Implementasi Dimensionality Reduced Local Directional Pattern Pada Aplikasi Pengenalan Wajah

Yuseti, Kevin Zulkarnain (2017) Implementasi Dimensionality Reduced Local Directional Pattern Pada Aplikasi Pengenalan Wajah. Undergraduate thesis, Institut Teknologi Sepuluh Nopember Surabaya.

[thumbnail of 5113100146-Undergraduate-Theses.pdf]
Preview
Text
5113100146-Undergraduate-Theses.pdf - Published Version

Download (2MB) | Preview

Abstract

Local Directional Pattern (LDP) adalah sebuah metode ekstraksi fitur yang digunakan untuk pengenalan pola pada gambar dan pengenalan wajah. Pada tugas akhir ini pengembangan dilakukan untuk mereduksi jumlah dimensi dari hasil ekstraksi fitur. Oleh karena itu dinamakan Dimensionality Reduced–Local Directional Pattern (DR-LDP). LDP mendapatkan deskriptor gambar dari operasi korelasi matriks antara piksel gambar dan tetangganya dengan Kirsch mask. Dari korelasi matriks didapatkan matriks baru dengan kode LDP sebagai deskriptor gambar. Reduksi dimensi pada matriks ini dilakukan dengan mengambil nilai LDP terbesar dari blok kode LDP sehingga didapati kode yang merepresentasikan satu blok kode LDP. Pengujian dilakukan pada dataset standar ORL dan Extended YALE-B. Percobaan pada dataset ORL menghasilkan akurasi 80% dan YALE-B 81% untuk kasus pencahayaan, 67% untuk kasus kemiringan kepala.
=================================================================
Local Directional Pattern (LDP) are feature extraction method used for pattern and face recognition. This undergraduate thesis purpose were to conduct dimensional reduction on feature extracted from LDP. Therefore called Dimensionality Reduced-Local Directional Pattern. Image descriptor from LDP came from matrix correlation between image pixel and neighboring pixels with kirsch mask. The result of matrix correlation resulting in LDP matrix as image descriptor. Dimension reduction done by pick maximum value from LDP coded block, resulting in single code representing the entire of LDP code block. Testing done by using this method on standard dataset ORL and extended YALE-B. Accuracy result for ORL dataset were 80% and YALE-B 81% for illumination variance, 67% for head tilt variance.

Item Type: Thesis (Undergraduate)
Additional Information: RSIf 006.42 Yus i
Uncontrolled Keywords: Pengenalan Wajah, Local Directional Pattern, Reduksi Dimensi, Face Recognition, Dimensionality Reduction
Subjects: Q Science > QA Mathematics > QA75 Electronic computers. Computer science. EDP
Divisions: Faculty of Information Technology > Informatics Engineering > 55201-(S1) Undergraduate Thesis
Depositing User: Kevin Zulkarnain Yuseti
Date Deposited: 16 Oct 2017 04:32
Last Modified: 08 Mar 2019 04:00
URI: http://repository.its.ac.id/id/eprint/46236

Actions (login required)

View Item View Item