3D Line Segments Extraction from Semi-dense SLAM

Despite much interest in Simultaneous Localization and Mapping (SLAM), there is a lack of efficient methods for representing and processing their large scale point clouds. In this paper, we propose to simplify the point clouds generated by the semi-dense SLAM using three-dimensional (3D) line segments. Specifically, we present a novel incremental approach for 3D line segments extraction. This approach reduces a 3D line segment fitting problem into two two-dimensional (2D) line segment fitting problems, which take advantage of both image edge segments and depth maps. We first detect edge segments, which are one-pixel-width pixel chains from keyframes. We then search 3D line segments of each keyframe along their detected edge pixel chains by minimizing the fitting error on both image plane and depth plane. By incrementally clustering the detected line segments, we show that the resulting 3D representation for the scene achieves a good balance between compactness and completeness. Our experimental results show that the 3D line segments generated by our method are highly accurate in terms of the location of their end points. Additionally, we also demonstrate that these line segments greatly improve the quality of 3D surface reconstruction compared to a feature point based baseline.