Ground Truth for Evaluating Time of Flight Imaging

摘要

In this work, we systematically analyze how good ground truth (GT) datasets for evaluating methods based on Time-of-Flight (ToF) imaging data should look like. Starting from a high level characterization of the application domains and requirements they typically have, we characterize how good datasets should look like and discuss how algorithms can be evaluated using them. Furthermore, we discuss the two different ways of obtaining ground truth data: By measurement and by simulation.