Accurate pedestrian localization is fundamental to urban monitoring systems such as crowd management and anomalous behavior detection. Multi-view systems combining CCTV and UAV imagery extend monitoring coverage, but pedestrian locations estimated from each domain must be consistent in a common world coordinate system to enable cross-domain integration and reliable pedestrian tracking. Despite this requirement, no systematic comparison of how to define a reference point for pedestrian localization across aerial and ground-level camera domains has been conducted. This study proposes ten reference point selection rules combining bounding box- and pose estimation-based methods across CCTV and UAV domains, and evaluates their localization errors under varying camera heights, UAV altitudes, distances from the camera, and pedestrian poses in a simulation environment. Results show that using the bilateral toe tip midpoint as the reference point achieved the lowest overall localization error of 0.169 m, as toe tips are the body part closest to the ground. However, foot center- and ankle-based rules outperformed it under poses where the toe tips are lifted and at far distances, suggesting that condition-aware reference point selection may yield better overall performance. To the best of our knowledge, this is the first systematic comparison of reference point selection rules for pedestrian localization in aerial-ground multi-view systems, providing a quantitative foundation for reference point selection in practical monitoring system design.