Variable-Scale Visualization of High-Density Polygonal Buildings on a Tile Map

Abstract

To better satisfy user’s needs for the accurate visualization of massive amounts of geographic data, the variable-scale expression of map content based on multilevel data organization has attracted increasing attention. Traditional methods based on vector data usually cannot handle tile data in the form of a grid on the network. Therefore, this paper proposes a variable-scale visualization method for high-density buildings based on a raster tile map. First, the buildings on a tile map are typified on the basis of linear spectral clustering (LSC) superpixel segmentation to reduce the number of buildings. Then, the shapes of buildings are simplified using the minimum bounding rectangle method. Lastly, the designed focus + glue + context (F + G + C) variable-scale model is used for visual output. The OpenStreetMap tile data are used to perform experiments. Compared with traditional methods, the proposed variable-scale visualization method in this paper considers the spatial distribution, quantity, and shape characteristics of buildings, reduces the clutter of data, and has a better (average value of building quantity, area and density is 57%) visual effect. Variable-scale visualization can be applied to unstructured map data sources and extended to grid data sources to improve the readability and recognizability of high-density buildings.

Keywords

• variable visualization
• variable-scale map
• architectural typification extraction
• high-density buildings
• LSC superpixel segmentation
• hierarchical data organization