Incremental reconstruction of planar B-Rep models from multiple point clouds

Title data

Sand, Maximilian ; Henrich, Dominik:
Incremental reconstruction of planar B-Rep models from multiple point clouds.
In: The Visual Computer. Vol. 32 (2016) . - pp. 945-954.
ISSN 1432-2315
DOI: https://doi.org/10.1007/s00371-016-1247-7

Project information

Abstract in another language

Boundary representation (B-Rep) models are surface representations containing geometric and topological information that are used in many fields such as reverse engineering, simulation and optimization. To create a B-Rep model from point clouds of different viewpoints, the data are usually merged into one global point cloud and subsequently a surface model is reconstructed. The drawbacks of this approach are high processing times and the fact that all data must be available at once. We present an online approach for building planar B-Rep models from multiple organized point clouds of different viewpoints. Our procedure consists of two steps: B-Rep reconstruction and B-Rep fusion. For each viewpoint, a partial, incomplete boundary representation model is instantly reconstructed from the depth image. Our incremental fusion algorithm incorporates this partial B-Rep model into a global model built from previous viewpoints, making the model more complete and reducing uncertainty. In contrast to existing approaches, our method can be performed online and a valid B-Rep model of all previous point clouds is always available since our fusion is incremental. This enables an application to terminate the reconstruction if the model contains enough information for a given task. We evaluated our approach with an application in the field of robotics, where a depth sensor is attached to a robot arm that observes its workspace.