<p>The technology of locating magnetic anomaly targets via geomagnetic field measurements has been increasingly widely applied, with multiple magnetic anomaly target localization emerging as a critical research direction. However, when two magnetic anomaly targets are horizontally close but vertically separated, traditional clustering-based localization methods tend to omit the deeper target. To address this issue, we propose an improved clustering-based localization method for multiple magnetic anomaly targets, which integrates two core innovations: the introduction of a reference target to establish a benchmark for normal magnetic moment distribution, and the utilization of spatial distribution characteristics of magnetic moment estimates to judge the presence of omitted targets. Simulation results demonstrate that the proposed method not only achieves accurate localization of conventional targets but also effectively identifies the omission of deeper targets, providing a reliable basis for determining whether supplementary localization steps are required.</p>