Detection of Camouflaged Targets in Hyperspectral Images

Abstract

This paper presents a study of a deterministic camouflaged target detection using hyperspectral images. Hyperspectral imagery has hundreds of spectral bands. This plethora of spectral information enables the detection of targets concealed by tree canopy of camouflage net. This problem cannot be addressed using high-resolution visible data alone because ‘shape information’ provided by such imagery cannot be used reliably in case of concealed targets. Hence, this problem is modeled as a target detection problem using ‘spectral information’ present in hyperspectral imagery. Moreover, the presence of concealing material such as canopy or cam net over the target necessitates the use of ‘sub-pixel’ target detectors.