SELF: Semantically Enriched Line simpliFication

Abstract

Linear features are represented on paper or digital maps with polyline geometries. Sampling, discretization, and generalization processes result in polylines of a length smaller than that of the actual features. In addition, semantics associated to the original line features may be lost. This becomes more significant for coarse sampling and/or high degree of generalization. This paper introduces a data structure that can alleviate this problem, by preserving the attributes and semantic characteristics associated to the original features in cartographic representation. The structure can handle both linear features and polygon outlines. Various compression methods have been examined. The structure has been implemented and tested with both synthetic and real datasets. Extensions to spatiotemporal features, like trajectories, have also been considered.