Zijia Zhang, Yaoming Cai, Wenyin Gong, Pedram Ghamisi, Xiaobo Liu, and Richard Gloaguen
Subspace clustering methods have become a powerful tool to cluster hyperspectral imaging data (HSI) as they ensure theoretical guarantees and empirical success. However, existing methods simply explore subspace representation in the Euclidean domain, meaning that high-order structural information in HSI is ignored, which may lead to poor robustness. This paper presents a simple yet effective method, to extend subspace clustering into the non-Euclidean domain entitled Hypergraph Convolutional Subspace Clustering (HGCSC). Instead of treating HSI as data only, we represent all the intra-class relations as hyperedges in a hypergraph. With this representation, we can recast the classic self-expression as a hypergraph convolutional self-representation model. To explore the long-range neighboring relation, we introduce a multi-hop hypergraph convolution process into the method by collapsing the repeated multiplications into a single matrix. HGCSC adopts the Frobenius norm to ensure a closed-form solution. We assess the performance of HGCSC on five real HSI datasets and show that HGCSC significantly outperforms competitors in terms of clustering accuracy.
IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 15, 676-686, 2021-12-21.