Jonas Gutknecht kindly sent me a video presentation of his MSc Thesis “Multi-Layer-ToF: 3D-ToF-Camera with multiple object distances" from Institute for Signal Processing and Wireless Communications of ZHAW school of engineering in Winterthur, Switzerland, in cooperation with ESPROS Photonics.
Abstract:
"For the three-dimensional acquisition of a scene, 3D cameras based on the Time of Flight (ToF) method have proven themselves for many applications. However, Multi Path Interference (MPI) is widely spread in practice and represents a significant source of error for the ToF method with CW and PN modulation. The aim of this work is not only to suppress this error, but also to separate the individual signal components of real measurements and further determine the distances of the different signal paths.
In a first step a simulation tool was developed with which CW and PN based ToF measurements can be approximated. The simulation takes into account the main properties of a real 3D ToF System and provides reproducible data for any scene. In a second step, besides a closed form method, the two iterative algorithms OMP and PSO were implemented to separate the different signal paths. These three methods are based on a CW modulation and several measurements with variable modulation frequency. By means of different simulations and measurements the suitability of these algorithms for practical use was tested and the performance was compared.
With the three implemented methods the MPI influence can be greatly reduced and the individual signal paths separated. However, the closed form method has a high sensitivity to measurement noise. The PSO method is computationally intensive compared to the other methods and the results show a considerable amount of noise. In contrast, the OMP method has proven itself in practice and has prevailed over the other methods. With an application based on the OMP method, the process can be demonstrated in real time.
With this thesis it could be shown that MP separation for ToF measurements is possible in practice and the influence of MPI can be suppressed. For example, errors caused by stray light can be corrected or the contrast of the distance image can be increased. In addition, several distances per pixel can be resolved and thus semi-transparent objects can be measured. The presented method provides reliable data in difficult conditions and extends the application range of the 3D ToF method."
0 Response to "ToF Multipath Reduction Thesis"
Post a Comment