Bridge detection in airborne laser scanner data
Bridge detection in airborne laser scanner data
复制标题
DOI：
10.1016/j.isprsjprs.2006.07.004
复制
发表时间：
2006-10
影响因子：
12.7
通讯作者：
G. Sithole;G. Vosselman
中科院分区：
文献类型：
--
作者：
G. Sithole;G. Vosselman
在景观中，桥梁的特征是模糊的，因为它们是裸露大地的延伸，但它们也是人造的。因此，在数字地形模型 (DTM) 生成中，根据应用需要保留或移除桥梁。因此，必须明确检测它们。提出了一种几何桥梁检测算法，该算法基于可以根据景观中的横截面（轮廓）检测景观中的特征的思想。横截面中包含的拓扑信息用于识别种子桥点。然后使用种子桥点来检测各个桥。该桥梁检测算法的新颖之处在于该算法对桥梁的形状是盲的并且能够适应不同的桥梁设计。因此，桥不需要具有平行的边缘或均匀的宽度。此外，桥梁可以弯曲、分裂成多个部分并且包含孔。桥梁检测算法已经在各种数据集上进行了测试，发现在桥梁检测方面效果非常好。
In a landscape bridges are ambiguous features in that they are extensions of the bare earth but they are also man made. Therefore, in a Digital Terrain Model (DTM) generation, depending on application it is necessary to keep or remove bridges. For this reason they have to be detected explicitly. A geometric bridge detection algorithm is proposed that is based on the idea that features can be detected in a landscape based on cross-sections (profiles) in a landscape. The topological information contained in the cross-sections are used to identify seed bridge points. The seed bridge points are then used to detect individual bridges. The bridge detection algorithm is novel in that the algorithm is blind to the shape of a bridge and is adaptable to different bridge designs. Therefore, bridges need not have parallel edges or uniform width. Furthermore, bridges can curve, split into parts, and contain holes. The bridge detection algorithm has been tested on various data sets and found to work very well in detecting bridges.