200 likes | 397 Views
Paper Reading. Author: Zuchao Wang, Min Lu, Xiaoru Yuan, Junping Zhang, Huub van de Wetering. Visual Traffic Jam Analysis Based on Trajectory Data. A programmer’s perspective. Explainer: Yabo Su. Outline. Paper information Paper structure Work flow and its details Proprocessing part
E N D
Paper Reading Author: Zuchao Wang, Min Lu, Xiaoru Yuan, Junping Zhang, Huub van de Wetering Visual Traffic Jam Analysis Based on Trajectory Data A programmer’s perspective Explainer: Yabo Su
Outline • Paper information • Paper structure • Work flow and its details • Proprocessing part • Visualization part
Paper Information • Conference Paper • IEEE Transactions on Visualization and Computer Graphics (VAST'13) • Key Point • Traffic jam propagation
Before we moving forward • Key problem to be solved: • How traffic jam propagates from one road to other road? • Questions: • What factors make this problem interesting? • How would you solve it qualitatively if you are provided historical taxi trajectory data and road network data? • How you define traffic jam using the information in the datasets? • How you detect traffic jam accroding to your defination? • How you visualize the traffic jam you have detected?
Paper structure • 1 INTRODUCTION • 2 RELATEDWORK • 2.1 Traffic Event Detection • 2.2 Traffic Visualization • 2.3 Propagation Graph Visualization • 3 OVERVIEW • 3.1 Design Requirement • 3.2 Description of Input Data • 3.3 Traffic Jam Data Model • 3.4 Work Flow • 4 PREPROCESSING • 5 VISUALIZATION DESIGN • 6 VISUALIZATION RESULTS ANDCASE STUDY • 7 DISCUSSION • 8 CONCLUSION ANDFUTUREWORK • APPENDICES Why we build this system How the system is built How you use the system to get information we want to privide to you?
Objective: • Extract all drivable ways • Ensure all roads are connected How to? Premise: You should be familiar with the road network data(OpenStreetMap). How to next? Easy, no particular algrothms!
Objective: • Remove the useless trajectory data What kinds of trajectory data should be removed? Actions in the paper: Unrealistic coordinates Duplicated Time Stamp High Speed Long Distance Long Time Parking Waiting for Passenger Tiny Trajectory One point constraint Two points constraint multi points constraint A stop dection algorithm[37]
Objective: • Map trajectory points to the road network How to? Actions in the paper: It’s a complicated probem. ST-matching algorithm[30] Result: <dWay, List<trajectory point>>
Objective: • Map trajectory points to the road network How to? Actions in the paper: Removing the taxis with exeptionally high speed road speed = an average of the average speeds on the remaining taxis Record “support” Result: <dWay, road speed, support>
How to? • Objective: • Detect traffic jam events Actions in the paper: Sort all valid speeds for a dWay in ascending Pick the speed value at the percentage F% position With a valid speed less than percentage C% of the free flow speed, is said to have a low speed Result: List<traffic jam event> traffic jam event - <d, t0, t1>
How to? • Objective: • Find the events propagation Actions in the paper: Assumption: e1->e2 <=> e1.t0<= e2.t0<=e1.t1 && e1.d->e2.d 2. Use the STOTree algorithm [29] construct the progation graphs Result: List<progation graph> progation graph - <V, E> direct, acyclic number of events, time span, total distance
Visualization design • 1. Pixel Based Road Speed View • 2. Graph List View • 3. Graph Projection View • 4. Spatial View • 5. Multi-faceted Filter View