Online pattern recognition in the CTD is performed in
two dimensional projections.
Hits in the transverse r- 
projection are obtained by
knowing which axial sense wires were hit and using a planar
drift approximation.
Although the stereo layers will give precise measurements of
track polar angles ( 
) and beam-line vertices (z)
this information cannot be decoded until the track finding
in the r- projection is completed.
This is too time consuming for the trigger so some axial
layers have been instrumented with z-by-timing
electronics enabling r-z information to be available to
the trigger.
The CTD SLT works in the r- plane to determine
the number of tracks in the event, their 
, origin,
and the event vertex, while the FLT supplies rudimentary
r-z hit patterns.
The SLT two dimensional r- information will be
correlated with the FLT z-information for full
three dimensional pattern recognition or consistency checks
between the two trigger levels.
At present the CTD-SLT pattern recognition works
exclusively with the axial r- projection but will
eventually include z-information.
Parallel processing is exploited in the r-
track finding.
The layered structure of the wires in the CTD allows the
eight sense wires in an axial superlayer to be used to form
track segments which are labeled by their position and
direction, these are called vector
hits
.
Segment finding is a local operation and can be performed in
each cell independently.
By exploiting parallelism the method has the
advantage of reducing the amount of data that must be
handled in each pattern recognition step.
The CTD-SLT track finding algorithm is thus split into two stages.
In this note we will discuss in detail the segment finding algorithm, while the segment matching is only briefly mention and will be described in a future note. Figure 2 show the Entity Relationship Model diagram of the CTD-SLT system. The actual online data formats proposed for the system are described in reference [3].
Figure 2: Entity Relationship Model for CTD-SLT
system.