New low-rank model improves hyperspectral anomaly detection speed and flexibility
Hyperspectral Anomaly Detection via Group Sparse Low-Rank Tensor Factorization With Automatic Anomaly Grouping
Computer Vision and Pattern Recognition
Summary
Detecting unusual materials or objects in hyperspectral images is hard because existing methods are slow and don’t handle groups of anomalies well. The authors propose a method that simplifies the math using group sparsity, making detection faster and better at finding clusters of anomalies automatically. They combine information from both the spectral details and the image layout to improve results. Their approach works better than current methods on several real-world hyperspectral datasets.
What this means in practice
- •For remote sensing professionals: Detect and group unusual materials or objects in hyperspectral satellite images faster and more reliably than existing tools.
- •For environmental monitoring teams: Identify spatial clusters of anomalies in hyperspectral images for better tracking of pollution or land changes without manual grouping.
Authors
Quan Yu, Yu-Hong Dai, Xiongjun Zhang
Abstract
Low-rank tensor modeling has become an effective tool for hyperspectral anomaly detection. However, existing methods still suffer from high computational cost and limited flexibility in characterizing spatially structured anomalies. To address these issues, this paper proposes a hyperspectral anomaly detection method based on group sparse low-rank tensor factorization with automatic anomaly grouping (GSAA). Specifically, the low tubal rank background is characterized by imposing group sparsity on tensor factors, which provides an efficient alternative to direct tensor rank regularization. For anomaly modeling, a latent grouping map is introduced to build an automatic anomaly grouping penalty, allowing anomaly groups to be adaptively inferred from the data rather than predefined at the pixel level. To further exploit complementary spectral and spatial information, GSAA is applied in both domains, and the resulting detection maps are fused to form a spectral--spatial version of GSAA, termed GSAA-SS. An efficient linearized alternating direction method of multipliers algorithm with convergence guarantee is developed to solve the resulting model. Experimental results on five real hyperspectral datasets demonstrate that the proposed method achieves superior detection performance and competitive computational efficiency compared with several state-of-the-art methods.