UNSUPERVISED INTERFERENCE REJECTION APPROACH TO TARGET DETECTION AND CLASSIFICATION FOR HYPERSPECTRAL IMAGERY

A widely used approach to hyperspectral image classification is to mod el a mixed-pixel vector as a linear superposition of substances reside nt in a pixel with additive Gaussian noise. Using this linear mixture model many image processing techniques can be applied, such as linear unmixing or orthogonal subspace projection. However, a third source no t considered in this model, called interference (clutter or structured noise), may sometimes give rise to more serious signal deterioration than the additive noise. We address this issue by introducing the inte rference into the linear mixture model. Including interference in the model enables us to treat the interference as another undesired source , like a passive jammer, so that it can be eliminated prior to detecti on and classification. This is particularly useful for hyperspectral i mages, which tend to have a high SNR but a low signal-to-interference ratio with the interference difficult to identify. To find and reject interference, we propose an unsupervised vector quantization-based int erference rejection (UIR) approach in conjunction with either an ortho gonal subspace projection (OSP) or an oblique subspace projection (OBS P) to simultaneously project a pixel into signature space as well as t o null out interference. Since there is no prior knowledge about the i nterference, the UIR is implemented in an unsupervised manner to gener ate the desired interference clusters so that they can be annihilated by the OSP or OBSP. The proposed approach is shown by evaluation with Hyperspectral Digital Imagery Collection Experiment (HYDICE) data to e xhibit considerable improvement in comparison to linear unmixing or th e OSP where interference is not considered, (C) 1998 society of Photo- Optical Instrumentation Engineers.[S0091-3286(98)00103-2].