UVM Theses and Dissertations
Format:
Print
Author:
Ouyang, Xi
Dept./Program:
Electrical Engineering
Year:
2011
Degree:
MS
Abstract:
The thesis consists of the development of three Labs for undergraduate and graduate students in the School of Engineering. The Labs utilize hardware, software and tracking algorithms.
Labs are as follows: 1. Color based red ball tracking. 2. Viola-Jones classifier-based face tracking and improvement with Kalman filtering. 3. Microsoft 3-D infrared-based Kinect camera contour detection.
The red ball tracking Lab covers the basic concepts of object detection and tracking. It has three main parts: a camera, two servo motors and a mirco-controller. A color based algorithm is used to follow a red ball. The face tracking Lab introduces a more advanced algorithm -the well known Viola-Jones classifier -to track more complex object such as a human face. Since the tracking trajectory is unstable, Kalman filtering is utilized to smooth the face bounding rectangle. In the Kinect camera Lab, the principle of the Microsoft Kinect camera is introduced. Depth perception with an infrared grid is illustrated through contour detection with OfxKinect open source toolkit.
Labs are as follows: 1. Color based red ball tracking. 2. Viola-Jones classifier-based face tracking and improvement with Kalman filtering. 3. Microsoft 3-D infrared-based Kinect camera contour detection.
The red ball tracking Lab covers the basic concepts of object detection and tracking. It has three main parts: a camera, two servo motors and a mirco-controller. A color based algorithm is used to follow a red ball. The face tracking Lab introduces a more advanced algorithm -the well known Viola-Jones classifier -to track more complex object such as a human face. Since the tracking trajectory is unstable, Kalman filtering is utilized to smooth the face bounding rectangle. In the Kinect camera Lab, the principle of the Microsoft Kinect camera is introduced. Depth perception with an infrared grid is illustrated through contour detection with OfxKinect open source toolkit.