Electronic Theses and Dissertations

Identifier

910

Date

2013

Document Type

Thesis

Degree Name

Master of Science

Major

Computer Science

Committee Chair

King-Ip Lin

Committee Member

Vasile Rus

Committee Member

Bonny Banerjee

Abstract

This thesis proposed a system for real time detection of facial expressions those are subtle and are exhibited in spontaneous real world settings. The underlying frame work of our system is the open source implementation of Active Appearance Model. Our algorithm operates by grouping the various points provided by AAM into higher level regions constructing and updating a background statistical model of movement in each region, and testing whether current movement in a given region substantially exceeds the expected value of movement in that region (computed from statistical model). Movements that exceed the expected value by some threshold and do not appear to be false alarms due to artifacts (e.g., lighting changes) are considered to be valid changes in facial expressions. These changes are expected to be rough indicators of facial activity that can be complemented by contexual driven predictors of emotion that are derived from spontaneous settings.

Comments

Data is provided by the student.

Library Comment

Dissertation or thesis originally submitted to the local University of Memphis Electronic Theses & dissertation (ETD) Repository.

Download

Share

COinS