Selection of Image Processing Algorithms for Evaluation of Pervious Pavement Pore Network Properties

Abstract

Digital Image Processing (DIP) algorithms are often required as a precursor to measure the internal characteristics of pavement structures during X-ray computed tomography (XRCT) based non-destructive evaluation (NDE) of pavement materials. The improper use of DIP algorithms can result in the significant under- or over-estimation of internal pavement characteristics, thereby affecting pavement design and maintenance strategies. Past research studies highlighted the significance of threshold segmentation algorithms and binarization of greyscale images on the porosity and permeability characteristics of pervious pavement mixtures. In addition, the use of a watershed segmentation algorithm was introduced to separate interconnected pore network structure into multiple pores. However, isolated pores were not removed in past analyses found in the literature due to a lack of consideration in using ungrouping algorithm to segregate connected and isolated pores. The main objective of this study is to select the appropriate DIP algorithms that can be used to evaluate pervious pavement pore network properties from three-dimensional XRCT based images. In this paper, a key microstructural pore parameter was investigated using various DIP algorithms for different pervious pavement mixtures and recommendations are made. It is expected that the results presented in this paper can help researchers understand the importance of DIP algorithms on XRCT-based pavement evaluation studies.