Detection of occluded targets using thermal imaging spectroscopy

Automatic detection of occluded targets from a sequence of images is an interesting area of research for defense related application. In this paper, change detection methods are investigated for the detection of buried improvised explosive devices (IED) using temporal thermal hyperspectral scenes. Specifically, the paper assesses the detection of buried small aluminium plates using the TELOPS Hyper-Cam sensor and by applying two change detection algorithms: multivariate statistical based method (Cross-Covariance (CC)) and class-conditional change detector (QCC). It was found that spectral based change detection is a good method for the detection of buried IED under disturbed soil. Moreover, the Cross-Covariance (CC)) and the class-conditional (QCC) change detector were able to detect changes using short temporal sequences as long temporal sequences pairs.