Entropy-Based Similarity Evaluation and Visualization of Cartographic Symbol Sets

In cartographic symbology, the evaluation of a symbol set´s visual similarity is an important and frequently required task. Usually, after a set of map symbols are designed, the visual similarity of symbols need to be manually examined. To fully automate this task, in this paper, we propose two approaches based on entropy calculated on SOM surface for quantifying and visualizing the visual similarities of cartographic symbol sets. Using our approaches, the visual similarities of multiple symbol sets can be compared efficiently. Apart from evaluating the visual similarities of cartographic symbol sets, both of our methods are also suitable for applications where comparing the visual similarities of multiple image datasets are required.