During heavy rain, sparse short-lived points appear in an outdoor monitoring zone. I want to reduce nuisance detections without delaying detection of an approaching vehicle.
Distinguishing rain returns from a real object
◉ 77 · ◇ 1
1 replies
Compare temporal persistence, cluster size and motion consistency instead of rejecting all isolated points with one threshold. Test against the smallest legitimate target so the weather filter does not remove the event you need to detect.