Technical SectionAutomatic generation of puzzle tile maps for spatial-temporal data visualization
Graphical abstract
Introduction
Maps can be classified into two categories, namely, topographic maps and thematic maps. Topographic maps accurately display the location of an object in a space, and thematic maps focus on combining specific themes with maps [1]. With the development of information visualization techniques, thematic maps have elicited increasing attention in visualization and computer graphics. Tile maps are important tools in thematic cartography, and their characteristics are distinct from those of well-known tools, such as choropleths, cartograms, and symbol maps [2]. The tiles in a tile map are placed in a grid at positions that approximate their geographic positions so that each tile has an identical shape and size. Tile maps provide an easy means to visualize the level of variability within a region and how a measurement varies across a geographic area. Furthermore, complex data can be shown on tiles in a consistent format that allows users to intuitively and easily compare data among tiles, as shown in Fig. 1. Compared with choropleths, cartograms, and symbol maps, tile maps have the clear and simple appearances.
Large amounts of time-series data exist in our daily life, and these data are generally visualized by diagrams, such as line charts and heat maps [3]. Most of these time-series data contain geographic information, which is referred to as spatial-temporal data. Spatial-temporal data cannot be exposed by diagrams. Geographic information plays an important role in this type of data. Thus, how to intuitively and clearly display spatial-temporal data is a challenge that requires urgent attention. 2D choropleths with time sliders [4] and 3D choropleths [5], [6] are common means to visualize spatial-temporal data with the aid of a geographical map. Time-series data are displayed in a geographical map with the aid of time sliders to manually display time-varying quantities or 3D graphical widgets and to represent the profiles of time-varying quantities in 3D space. In their study on information availability in 2D and 3D displays [7], Smallman et al. concluded that the readability of 2D displays generally outperforms that of 3D displays because of the simplicity of 2D displays and the occlusion problem in 3D displays. Therefore, in this study, 2D graphical widgets, that is, 2D puzzle pieces, are used in tile maps instead of time sliders or 3D widgets in a geographic map. A tile with puzzle pieces in a map is called a puzzle tile map, which can show spatial-temporal data in a single and static 2D diagram.
The proposed method for generating a puzzle tile map includes algorithms to optimize district-to-tile mapping according to geographic positions and orientations and to place puzzle pieces in a tile for time-series data visualization. The resulting puzzle tiles can display spatial-temporal data in a 2D map, reveal time-varying quantities in a sequence of puzzle pieces, and facilitate comparisons of time-varying quantities among puzzle tiles. Compared with related tile map generation study [2], [8], [9], this study proposes the use of puzzle pieces to display time-series data of a region and establishes an algorithm to optimize district-to-tile mapping. The remainder of this paper is organized as follows. Section 2 reviews related work. Section 3 introduces the overview of puzzle tile map. Section 4 describes the tile map generation. Section 5 demonstrates the puzzle representation. Section 6 discusses the experimental results and user studies. Section 7 presents the conclusions and future work.
Section snippets
Related work
Several spatial-temporal data visualization methods have been proposed, and they can be classified into three categories, namely, 2D static, 2D dynamic, and 3D visualization methods, on the basis of display dimensionality and data representation. 2D static methods were developed for temporal data visualization. For example, the heat map proposed by Suematsu et al. [3] visualizes multiple time-series data by using multiple color belts. A belt is filled with colors to represent time-sequence
The overview of puzzle tile map
The tile map generation, which is illustrated in Fig. 2, consists of four main steps, namely, data preprocessing, mosaic filtering, tile position correction, and mapping optimization. It represents spatial information. The input to the system is spatial-temporal data with a corresponding local map. In the first step, preprocessing is performed to simplify the input geodata and separate each connected object from the input map. This step accelerates the map generation and deals with maps
Data preprocessing
The input geodata format is GeoJSON, which supports the following geometry types: points, line segments, polygons, and collections of features. Text-based GeoJSON may lead to poor performance because of the duplicated region boundaries. To consider system performance, the input GeoJSON data is simplified. Buttenfield [21] identified the line simplification problem as a part of linear features, and there are several classic algorithms for the manipulations of lines, such as Peucker’s study [22].
Puzzle representation of tile
To visualize time-series data on tile maps, an aesthetic pattern, namely, a puzzle, is adopted to represent time-varying data. The input time-varying data can be sliced according to the users demand, and the appropriate time slices are 8, 12, 16, and 20. A puzzle consists of five characteristics, including tile position, puzzle pieces, piece color, puzzle connection point, and event point, as shown in Fig. 5. The proposed map based on puzzle tiles can visualize the information of geographical
Experimental results and user study
The proposed method is tested on a PC with an Intel Core i7 (1.7GHz) processor and 8 GB of memory. The code is written in JavaScript, and the one-to-one mapping problem is solved with the Hungarian algorithm. The computational time for puzzle tile map generation is 19.19 milliseconds. In addition, we provide several sub-panels to show the details of the selected tile (i.e., a summary statistic and focus tile).
A screenshot of the interface is shown in Fig. 8 and it is an overview-plus-detail
Conclusions, limitations, and future work
The puzzle tile map is a novel visualization technique that combines various characteristics into 2D space without sacrificing one of the characteristics of spatial-temporal data. A tile is composed of puzzle pieces and can integrate the temporal information into the spatial data. The proposed spatial-temporal visualization represented by 2D diagram can focus on time-series data in a specific region from the geographic perspective. Puzzle tile map has the ability to visualize numeric data in
Acknowledgments
The authors would like to thank the anonymous reviewers for their valuable comments and suggestions to improve the manuscript. This research was supported in part by the Ministry of Science and Technology (contracts MOST-106-2221-E-019-069-MY2, MOST-107-2119-M-006-027, and MOST-104-2321-B-019-009), Taiwan.
References (30)
- et al.
Drawing road networks with mental maps
IEEE Trans Vis Comput Gr
(2014) - et al.
Generating tile maps
Comput Gr Forum
(2017) - et al.
A heatmap-based time-varying multi-variate data visualization unifying numeric and categorical variables
Proceedings of the eighteenth international conference on information visualization
(2014) - et al.
Dynamic queries and brushing on choropleth maps
Proceedings of the fifth international conference on information visualization
(2001) - et al.
A 3d visualization of multiple time series on maps
Proceedings of the fourteenth international conference on information visualization
(2010) - et al.
3d information visualization for time dependent data on maps
Proceedings of the ninth international conference on information visualization
(2005) - et al.
Information availability in 2d and 3d displays
IEEE Comput Gr Appl
(2001) - et al.
Small multiples with gaps
IEEE Trans Vis Comput Gr
(2017) - et al.
Improved grid map layout by point set matching
Int J Comput Geom Appl
(2015) - et al.
Data vases: 2d and 3d plots for visualizing multiple time series
Proceedings of the fifth international symposium on advances in visual computing: part II
(2009)
What about people in regional science?
Pap Reg Sci Assoc
Interactive analysis of event data using space-time cube
Proceedings of the eighth international conference on information visualization
Data visualization strategies using tile grid maps
A semi-automatic way to create your own grid map
Good data visualization practice: tile grid maps
Cited by (5)
A Note from the Editor in Chief
2019, Computers and Graphics (Pergamon)Positioning and Mapping Technology for Substation Inspection Robot Based on Gaussian Projection
2023, 2023 29th International Conference on Mechatronics and Machine Vision in Practice, M2VIP 2023Development of JavaScript library for visualization of operational environmental monitoring data
2021, AIP Conference ProceedingsData visualization system based on big data analysis
2020, Proceedings - 2020 International Conference on Robots and Intelligent Systems, ICRIS 2020