Sari, Afifah Asmar (2018) Implementasi AdaBoost Learning Menggunakan Fitur Multi-scale Block Local Binary Pattern Pada Pengenalan Wajah. Undergraduate thesis, Institut Teknologi Sepuluh Nopember.
Preview |
Text
05111440000154-Undergraduate_Thesis.pdf - Accepted Version Download (2MB) | Preview |
Abstract
Pengenalan wajah masih menjadi topik yang hangat untuk diteliti sejak diperkenalkan 20 tahun silam. Berbagai teknik pengenalan wajah telah dikembangkan guna mendapatkan sebuah sistem yang tangguh. Salah satu metode ekstraksi fitur wajah yang sering digunakan adalah Local Binary Pattern (LBP). Jika LBP memanfaatkan piksel sebagai fitur, pada pengembangan selanjutnya Multi-scale Block Local Binary Patten (MB-LBP) memanfaatkan area piksel sebagai fitur.
Pada Tugas Akhir ini diimplementasikan AdaBoost Learning menggunakan fitur MB-LBP untuk pengenalan wajah. Terdapat empat tahap utama yang dijalankan pada sistem pengenalan wajah untuk Tugas Akhir ini yakni pra-proses, ekstraksi fitur menggunakan metode MB-LBP, seleksi fitur menggunakan metode Statistically Effective MB-LBP (SEMB-LBP), dan klasifikasi menggunakan metode AdaBoost. Data yang digunakan dalam proses uji coba berjumlah 3059 citra wajah frontal yang diperoleh dari Collection of Facial Image : Faces94.
Dari hasil uji coba, diperoleh nilai akurasi terbaik sebesar 95,42% untuk MB-LBP dengan seleksi fitur menggunakan skala kombinasi 3x3, 9x9, dan 15x15. Sedangkan nilai akurasi terbaik untuk MB-LBP tanpa seleksi fitur adalah sebesar 98,50% dengan skala 9x9. ============================================================================================================
Facial recognition is still a hot topic to research since its introduction 20 years ago. Various facial recognition techniques have been developed to obtain a robust system. One of the most commonly used facial feature extraction methods is the Local Binary Pattern (LBP). If LBP utilizes pixels as a feature, the next development Multi-scale Block Local Binary Patten (MB-LBP) utilizes the area as a feature.
This Final Project implemented AdaBoost Learning using MB-LBP feature for face recognition. There are four main stages of the facial recognition system in this Final Assignment, pre-processing, feature extraction using MB-LBP method, feature selection using the Statistically Effective MB-LBP (SEMB-LBP) method, and classification using the AdaBoost method. The data used in the trial process amounted to 3059 frontal face images obtained from the Collection of Facial Image: Faces94.
From the experimental results, the best accuracy value is 95,42% for MB-LBP with feature selection using 3x3, 9x9, and 15x15 combination scale. While the best accuracy value for MB-LBP without feature selection is 98,50% with 9x9 scale.
| Item Type: | Thesis (Undergraduate) |
|---|---|
| Additional Information: | RSIf 006.42 Sar i-1 3100018076222 |
| Uncontrolled Keywords: | Pengenalan Wajah, Multi-scale Local Binary Pattern, Statistically Effective Multi-scale Local Binary Pattern, AdaBoost Learning. |
| Subjects: | T Technology > T Technology (General) > T57.5 Data Processing T Technology > TA Engineering (General). Civil engineering (General) > TA1637 Image processing--Digital techniques T Technology > TA Engineering (General). Civil engineering (General) > TA1650 Face recognition. Optical pattern recognition. |
| Divisions: | Faculty of Information and Communication Technology > Informatics > 55201-(S1) Undergraduate Thesis |
| Depositing User: | Afifah Asmar Sari |
| Date Deposited: | 31 Dec 2018 02:14 |
| Last Modified: | 12 Oct 2020 03:23 |
| URI: | http://repository.its.ac.id/id/eprint/53398 |
Actions (login required)
 |
View Item |