Date of Award

2020

Document Type

Thesis

Degree Name

M. S.

Department

Mechanical Engineering

First Advisor

Paul Rullkoetter

Second Advisor

Chadd W. Clary

Third Advisor

Barry Zink

Keywords

Kinect, Marker-less, Motion capture, Validation

Abstract

The ability to understand human movement is beneficial for deciding surgical procedures, tracking disease progression over time and helping with patient rehabilitation. The current gold-standard for collecting human movement is the use of 3-dimensional marker-based systems. Several studies have presented the many limitations to the current gold-standard that reduces the number of people who are able to benefit from a gait analysis. Those limitations in the current gold-standard include the requirements of large laboratory space, costly equipment, long instrumentation and collection time, and the potential for motion artifact from markers being placed on the skin. The purpose of this study is to create a marker-less motion capture system using the newly–released Kinect Azure cameras from Microsoft. The study aims to validate the new system against the gold-standard. A validation of a four Kinect Azure camera system was conducted with 10 subjects completing over ground walking trials at a self-selected pace, sit-to-stand, lunge, and step up/down while Kinect and 3D marker-based data were collected simultaneously. The data was synchronized and cut to a single activity cycle where joint angles and spatio-temporal measures were compared between the two systems. Walking speed and stride length were highly correlated between the two systems with r-values >0.9 and p-values <0.001. The average difference in maximum knee flexion angle between the two systems is 2.84° with a r=0.785 and p-value <0.001. A 3D point cloud was generated from the four Kinect Azure camera system to generate a surface mesh. The 3D mesh was used to provide a better understanding of body habitus than current BMI.

Publication Statement

Copyright is held by the author. User is responsible for all copyright compliance.

Provenance

Received from ProQuest

Rights holder

Abigail Eustace

File size

110 p.

File format

application/pdf

Language

en

Discipline

Biomechanics, Biomedical engineering

Share

COinS