Marker detection and trajectory generation algorithms for a multicamera based gait analysis system

Shafiq M., Tumer S. , Guler H.

MECHATRONICS, vol.11, no.4, pp.409-437, 2001 (Journal Indexed in SCI) identifier identifier

  • Publication Type: Article / Article
  • Volume: 11 Issue: 4
  • Publication Date: 2001
  • Doi Number: 10.1016/s0957-4158(00)00026-x
  • Title of Journal : MECHATRONICS
  • Page Numbers: pp.409-437


The current gait analysis methodology requires tracking of markers placed on the body segments of the lower extremities, by using multiple camera images while tl-ie subject is walking. This study was aimed at the development of application software for the multicamera marker based gait analysis system established at the Middle East Technical University. The main functions of this software are pixel grouping. marker detection and generation of three-dimensional (3D) marker trajectories. The software removes noise and artefacts, and effectively detects the physical markers which may be closely spaced and even overlapped on some camera images. Then, the images in different cameras are matched by utilising the concept of epipolar lines, and 3D co-ordinates of markers are determined for each matching group at every field. Finally, matching of each marker in successive fields is accomplished by employing a set of extrapolation algorithms. thus yielding the required 3D marker trajectories. This paper presents theory, algorithms and implementation of the developed software, as well as an assessment of its performance. In addition to gait analysis for clinical and research purposes. the developed algorithms can find application in sports. 3D computer animation and motion analysis of mechanical constructions. (C) 2000 Elsevier Science Ltd. All rights reserved.