Optimizing Trajectory Visualization: From Data Collection to Rendering

1. Why Trajectory Needs "Decoration"

Mobile devices enable city management to track personnel and vehicle trajectories, but achieving a smooth and accurate trajectory line is challenging due to factors such as coordinate collection frequency, keep‑alive, accuracy, noise removal, coordinate conversion, line connection, and playback.

2. Optimization Starting from Coordinate Collection

Three primary issues are addressed: (a) intelligent adjustment of GPS sampling frequency based on movement patterns, (b) keeping the collection service alive—illustrated with the "GuDong" app’s mechanisms such as foreground notifications, dual‑process guarding, and lock‑screen broadcast handling—and (c) improving satellite positioning accuracy by monitoring signal strength, recording weak‑signal points, and switching to alternative providers like Baidu when needed.

2.1 Coordinate Sampling Frequency Optimization

Default frequencies differ for personnel and vehicles; they are slowed when points are static and accelerated when movement exceeds a threshold.

2.2 Collection Service Keep‑Alive Optimization

Key techniques include disabling the back button, continuously updating a foreground notification, auto‑restart via a watchdog service, persisting state in configuration files, and reacting to lock‑screen broadcasts.

3. Fitting – Improving Trajectory Point Quality

Four fitting strategies are compared:

Road‑network matching aligns points to the nearest road segment, producing visually clean trajectories but relying on complete road data.

Spatio‑temporal clustering aggregates similar points while preserving time order, reducing point clutter.

Kalman filter smoothing predicts and corrects positions for a smooth curve, though excessive smoothing can obscure real features.

A final combined scheme filters out extreme errors, removes "flying" points, applies controlled Kalman smoothing, aggregates via clustering, and uses road‑matching when reliable map data exists.

4. Animation – Giving Trajectories a Cinematic Feel

Dynamic visual effects include:

Smooth moving icons with 16 directional sprites that rotate based on bearing and adjust animation speed according to distance between points.

Flowing arrowheads that appear at appropriate zoom levels to indicate direction without overcrowding the map.

Multi‑trajectory displays similar to ride‑hailing apps, featuring distinct icons for different vehicle types, license‑plate overlays, video links, and IoT device integration.

2‑3D integrated visualizations that place trajectories in a three‑dimensional scene for immersive interaction.

5. Conclusion

By integrating the above optimizations with a big‑data architecture—NoSQL storage, distributed messaging, and stream‑processing engines—a product comparable to major ride‑hailing trajectory displays was launched, successfully deployed in real‑world systems such as Chengdu dust‑control (1.6 k vehicles, 24‑hour real‑time tracking) and Longyan waste‑transport monitoring, with plans for further research in trajectory mining and 2‑3D visualization.