A novel UWB Radar super-resolution object recognition approach for complex edged objects

In this paper a novel approach for object recognition (OR) by backscattered Ultra Wideband (UWB) signals on the basis of a reference data set is presented. The aim of this robust OR is succeeded by UWB imaging, classifying and finally applying a maximum probability recognition algorithm. To provide super-resolution even under multi-scattering conditions, an advanced wavefront detection algorithm with subsequent high resolution imaging technique is performed. After postprocessing a joint moment based feature, texture feature and geometrical feature recognition algorithm is applied. The simulation-based performance evaluations show a very precise imaging and a perfect maximum probability recognition rate with additional outstanding robustness. First tests using an m-sequence UWB Radar indicate the feasibility of this concept.