Deep learning in automatic map generalization: achievements and challenges

Abstract

Map generalization has always been a hot topic in the field of Geographic Information Science (GIS) over the past decades. Scholars have been dedicated to utilizing opportunities offered by technological advancements to drive the rapid transformation of map generalization from manual to interactive modes, with an extension toward automatic mode. Deep Learning (DL), known for powerful data-processing and pattern recognition capabilities, has introduced new possibilities for automatic map generalization. Novel studies eagerly adopt DL methods and explore their mechanisms to enhance the level of automation of map generalization. However, current research on this topic remains relatively scattered and thus a systematic summary and in-depth analysis are required. This study presents an overview of the achievements in addressing map generalization task using DL, with emphasis on the progress in the past five years, covering the aspects of pattern recognition, algorithm design, process control, and result evaluation. Furthermore, we examined the latest development trends of advanced DL methods, specifically large models (LMs), in the context of map generalization and identified potential future research directions. We anticipate that this work will catalyze a new wave of technological advancements in the field of automatic map generalization.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 16 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.