Map Matching Based on Conditional Random Fields and Route Preference Mining for Uncertain Trajectories

Algorithm 1

Map matching algorithm based on CRF and route preference mining.

Input: a trajectory of vehicle , CRF model, IDT of vehicles

Output: the matched path (road segment sequence)

(1)  Check the average sampling interval of trajectory, if lower than threshold, goto (2); otherwise  goto (9);

(2) for   to   do

(3)   for  each candidate road in candidate road set of GPS observation of time   do

(4)    Count the number of paths passing through

(5)     for  each candidate road in candidate road set of GPS observation of time   do

(6)       Count the number of paths passing through to in order

(7)       Calculate the route preference of driver using formula (10)

(8)       Calculate new transmission features using formula (11)

(9) Calculate the maximum posterior probability using Viterbi algorithm.