We present a method for analyzing short duration optical tracks of Earth orbiting objects and combining them into longer duration tracks. This allows for a high quality initial orbit determination fit and improved observation association, particularly when working with tracks on closely spaced objects. The method takes advantage of the locally linear behavior of the measurements in each track to identify pairs of tracks that fit well together. These pairings are used to build a directed acyclic graph (DAG) of all combinations of the tracks being analyzed. The longest path through the DAG identifies a set of short tracks that can be combined into a longer one. These tracks are then removed from consideration and the process is repeated. The method is in operational use in the Commercial Space Operations Center (ComSpOC) where it is applied to data from multiple sensor networks. Results from interesting operational cases such as the SJ-17 proximity operations will be presented.