Stereo Light Stripe Profiler with Redundancy Check for Robust 3-D Modeling
A Laser Stripe Profiler (LSP) is a range-sensing device for 3-D modeling by active (laser) optical triangulation. A focused plane of laser light illuminates a stripe when colliding with the surface of an object. A digital camera records its reflection. Reconstruction is done by triangulation, i.e., intersecting the laser plane with the rays of sight corresponding to the laser stripe projection in the image frame. A complete 3-D model is then acquired by sweeping (i.e. scanning) the projected laser stripe over the surface of an object.
The DLR-LSP is a general implementation of a LSP. On the one hand, two laser beams in different orientations are implemented for faster 3-D scanning. On the other hand, two digital cameras are rigidly attached to them, synchronized and calibrated. In addition, the digital cameras are not optically filtered to laser light frequency, as this would preclude from obtaining e.g. object appearance (texture).
The absence of optical filters at the DLR-LSP constitutes however a challenge to robust laser stripe detection by image processing. In spite of advanced measures against these errors, mismatches may still occur unless image processing is continuously tuned to current camera parameters and outer conditions (e.g. the scene and its illumination).
In this internship the student's task is to extend the current DLR-LSP by redundant operation of the second, remaining camera to independently detect the same laser stripe being used by the original camera. Even though this second detection stage may still yield erroneous data after triangulation, the probability that both range sensors deliver consistent erroneous 3-D data is extremely low. By cross checking their 3-D results, highly robust operation of the LSP can be attained. In fact, parallel operation of two LSPs in separate threads effectively comes for free when using customary hardware. In detail:
- Initial training on the DLR-LSP operation and its methods for robust laser stripe detection.
- Literature review on point-wise, 3-D data fusion.
- Extension of the DLR-LSP software for dual operation in two parallel threads using different cameras.
- Design and implementation of a method for 3-D data fusion from the last raw results.
- Testing, evaluation, and explicit parametrization of the implementation.
Undergraduate studies in computer science or electrical engineering degrees. Preferentially prior knowledge of computer vision and robotics. Strong knowledge of mathematics and programming (C++, Matlab).
K. H. Strobl, W. Sepp, E. Wahl, T. Bodenmüller, M. Suppa, J. F. Seara, and G. Hirzinger. "The DLR Multisensory Hand-Guided Device: The Laser Stripe Profiler." Proceedings of the IEEE International Conference on Robotics and Automation (ICRA 2004), New Orleans, LA, USA, pp. 1927-1932, April 2004.
M. Suppa, S. Kielhoefer, J. Langwald, F. Hacker, K. H. Strobl, and G. Hirzinger. "The 3D-Modeller: A Multi-Purpose Vision Platform." Proceedings of the IEEE International Conference on Robotics and Automation (ICRA 2007), Rome, Italy, pp. 781-787, April 2007.