Effective Very-wide-area 3D ToF Sensing
Alvaro Lopez Paredes, Miguel Heredia Conde, Otmar Loffeld
Abstract
In this work, we propose a conceptually simple ToF sensing scheme which tackles the challenge of obtaining high lateral resolution, and long ranges. We present a rotary Time-of-Flight (ToF) 3D imaging system which avoids motion artifacts by drastically reducing the exposure time and leverages the sparse nature of the underlying scene response functions to reconstruct the 3D geometry of the scene by extending a conceptually simple greedy algorithm, such as Orthogonal Matching Pursuit (OMP). The sensing and recovery strategies consist of two steps: Firstly, the spatial domain is sliced in several partitions and the signals are preliminary localized within them. Then, OMP is applied to recover depth and amplitude values in the refined spatial domain. Observation directions are defined over a regular 2D angular grid to yield a high-density regular point cloud. Thorough numerical simulations show that panoramic 3D reconstruction is possible from few measurements and that the proposed methodology performs better compared to classical OMP for equal number of measurements.