Heterogeneous structure fusion for Target Recognition in infrared imagery

We study Automatic Target Recognition (ATR) in infrared (IR) imagery from the perspective of feature fusion. The key to feature fusion is to take advantage of the discriminative and complementary information from different feature sets, which can be represented as internal (within each feature set) or external structures (across different feature sets). Traditional approaches tend to preserve either internal or external structures via certain feature projection. Some early attempts consider both structures implicitly or indirectly without revealing their relative importance and relevance. We propose a new unsupervised heterogeneous structure fusion (HSF) algorithm that is able to jointly optimize two kinds of structures explicitly and directly via a unified feature projection. The objective function of HSF integrates two feature structures in a closed form which can be optimized alternately via linear programming and eigenvector methods. The HSF solution provides not only the optimal feature projection but also the weight coefficients that encode the relative importance between two kinds of structures and among multiple feature sets. The experimental results on the COMANCHE IR dataset demonstrate that HSF outperforms state-of-the-art methods.