Deep Spatially-Regularized and Superpixel-Based Diffusion Learning for Unsupervised Hyperspectral Image Clustering

TL;DR

This paper introduces an unsupervised hyperspectral image clustering framework that combines masked autoencoder-based representation learning with diffusion-based clustering, utilizing superpixels and spatial regularization for improved accuracy.

Contribution

It extends the S^2DL algorithm by integrating deep representation learning with a novel diffusion graph construction in the latent space.

Findings

Demonstrates improved clustering accuracy on Botswana and KSC datasets.

Leverages a Vision Transformer backbone for effective spectral and spatial feature extraction.

Uses a diffusion graph in the latent space to better reflect data geometry.

Abstract

An unsupervised framework for hyperspectral image (HSI) clustering is proposed that incorporates masked deep representation learning with diffusion-based clustering, extending the Spatially-Regularized Superpixel-based Diffusion Learning ($S^2DL$) algorithm. Initially, a denoised latent representation of the original HSI is learned via an unsupervised masked autoencoder (UMAE) model with a Vision Transformer backbone. The UMAE takes spatial context and long-range spectral correlations into account and incorporates an efficient pretraining process via masking that utilizes only a small subset of training pixels. In the next stage, the entropy rate superpixel (ERS) algorithm is used to segment the image into superpixels, and a spatially regularized diffusion graph is constructed using Euclidean and diffusion distances within the compressed latent space instead of the HSI space. The proposed algorithm, Deep Spatially-Regularized Superpixel-based Diffusion Learning ($DS^2DL$), leverages more faithful diffusion distances and subsequent diffusion graph construction that better reflect the intrinsic geometry of the underlying data manifold, improving labeling accuracy and clustering quality. Experiments on Botswana and KSC datasets demonstrate the efficacy of $DS^2DL$.