Indexing Method for Hyperspectral Data Fast Retrieval by Pyramid Technique

Hyperspectral data indexing and corresponding query processing techniques are of particular interest in a wide range of application such as content-based remote sensing image retrieval system, ground feature discrimination, classification and recognition in remote sensing images. The Pyramid Technique is an effective high-dimensional data mapping method, by which high-dimensional space point can be mapped to one-dimensional space to exploit one-dimensional indexing structure such as the B⁺-tree. In this paper, we present a new approach for fast indexing the hyper-spectral data based on the Pyramid Technique, and further propose new algorithms to process k Nearest Neighbor queries and spectral partial feature queries. Experiment results on hyper-dimensional spectral database demonstrated that our hyperspectral data indexing scheme can efficiently implement various querying tasks and can be widely employed in many real-time hyperspectral remote sensing applications.