Application of Novel Radial Thresholding Method for the Segmentation of Different Phases from X-Ray Microtomographic Images of Concrete

Abstract

Concrete is a composite material, which can be segmented into three major phases, namely voids, aggregates and mortar. This paper presents the digital image processing techniques based on grey value thresholding and a novel radial thresholding approach to segregate the three phases of concrete. In this context, the 8-bit images of concrete specimen obtained from X-ray microtomography (XRT) scanning of cylindrical specimen are operated. The non-local means denoising filter is used to remove the unwanted noise from the original images and enhance their clarity without losing any details. There is a clear distinction in the grey values of air voids from that of aggregates and mortar. The threshold grey value of air voids is determined by observing the variation in grey value profile near the edges of the air voids, and using this threshold grey value, air voids are segmented assuredly. However, the segmentation of phases using this thresholding technique doesn’t suffice to isolate the aggregates from mortar because of the overlap of their grey values. Hence, a radial thresholding method is proposed for the detection and determination of the phases, which works similar to our eyes. The grey value vs radius graph exhibits sudden jumps, which represent the change in contrast, that is, phase. The change in phases is evaluated by using a simple function, ∥GV(n)−GV(n−1)|−L|>0 which is considered the radial variation for every degree rotation. The estimated air voids and aggregates content are 0.91 and 49.19%, respectively. The error in the detection of aggregates content is only 0.6%.