RGB-D 영상을 이용한 실내 레이아웃의 빠른 복원
- 원문 URL
This thesis presents a fast recovering method that estimates an indoor layout from a single RGB-D image of Kinect in a second. We recover the layout, consisting of room structural objects and cuboid-like objects, by optimizing the combination of axes-aligned cuboids from candidates. For accelerating, we develop fast polygon extraction method and generate candidate cuboids based on extracted polygons. With these approaches, a computation of optimization process is dramatically reduced as acquring a small number of candidate cuboids. In the optimization step, we formulate the mixed integer programming model for selecting optimal cuboids that suitably represent the indoor scene. Experimental results demonstrate that our method recovers the layout fast, also quality of result is reliable.