The tracking approach for small target with complex background based on spectral features

Abstract

Hyperspectral data provide abundant information about materials of targets. The characteristics of different materials can be reflected by the different spectral absorption features of different materials, referred to as spectral features in this paper. Therefore, with the inherent differences of the absorption features of different materials, a tracking approach based on spectral features achieves the precise target detection and has the anti-jamming capability so that the small target under complex environment can be tracked effectively. Firstly, the stable spectral features are extracted from hyperspectral data and then the histogram extension of feature space dimension is performed, by which a parameterized model of stable spectral features of targets is established. Working with the parameterized model, a nonparametric kernel density estimator is proposed to track the target. Experimental results prove that the proposed approach can improve the tracking accuracy and anti-jamming capability, by which the small target with complex background can be tracked effectively.