Sammendrag
This paper presents a real-time contour tracking and object segmentation algorithm for 3D range images. The algorithm is used to control a novel micro-mirror based imaging laser scanner, which provides a dynamic trade-off between resolution and frame rate. The micro-mirrors are controllable, enabling us to speed up acquisition significantly by only sampling on the object that is tracked and of interest. As the hardware is under development, we benchmark our algorithms on data from a SICK LMS100-10000 laser scanner mounted on a tilting platform. We find that objects are tracked and segmented well on pixel-level; that frame rate/resolution can be increased 3–4 times through our approach compared to scanners having static scan trajectories, and that the algorithm runs in 30 ms/image on a Intel Core i7 CPU using a single core.
