Characterization of coal permeability considering fracture shape—Using the MP-Otsu threshold segmentation algorithm

作者：Gang Wang, Xuechang Chen, Jiangfeng Liu, Zhen‐Yu Zhang, Qiming Huang, Hao Xu · 发表于：Journal of industrial safety. · 年份：2024 · DOI：10.1016/j.jinse.2024.100021 · 被引用次数：11 · 研究领域：Coal Properties and Utilization、Rock Mechanics and Modeling、Geoscience and Mining Technology

Coal seepage simulation based on digital images is capable of visualizing the three-dimensional seepage process and providing permeability data. However, Otsu's conventional thresholding algorithm is prone to image mis-segmentation for extracting pores and fractures, which makes the reconstructed model unable to reestablish the coal structure truly. The present work first explains the failure mechanism of the conventional Otsu algorithm for image segmentation based on computed tomography (CT) images of coal. Three weighting factors, including the slope of the mineral content fitting curve against the Otsu threshold and the approximate proportion of the target in the image, are introduced to enhance the accuracy of the threshold relation. In continuing, the MP-Otsu thresholding algorithm is proposed. Finally, the 3D fracture model of the coal is reconstructed from the CT image via the MP-Otsu thresholding algorithm. Focusing on the shape of the model, the permeability is then calculated on the basis of the pipe flow model and the parallel plate model. The contribution of fractures with various radii in the pipe model and different apertures in the parallel plate model is appropriately incorporated into the permeability evaluation. The obtained results reveal that the presence of mineral components and the low porosity of coal are the chief reasons for the failure of the conventional Otsu algorithm in the digital image segmentation of coal. The porosities calculated by the impr...