**5. Conclusions and further research**

The KK-Distribution was embedded within a SIRP in the complex domain, so that the Neyman-Pearson optimal detector could be constructed for multilook detection. A GLRT solution was then constructed, and this, together with the WMF suboptimal decision rule, were compared using ROC curves. These ROC curves were generated using clutter parameter estimates based upon the Ingara data set. For the horizontally polarised case, it was shown that the GLRT produced a very good probability of detection for a Swerling 1 target model. In the case of vertically polarised returns, the WMF often matched the performance of the GLRT. Hence, in such cases, the WMF can be used as a simpler suboptimal detector.

Further work will be spent on trying to find suboptimal decision rules based upon simplification of the GLRT solution . Additionally, performance analysis of detectors for other target models will be undertaken .
