Traffic Congestion Identification and Prediction based on Image Processing and Deep Learning Methods

Traffic congestion is one of the major issues that most metropolises are confronted with. Many measures have been taken to mitigate congestion. It is believed that measurement of congestion characteristics is the first step for mitigation efforts since it can provide guidance for selecting appropriate measures. Therefore, implementing the image processing techniques, this project first focuses on identification of traffic congestion and extraction of congestion features from probe data. Second, from detectors on the road, abundant traffic data (flow, velocity, occupancy) can be achieved to depict traffic states. Fuzzy logic method is leveraged to derive a more accurate congestion index than a single measurement. The congestion states over the whole road network at one time can be snapshot as a static image, the evolution of network congestion is therefore regarded as a motion. Conventional prediction models are no longer able to deal with the motion prediction issue. Motivated by predominance of deep learning in motion prediction, this project will represent congestion levels of a traffic network as an image, then introduce an image-based deep learning approach for congestion forecasting.

Language

• English

Project

• Status: Completed
• Funding: $57500
• Contract Numbers:

  69A3551747104

• Sponsor Organizations:

  Office of the Assistant Secretary for Research and Technology

  University Transportation Centers Program
  Department of Transportation
  Washington, DC  United States  20590
• Managing Organizations:

  Southeastern Transportation Research, Innovation, Development and Education Center (STRIDE)

  University of Florida
  365 Weil Hall
  Gainesville, FL  United States  32611
• Project Managers:

  Tucker-Thomas, Dawn

• Performing Organizations:

  Jackson State University, Jackson

  Department of Civil and Environmental Engineering
  Jackson, MS  United States  39217-0168
• Principal Investigators:

  Whalin, Robert W.

• Start Date: 20200101
• Expected Completion Date: 20211231
• Actual Completion Date: 20220616
• USDOT Program: University Transportation Centers Program

Subject/Index Terms

• TRT Terms: Detectors; Fuzzy logic; Image processing; Machine learning; Probe vehicles; Traffic congestion; Traffic data; Traffic surveillance
• Subject Areas: Data and Information Technology; Highways; Operations and Traffic Management; Planning and Forecasting;

Filing Info

• Accession Number: 01729951
• Record Type: Research project
• Source Agency: Southeastern Transportation Research, Innovation, Development and Education Center (STRIDE)
• Contract Numbers: 69A3551747104
• Files: UTC, RIP
• Created Date: Jan 31 2020 4:49PM