A GRAPHICAL PROCEDURE FOR ANALYSING PARTIAL REGISTRATION-PLATE DATA
David Paul Watling, Mike J. Maher · Traffic engineering & control · 1988
The problem of determining genuine vehicle matches from two lists of partial registration numbers has been addressed in the past by deriving an approximate distribution for the number of spurious matches generated by some matching scheme; or, more recently, using direct maximum likelihood estimation of the number of true matches. However, thus far, the fact that vehicle passage times are usually available has not been fully exploited; intuitively, their inclusion should improve the accuracy of any matching algorithm, as well as giving an insight into the distribution of journey times. The analysis proposed is 'graphical' in that it was inspired by a plot (for simulated data) of passage time at first observation point (s) versus time of arrival at second observation point (t) for all possible vehicle matches. The result is a clustering of points about the line t = s + mu (where mu is the true mean journey time) corresponding to genuine trip-makers, with background noise caused by spurious matches.