Point set registration algorithms such as Iterative Closest Point (ICP) are commonly utilized in time-constrained environments like robotics. Finding the nearest neighbor of a point in a reference 3D point set is a common operation in ICP and frequently consumes at least 90% of the computation time. We introduce a novel approach to performing the distance-based nearest neighbor step based on Delaunay triangulation. This greedy algorithm finds the nearest neighbor of a query point by traversing the edges of the Delaunay triangulation created from a reference 3D point set. Our work integrates the Delaunay traversal into the correspondences search of ICP and exploits the iterative aspect of ICP by caching previous correspondences to expedite each iteration. An algorithmic analysis and comparison is conducted showing an order of magnitude speedup for both serial and vector processor implementation.

No abstract

In computer vision and robotics, point set registration is a fundamental issue used to estimate the relative position and orientation (pose) of an object in an environment. In a rapidly changing scene, this method must be executed frequently and in a timely manner, or the pose estimation becomes outdated. The point registration method is a computational bottleneck of a vision-processing pipeline. For this reason, this paper focuses on speeding up a widely used point registration method, the iterative closest point (ICP) algorithm. In addition, the ICP algorithm is transformed into a massively parallel algorithm and mapped onto a vector processor to realize a speedup of approximately an order of magnitude. Finally, we provide algorithmic and run-time analysis.