A method for partitioning very small targets that accounts for crossing point constraints

Very small targets (VSTs) are common elements of national geographical condition data, and the integration of these targets directly affects the quality of results synthesized from these data. Most conventional methods use amalgamation or aggregation to merge VSTs with their proximal patches, but these approaches tend to neglect the competitiveness of each proximal patch. To address this gap, we propose a method of partitioning VSTs that accounts for crossing point constraints. We first analyze how surface area, semantic proximity, length of shared edges, and regional importance affect the splitting ability of a proximal patch. Then, we use the analytic hierarchy process to construct a hierarchical model of these factors, in which the weights of each factor are calculated. Finally, a comprehensive assessment of the splitting ability of each proximal element is performed, and the skeletons of VSTs are amended accordingly, thus realizing the partitioning of VSTs. The viability and effectiveness of the method proposed in this work is validated in experiments using real data.