Described is system for optimizing visual inspection of radar targets. The system detects radar hits with a radar system comprising radar and a radar-cued camera. The camera has a current state comprising a current slew position and a current zoom level. The radar hits are stored, and a set of metrics are determined for each radar hit. A track value (TV) metric is determined that combines the set of metrics for each zoom level of the camera. The camera sends a set of commands based on the TV metric, such that the camera slews from the current state to a new state. The new state is a position of a radar hit with the largest TV metric and a corresponding zoom level. Given the new state, an image is captured and processed to generate captured tracks. The current state of the camera is updated to reflect the new state.