Estimation of underground fluid state has been paid great attention in subsurface explorations (e.g. oil reservoir development, carbon capture and storage). Permeability is the most important parameter in considering subsurface fluid flow, and seismic-wave velocity is the most popular and trusted parameter derived from geophysical surveys. The objective of this study is estimating permeability from seismic velocity by revealing the relationship between permeability and seismic velocity. Although these two parameters have no direct relationship, the pore geometry of rock can be a bridge of them because it is dominant factor to govern permeability and seismic velocity. Since pore geometry of rock mass is highly complicated, two rock models (cracked rock model and granular model) are adopted for the research. For the calculation of permeability, lattice Boltzmann simulation is conducted in this research. Self-consistent approximation and finite element method are applied to calculate seismic velocity on cracked model and granular model, respectively. As a consequence of the research, permeability can be estimated from seismic velocity using the information of pore geometry: (1) crack aspect ratio and intensity for cracked model and (2) grain-size sorting for granular model.