An efficient image processing methodology for rapid 3D size and shape parameter extraction of multiple particles

Abstract

Accurate, fast, and automatic measurement of 3D size and shape parameters for multiple particles is essential for understanding their characteristics and behavior across different site conditions. This study proposes an efficient methodology for determining particle size and shape parameters using image processing techniques. The approach allows for quick measurement of multiple particles, with an easy setup for automatic segmentation and extraction the 3D parameters, while also exploring influencing factors and optimal setting. The method involves four key processes: image capturing, 3D models reconstruction, single-particle segmentation, and data acquisition. It combines computer vision algorithms and point cloud processing technology to extract 3D data from multiple particle images captured using a smartphone camera. By comparing the results from a 3D laser scanner, the method for 25 particles is assessed using error percentage in the laboratory under various influencing factors, including particle size, shooting angle, object-to-camera distances, and the number of captured images. Additionally, a site investigation of the methodology is executed and contrasted with sieve analysis. The results show an error percentage of less than 10% when using a 12-megapixel smartphone camera at a 45° capturing angle with 36 images. With rapid data acquisition, semi-automatic analysis, a simple, low-cost setup, and minimal error, the method proves efficient in capturing 3D size and shape parameters for multiple particles.