Evaluation of Different Structural Models for TargetDetection in Hyperspectral Imagery

摘要

Target detection is an essential component for defense, security and medical applications of hyperspectral im-agery. Structured and unstructured models are used to model variability of spectral signatures, for the design ofinformation extraction algorithms. In structured models, spectral variability is modeled using different geometricrepresentations. In linear approaches, the spectral signatures are assumed to be generated by the linear combi-nation of basis vectors. The nature of the basis vectors, and its allowable linear combinations, define differentstructural models such as vector subspaces, polyhedral cones, and convex hulls. In this paper, we investigatethe use of these models to describe background of hyperspectral images, and study the performance of targetdetection algorithms based on these models. We also study the effect of the model order in the performanceof target detection algorithms based on these models. Results show that model order is critical to algorithmperformance. Underfitting or overfitting result in poor performance. Models based on subspace are of lowerorder than those based on polyhedral cones or convex hulls. With good target to background contrast all modelsperform well.