Development of a Video Image Processing-Based Micro-level Data Extractor for Non-lane-Based Heterogeneous Traffic Conditions

Limited studies are conducted to develop trajectory-level data extraction tools for non-lane-based heterogeneous traffic conditions prevalent in developing countries, like India. Comprehending this research gap, the development of an automation-based conceptual methodology is crucial, that would help researchers to understand and assess heterogeneous, non-lane-based traffic at micro-level. Microscopic parameters like individual vehicular trajectory and inter-vehicular spacing, in lateral as well as longitudinal dimensions, play significant roles in understanding traffic dynamics. A major hindrance in acquiring the huge data required for this purpose is the high demand of manpower and time in extracting traffic data manually with desirable accuracy. Implementing automated systems can mitigate the burden of data acquisition. One of the promising ideas for effective traffic analysis is video-based image processing, using which highly accurate data can be obtained by suitably calibrating the threshold values, thereby optimizing it for a given video. In this study, an attempt has been made to develop an automated image processing tool using MATLAB, to first classify vehicles under varying roadway and traffic conditions, and then to obtain lateral as well as longitudinal spacing maintained, based on the detected positions of vehicles on the road over time. The developed traffic data extractor also provides output on individual vehicle trajectory, and, hence, travel times and speed profiles of different vehicle categories. Evaluation results showed a mean absolute percentage error (MAPE) of less than 13%, suggesting its reliability under varying mixed traffic conditions. The logic developed in this research is expected to cater well for similar traffic conditions in other Asian countries.

Language

  • English

Media Info

Subject/Index Terms

Filing Info

  • Accession Number: 01717575
  • Record Type: Publication
  • Files: TRIS
  • Created Date: Aug 2 2019 4:24PM