Phänomenologische Modellierung und multistatische Simulation von Time-of-Flight 3D PMD Kameras

One of the central problems of automation, control and safety technology is fast and accurate detection of moving objects and their spatial orientation. Conventional optical sensors providing only 2D images with color information are not fully suitable for this. In conventional methods, according to the principle of stereoscopy, two images from the field of interest from different viewpoints are required to capture the object surface. An efficient real-time capable solution of the correspondence problem associates with high computational costs and is for some object types only partially solved.

In contrast, PMD sensors, whose operating principle is based on the measurement of the light propagation time, provide without further processing three-dimensional vector measurements of the environment and thus allow a direct 3D visualization and localization of objects. Another advantage over other image sensors is due to the self-illuminating construction the ability to suppress the most of the background light coming from artificial lighting or the sun and to increase thus the dynamic range.

The newer generations of PMD sensors have been greatly improved in terms of the measurement accuracy. However, the numerous studies have shown that the measured distance values are superposed with systematic errors (noise, motion artifacts, flying pixels, measurement error due to the multistatic constellation of the camera, etc.). Simulations are a very effective tool in the investigation of such errors. They can support algorithm developers and system designers for generating synthetic reference data of the PMD sensor hardware. The results can help to classify the errors and to improve the existing processing and coregistration algorithms. By using synthetic test data algorithms can previously be tested for their ability to function and system designs can be appraised for their suitability for a particular application.

This work deals with the modeling and development of such a universal PMD simulator.