Unsupervised Diffusion and Volume Maximization-Based Clustering of Hyperspectral Images

Abstract

Hyperspectral images taken from aircraft or satellites contain information from hundreds of spectral bands, within which lie latent lower-dimensional structures that can be exploited for classifying vegetation and other materials. A disadvantage of working with hyperspectral images is that, due to an inherent trade-off between spectral and spatial resolution, they have a relatively coarse spatial scale, meaning that single pixels may correspond to spatial regions containing multiple materials. This article introduces the Diffusion and Volume maximization-based Image Clustering (D-VIC) algorithm for unsupervised material clustering to address this problem. By directly incorporating pixel purity into its labeling procedure, D-VIC gives greater weight to pixels corresponding to a spatial region containing just a single material. D-VIC is shown to outperform comparable state-of-the-art methods in extensive experiments on a range of hyperspectral images, including land-use maps and highly mixed forest health surveys (in the context of ash dieback disease), implying that it is well-equipped for unsupervised material clustering of spectrally-mixed hyperspectral datasets.

Multi-View Scene Classification Based on Feature Integration and Evidence Decision Fusion

“…Unsupervised feature-based learning focuses on learning the basis functions for feature encoding from manually created feature descriptors and generating learned features such as raw component analysis, sparse coding, or autoencoders [11]. This creates a new scheme in the place of human-based features, and a large number of current scene classification methods based on unsupervised learning are gradually appearing-e.g., using weighted inverse convolutional networks to learn features from the remote sensing data itself and mapping [12], unsupervised remote sensing analysis tasks based on superpixels and spatially regularized diffusion learning (S2DL) [13], unsupervised representation learning based on multilayer feature fusion fused with Wasserstein GAN [14], and unsupervised material clustering based on diffusion-and volume-maximization-based image clustering (D-VIC) for the task of classifying vegetation and other materials [15]-and have made substantial progress in scene classification. However, these features cannot fully utilize scene recognition information, resulting in poor performance in classification and recognition tasks [16].…”