Perception Recognition Study on Long and Narrow Arcs of Polygons in Maps using Intelligent Skeleton Decomposition

Abstract

The paper presents an automated method for identifying elongated and narrow sections of polygons, which may arise due to uncertainties inherent in map data production and updates. It leverages a skeleton-based shape decomposition approach, encompassing the creation of skeletons and end-point sub-polygons based on the polygon boundary-constrained triangulation. Subsequently, it identifies a pair of break points using concave vertices of polygon boundaries based on the nearest neighbor principle. The optimal segmentation of the end-point sub-polygons is then determined by evaluating weighted base-height ratios, which are added to a set of candidates long and narrow arcs. Finally, the model selects the long and narrow arcs based on their base-height ratios and shape indexes. The paper reports the results of an experiment using land parcel data from Jinjiang in Fujian Province, demonstrating that the proposed method produces results that align with human perception and outperforms the internal and external buffering method. In summary, the innovative method exhibits promise in accurately identifying elongated and narrow sections of polygons, Its applicability spans diverse domains such as map data analysis and updates, where precision in delineating such features is crucial.