Abstract
The method described here for three-dimensional (3D) reconstruction of macroscopic features in geological materials requires the features of interest to be macroscopically distinguishable by their color. Samples are subject to cycles of precision serial lapping and subsequent data acquisition by a flatbed color scanner. The yielded true color images of the sample surfaces are separated into red, green and blue component images. Spectral characteristics of features subject to 3D reconstruction are defined by training sets which control batch-supervised image classification by the maximum-likelihood algorithm. To recover the 3D feature geometries, the classified images are stacked in a voxel array enabling efficient analysis and visualization of discrete, as well as continuous, spatial variation. Feature classification, 3D reconstruction, 3D model handling and visualization are demonstrated using two samples, a doleritic basalt and a fossiliferous limestone.
Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
