Image Segmentation for Quantification of Air-Water Interface in Micro-CT Soil Images

Soils are complex environments comprising various biological (roots, water, air etc) and physical constituents (minerals, aggregates, etc). Synchrotron radiation based X-ray microtomography (XMT) is widely used in extracting qualitative and quantitative information regarding spatial distribution of biological and physical soil constituents. Segmentation of these micro-CT soil images is of interest to geologists, hydrologists, civil and petroleum engineers and soil scientists. In this present work, we study and implement segmentation algorithms for microhydrology studies, specifically for soil water conductivity. Three well-known image segmentation algorithms are studied for evaluating their performance for the task. We demonstrate the problems and ways to segment XMT images and extract data for evaluating the air pressure in the soil pores to promote soil hydrology studies. To this end we take the recommended in the literature approach to differentiate textures and segment images using Fuzzy C-means Clustering (FCM). Secondly, we demonstrate the performance of two state-of-the-art level-set based active contours methods followed by curve fitting for radii detection and air pressure calculation