Full text loading...
-
Unsupervised Artificial Neural Networks for Classifying Lidar Datasets
- Publisher: European Association of Geoscientists & Engineers
- Source: Conference Proceedings, 77th EAGE Conference and Exhibition 2015, Jun 2015, Volume 2015, p.1 - 5
Abstract
Classifying and interpreting digital outcrop data from lidar and other digital source can be very time consuming. A variety of automated approaches have been used in the past, but here unsupervised Artificial Neural Networks are proposed as a method for classifying digital outcrop data. Using attributes derived from the surface topography a digital outcrop model can be classified into discrete planar or fracture sets with minimal intervention from the geoscientists, maximising the amount of useful data extracted in the minimum time. This allows larger geostatistical datasets to be extracted from the outcrop models. Care must be taken, as with any automated approach, to ensure the results are valid, but ANNs provide a useful approach to classifying large lidar datasets.