3D Octave and 2D Vanilla Mixed Convolutional Neural Network for Hyperspectral Image Classification with Limited Samples

• Published in: Remote sensing (Basel, Switzerland) Vol. 13; no. 21; p. 4407
• Main Authors: Feng, Yuchao, Zheng, Jianwei, Qin, Mengjie, Bai, Cong, Zhang, Jinglin
• Format: Journal Article
• Language: English
• Published: Basel MDPI AG 01.11.2021
• Subjects: 3D octave and 2D vanilla mixed convolutions; 3D-2D mixed convolutional neural network (MCNN); Artificial neural networks; Classification; Computer applications; Computing costs; Deep learning; Discriminant analysis; Feature extraction; Feature maps; Homology; homology shifting; hyperspectral image (HSI) classification; Hyperspectral imaging; Image classification; Neural networks; principal component analysis (PCA)

Owing to the outstanding feature extraction capability, convolutional neural networks (CNNs) have been widely applied in hyperspectral image (HSI) classification problems and have achieved an impressive performance. However, it is well known that 2D convolution suffers from the absent consideration...