A fast curtain-removal method for 3D FIB-SEM images of heterogeneous minerals.

The focused ion beam-scanning electron microscope (FIB-SEM) system plays a crucial role in the research of shale reservoirs. It enables visualise nano-scale pores and helps characterise unconventional reservoirs. In this system, FIB removes a thin layer and SEM generates a high-resolution grey-scale image of this fresh surface. By iteratively using FIB and SEM, we can create a series of 2D images. Through stacking these images, a 3D model of a rock sample is generated. However, curtain noise of varying intensity often appears in FIB-SEM data. Its presence can cause a severe effect on estimation of rock properties, such as porosity and permeability, and lead to incorrect interpretation of the FIB-SEM image. Because curtain noise can be falsely identified as needle-shaped pore throats based on grey-scale image segmentation. Thus it is imperative to decrease curtain noise before segmentation in order to obtain a better understanding of a rock sample. In this paper, we propose a novel approach considering mineral density to decrease curtain noise and compare its results with several conventional used methods.

LAY DESCRIPTION: The focused ion beam-scanning electron microscope (FIB-SEM) system plays a crucial role in the research of shale reservoirs. It enables visualise nano-scale pores and helps characterise unconventional reservoirs. In this system, FIB removes a thin layer and SEM generates a high-resolution grey-scale image of this fresh surface. By iteratively using FIB and SEM, we can create a series of 2D images. Through stacking these images, a 3D model of a rock sample is generated. However, curtain noise of varying intensity often appears in FIB-SEM data. Its presence can cause a severe effect on estimation of rock properties, such as porosity and permeability, and lead to incorrect interpretation of the FIB-SEM image. Because curtain noise can be falsely identified as needle-shaped pore throats based on grey-scale image segmentation. Thus it is imperative to decrease curtain noise before segmentation in order to obtain a better understanding of a rock sample. In this paper, we propose a novel approach considering mineral density to decrease curtain noise and compare its results with several conventional used methods.