Estimation and documentation of the grain size distribution of soil using image analysis and low-cost optical sensors

With the progress of the mining industry, the rapid and efficient characterization of geotechnical properties of excavated soil has become a key component in mining operations evaluation, planning, and particularly in the management of post-mining landscapes. For landfills and mining waste dumps, the grain size distribution of soil strongly influences compressibility, liquefaction potential, and other critical parameters. Image analysis techniques represent a low-cost and sustainable alternative for soil texture characterization, enabling effective evaluation of soil particle size and shape without the need for extensive laboratory work. This study investigates the capability of image analysis to characterize waste sands for assessing sand liquefaction potential and supporting dump model evaluation. An image dataset of several representative waste sands was collected, documented, and processed, forming the basis for a spatial image database to support future research and operational monitoring. The analysis yielded quantitative measurements of soil grains—such as diameter, area, perimeter, and shape factors. Comparison with traditional laboratory-based sieve analysis demonstrated that the computed grain size distribution curves achieved an approximately 90% correspondence. The integration of advanced quickshift segmentation with image polygon property calculations provided accurate estimations of real grain sizes, offering an efficient, low-cost, and well-documented approach with minimal material and time requirements.