Object CFAR detection in gamma-distributed textured-background images

The authors introduce a new non-Gaussian CFAR detector, specifically a CFAR detector for gamma-distributed textured backgrounds. The design and analysis of the detector is carried out in several steps: basic design, analysis in only noise conditions, bias removal and analysis in the presence of correlation. The authors also discuss the possibility of applying a data prewhitening technique to control the false alarm rate in correlated textured patterns. In terms of detection, they analyse the performance of their detector for a certain model of target. They compare its performance to that of the ideal detector, and quantify in which conditions the former behaves closely enough to the latter.