Ryan A. Manzuk
Dept. of Geosciences, Princeton University, Princeton, New Jersey 08544, USA, rmanzuk@princeton.edu
Devdigvijay Singh
Dept. of Geosciences, Princeton University, Princeton, New Jersey 08544, USA
Akshay Mehra
Dept. of Geosciences, Princeton University, Princeton, New Jersey 08544, USA, and Dept. of Earth Sciences,
Dartmouth College, Hanover, New Hampshire 03755, USA
Emily C. Geyman
Dept. of Geosciences, Princeton University, Princeton, New Jersey 08544, USA, and Division of Geological
and Planetary Sciences, California Institute of Technology, Pasadena, California 91125, USA
Stacey Edmonsond
Dept. of Geosciences, Princeton University, Princeton, New Jersey 08544, USA, and School of Earth and Ocean
Sciences, University of Victoria, Victoria, British Columbia V8W 2Y2, Canada
Adam C. Maloof
Dept. of Geosciences, Princeton University, Princeton, New Jersey 08544, USA
Abstract
Accurately assessing the shape, size, and modality of features in rock samples is a longstanding problem in
geology. Recent advances in machine learning have introduced the possibility of performing these tasks
through automated image analysis. To leverage these methods for geological and paleontological applications,
we first need a way to acquire high-resolution images of polished slabs and thin sections with a field of
view large enough to fit samples containing crystals, fossils, bedforms, etc. We describe a new
multispectral setup that can acquire images at ~3.76 μm per pixel spatial resolution over a 21
cm2 field of view, equipped with 8-band (470–940 nm) spectral resolution, plus a band for
ultraviolet (365 nm) fluorescence. Additionally, we present a 5-band (470–940 nm) light table with automated
rotating polarizers, which allows use of the camera as a high-throughput transmitted light thin section
imager. The use of color bands outside the visible spectrum, as well as the registration of multiple
cross-polarized rotations, encode rock properties that heighten image contrast and improve the accuracy of
machine learning models. Our setup and methods provide an efficient way to (1) build reproducible image
archives of rock specimens to complement field observations, (2) classify and segment those images, and (3)
quantitatively compare lithofacies and fossil assemblages.
Manuscript received 21 Jan. 2022. Revised manuscript received 29 Apr. 2022.
Manuscript accepted 3 May 2022. Posted 6 June 2022.
© The Geological Society of America, 2022. CC-BY-NC.
https://doi.org/10.1130/GSATG533A.1