Author: Biswas Lohani V K
Lohani V K, Biswas, 2017 Automated camera tracking system in Gait analysis, Flinders University, School of Computer Science, Engineering and Mathematics
Terms of Use: This electronic version is (or will be) made publicly available by Flinders University in accordance with its open access policy for student theses. Copyright in this thesis remains with the author. You may use this material for uses permitted under the Copyright Act 1968. If you are the owner of any included third party copyright material and/or you believe that any material has been made available without permission of the copyright owner please contact copyright@flinders.edu.au with the details.
People have numerous walking abnormalities due to the certain illness or medical condition. The main reason is due to the muscle or neurological issues. Hence, the scope of Clinical Gait Analysis (CGA) arises to measure and estimate the gait biomechanics, which makes it easier to recognize the abnormal appearances during the Gait Analysis and it also helps to make the accurate clinical decision regarding orthopedic surgery and rehabilitation for the clinician.
There are different techniques to collect the motion related information during the Gait Analysis. In which, recording of good quality video has the significant role in the CGA process which visualize the walking pattern of the patient immediately and the recorded video data can be
utilized a number of times during the gait analysis process as well as it can be used as reference video information for the future assessment of the same patient.
Therefore, the project “Automated Camera Tracking System in Gait Analysis” is related to the movement of the video recording camera along with the patient so that it could capture the patient’s gait in different time intervals. This project is the advancement of the current system in the Gait Lab of Repartition General Hospital, South Australia. The present system is the still camera that has been used to capture the video of the walking patient from the side and front in the walkway. This camera is not able to record more than a stride of a patient during the process which is not sufficient to make an accurate clinical decision.
Consequently, in this master thesis, it has been proposed to build an automated four wheeled drive robot which is operated to keep the object in sight by maintaining a steady perpendicular distance between the system and the patient. Furthermore, different object detection and tracking methods are implemented to focus a patient in the video with the automated camera tracking system. Furthermore, the experimental analysis of the automated tracking system by using the object detection and tracking method in image processing shows that the Histogram of Oriented Gradient Features (HOGs) extraction methods for object detection and tracking is more effective in this project because it gives more precise location data due to the locally normalized histogram of gradient orientation features of an object which is also not affected by the environmental condition ( light, temperature etc.) in the Gait Lab and it also has less variation in its speed so probability of losing patient from the focus is less in this method and the system can move smoothly. On the other hand, the color object detection and tracking method is relatively less effective in comparison with HOGs method of object detection and tracking due to the fluctuation that occurs in object detection. This method not able to identify the exact location of the object when used in the poor light conditions inside the Gait Lab but due to the low computational cost it is implemented in the system by improving the brightness of light source in the current Gait Lab.
Keywords: Tracking System, Gait lab, Object detection, Object tracking
Subject: Engineering thesis
Thesis type: Masters
Completed: 2017
School: School of Computer Science, Engineering and Mathematics
Supervisor: Dr. Sherry Randhawa