A parabolic spatial filtering approach for route search optimization

Web-based transit information system, aiming to advise on the choice of public transport routes, is receiving popularity. Route matching over a large network is endowed with lots of practical considerations especially for multi-criterion trip planning such as least time, least fare, least and convenient interchanges. Most passengers like to be provided with more than one reasonable choice. However, in a large city like Hong Kong with complex fare structure, route computation for least fare using conventional path finding methods may result in illogical options in terms of travel behavior. This paper proposes a parabolic spatial filtering method to eliminate spatially unacceptable results from existing path finding algorithms. Using results from the Hong Kong eTransport system, the method can effectively eliminate quite a number of routes that traverse a long and winding path between the origin and destination, thereby only those routes commonly accepted by public transport commuters will remain.