Amedeo Capozzoli

Learn More
We present an ω-k approach based on the use of a 1D NonUniform FFT (NUFFT) routine, of NER (Non-Equispaced Results) type, programmed on a GPU in CUDA language, amenable to realtime applications. A Matlab main program links, via mex files, a compiled parallel (CUDA) routine implementing the NUFFT. The approach is shown to be an extension of an already(More)
We discuss the performance of two acceleration data structures for electromagnetic ray-tracing on GPU using the CUDA language, namely the kD-tree and the SBVH. Our implementations have been based on the approach by NVIDIA which accounts for the programming optimizations possible for the latest version of CUDA and NVIDIA GPU architectures.
We develop an approach for the fast and accurate determination of geometrical optics solutions to Maxwell's equations in inhomogeneous 2D media and for TM polarized electric fields. The eikonal equation is solved by the fast marching method. Particular attention is paid to consistently discretizing the scatterers' boundaries and matching the discretization(More)
We address the problem of extracting the maximum amount of information on an electromagnetic field over a domain D<inf>O</inf> from field samples on a domain D<inf>I</inf>, with a priori information on the radiating source. The problem is faced in two steps. In the first one, the source reconstruction is dealt with by taking into account the available a(More)
The paper deals with the problem of reconstructing the height of forests from polarimetric/multi-baseline SAR data. The approach consists of optimizing an objective functional defined as the distance between the measured data and the data predicted by the model at the actual estimate of the unknowns. We indicate the role of global optimization on the(More)
This paper presents a fast and robust approach to evaluate the singular values of a very large number of small matrices on GPUs. The timings and the accuracy obtained are compared to those achieved by "state-of-the-art" libraries. The results show that, by suitably tuning the trade-off between computational speed and accuracy, a reasonable speedup over a(More)
We present an algorithm for the fast tomography of vegetation, based on a Radon mathematical setting and on the combined use of advanced processing algorithms (NonUniform FFTs) and hardware resources (Graphic Processing Units GPUs). The algorithm performance is firstly numerically estimated, showing the favorable trade off between faithfulness and speed,(More)