Evaluation of Pervious Concrete Pore Network Properties Using Watershed Segmentation Approach

Abstract

Pervious concrete is widely used as pavement surfaces as means to increase water infiltration for water storage or conservation purposes or to reduce surface runoff (and increase skid resistance) for roadway safety. A proper evaluation of pervious concrete pore network properties is important to ascertain the ability of the material to serve the intended purposes and X-ray computed tomography (CT) scan is one method that allows for the non-destructive evaluation of the pervious concrete specimens. Pore network structures can be derived from X-ray CT scan images through the use of segmentation algorithms. Current image processing-based segmentation algorithms, however, can yield significant errors when deriving pervious concrete pore network properties. This paper describes the use of the watershed segmentation algorithm on X-ray CT scans of pervious concrete pavement mix and evaluate essential pore network properties such as pore volume, flatness, elongation, and shape factor distributions. First, the fundamentals of the watershed segmentation algorithms are described. The paper next presents on the experimental program in testing pervious concrete mix and the use of X-ray CT scans in deriving images of the samples. The watershed algorithm of different elevation functions are then applied to derive the pore network properties and the results are presented. Finally, the advantages of this algorithm over existing image processing techniques are discussed.