Min-max detection fusion for hyperspectral images

We propose an alternative formulation for an already known approach of continuum fusion of detectors. Our approach is based on min-max and max-min detectors that turn out to be equivalent to continuum fused detectors. The new formulation is easier for analytical purposes. We prove a theorem about the relationship between min-max and max-min detectors that would have been difficult to obtain using continuum fusion. For the case of a simple background space, when the two detectors simplify to the max-type detector, we compare the performance of that detector with the generalized likelihood ratio (GLR) detector, and show cases when either of the two outperforms the other detector.