Abstract: Traditional methods in image processing have notenjoyed an easy transition to radar data because mostof these techniques are phase insensitive andgeneralizations have often not led to unique results.The desire to develop image reconstruction algorithmswhich are not `phase blind' has well-recognizedresolution and superresolution consequences since thesealgorithms are typically based upon complex Fouriertechniques. In addition, standard radar imaging methodshave employed a linear `weak scatterer' target model tomake a simple connection between target and scatteredfield - a model that is not always appropriate andwhich can cause deleterious image artifacts. Clearly,the accuracy of follow-on model appraisal requires morethan simple `resolution' analysis. Recently, severalimportant ideas have been developed which help tobridge the gap between algorithmic image resolution`enhancement' processing and usual radar imageappraisal methods. We present a coordinated overview ofsome of the more promising of the techniques, includingnonlinear Backus-Gilbert restoration and complex targetmodeling.!12
展开▼
