A Novel Approach for Solving MPI for Multi-Target ToF Imaging using Subdivision-based Nested Compressed Sensing

Time-of-Flight (ToF) cameras have become a popular imaging modality for macroscopic scene detection. In particular, amplitude-modulated continuous wave (AMCW) ToF cameras use the phase difference between sent and received signals for object depth reconstruction. However, various sources of multipath reflections exist in practice, causing each ToF pixel to erroneously receive a superposition of multiple reflections instead of a single bounce. This leads to distortions in the phase difference and consequently, errors in the depth maps. Compressed Sensing methods have emerged as an effective approach to solve this multipath interference (MPI) problem. However, it has two major disadvantages-large sensing matrix size leading to high computational load, and a high mutual coherence causing reconstruction failure. This paper introduces a subdivision-based nested compressed sensing algorithm that aims to alleviate these known disadvantages. Measurements at multiple modulation frequencies are used to isolate the k interfering signals in the time domain. Simulation results are presented with a noise performance analysis and results based on real multi-target measurement data are also discussed.