Combined sparse and collaborative representation for hyperspectral target detection

A novel algorithm that combines sparse and collaborative representation is proposed for target detection in hyperspectral imagery. Target detection is achieved by the representation of a testing pixel using a target library and a background library. Due to the fact that sparse representation encourages competition among atoms while collaborative representation tends to use all the atoms, the testing pixel is sparsely represented by target atoms because the pixel can include only one target; meanwhile, it is collaboratively represented by background atoms since multiple background atoms may be present in the pixel area. The detection output is simply generated by the difference between the two representation residuals. Experimental results demonstrate that the proposed algorithm outperforms the existing target detection algorithms, such as adaptive coherence estimator and pure sparse representation-based detector. HighlightsA combined sparse and collaborative representation-based detector is proposed.Target detection can be achieved by evaluating the difference of the two residuals.Residuals are from target sparse representation and background collaborative one.The background collaborative representation is enhanced by a Tikhonov regularization.