Miracle: Facial feature extraction using active appearance model

Date of Publication

2011

Document Type

Bachelor's Thesis

Degree Name

Bachelor of Science in Computer Science

College

College of Computer Studies

Department/Unit

Computer Science

Thesis Adviser

Jocelynn W. Cu

Defense Panel Chair

Ethel Chua Joy Ong

Defense Panel Member

Arnulfo P. Azcarraga
Arturo Caronongan

Abstract/Summary

Emotion detection systems that use facial features as input have been around for quite some time. Determining emotions based on facial features sometimes mistake the emotion for another. The Active Appearance Model (AAM), which is a new technology, shows promising facial feature recognition that could be used to classify emotions based on facial input. Full frontal capture of images was taken as input, and the system identified the region of interest, where the face will be located, which was processed. Once the frames have been fed, it went through a series of image pre-processing. Each image in the AAM requires a total of 68 facial points, which are all relevant. The relevant points were taken from previous system, and other readings, which were used to prove AAMs capabilities over its predecessor, Activate Shape Model (ASM). These relevant facial points will make-up the relevant facial features which will be used for classifying emotions for future systems. Tests were also done, to show other capabilities and limitations of the AAM. The AAM does a better job in face tracking compared to ASM and the use of new features improved the accuracy of emotion classification.

Abstract Format

html

Language

English

Format

Print

Accession Number

TU18498

Shelf Location

Archives, The Learning Commons, 12F, Henry Sy Sr. Hall

Physical Description

1v. various foliation : illustrations (some colored) ; 28 cm.

This document is currently not available here.

Share

COinS