Trajectory based vehicle counting and anomalous event visualization in smart cities

Fozia Mehboob, Muhammad Abbas, Richard Jiang, Abdul Rauf, Shoab Khan, Saad Rehman

Research output: Contribution to journalArticlepeer-review

7 Citations (Scopus)

Abstract

Motion pattern analysis can be performed automatically on the basis of object trajectories by means of tracking videos; an effective approach to analyse and to model the traffic behaviour; is important to describe motion by taking the whole trajectory whereas it’s more essential to identify and evaluate object behaviour online. In this paper, pattern detection approach is presented which takes spatio-temporal characteristic of vehicle trajectories. A real time system is built to infer and track the object behaviour quickly by online performing trajectory analysis. Every independent vehicle in the video frame is tracked over time. As the anomaly behaviour occurs, glyph is generated to show it occurrences. Vehicle counting is done by estimating the trajectories and compared with Hungarian tracker. Several surveillance videos are taken into account for the performance checking of system. Experimental results demonstrated that proposed method in comparison with the state of the art algorithms, provides robust vehicle density estimation and event information i.e., lane change information.
Original languageEnglish
Pages (from-to)443-452
JournalCluster Computing
Volume21
Issue number1
Early online date5 May 2017
DOIs
Publication statusPublished - 1 Mar 2018

Fingerprint

Dive into the research topics of 'Trajectory based vehicle counting and anomalous event visualization in smart cities'. Together they form a unique fingerprint.

Cite this