FACIAL MOTION CAPTURE MENGGUNAKAN ALGORITMA INVERSE COMPOSITIONAL IMAGE ALIGNMENT PADA AAM

UPT. PERPUSTAKAAN Institut Teknologi Sepuluh Nopember Surabaya Kampus ITS Sukolilo - Surabaya 60111
Phone	:	031-5921733 , 5923623
Fax	:	031-5937774
E-mail	:	libits@its.ac.id
Website	:	http://library.its.ac.id
Support (Customer Service) :
timit_perpus@its.ac.id

Welcome..guys!

Have a problem with your access?
Please, contact our technical support below:

LIVE SUPPORT
08.30 s/d 15.30 WIB

Taufik

Dewi

Ansi

Tondo

! ATTENTION !

To facilitate the activation process, please fill out the member application form correctly and completely

Registration activation of our members will process up to 3 x 24 hours (confirm by email). Please wait patiently

Still Confuse?
Please read our User Guide

POLLING

Bagaimana pendapat Anda tentang layanan repository kami ?

Pengunjung hari ini : 375

Total pengunjung : 6508

Hits hari ini : 17196

Total Hits : 83089

Pengunjung Online: 35

• You are connected from 38.107.179.210

OR You May Search Our Knowledges Using Google Customize Search :

Custom Search

Detail Collection

ITS » Master Theses » Teknik Elektro S2
Posted by dee@its.ac.id at 17/11/2011 09:35:39 • 161 Views

FACIAL MOTION CAPTURE MENGGUNAKAN ALGORITMA INVERSE COMPOSITIONAL IMAGE ALIGNMENT PADA AAM

FACIAL MOTION CAPTURE USING INVERSE COMPOSITIONAL IMAGE ALIGNMENT ALGORITHM ON AAM

Created by :
SUARDINATA, I WAYAN ( 2209205027 )

Subject:	Pengolahan data elektronis
Alt. Subject :	Image processing--digital techniques face perception--data processing
Keyword:	Aam Facial motion capture Inverse compositional image alignment Fitting method

[ Description ]

Teknologi motion capture dibutuhkan dalam berbagai aplikasi khususnya animasi yang terus berkembang pesat. Teknik yang digunakan dapat menggunakan penanda maupun tanpa penanda (markerless). Metode AAM mampu melakukan capture titik-titik landmark pada wajah dengan baik, tetapi implementasi pemrosesan masukan secara real time masih perlu ditingkatkan. Penelitian ini diarahkan untuk mengembangkan teknik markerless motion capture dengan menggunakan metode Active Appearance Model (AAM) pada wajah secara real time. Penelitian dimulai dengan memilih metode AAM yang cepat dan akurat untuk capture titik-titik pada wajah. Penelitian ini membandingkan dua metode AAM yaitu AAM Basic dan inverse compositional image alignment (ICIA). Sumber data yang digunakan adalah 4 dataset yaitu BIOID, AGING, FRANCK dan Jia Pei. Metode evaluasi menggunakan empat kriteria yaitu waktu pencocokan, jumlah iterasi yang diperlukan, nilai overlap yang menunjukkan perbandingan jumlah titik yang berada pada tempat yang benar dengan keseluruhan jumlah titik serta jarak penyimpangan yang menunjukkan jarak Euclidean titik hasil pencocokan dari posisi yang seharusnya. Pengujian juga menggunakan 10K-Fold Cross Validation menguji kriteria diatas. Hasil eksperimen menunjukkan bahwa rata-rata waktu pencocokan CMUICIA adalah 0,134 detik lebih baik daripada AAM_BASIC yang rata-waktu pencocokannya adalah 1,545 detik. CMUICIA membutuhkan rata-rata 7,236 iterasi sedangkan AAM_BASIC membutuhkan rata-rata 8,378 iterasi. Sedangkan dari segi akurasi, kedua metode menunjukkan hasil yang tidak jauh beda dimana nilai rata-rata overlap CMUICIA adalah 0.896495412 sedangkan AAM_BASIC adalah 0.884339402. Nilai rata-rata jarak penyimpangan CMUICIA adalah 5,737 sedangkan AAM_BASIC adalah 6,749.

Alt. Description

Motion capture technology is needed in many applications especially in animation that continues to grow rapidly. The technique used can be using marker or markerless. AAM method is able to capture landmark points on the face well, but implementation in real-time processing still needs to be improved. This study aimed to develop markerless motion capture technique using the active appearance model (AAM) on the face in real time. The study begins by selecting the AAM method that quickly and accurately to capture the points on the face. This study compared two methods of AAM i.e. AAM BASIC and inverse compositional image alignment (ICIA). Source data used are 5 datasets namely BIOID, AGING, FRANCK, Jia Pei and the writer himself. Evaluation method using four criteria:matching time, the number of iterations required, the overlap shows the comparison number of points which are on the right place with the overall number of points and the istance deviation indicates the Euclidean distance from point matches and the position should be. Evaluation also uses 10K-Fold Cross Validation to test the above criteria. The experimental results showed that the average time for CMUICIA is 0.134 better than the average time of AAM_BASIC. CMUICIA require an average of 7.236 iterations while AAM_BASIC require an average of 8.378 iterations. In terms of accuracy, both methods showed results that are not much different where the average overlap value of the CMUICIA is 0.8964 while the AAM BASIC is 0.88433. The average value of the deviation distance for CMUICA is 5.737 whereas AAM BASIC is 6.749.

Contributor	:	Mochamad Hariadi, ST., MSc., Ph.D.
Date Create	:	09/08/2011
Type	:	Text
Format	:	pdf
Language	:	Indonesian
Identifier	:	ITS-Master-3100011043714
Collection ID	:	3100011043714
Call Number	:	RTE 006.42 Sua f

Source :
Master Thesis of Electrical Engineering, RTE 006.42 Sua f, 2011

Coverage :
ITS Community

Rights :
Copyright @2011 by ITS Library. This publication is protected by copyright and per obtained from the ITS Library prior to any prohibited reproduction, storage in a re transmission in any form or by any means, electronic, mechanical, photocopying, reco For information regarding permission(s), write to ITS Library

Share this information to :

Google Bookmarks | Technorati | Del.icio.us | Digg

[ Free Download - Free for All ]