Offsite Aerial Path Planning for Efficient Urban Scene Reconstruction

摘要

With rapid development in UAV technologies, it is now possible to reconstructlarge-scale outdoor scenes using only images captured by low-costdrones. The problem, however, becomes how to plan the aerial path for adrone to capture images so that two conflicting goals are optimized: maximizingthe reconstruction quality and minimizing mid-air image acquisitioneffort. Existing approaches either resort to pre-defined dense and thus inefficientview sampling strategy, or plan the path adaptively but require twoonsite flight passes and intensive computation in-between. Hence, usingthese methods to capture and reconstruct large-scale scenes can be tedious.In this paper, we present an adaptive aerial path planning algorithm thatcan be done before the site visit. Using only a 2D map and a satellite imageof the to-be-reconstructed area, we first compute a coarse 2.5D model forthe scene based on the relationship between buildings and their shadows.A novel Max-Min optimization is then proposed to select a minimal set ofviewpoints that maximizes the reconstructability under the the same numberof viewpoints. Experimental results on benchmark show that our planningapproach can effectively reduce the number of viewpoints needed than the previous state-of-the-art method, while maintaining comparable reconstructionquality. Since no field computation or a second visit is needed, and theview number is also minimized, our approach significantly reduces the timerequired in the field as well as the off-line computation cost for multi-viewstereo reconstruction, making it possible to reconstruct a large-scale urbanscene in a short time with moderate effort.