Purdue University Graduate School
Browse

File(s) under embargo

1

year(s)

9

month(s)

7

day(s)

until file(s) become available

Lane-based Weaving Area Traffic Analysis Using Field Camera Data

thesis
posted on 2024-01-03, 15:54 authored by Wei LinWei Lin

Vehicle weaving describes the lane-changing actions of vehicles, which is a critical aspect of traffic management and road design. This study focused on the weaving behavior of vehicles occurring between ramp merge and diverge areas. Weaving in these areas causes congestion and increases the risk of accidents, especially during heavy traffic. Redesigning such areas for enhanced safety requires a comprehensive analysis of the traffic conditions. Obtaining the weaving pattern is a challenge in the traffic industry. To address this challenge, we leveraged AI and image processing technology to develop algorithms for quantitative analysis of weaving using surveillance videos at the consecutive ramp merge and diverge areas. This approach can also determine the weaving patterns of passenger cars and trucks respectively. The experimental results captured the lane-based weaving behavior of around 30% of vehicles in the favorable areas. The captured weaving data is used as weaving data samples to derive an overall analysis of a weaving location. Remarkably, our approach can reduce the manual processing time for weaving analysis by more than 90%, making this highly practical for use.

Funding

INDOT/JTRP SPR-4738

History

Degree Type

  • Master of Science

Department

  • Computer Science

Campus location

  • Indianapolis

Advisor/Supervisor/Committee Chair

Mihran Tuceryan

Additional Committee Member 2

Stanley Chien

Additional Committee Member 3

Rajeev Raje

Additional Committee Member 4

Lauren Christopher

Usage metrics

    Licence

    Exports

    RefWorks
    BibTeX
    Ref. manager
    Endnote
    DataCite
    NLM
    DC