Abstract
Advances in remote sensing technologies made in the last two decades have transpired to a broad range of geoscience and engineering applications. LiDAR and photogrammetry tools allow users to measure a wide variety of features and components along rock outcrops, slopes, mine tunnels and shafts. Great progress has been achieved in automating feature recognition. Such applications, prevalent in the mining industry, include fragmentation distribution analysis, rock mass structural description, and identification of instability indicators and rock slope failure events. The increasing demand for specific applications has led to the emergence of a great number of processing tools and software to assist users in obtaining the information needed. Software reliance invariably leads to dissociation between the processing methods and the results, causing a “black box” effect with the tools used. In conjunction, a growing number of open source codes and applications have been developed by the booming geoscience and computer community. These tools present the added advantage of transparent processing methodology and builds upon a continuous improvement strategy. The present study focuses on open access, intuitive tools and processes to facilitate geomechanical feature recognition.
This work investigates the complementarity of three fundamental notions associated with structural feature recognition. These aspects are commonly available in most open source point cloud processing tools, libraries and software packages. These approaches are (i) point cloud surface roughness, (ii) mesh curvature, and (iii) facet analysis. The study first outlines advantages and describes readily available results that can be obtained from the considered methods taken individually. Special considerations are given the key inputs that need to be adjusted. Recommendations are made for systematic parameter testing to ensure optimal results. The methods are applied to a number of underground tunnel faces and compared to highlight the limitations of each methods. Surface roughness is incorporated into the mesh analysis to facilitate edge recognition and family grouping. The result is compared to automated facet analysis. Surface roughness is then incorporated to the facet recognition method to assist in joint grouping. Finally, sample joint surfaces are marked manually and analyzed in terms of roughness and mesh curvature. The resulting parameters are used as reference baselines for facet recognition and joint grouping. Special considerations are given to the influence of the scale of measurement throughout the analysis, and how systematic scale variation can be leveraged.