Shape-adaptive geometric simplification of heterogeneous line datasets

Line generalization is an essential data processing operation in geographic information systems and cartography. Many point reduction and simplification algorithms have been developed for this purpose. During the last three decades, several attempts have been made to develop approaches of generalization that adapt to the geometric properties of the processed lines. They typically incorporate line segmentation based on quantitative descriptions of the line shape. Little attention has been paid to the cases in which heterogeneous lines of different geometric character are mixed in one dataset. One common example is administrative borders, which often contain natural and artificial, smooth and sharp, schematic and non-schematic, and regular and irregular shapes. The tuning of the simplification algorithm based on a quantitative description of the shape would be ineffective here, since different algorithms must be applied to lines of different characters. In this article, we present a general method and generalization model for the simplification of such datasets. The properties of schematism, smoothness and regularity are used to differentiate various line characters. Three line characters, including irregular non-schematic, irregular schematic and regular orthogonal schematic, were selected for the current study. The developed generalization model consists of preprocessing, processing and postprocessing stages. Line segmentation based on the detection of different characters is performed during the preprocessing stage. Then, Li–Openshaw, Douglas–Peucker and orthogonal simplification are applied to the appropriate segments in the processing stage. Postprocessing enables the addition of extra regularity to the simplified shape. A visual and quantitative assessment of the results is provided and demonstrates the effectiveness of the developed approach in comparison with the global application of a single simplification algorithm.