作者: Eric Turner , Avideh Zakhor
关键词: Scalability 、 Artificial intelligence 、 Range (statistics) 、 Representation (mathematics) 、 Process (computing) 、 Computer science 、 Floor plan 、 Computer vision 、 Point cloud 、 Laser 、 Dimensionality reduction 、 Computer graphics (images)
摘要: In this paper we describe an approach to automatically generate a floor plan for the interior environment of building using laser scan data. This is meant accurately indicate positions walls within area interest in building. proposed algorithm separates floors scan, selects representative sampling wall scans each floor, and triangulates these samples develop watertight representation scanned areas. Curves straight line segments are fit walls, order mitigate any registration errors from original scans. method not dependent on scanning system can successfully process noisy with non-zero error. Most processing performed after dramatic dimensionality reduction, yielding scalable approach. We demonstrate effectiveness our three-story point cloud commercial as well lobby hallways hotel.