Multi-View Planning for Simultaneous Coverage and Accuracy Optimisation

摘要

View planning for three-dimensional (3D) reconstruction and inspection solves the problem of finding an efficient sequence of views allowing complete and high quality reconstruction of complex objects. To fulfil this task, view planning methods need to deal with sensor limitations and satisfy predefined goals. Our objective is to jointly evaluate accuracy requirements and coverage during planning, to optimise the reconstruction procedure, as well as to explicitly take configuration space constraints into account. We present a novel model-based approach, which at the same time optimises accuracy and coverage based on an existing model (e.g. CAD or preview scan). For accuracy optimisation we extended the statistical E-criterion to model directional uncertainty. Coverage is maximised while taking configuration space and sensor characteristics into account. We validate our approach through experimental evaluation using the Next-bestview (NBV) benchmark framework and a robot mounted stereo fringe projection sensor.