Segmentation of X-ray tomography images of compacted soils

X-ray computed tomography (X-ray CT) in tandem with mercury intrusion porosimetry (MIP) has gained importance among researchers for examining the internal structure of geomaterials owing to the wide scale of coverage. The success of the X-ray CT lies in the proper segmentation of the acquired images during image processing. This study proposes a novel methodology for finding out the most probable threshold number for the segmentation of X-ray CT images of compacted soils as well as the quantification of small and large pores beyond the detection range of the MIP test. The methodology was developed based on total void ratio, tomographic void ratio and total cumulative mercury intruded void ratio obtained from vernier caliper measurements, analysis of X-ray CT images and MIP data of compacted soil specimen, respectively. The threshold number obtained was evaluated by visual observations of X-ray CT images and their corresponding binary images. The evaluation results showed that the threshold number obtained from the proposed methodology could precisely separate the soil particles from the voids in X-ray CT images and also gave a complete range of different pore sizes in the compacted soil specimen. Thus this research finds its significance in different image segmentation applications in the areas of geotechnical and geoenvironmental engineering. Also, the study on the variation of threshold number with different parameters showed that the threshold number is directly proportional to the compaction energy and sand content whereas it is inversely related to the size of the specimen. Article Highlights A new methodology for accurate thresholding of X-ray CT images of compacted soils is proposed. Exact quantification of small and large pores outside the range covered by MIP tests is attained by the proposed method. Variation of most probable threshold number with sand content, compaction energy and specimen size is examined.