多源数据多语义岩体结构面提取方法

实现岩体场景下复杂岩体结构面提取是边坡、隧道等岩体工程稳定分析的关键。本文针对岩体结构面纹理与形态等特征复杂且不规则造成提取困难的问题，首先通过构建TLS点云与数码影像的映射关系，融合二维像素特征、三维方位及形态特征建立多重语义区，提取具有相似地质属性的同质区单元，然后利用产状与粗糙度特征对其进行合并聚类，有效实现复杂多属性岩体结构面的准确提取。试验结果表明，本文方法提取的岩体结构面具有较好的完整性和较高的准确性。

Rock surface extraction method with multi-semantic of multi-source data

The extraction of complex rock surface in the rock mass environment is the key to the stability analysis of rock mass engineering such as slopes and tunnels. The texture and shape of the rock surface are complicated and irregular, which makes it difficult to extract. This paper proposes a mapping relationship construction method for TLS point cloud and digital images. Multi-geological semantics features are first integrated for homogeneous units extraction, and then homogeneous units clustering based on geological semantics is proceeded. The experimental results show that the method in this paper can achieve rock surface extraction completely and accurately.