A Combined Signal Subspace Projection and Partial Filtering Approach to Target Detection for Hyperspectral images

摘要

In the study, a novel signal subspace projection (SSP) approach is first proposed to detect and extract target signatures in unknown background. The weights of SSP are equivalent to optimal weights given by Wiener-Hopf equations. To further reduce the computation complexity in SSP, we implement the SSP-based classifier by an adaptive filter combined with some partial filters, called CSSPPF. In CSSPPF, we partition the image into several groups according to their associated band correlation. By the way, we transfer the design of a large fully filter into several small partial filter designing. Simulation results performed on AVIRIS images have demonstrated the efficiency of the proposed approaches.