To more accurately describe the mechanical properties of deep rock under impact loads, an increasing number of studies are considering the influence of microscale anisotropy on the dynamic mechanical performance of rocks. The microstructures of rock, such as pores and joints, are significant factors causing the anisotropy observed in rock. Most studies on the anisotropy of rock focus on systematically distributed pores or joints, while there is relatively less investigation into the influence of randomly distributed pores on the dynamic mechanical properties of rock. To address this issue, this study developed an algorithm for generating three-dimensional models of rock samples containing randomly distributed ellipsoidal pores. The effectiveness of this three-dimensional model construction method was partially validated through numerical simulations using the Split Hopkinson Pressure Bar (SHPB) technique. The specific process involved: (1) A MATLAB algorithm, which enable to create rock sample three-dimensional models with both size and distribution randomness of pores, was developed; (2) based on the fundamental principles of the mesh mapping method commonly used for irregular shape meshing, a Ls-Dyna secondary algorithm was developed for meshing samples containing irregular ellipsoidal pores; (3) A series of SHPB compression numerical simulation experiments were conducted on rock samples with different porosities. The result exhibits a high degree of consistency with previous results: as the porosity of the rock increases, the uniaxial compressive strength of the rock under impact loads gradually decreases, which reflects the effectiveness of this model construction method in the field of rock mechanics investigation.