本文已被:浏览次 下载次 |
码上扫一扫! |
|
|
Pore structure characterization of shales using SEM and machine learning-based segmentation method |
LIU Xuefeng1, ZHANG Xiaowei1, ZENG Xin2, CHENG Daojie3, NI Hao1, LI Chaoliu4, YU Jun4, HU Falong4, LI Changxi4, WEI Baojun1
|
(1.College of Science in China University of Petroleum (East China), Qingdao 266580, China;2.Development and Reform Bureau of Xingwen County, Yibin 644400, China;3.China Petroleum Logging Company, Xi 'an 710077, China;4.PetroChina Research Institute of Petroleum Exploration & Development, Beijing 100083, China)
|
Abstract: |
Three shale samples from Wufeng-Longmaxi formation were selected to acquire two-dimensional (2-D) and three-dimensional (3-D) grayscale images by modular automated processing system(MAPS) and focused ion beam scanning electron microscopy(FIB-SEM) techniques, respectively. The grayscale images are segmented into pores, organic matter, inorganic matrix, and pyrite by the machine learning-based algorithm. The identified pores are classified into organic pores, inorganic pores, and micro-fractures according to their location and shape. The contents and size distributions of three pore type are then calculated. The results show that compared with threshold-based segmentation algorithms, the pixels located in the anomalous contrast area are classified properly by the machine learning-based method. The pore of shales with radius ranging from 10 nm to 200 nm are categorized accurately. The total porosities and volume fractions of organic matter calculated from SEM images are slightly different with those measured in lab. |
Key words: shales pore structure SEM image segmentation machine learning |
|
|