The algorithm described in Section 18.4.4 is only applicable for extending tracks which already exist in the track file. The creation of new entries is done by a separate track initiation function.
The track initiator involves little more than searches for nearly colinear triples of data over the last three scans. A triplet
where the cutoff is generally fairly loose (e.g., ). In addition, a number of simple heuristic cuts (maximum speed) are applied.
The initiator searches for all approximately linear triples over the last three scans, subject to the important additional restriction that no initiations to a particular item of the current data set are attempted if any established track (minimum age cut) already exists ending at that datum. The nominal complexity of the initiator is reduced to approximately by exploiting the sorted dature on the incoming data sets.