A novel method to determine the optimal threshold of SEM images

Abstract

The determination of threshold is crucial for the binarization of scanning electron microscopy (SEM) images. In this paper, a new method for determining the optimal threshold considering the three-dimensional characteristics of shale samples is proposed based on GIS. Firstly, the SEM images of seven shale samples at six magnifications are 3D reconstructed and visualized using ArcMap and ArcScene. Relevant parameters are obtained to calculate the 3D porosity φ3D. Secondly, the effects of magnification and segmentation size on φ3D are considered. By establishing the relationship between threshold and porosity, the optimal magnification, optimal segmented size and optimal threshold are obtained for the seven shale samples, respectively. Finally, the optimal threshold obtained by this method is applied to the finite element model of the pore structure and fractal dimension, respectively. The results show that the overall effect of 3D reconstruction and visualization is good. The optimal thresholds of seven shale samples are 94, 93, 105, 93, 81, 98 and 88, respectively. The porosity obtained from the finite element model of the pore structure based on the optimal threshold is suitable for SEM images with larger sizes. The fractal dimensions obtained based on the optimal threshold are all larger than those obtained at six magnifications, which can characterize a more complex pore structure and a more inhomogeneous pore size distribution. This method fully considers the three-dimensional characteristics of different samples and improves the accuracy of the threshold study of SEM images.