A novel multiphase segmentation method for interpreting the 3D mesoscopic structure of asphalt mixture using CT images

Abstract

The three-dimensional reconstruction of asphalt mixture using X-ray computed tomography (CT) images is beneficial to establish the correlation between its mesoscopic structure and its macroscopic properties. Segmentation of CT images that distinguish the mineral aggregates and air voids in asphalt mixture is an essential step for this purpose, wherein up-to-date methods are often limited due to complex image quality and failed to identify the bituminous mortar with gray levels ranging between void and aggregate. In this paper, we propose a local threshold based and Monte-Carlo incorporated segmentation method to address the above problem effectively. The probability of each pixel identified as aggregate could be calculated in a CT image. Notably, the proposed method is able to identify the multiphase components of asphalt mixture and build the three-dimensional mesoscopic structure model to investigate the spatial variation of different components. The proposed method has been verified in a case study, wherein 400 sequenced CT images of an asphalt mixture sample are segmented to interpret its mesoscopic structure. Results show that the proposed method performs well to distinguish the multi-phase components of the asphalt mixture, particularly the bituminous mortar that is usually miscategorized as void or aggregate by previous binarization methods.