Robust Detector of Multi-Pixel Targets Using a Sequence of Images

摘要

We design and assess an algorithm to detect a multi-pixel target of an unknown spatial size, shape and position in a sequence of images in the presence of an additive Gaussian background clutter and a channel noise. The presence of the target decreases the background plus noise power that hence may be different under the null and alternative hypotheses. We use the generalized likelihood ratio (GLR) approach to derive a modified multi-pixel matched subspace detector (MMMSD) that is sensitive to both energy in the target subspace and reduced energy in the orthogonal subspace. The derived algorithm combines the multi-pixel matched subspace detector and multi-pixel background-plus-noise power change detector in a unique scheme. The crucial characteristic of the proposed detector is that prior knowledge of the target size, shape and position is not required. The designed detector is theoretically proved and numerically evaluated. Numerical simulations attest the validity of the theoretical analysis and show that the proposed MMMSD outperforms the known detector in the case of unknown spatial parameters of the target.