Compressive Hyperspectral Imaging and Anomaly Detection
Abstract
We have developed and tested state-of-the-art target detection/template matching methods based on LI minimization. Given the spectral signature of a material, we are able to identify the pixels in a hyperspectral image, even for very noisy data, that contains the material. The speed of our unmixing algorithm is now much faster than any previous methods. We have also expanded the use of the Bayesian dictionary learning and sparse reconstruction method by utilizing spatial inter-relationships between different components in images and incorporating sparsity of spectral vectors in terms of sparse representation by endmembers into reconstruction.
Document Details
- Document Type
- Technical Report
- Publication Date
- Feb 01, 2010
- Accession Number
- ADA521108
Entities
People
- Kevin F. Kelly
- Pradeep Thiyanaratnam
- Stanley Osher
- Susan Chen
- Wotao Yin