Abstract:

In order to correctly describe and recognize the configuration of a time series crave （TSC） , five morphemes and one wildcard are defined according to the curve segment types, and the corresponding morpheme vector and the wildcard vector are defined for the hierarchical description of the curve. Then, one TSC is transformed into a 2D table. The second colmnn of the table makes up of a string that can effect the elementary pattern matching and the rows of the table represent the attribute vectors that can enhance the morpheme-relational operation, thus reali- zing a more complex pattern matching at the next step. Moreover, a string matching arithmetic with attribute re- straints and wildcard strings is proposed based on the classical backtracking method. The method is finally em- ployed to classify vehicles using the TSCs collected from inductive loops in a real traffic scene. The results show that the proposed method is effective and reasonable.

Key words: time series curve, configuration description, configuration recognition, string matching, vehicle classification