Date of Graduation


Document Type


Degree Type



Statler College of Engineering and Mineral Resources


Lane Department of Computer Science and Electrical Engineering

Committee Chair

Arun A. Ross.


This thesis explores the possibility of incorporating human body measurements in a biometric framework. While metrological features have been used for identifying persons in the late 19th century, there is limited work in automating this process for surveillance applications. We first establish the relevance of using metrological features in biometric systems by studying two anthropometric data-sets (NASA and NHANES). We then propose a technique to automatically extract a subset of these measurements from a video sequence. A robust segmentation technique (HMMF) to detect moving pixels corresponding to human objects is used in the first stage. Next, we use Active Contours to obtain a precise contour of the human body. Finally, we design a technique to extract the measurements of human body, viz., height, width of the head and the torso, from the segmented image. We show that the measurements extracted in this manner bear close resemblance to manual measurements in terms of their pixel count. To validate the procedure outlined here, we extract these measurements from different videos containing human objects and check for consistency across multiple stand-off distances between the subject and the camera. Data pertaining to 9 different individuals (3 video sequences each) was used in this research.