Label Consistent Transform Learning for Hyperspectral Image Classification

TL;DR

This paper introduces Label Consistent Transform Learning (LCTL), a new supervised unsupervised representation method tailored for hyperspectral image classification, demonstrating superior performance over existing techniques.

Contribution

The paper presents LCTL, a novel transform learning approach with label consistency, specifically designed for hyperspectral images, and shows it outperforms current state-of-the-art methods.

Findings

LCTL achieves more than 0.1 improvement in Kappa coefficient over existing methods.

LCTL is effective with fewer training samples.

LCTL outperforms label consistent KSVD, autoencoders, DBN, and CNN.

Abstract

This work proposes a new image analysis tool called Label Consistent Transform Learning (LCTL). Transform learning is a recent unsupervised representation learning approach; we add supervision by incorporating a label consistency constraint. The proposed technique is especially suited for hyper-spectral image classification problems owing to its ability to learn from fewer samples. We have compared our proposed method on state-of-the-art techniques like label consistent KSVD, Stacked Autoencoder, Deep Belief Network and Convolutional Neural Network. Our method yields considerably better results (more than 0.1 improvement in Kappa coefficient) than all the aforesaid techniques.