由于缺乏有效并具有针对性的方法以及传统遥感影像的2维局限性,城市立交桥点的识别与轮廓提取存在较大困难。本文基于机载LiDAR点云数据提出一种城市立交桥的提取方法。首先,通过基于局部平面约束的剖面点分割方法,利用多方向扫描线建立改进的剖面邻域点云组织结构,从而提取出立交桥种子点。然后,利用Alpha-shape算法定位立交桥轮廓,并获取立交桥初始点。最后,利用区域增长的方法提取出完整结构,同时获取立交桥轮廓。实验结果表明,从提取结果上获取的立交桥长度、宽度以及形态等指标与人工量测结果相近,具有较高的准确度。

It is difficult to detect city cloverleaf, which is arised from shortcomings in traditional remote sensing image and the lack of appropriate methods. This paper presents an effective approach to extract city cloverleaf interchange footprints and boundary based on laser scanning data. We developed a plane-based approach to segment the data in the profile area and built a point organization structure called advanced scanline profile neighborhood system to identify the seed points. Alpha-shape algorithm was introduced to segment the seed points and generate the point sets boundaries. The cloverleaf interchange boundary was located by three-dimensional contour analysis based on model description. The complete cloverleaf footprints were determined by region-growing algorithm in the end, and meanwhile, the final cloverleaf boundary was obtained. The experiment was performed on the LiDAR data collected on a urban city. The result shows integrity and validly of our function that can get accurate length, width and morphology measurements comparing with the manual measurements from the high resolution ortho image.