Grain segmentation in sandstone thin-section based on computer analysis of microscopic images

Abstract

The segmentation of minerals visible in sandstone thin-sections is a necessary part of microscopic estimation of mineral composition, the most important part of petrographic and sedimentological studies. This process is often a hard task due to the difficulty in determining the exact boundaries between grains, mainly caused by secondary alteration in minerals, etching of grains and pore-space filling by cements. Structural features play a very important role in mineral identification and undoubtedly without their use mineral recognition in thin sections gives many misclassification results. Calculation of each grain area is an important part of the process that is very time-consuming if done by-hand. Presented method provides a precise solution while mixing automated and non-automated approach. Photos of a sandstone thin section were taken using a Nikon Eclipse LV100N POL polarizing microscope, at 200x magnification, in transmitted light, with crossed polarizers. Then to determine the borders and therefore area of distinct grains, a sample image has been chosen. Initial contours have been created by-hand inside graphic application on a tablet device, as the non-automated part of the presented method. Afterward, the layer with marked boundaries was analyzed by computer software. In the automated part of the method, marked contours of grains were detected inside the algorithm with areas calculated afterward.