Keywords

1 Introduction

Before smartphones got popular, bicyclists needed dedicated GPS devices [1] for position tracking, and the recording and sharing of the users’ GPS trajectory were limited. Currently, the smartphones make it possible for them to easily collect and share the user’s daily position logs with built-in GPS module and tracking applications. The tracking applications of the smartphones record positions in two- (latitude and longitude), or three-dimensional (latitude, longitude and elevation) coordinates and overlays the trajectories on the map. They also provide additional activity information such as average speed, acceleration, pace and calories burned which were retrieved from the trajectories.

The bicyclists have tendency to share their riding experiences with others [2]. Recently, GPS trajectories sharing on the website is one of the most popular activities in the online bicycle communities, so the number of the shared bicycle GPS data is steadily increasing. The route choice of bicyclists differs from the motorists [3, 4]. While the latter tends to choose the shortest path, former prefers to ride on the segregated bicycle facilities such as bicycle tracks or lanes. Therefore, the distribution of the bicycle GPS trajectories is tends to concentrate on the specific locations.

The aggregated trajectories have been studied in various fields. In the fields of intelligent transportation system, massive GPS trajectories are used to create and modify the road maps in real-time [5]. They also have been applied to the map-matching techniques to enhance accuracy [6]. For the visualization of the spatial data, the GPS trajectories provide the base elements for drawing: vertices and lines. With these primitive elements, the trajectories can simply be displayed as sequences of line segments.

Various methods and techniques were proposed to provide more insight into the visualized trajectories. Visualizing GPS trajectories based on the kernel density estimation provides intensity of the scattered data with proper range of the distribution [7]. Scheepens et al. [8] proposed a visualization system which consider the moving objects as multivariate time series and shown the architecture of the system. The system is based on the density map and applied multiple density fields to the visualization.

Up to date, most of the trajectories visualization methods are focused on the high-speed transportations such as automobiles, airplanes and vessels [7, 8]. The visualization methods for bicycle trajectories or pedestrians [9] are relatively hard to find. In this paper we propose a simple color coding method for the visualization of bicycle trajectories. The motivation of the method is based on the behavior of the bicycling. The bicycle trajectories on the segregated bicycle lanes are less concentrated on one side of the road compare to the trajectories of automobiles because bicyclists cross the centerline of the road much more often than motorists. This symptom creates overlapping of the trajectories and makes display hard to understand.

Figure 1 shows the bicycle trajectories collected from segregated bicycle tracks alongside Han River in Seoul. The bicycle track consists of two-way bicycle path, thus the trajectories were collected from both side of the roads. The overlapping of the trajectories can be found from the trajectories of automobiles. However, while the automobile trajectories overlap among their own ways, the bicycle trajectories overlap in both ways. To solve this problem, the proposed method changes color of the trajectory based on its vector angle and provides the direction and flow.

Fig. 1
figure 1

Trajectories of bicycles collected from two-way bicycle path

Full size image

In Sect. 2, the detailed design of the color coding method will be explained. The experimental results of the proposed method will be shown in Sect. 3. In Sect. 4, we draw conclusions and suggest future work.

2 Method

The trajectory of a moving object is a gathering of the position sequence according to the timeline. In the aspect of visualization, it can be considered as the sequence of line segments. Each line segment consists of two neighboring points, and by connecting adjacent line segments, a trajectory can be visually created. The proposed color coding method maps each point of the trajectories with different colors to provide effective visual information.

The color of the point is based on the angle between the vector of the line segment and a vector directed to the north. Since the angle ranges from 0o to 360o, the color of the point can be gradually changed according to the direction of the line segment. Based on the vector angle of the line segment, elements of the color such as chroma and hue were controlled. Basic idea of color change can be simply described by using an edge of a circle. Figure 2 shows the chroma transition of the point according to the angle. As the angle closes to 0o or 360o color gets brighter, and it gets darker when the angle closes to 180o.

Fig. 2
figure 2

Chroma transition and angle of the line segment

Full size image

By applying the chroma transition, the line segments around 0o vector angle and 180o can be divided. However, the line segments having 90o vector angle has same chroma value with the line segments having 270o. To differentiate these line segments, each half of the circle has to have different hue. Figure 3 shows the hue variations pairs. Each pair consists of complementary colors except number (6) which consists of two primary colors.

Fig. 3
figure 3

Hue variation pairs for line segments

Full size image

3 Experimental Results

The proposed color coding method was applied to the bicycle trajectories collected from abovementioned bicycle tracks. Over one thousand GPS trajectories were recorded by using smartphone applications and the dedicated GPS devices. Invalid and erroneous data were excluded during preprocessing.

Figure 4 shows the visualized bicycle trajectories. The hue variation pairs in Fig. 3 were applied to visualizations in Fig. 4 according to the number shown in the figures. The visualized trajectories in the figure show their direction and transition of flow with gradually changing color of the line segments.

Fig. 4
figure 4

Bicycle trajectories visualized with color coding

Full size image

4 Conclusions and Future Work

In this paper, we proposed a color coding method for visualization of bicycle trajectories. The method was motivated from the bicyclists’ behaviors which differ from the motorists’ counterparts. The purposes of the method are to differentiate opposing trajectories and provide visual cues to the user. To evaluate the color coding, trajectories collected from segregated bicycle paths were visualized. As a result, applying color coding provided insights among the numerous overlapping trajectories.

In future research, we are planning to analyze principal component of the bicycle trajectories. The principal components such as principal curves can extract the spatial summary of the data. With the principal components, density estimation will be applied to enhance the visualization of the trajectory. We expect applying these methods to visualization will reveal the characteristics of bicycle trajectories.