<p>Pore network modeling is a critical tool in material science, geology, and chemical engineering for predicting fluid flow and transport properties in porous media. Traditional pore network extraction methods often depend on parameter tuning and require separate post-processing, resulting in high computational costs. This paper introduces NetXtractor, an efficient algorithm that streamlines the extraction of pore networks from 3D images by unifying marker generation, watershed segmentation, and structural property computation into a single-pass process. By integrating intuitive depth-first search-based marker detection, an efficient bucket queue-driven watershed segmentation, and concurrent structural property computation into a single-pass process, NetXtractor eliminates the need for extensive parameter tuning and additional post-processing. Benchmarking against state-of-the-art methods such as SNOW and the Maximal Ball algorithm, NetXtractor demonstrates substantial improvements in computational efficiency. For instance, in large-scale 3D images, it achieves up to a 19-fold speedup and uses up to 5.6 times less RAM compared to SNOW, while providing comparable accuracy in predicting effective diffusivity and permeability, correlating well with lattice Boltzmann simulations and experimental data. This comprehensive approach not only accelerates the extraction process but also ensures reliable pore network representation by consistently capturing both fine-scale features and large-scale connectivity without parameter tuning. These results highlight the potential of NetXtractor as a scalable, high-fidelity tool for advanced porous media analysis in both research and industrial applications.</p>