Capturing videos of trains under ambient lighting conditions for computer vision analysis

Gopalakrishnan, Suchithra

Permalink

https://hdl.handle.net/2142/42223 Copy

Description

Title

Capturing videos of trains under ambient lighting conditions for computer vision analysis

Author(s)

Gopalakrishnan, Suchithra

Issue Date

2013-02-03T19:28:23Z

Director of Research (if dissertation) or Advisor (if thesis)

Ahuja, Narendra

Department of Study

Electrical & Computer Eng

Discipline

Electrical & Computer Engr

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

M.S.

Degree Level

Thesis

Keyword(s)

• Machine Vision system
• railroad monitoring

Abstract

Intermodal trains are usually the fastest freight trains in North America. Fuel con- sumption in these trains is high because of their aerodynamic characteristics. If the loads on these railcars were placed in better con gurations, it would be possible to reduce fuel consumption. In order to inspect their trains for their aerodynamic char- acteristics, Burlington Northern Santa Fe (BNSF) has sponsored a project to build a wayside machine vision (MV) system in Sibley, MO, a busy intermodal freight train route called TRANSCON. Our research group at the Beckman Institute, in collab- oration with the Civil Engineering Railroad program, has developed a wayside MV system that captures videos of trains passing by. It consists of various train detection sensors, personal computers, and camera and lighting towers. After the video data is collected, it is analyzed using the Train Monitoring System (TMS) and Train Scor- ing System (TSS) algorithms. Finally, each car in the train is given an aerodynamic score. Currently, the system is functional in analyzing videos that have been acquired in daylight. This work analyzes the requirements of the TMS algorithm, improves TMS per- formance on daytime trains, and reports the design of a lighting system that will be used to provide su cient lighting for the scene at night or when there is not enough daylight. This work investigates the causes of the problems with exposure, and uses non-realtime adjustment of the camera to acquire properly exposed videos. The re- sults show improvement in TMS results. Also discussed is the groundwork done for the lighting sub-system.

Graduation Semester

2012-12

Permalink

http://hdl.handle.net/2142/42223

Copyright and License Information

Copyright 2012 Suchithra Gopalakrishnan

Owning Collections