<p>Frequent and reliable observation of ground targets is critical for an Earth observation satellite constellation. Limited satellite numbers often result in long revisit intervals. This study aims to minimize the worst-case revisit time in satellite constellation design. The paper proposes an integer linear programming (ILP)-based framework under common ground track constraints. The framework optimizes seed orbital elements and discretizes possible satellite phases for inclusion in the ILP. In a low-Earth orbit case study, the approach reduces the worst-case revisit time by over 60% compared to a symmetric constellation. The ILP-based method provides a systematic, computationally efficient solution for minimizing worst-case revisit time in constellation pattern design. Future work can extend the framework to account for multiple ground targets and incorporate additional measures of effectiveness.</p>