Degree Name

Master of Science (Hons.)


Department of Biomedical Science


The dimensional quantitative analysis of human motion requires the reproduction of three dimensional coordinates from multiple camera images. In current photogrammetric systems the ability to identify and track individual camera image points imposes limitations on the accuracy and complexity of human motion analysis. Current photogrammetric systems are limited by the number of cameras, three dimensional segments, markers per segment, and the complexity of movement possible due to the increased difficulty and time required to reproduce three dimensional coordinates. The automated reconstruction and tracking of three dimensional coordinates may overcome these limitations by removing the necessity to track and identify two dimensional camera coordinates. The aims of the present research were firstly, to identify limitations and practical problems associated with the use of conjugate imagery in the reproduction of three dimensional coordinates, and secondly, to identify and implement techniques involving conjugate imagery for the automated reproduction of three dimensional coordinates. The methods and procedures developed in the current research provides the basis for future research into automated three dimensional tracking. Four criterion measures, i) Conjugate Point Error, ii) Lab Point Standard Error, iii) Lab Point Error, and iv) Lab Point Paired Error, were established for determining the validity of conjugate image points. Based on the criterion measures an algorithm was developed which accurately reproduced three dimensional coordinates and conjugate image points for a 55 point marker system viewed in four cameras (digitisation error < 0.2%, laboratory point separation > 6cm.). The success of the algorithm was dependent on the digitisation error, laboratory point separation, and the number of laboratory points appearing in two camera images. The present research has shown the applicability of conjugate imagery in the automated reproduction of three dimensional coordinates from multiple camera images, a well as the viability of this approach in the automated three dimensional tracking of camera image data to achieve an increase in accuracy and complexity of human movement analysis.



Unless otherwise indicated, the views expressed in this thesis are those of the author and do not necessarily represent the views of the University of Wollongong.