Graph Regularized Low-Rank Representation for Hyperspectral Anomaly Detection

One of the most crucial uses of hyperspectral images is anomaly identification, which seeks to find items that deviate significantly from their surroundings. Many other approaches to anomaly detection have been suggested in the past. However, given the spectral connection of all the pixels, the majority of them never reach a climax. In recent years, there has been a lot of interest in Low Rank Representation (LRR) as a viable model for hyperspectral images, which frequently contain a global structure made up of a few groundcover signatures.In this research, we offer a new hyperspectral anomaly detection approach based on the Graph Regularized LRR (GLR), which we combine with the graph regularization into the LRR formulation. The combination of the global and local structure in the suggested algorithm is a key benefit.