A Feasibility Study of Vehicle-type Classification based on Symbolic Representation

For automated traffic monitoring and analysis system, the information about vehicletype is very important. It is necessary to develop a technique of vehicle-type classification which requires less computation power and is robust against small variations in appearance such as translation, expansion, difference of color, and so on. We have been considering developing a vehicle-type classification based on symbolic representation which focuses attention on the structure of vehicles. In this paper, we introduce points to be considered in determining the feasibility of such a classification technique.