challenges under few-spot conditions.
HUANG Chuan , LI Yaqin , QI Yueran , WEI Xiaoyan , SHAO Yuanzheng . A hyperspectral unmixing and few-shot classification method based on 3DCAE network[J]. Remote Sensing for Natural Resources, 2025 , 37(1) : 8 -14 . DOI: 10.6046/zrzyyg.2023260
图2 基于3D-CAE的高光谱分类流程Fig.2 The hyperspectral classification workflow based on 3D-CAE |
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 |
表2 模型预测不同端元的丰度值与真实值Tab.2 Model prediction of the abundance from different endmembers compared to the true values |
| 类型 | 沥青 | 草地 | 树木 | 屋顶 | 金属 | 土壤 |
|---|---|---|---|---|---|---|
| 预测丰度 | ![]() | |||||
| 真实值 | ||||||
| RMSE | ||||||
| 图例 | ![]() | |||||
图3 不同模型预测分类结果与真值对比图Fig.3 Comparison chart of classification results predicted by different models versus ground truth |
表3 不同模型预测精度对比表Tab.3 Comparison table of prediction accuracy for different models |
| 模型 | 总体分 类精度 | 精确度 | 召回率 | F1得分 | mIoU |
|---|---|---|---|---|---|
| 3D-CAE-SVM | 0.927 | 0.936 | 0.883 | 0.905 | 0.829 |
| PCA-SVM | 0.912 | 0.928 | 0.854 | 0.884 | 0.796 |
| MNF-SVM | 0.895 | 0.913 | 0.847 | 0.873 | 0.779 |
| FCLS-SVM | 0.850 | 0.832 | 0.775 | 0.778 | 0.657 |
| SVM | 0.916 | 0.929 | 0.856 | 0.885 | 0.799 |
| SAM | 0.847 | 0.870 | 0.831 | 0.821 | 0.777 |
| SID | 0.712 | 0.652 | 0.771 | 0.625 | 0.513 |
| [1] |
杜培军, 谭琨, 夏俊士. 高光谱遥感影像分类与支持向量机应用研究[M]. Beijing: Science Press, 2012,1-160.
|
| [2] |
|
| [3] |
刘银年, 孙德新, 胡晓宁, 等. 高分五号可见短波红外高光谱相机设计与研制[J]. 遥感学报, 2020, 24(4):333-344.
|
| [4] |
|
| [5] |
|
| [6] |
|
| [7] |
|
| [8] |
|
| [9] |
|
| [10] |
|
| [11] |
|
| [12] |
|
| [13] |
|
| [14] |
|
| [15] |
|
| [16] |
|
| [17] |
|
| [18] |
|
| [19] |
|
| [20] |
|
| [21] |
|
| [22] |
韩竹, 高连如, 张兵, 等. 高分五号高光谱图像自编码网络非线性解混[J]. 遥感学报, 2020, 24(4):388-400.
|
| [23] |
朱玲, 秦凯, 李明, 等. 改进的堆栈稀疏自编码矿物高光谱端元识别研究[J]. 光谱学与光谱分析, 2021, 41(4):1288-1293.
|
| [24] |
朱治青, 苏远超, 李朋飞, 等. 双约束深度卷积网络的高光谱图像空谱解混方法[J]. 信号处理, 2023, 39(1):128-142.
|
| [25] |
|
| [26] |
|
| [27] |
|
| [28] |
|
| [29] |
|
| [30] |
|
| [31] |
|
| [32] |
杜培军, 夏俊士, 薛朝辉, 等. 高光谱遥感影像分类研究进展[J]. 遥感学报, 2016, 20(2):236-256.
|
/
| 〈 |
|
〉 |