Hyperspectral targetdetection by using superpixels and signature based methods

Spectral signature based methods which form the mainstream in hyperspectral target detection can be classified mainly in three categories as the methods using background modeling, subspace projection based methods, and hybrid methods merging linear unmixing with background estimation. A common characteristic of all these methods is to classify each pixel of the hyperspectral image as a target or background while ignoring the spatial relations between neighbor pixels. Integration of contextual information defined over neighboring relations can, however, suppress the noise on individual pixels and yield better detection. The proposed methodology in this paper adapts the usage of superpixels defined over neighboring relations to the mentioned three classes of target detection algorithms. In particular, ACE, DTDCA and HUD algorithms are selected for each class for the proposed methodology. The experiments reveal that using superpixels for target detection improves the detection performances compared to the baseline methods using only pixels.