Currently sorting through the mechanization of the other algorithm mentioned at the bottom. I worry about the rejection of false positives when there are no horizons in frame--this might be possible through a simple threshold on the minimum possible mahalanobis distance to register as a hit.
Currently sorting through the mechanization of the other algorithm mentioned at the bottom. I worry about the rejection of false positives when there are no horizons in frame--this might be possible through a simple threshold on the minimum possible mahalanobis distance to register as a hit.