主流喉道是致密砂岩中决定储层渗透率、沟通不同类型孔喉的桥梁，因此其三维结构表征对于理解油藏微观驱油机理有重要意义。micro-CT图像作为储层表征的重要手段，受限于分辨率，表征致密砂岩主流喉道难度大。本研究以micro-CT以及高精度扫描电镜成像为基础，以深度学习技术为主要工具，开展micro-CT图像超分辨率重建和高精度图像自动分割，实现致密砂岩主流喉道的三维表征。结合水驱油、气驱油、渗吸实验后的micro-CT三维流体分布成像，阐明主流喉道结构特征对不同驱油方式下驱油路径、驱油效率的影响。项目的成功实施，对于提高致密砂岩储层表征精度，揭示其微观驱油机理，优化其评价指标以及开发技术政策，并最终提高油藏采收率有重要意义。

The mainstream throats in tight sandstones govern permeability and connect different pores, therefore their three dimensional properties are critical to the oil-displacement mechanism. Micro-CT is an effective tool for the reservoir characterization, however, still lack enough resolution for the mainstream throats in tight sandstones. Micro-CT images and high resolution SEM (MAPS and EDS-SEM) images are integrated for the super-resolution reconstruction of micro-CT images and automatic image segmentation based on the deep learning methods. The reconstructed images are then used for the dominant throats characterization and when combined with the oil drainage experiments (water flooding, gas drive and imbibition) and fluid distribution recorded by the micro-CT, the effects of the mainstream throats on the drainage path and drainage efficiency can be understood. This project’s successful implementation contributes to the reservoir characterization of tight sandstones and the understanding of oil-displacement mechanisms. In addition, it will lead to better reservoir evaluation parameters and production strategies, which will contribute to the increase of final oil recovery of the tight sandstone reservoirs.