Comparison of directly measured to derived polarization imagery using an adaptive signature detection algorithm

Directly measured linear polarization images are shown to be more effective in target detection compared with derived imagery using a constant false alarm rate (CFAR) detection algorithm. The CFAR algorithm is derived from a maximum likelihood ratio test and is used to compare two pairs of inputs. One pair is directly measured imagery: an image with reflectivity/emissivity and a linear polarization and another with reflectivity/emissivity and a linear polarization perpendicular to the first image. The other pair is the first two Stokes images (S0, S1): a linear polarization image and a reflectivity/emissivity image. Detection using the directly measured pair is shown to be consistent with detection using the derived pair. Furthermore, using the directly measured pair is computationally simpler, and for target detection on natural backgrounds, does not increase the false alarm rate.