Effect of Global Thresholding Algorithms on Pervious Concrete Pore Network Properties Using XRCT-Based Digital Image Processing

Digital image processing of the X-ray computed tomography images involves the crucial step of image segmentation which affects the subsequent pore structure quantitative analysis. The main objective of this study is to investigate the effect of ten different global thresholding algorithms based on the grey scale histogram, clustering, entropy and laboratory volumetric characteristics on the internal pore structure properties of the pervious concrete. The key microstructural parameters of the pervious concrete air voids such as porosity, tortuosity, throat number, pore coordination number and distributions of pore volume, throat area, pore sphericity, shape factor and throat eccentricity were analyzed for different thresholding algorithms. It was found from the analysis that the nine histogram, clustering and entropy based algorithms are found to be either under or over estimating the air void voxels compared to the volumetric segmentation method. And as the threshold value increases, effective porosity and number of throats increases and isolated porosity and tortuosity decreases due to the increase of air void voxels and pore connectivity. Overall, it is expected that the present study will help in understanding the importance of threshold segmentation in the field of pavement image processing.