GPU‐accelerated parallel forward method to establish parametric scattering center models for complex targets

Forward method to establish parametric scattering center models is highly efficient for radar feature extraction of target. Nevertheless, scattering center extraction especially for multiple reflections is still time‐consuming due to path tracing, ray clustering, and parameter determination from electrically large and complex targets. As the multiple reflection items add, the time consumption of forward parametric scattering center modeling method increases dramatically, which can hardly meet the practical requirements. In this work, we propose a graphics processing unit‐accelerated forward parametric scattering center modeling method together with stackless KD‐Tree traversal technique that accelerates path tracing, clustering, and parameter determination to extract parametric scattering centers. Concurrently, the parameters of the scattering centers arising from triple (usually the maximum number of reflections) or above reflection items are forward determined for obtaining feature information as much as possible from complex target. Finally, the reconstructed radar characteristics from parametric models have been verified by comparison with exactly simulated results. Meanwhile, those examples using this improved method have achieved great speedup.