Scan segments matching for pairwise 3D alignment

Abstract

Abstract-This paper presents a method for pairwise 3D alignment which solves data association by matching scan segments across scans. Generating accurate segment associations allows to run a modified version of the Iterative Closest Point (ICP) algorithm where the search for point-to-point correspondences is constrained to associated segments. The novelty of the proposed approach is in the segment matching process which takes into account the proximity of segments, their shape, and the consistency of their relative locations in each scan. Scan segmentation is here assumed to be given (recent studies provide various alternatives [10], [19]). The method is tested on seven sequences of Velodyne scans acquired in urban environments. Unlike various other standard versions of ICP, which fail to recover correct alignment when the displacement between scans increases, the proposed method is shown to be robust to displacements of several meters. In addition, it is shown to lead to savings in computational times which are potentially critical in real-time applications.

Joint self-localization and tracking of generic objects in 3D range data

“…When using dense data, the association of measurements can be ambiguous especially when several detections per object exist. To overcome ambiguities, solutions like fuzzy segmentation [21], segment matching [8] or appearance learning [10] have been proposed. This two-stage approach has been applied to various kind of sensors, from 2D laser scanners [17] over 3D laser scanners [12], [19] up to time-of-flight cameras [9].…”

FFT-based Terrain Segmentation for Underwater Mapping

“…The lower the value produced by the metric, the crisper the point cloud and in turn the more accurate the alignment. The rationale for using this metric as opposed to the ICP residual is developped in [7]. The voxel resolution used in all the experiments presented here is 2cm.…”

“…The Spin Image [10] and the NARF [20] features, for instance, have this requirement. An alternative to point feature based approaches are approaches which attempt to match segments of the data [7]. This makes them more directly applicable to scan pairs of different densities.…”

“…This makes them more directly applicable to scan pairs of different densities. The alignment method in [7] is used here and will be be referred to as S-ICP ('S' for segmentation-based).…”

“…Interpolation in the data sets processed here is not likely to modify the frequency content of the original elevation signal since empty cells represent at most 10% of the elevation grid. However, this is not true in more sparse scans, such as side-looking Velodyne scans [7] where the polar pattern of the scans induces a significant number of empty cells in a regular grid. Non-trivial interpolation mechanisms would be required to estimate the height in these cells.…”