Deep Convolutional Asymmetric Autoencoder-Based Spatial-Spectral Clustering Network for Hyperspectral Image

Table 4

The evaluation contradistinction of proposed clustering method and other typical methods.

Method	RI	FMI	AMI	DB
(a)	0.9905	0.9888	0.9094	1.3924
(b)	0.9920	0.9907	0.9271	0.7089
(c)	0.9363	0.9236	0.7299	0.9643
(d)	0.8954	0.8871	0.7351	0.9045
(e)	0.8992	0.8759	0.7808	0.8140
(f)	0.9960	0.9954	0.9536	0.8359

Note: (a) PCA + K-Means, (b) AE + K-Means, (c) CAE + K-Means, (d) CAE + Over-clustering K-Means (e) DCAAES² + K-means, (f) DCAAES²C-Net, (g) Ground truth.